Volume 20 Issue 5 by IFAMA

International Food and Agribusiness Management Review

Official Journal of the International Food and Agribusiness Management Association

Volume 20 Issue 5 2017

EVERY GENERATION NEEDS ITS LEADERS. As a farmer-owned cooperative, we help lead the agricultural industry by promoting the most innovative, responsible farming practices available today. And we want to make sure that the next generation of leaders is well trained for tomorrow. That’s why we proudly support youth leadership programs like FFA. It’s also why we’ve invested in scholarships and countless land-grant university initiatives across the country. At CHS, we know every generation needs its leaders. And every potential leader needs a little help becoming one. Visit chsinc.com/stewardship to learn more about our ongoing commitment to leadership.

International Food and Agribusiness Management Review

Editorial Staff Executive Editor

Gerhard Schiefer, University of Bonn, Germany

Regional Managing Editors Asia, Australia, and New Zealand

Derek Baker, UNE, Australia Kim Bryceson, University of Queensland, Australia Kevin Chen, IFPRI-Bejing, China Jeff Jia, University of Exeter, United Kingdom Nicola M. Shadbolt, Massey University, New Zealand

Europe

Pegah Amani, Technical Institute of Sweden, Sweden Vera Bitsch, Technical University of Munich, Germany Laura Carraresi, University of Bonn, Germany Alessio Cavicchi, University of Macerata, Italy Hans De Steur, Ghent University, Belgium Loic Sauvee, UniLaSalle, Beauvais, France Cristina Santini, University San Raffaele, Italy Jacques Trienekens, Wageningen University, The Netherlands

North America

Ram Acharya, New Mexico State University, USA Yuliya Bolotova, Clemson University, USA Michael Gunderson, Purdue University, USA William Nganje, North Dakota State, USA R. Brent Ross, Michigan State University, USA Aleksan Shanoyan, Kansas State University, USA David Van Fleet, Arizona State University, USA Nicole Olynk Widmar, Purdue University, USA Cheryl Wachenheim, North Dakota State University, USA

South America

Aziz da Silva Júnior, Federal University of Vicosa, Brazil Jose Vincente Caixeta Filho, University of Sao Paulo, Brazil Emilio Morales, University of New England, Australia

Africa

Ajuruchukwu Obi, University of Fort Hare, South Africa Nick Vink, Stellenbosch University, South Africa

Editorial Board Filippo Arfini, Universita’ di Parma, Italy Stefano Boccaletti, Universita’ Cattolica, Italy Michael Boehlje, Purdue University, USA Dennis Conley, University of Nebraska - Lincoln, USA Francis Declerck, ESSEC Business School, France Jay Lillywhite, New Mexico State University, USA Woody Maijers, INHOLLAND University, The Netherlands

Marcos Fava Neves, FEA / USP / PENSA, Brazil Onno Omta, Wageningen University, The Netherlands Hernán Palau, Buenos Aires University, Argentina Christopher Peterson, Michigan State University, USA Thomas Reardon, Michigan State University, USA Mary Shelman, (Retired) Harvard Business School, USA Johan van Rooyen, University of Stellenbosch, South Africa

The IFAMR (ISSN #: 1559-2448) is published quarterly and the archived library is available at http://www.ifama.org. For copyright and publishing information, please contact: Marijn van der Gaag, Administrative Editor Wageningen Academic Publishers • P.O. Box 220 6700 AE Wageningen • The Netherlands • Tel: +31 317 476511 Fax: +31 317 453417 • Email: ifamr@wageningenacademic.com • Web: http://www.wageningenacademic.com/loi/ifamr.

International Food and Agribusiness Management Review Volume 20 Issue 5, 2017

TABLE OF CONTENTS 1. 2. 3. 4.

When Amazon ate Whole Foods: big changes for Big Food

615

U.S. milled rice markets and integration across regions and types

623

The effects of price promotion on relative virtue and vice food products

637

Zero-inflated ordered probit approach to modeling mushroom consumption in the United States

655

Cooperation among Ugandan farmers: cultivating social capital

673

The mediated partnership model for sustainable coffee production: experiences from Indonesia

689

Kate Phillips-Connolly and Aidan J. Connolly

Man-Keun Kim, Hernan Tejeda, and T. Edward Yu

Josefa ParreĂąo-Selva, Francisco J. Mas-Ruiz, and Enar Ruiz-Conde

Yuan Jiang, Lisa A. House, Hyeyoung Kim, and Susan S. Percival

J.L. Morrow, Jr., Richard Patrick Joyce III, William J. McMahon, Antonio M. DeMaia, S. Caleb McVicker, Ashley E. Parsons, and Kristin Wilcox

Atika Wijaya, Pieter Glasbergen, and Surip Mawardi

Exploring the applicability of a sustainable smallholder sourcing model in the black soybean case in Java 709

August R. Sjauw-Koen-Fa, Vincent Blok, and Onno Omta

Understanding the determinants of adoption of enterprise resource planning (ERP) technology within the agri-food context: the case of the Midwest of Brazil

Caetano Haberli Jr, Tiago Oliveira, and Mitsuru Yanaze

729

An empirical investigation of patent and trademark ownership propensity and intensity in the U.S. food and drink industry 747

Jasper Grashuis and Stanley Kojo Dary

10. Analyzing job satisfaction and preferences of employees: the case of horticultural companies in Germany Stephan G.H. Meyerding

International Food and Agribusiness Management Review

765

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2017.0074 Received: 27 July 2017 / Accepted: 18 August 2017

When Amazon ate Whole Foods: big changes for Big Food INDUSTRY SPEAKS Kate Phillips-Connollya and Aidan J. Connolly

aLecturer,

Trinity College, University of Dublin, College Green, Dublin 2, Ireland; Director, Abhaile Consulting, 271 Beach Road N, Wilmington, NC 28403, USA bChief

Innovation Officer and Vice President, Corporate Accounts, Alltech, 3031 Catnip Hill Road, Nicholasville, KY 40356, USA; Associate Professor of Marketing, University College, Dublin, Ireland; Adjunct Agribusiness Professor, China Agricultural University, Beijing, China P.R.

Abstract The grocery store is ground zero in the tsunami of change facing Big Food. Consumers are changing how they relate to grocery stores, increasingly circling the perimeter, focusing on produce and preferentially choosing fresh, local, and new, even unknown, brands while spending less time in the processed food aisles in the center. The next generation, the millenials, are increasingly shunning traditional outlets when buying food. Traditional leading brands of processed food, backed by traditional marketing strategies (heavy advertising on traditional media, coupons, brand extensions, etc.) are failing to hold on to their customers. The challenges can be found throughout the food value chain, from new competitors for grocery providers to new delivery mechanisms, from changes in generational food preferences with social media platforms to express their preferences to farmers who increasingly can and want to communicate directly with the end-users who actually eat the food that they produce. This access to more information opens more options (and opportunities) to buyers and suppliers all along the food value chain. Barely 100 years old, the grocery store model is becoming obsolete, and with it the organization of the food value chain must be re-written. So what does that mean for Big Food and the food supply chain? What directions can the industry take to adjust to the new competitive realities? This paper offers direction and guidance for Big Food and other producers in the food supply chain. Keywords: Big Food, millenials, Amazon, Whole Foods, food supply chain JEL code: M00 Corresponding author: aconnolly@alltech.com

ÂŠ 2017 Phillips-Connolly and Connolly

Phillips-Connolly and Connolly

Volume 20, Issue 5, 2017

1. Introduction Amazon’s purchase of Whole Foods ended any doubt that Big Food is in a battle for survival and that the grocery store is the battlefront. Worse for Big Food was the sharp drop in share prices of leading grocery providers when the deal was announced (Thomas, 2017), with the market leaders in the US, UK and France falling from 5-19% in a single day (Shen, 2017), in effect a bet by the financial markets on who they think is winning the battle. It is an important story, but it is just the beginning: the changes facing food retailers will be rippling through the entire food value chain, profoundly affecting not just Big Food, but the entire chain.

2. Food retailing There are a number of obvious synergies for Amazon with the purchase of Whole Foods, including: a start toward becoming one of the top five grocers by 2025 (Rosen, 2017); a tidy tie in with AmazonFresh (providing fresh foods through online ordering), and access to an affluent demographic that is already voting for healthy, clean label food with their food dollars. The purchase also reflects demographic changes: Amazon is how millennials are used to shopping. Most millenials have no memory of life before Amazon, and it is ubiquitous in their lives. The two-thirds who have attended any form of college or university (Ryan and Bauman, 2016) came of age with free Amazon Prime for ordering their textbooks. In major cities, they are already accustomed to delivery in two hours through Prime Now. Yet Whole Foods has just 1.2% of the US grocery market, and Amazon just 0.2%, and overall online food and beverage purchases account for just 2% of the total food and beverage market in the US (The Economist, 2017). Given the small scale, the response of the marketplace might seem overly dramatic, but it reflects an awareness by the financial markets that the purchase signifies deeper changes than just the venue or delivery mechanism for buying food products. As one analyst noted, ‘I think this [deal] is certainly a game changer. Amazon’s strategy will make retailers like Target and Wal-Mart ask themselves questions about where retail is headed.’ (Thomas, 2017).

3. The Big Food landscape Consumers are changing how they relate to grocery stores, increasingly circling the perimeter, focusing on produce and preferentially choosing fresh, local, and new, even unknown, brands while spending less time in the processed food aisles in the center. The next generation, the millenials, are increasingly shunning traditional outlets when buying food (Connolly, 2016a), and those who do wander the grocery aisles are making different choices. Traditional leading brands of processed food, backed by traditional marketing strategies (heavy advertising on television and media, coupons, brand extensions, etc.) are failing to hold on to their customers. The food landscape is being torn apart, not by the food companies or government actions, but by consumers, led by prosumers (Connolly, 2016b; Havas, 2016). Since 2009, the lost market share of top U.S. food and beverage companies is equivalent to approximately $18 billion (Kowitt, 2015). Years of fast growth and consolidation into segment-dominating behemoths has created companies that are too big to react nimbly to changes in the marketplace. These super-sized companies were dubbed ‘Big Food’ in a nod to the market dominance of ‘Big Pharma’ that these companies aspired to. Instead they are feeling the heat of competition in a suddenly unfamiliar arena. Companies along the food value chain, from Archer Daniels Midland and Mondelez at one end, to CocaCola and Campbell Soup at the other, projected lower revenues and gross profits for 2016 (Higgins, 2017); sales in traditional supermarkets are projected to fall in parallel over the next several years (Ruddick, 2014). The shift in consumer preferences is profound. The breakfast cereal aisle, for 100 years home to a key part of breakfast, was recently called the most dangerous in the grocery store by celebrity chef Michael Ruhlman International Food and Agribusiness Management Review

616

Phillips-Connolly and Connolly

Volume 20, Issue 5, 2017

(Strom, 2017). Sales of orange juice, another staple of a ‘healthy breakfast’, dropped 13% in just four years (Peterson, 2016) as consumers realized that in terms of sugar content, orange juice is not much healthier than soft drinks, a category whose sales have declined by nearly 20% in six years (Sanger-Katz, 2015). Bottled water sales now surpass soft drinks (Allegri, 2017). The major soft drink manufacturers noticed this trend several years ago, and now own their own market-leading bottled water brands (e.g. Dasani by Coca-Cola and Aquafina by PepsiCo). Meanwhile, after decades of relatively slow growth, organic food sales more than tripled over the last decade (Kowitt, 2015). Even conservative, big box, price-conscious chains such as Aldi and Wal-Mart expanded their organic produce sections to meet demand: More than 80% of US households now buy some type of organic product (OTA, 2017). The trend toward local, organic and less processed foods has become large enough to sustain new (or in some cases renewed) food pathways. From farmers markets to home-grown food production companies, consumers increasingly seek out both convenient and clean foods (i.e. not laden with unnatural ingredients and chemicals) (The Hartman Group, 2016). Both Whole Foods and Trader Joe’s epitomize these efforts: healthy, fresh clean label food, an emphasis on own-label products, and simple meals packaged and ready to make. In the meantime, Blue Apron, Hello Fresh and their many competitors in the home delivery of ready-to-make meal kits are removing the need to even go to the market. The Amazon purchase of Whole Foods demonstrates this trend towards ‘food as a delivery service...what is technically known as ‘consumer convenience’ and what is commonly observed as ‘human sloth.’’ (Thompson, 2017). So, the boxed macaroni with powdered processed cheese (or ready to be microwaved) that many millenials grew up with is convenient; boxed macaroni and cheese made by a company started by a mom using all natural ingredients is just as convenient and ‘healthier’; store-made ready to eat macaroni and cheese at the salad bar in the grocery store is even more convenient and seems even ‘healthier’ yet. Finally, ‘home-made’ from a kit delivered to your door is the most convenient and ‘healthiest’.

4. Transformative change The challenge to the food value chain is not simply new competitors for grocery providers, or new delivery mechanisms. Nor is it merely changing food preferences or a new generation with its own ideas and social media platforms for expressing them. It is also farmers who increasingly want, and have the ability, to communicate directly with the end-users who actually eat the food that they produce. It is not just high end, but mid-level restaurants sourcing their ingredients directly from specialty producers and restaurants serving 3-D printed food (Ahmed, 2017). It is the availability of information to participants throughout the value chain, whether marketers mining social media data instead of crude demographics or farmers choosing crops to plant based on production in other parts of the world. In short, it is an industry in the middle of transformative change. Barely 100 years old (Ross, 2016), the grocery store model is becoming obsolete, and with it the organization of the food value chain gets re-written. Participants in the food value chain are now experiencing a period of rapid change rapid change and experimentation as the market finds a ‘new normal’. McGahan (2004) describes industries in this situation as being on an ‘intermediating trajectory’, one in which businesses must protect the value of their assets while restructuring their trading relationships. The dramatic increase in information availability and contact between all parts of the chain (including the rise of the prosumers), and the rise in ways to get food from producers to end consumers, puts the food value chain firmly on the ‘intermediating trajectory’. The ‘intermediating trajectory’ is hard to recognize initially because only individual segments of the value chain are affected, while the period of rapid change makes it particularly difficult to navigate (McGahan, 2004). Companies must identify which of their core assets continue to have value, while preparing to drop segments as they become unprofitable, while continuously re-evaluating trading relationships with both customers and suppliers. The core assets of Big Food and the food value chain (brands, knowledge, International Food and Agribusiness Management Review

617

Phillips-Connolly and Connolly

Volume 20, Issue 5, 2017

production and distribution systems, etc.) still have value, but all the participants will need to adjust to the new environment. During this period of change and transition to an as-yet unknown market structure, taking a ‘reactive segmentation’ (Phillips-Connolly, 2007) approach, at both operational and strategic levels, provides a framework for decision-making. Operationally, the key is to be prepared for volatility. Participants in the food chain need to assess their value chain modularly, to identify the relative profitability of each segment, and prepare to drop elements as they become unprofitable. Steps to take include: ■■ Pursue active, continuous information gathering, particularly on the changing nature of the competition, but also industry capacity, reduction of variance and forecasting errors. ■■ Review the value chain, tracking the relative profitability of core activities so that as soon as they are no longer profitable they can be outsourced or dropped. ■■ Manage existing relationships, being prepared to change or unwind them as the value chain segments and reconfigures, while staying particularly responsive to the interests of core customers. ■■ Improve efficiency and effectiveness through incremental process improvements (not through major investment) as long as business is profitable. ■■ Use supply chain management techniques such as reducing lead times, postponement and build to order, minimizing stock levels, and third-party logistics. Cost cutting through substitution, ‘shaving’ (using less expensive ingredients or packaging) and rationalization are familiar and well-developed tools within Big Food, and it is already deploying all of these approaches in reaction to changes in the marketplace. It can be difficult to recognize that even successful product areas may need to be jettisoned. Unilever recently decided that, despite owning the top two brands of margarine, the decline in the market for spreads (margarines, etc.) in developed markets is irreversible, and are no longer profitable enough to keep (Bray, 2017) and are selling that business unit.

5. Strategies In most segments of the food chain the dominant strategy has been to develop deeper relationships with trading partners, but in this period of rapid change it is important to reassess the terms of each of these relationships. This can be particularly challenging for firms whose relative power in the relationship is low (for example, a supplier to Wal-Mart). Points to consider during this period of volatility include: ■■ Segment buyers by profitability and longer-term prospects. ■■ Identify trading partners (suppliers and buyers) long term interests, as a basis for develop risk-andbenefit sharing approaches and incentives to motivate trading partners and align interests. ■■ Use alliances with new entrants to the industry to access knowledge about the evolving market paradigm, but be aware that the new paradigm may be using the resources and/or credibility of the old as a springboard to help itself grow – hence the volatility of these relationships, which rarely last. ■■ Assume that alliances with both existing and new trading partners (including exclusivity, access to needed assets through partnering. joint ventures, mergers, acquisitions, etc.) will be limited in both scope and duration. ■■ Develop strong internal-communication systems to manage information flows and to identify when to decouple activities. Big Food is responding to changing trends by adjusting recipes (for example, to reduce sugar or salt), being more ‘natural’, buying small ‘healthy’ food makers, and creating organic or natural own brands in an effort to stay ahead of the trend towards healthier food. For example, General Mills bought buying Annie’s, a small line of organic pastas and snacks primarily targeted at children; and Kroger’s successfully launched an own-label line of organic/healthy/natural food and non-food products (‘Simple Truth’). Unilever’s sale of its spreads business while simultaneously buying Sir Kensington’s (a maker of specialty condiments) International Food and Agribusiness Management Review

618

Phillips-Connolly and Connolly

Volume 20, Issue 5, 2017

reflects a re-focusing of their food segment away from foods now seen as ‘processed’ and prosaic and toward specialty condiments and desserts. Sir Kensington’s needed the market access that Unilever provides, and Unilever recognized that consumers are increasingly willing to try new, ‘clean’ label products and aimed to contain the ‘threat’ to legacy brands such as its market leading mayonnaise, Hellmann’s (Kell, 2017). In practice, though, most of these examples are simply doing more of the same. Unilever bought Ben & Jerry’s ice cream more than 15 years ago, saying that it represented a trend towards ‘enjoyment of life’ (Hays, 2000). The robust growth of the Aldi and Lidl ‘hard’ discount grocery chains can seem simply a more intense phase of normal competition, and their emphasis on carrying a limited product range of largely private label products, with very low prices (Choi, 2017) as extensions of the Sam’s Club or Costco model. On the ‘Intermediating’ trajectory, the linear understanding of the chain itself is under threat. General Mills, one of the giants of Big Food, has one of the best diagrams of the food chain. Direct communications between purple consumers and virtually all the other segments, growing direct communications and even direct trading relationships between orange purveyors and the blues and greens of early stage production are examples of the process of disintermediation engulfing the food value chain (Figure 1). As the examples above indicate, food companies are adjusting to the changing marketplace. The question is: do they see this as simply a normal course correction, or do they recognize the scale, volume and velocity of change that is underway?

6. Summary Transformative change can be hard to recognize, especially in the early stages. Big Food, along with many of the participants in the food value chain, might hope that building a social media presence and tweaking their offerings will be enough see them through the current changes. The fast growth of hard discounting grocery chains such as Lidl and Aldi may reassure other supermarkets that the role of supermarkets is still strong. But it is more likely that this is the swan song of the old order. Nobody knows how long the song will last, or what the new song will be, but as the market tumble from Amazon’s purchase of Whole Foods indicates, there are players who are already warming up. Using the steps outlined here can help Big Food and other participants in the food chain prepare for the big changes that are coming.

Agriculture: Growing crops

Transforming: Turning crops into food ingredients

Converting: Making products from raw ingredients

Packaging: Producing packaging materials

Shipping: Moving food to stores

Selling: Making food available for purchase

Figure 1. Food supply chain. Rendition of original image, courtesy of General Mills.

International Food and Agribusiness Management Review

619

Consuming: Enjoying food

Phillips-Connolly and Connolly

Volume 20, Issue 5, 2017

References Ahmed, R. 2017. 4 famous restaurants that use 3D printers. 3D Printing online, March 24. Available at: http://tinyurl.com/yc9dnz8s. Allegri, C. 2017. Americans drank more bottled water than soda in 2016. Reuters online, March 9. Available at: http://tinyurl.com/yaqojd9a. Bray, C. 2017. Unilever to sell its spreads business and restructure. New York Times online, April 6. Available at: http://tinyurl.com/yd5of23h. Choi, C. 2017, Europe’s Aldi and Lidl take on U.S. grocery giants with discount strategy. Seattle Times online, June 15. Available at: http://tinyurl.com/yaqbao2s. Church, J. 2015. How General Mills is advancing a sustainable supply chain. General Mills. Available at: http://tinyurl.com/y8zo7wqs. Connolly, A. 2016b. Living in the age of the prosumer. LinkedIn Pulse article published on September 30. Available at: http://tinyurl.com/y7o4pmhc. Connolly, A. 2016a. The future of the supermarket: a daycare for the elderly? LinkedIn Pulse article published April 8. Available at: http://tinyurl.com/ya5g42mk. Havas, 2016. Eaters digest: the future of food. Havas the Mag, July 26. Available at: http://tinyurl.com/ yauor4fy. Hays, C. L. 2000. Ben & Jerry’s to Unilever, with attitude. New York Times, April 13. Available at: http:// tinyurl.com/2a75qwj. Higgins, K.T. 2017. Poor sales and profits in 2016 for top food and beverage companies. Food Processing online, April 13. Available at: http://tinyurl.com/y9a67azj. Kell, J. 2017. The maker of Hellmann’s mayonnaise has bought a fancier upstart. Fortune online, April 20. Available at: http://tinyurl.com/yaagncfp. Kowitt, B. 2015. Special report: the war on big food. Fortune online, May 21. Available at: http://tinyurl. com/nkzt72r. McGahan, A.M. 2004. How industries change. Harvard Business Review online, October issue. Available at: http://tinyurl.com/yadpdl8l. OTA. 2017. Organic, big results available at small seeds. Organic Trade Association Infographic. Available at: http://tinyurl.com/y8xyvj4g. Peterson, H. 2016. orange juice is being called a massive scam – and now it’s disappearing at breakfast in America. Business Insider, October 13. Available at: http://tinyurl.com/y7hzhogc. Phillips-Connolly, K. 2007. A typology for strategic supply-chain management: bridging the gap between operations and strategy. Thesis from Trinity’s Access to Research Archive. Available at: http://tinyurl. com/y8gz8fsz. Rosen, E. 2017. Why Amazon bought whole foods. L2: Daily Insights, June 19. Available at: http://tinyurl. com/ya33jfdg. Ross, A. 2016. The surprising way a supermarket changed the world. Time Online, September 9. Available at: http://tinyurl.com/ybe4a6p9. Ruddick, G. 2014. Superstores could have just five more years on top. Telegraph online, June 29. Available at: http://tinyurl.com/ybv6addr. Ryan, C.L. and K. Bauman. 2016. Educational attainment in the United States: 2015. Available at: http:// tinyurl.com/l5r62f3. Sanger-Katz, M. 2015. The decline of ‘big soda’. New York Times online, October 2. Available at: http:// tinyurl.com/mabpu94. Shen, L. 2017. Amazon’s $13.7 billion whole foods deal just made grocery stocks plunge. Fortune online, June 16. Available at: http://tinyurl.com/ybhn42nv.

Strom, S. 2017. What’s new in the supermarket? A lot, and not all of it good. New York Times online, May 16. Available at: http://tinyurl.com/ya7m3uu9. The Economist. 2017. An industry shudders as Amazon buys Whole Foods for $13.7bn. The Economist June 16. Available at: http://tinyurl.com/yb85ogp6.

International Food and Agribusiness Management Review

620

Phillips-Connolly and Connolly

Volume 20, Issue 5, 2017

The Hartman Group. 2016. Is a reckoning at hand for big food companies? Forbes online, August 17. Available at: http://tinyurl.com/y7szxt2r. Thomas, L. 2017. Target, Wal-Mart endure the aftermath of Amazon’s Whole Foods buy; stocks still falling. Available at: http://tinyurl.com/y9eslbp3. Thompson, D. 2017. Why Amazon bought whole foods: the retailer’s $14 billion bet isn’t just about the future of food. It’s about the future of commerce – especially for rich urban consumers. The Atlantic online, Jun 16. Available at: http://tinyurl.com/y8r8cybs.

International Food and Agribusiness Management Review

621

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2016.0097 Received: 9 May 2016 / Accepted: 17 July 2017

U.S. milled rice markets and integration across regions and types RESEARCH ARTICLE Man-Keun Kim a, Hernan Tejedab, and T. Edward Yuc aAssociate

Professor, Applied Economics, Utah State University, 4835 Old Main Hill, Logan, UT 84322, USA

bAssistant

Professor and Extension Specialist, Agricultural Economics and Rural Sociology, Twin Falls Research and Extension Center, University of Idaho, 315 Falls Avenue East, Twin Falls, ID 83301, USA cAssociate

Professor, Agricultural and Resource Economics, University of Tennessee, 2621 Morgan Circle, Knoxville, TN 37996, USA

Abstract Rice is among the top seven U.S. major crops in terms of harvested acres – covering over 2.6 million acres – and sixth in terms of sales, with annual cash receipts around 3.1 billion dollars. This paper investigates whether U.S. milled rice markets are integrated across regions and whether these markets are integrated by rice types. Understanding dynamic relationships across regions and types provides important insights for risk management and policy making. Of the four major producing regions, three are in the South – ArkansasMissouri, Louisiana-Mississippi, and Texas – and the other is California. There are different rice types associated with a production region. California mainly produces short and medium grain; while Arkansas, Texas, and Louisiana primarily produce long and also medium grains. We determine the potential market integration of these rice markets by applying a Vector Error Correction Model and Directed Acyclic Graphs to monthly free on board milled rice price data from August 1986 to December 2015. Results suggest that Arkansas-Missouri region is the leading price reference in the long grain markets. Arkansas-Missouri medium grain also plays an important role in the medium grain markets. California medium grain markets are weakly exogenous in the short run, but affected by Arkansas-Missouri medium grain in the longer term. As anticipated, Arkansas-Missouri long grain milled rice markets are driven by rough rice futures price in the longer term. Interestingly, Arkansas-Missouri medium grain market has a sizable impact on long grain markets even though long and medium grains are not substitutes. This may be due to land competition to long grain rice production in Arkansas, a major area of long grain rice production. Keywords: rice markets, cointegration, impulse response functions JEL code: Q11, Q13, C32 Corresponding author: mk.kim@usu.edu

623

Kim et al.

Volume 20, Issue 5, 2017

1. Introduction Rice is considered as a staple commodity in major world markets â&#x20AC;&#x201C; especially Asia, Africa and Latin America. About 20% of the caloric intake of the world population is from rice (Giraud, 2013). Rice also plays a major role in U.S. agriculture and is produced in four distinct regions, i.e. Arkansas-Missouri, Mississippi Delta (parts of Mississippi, Missouri, Louisiana and Arkansas), Texas-Southwest Louisiana, and California (mainly Sacramento Valley). It is among the top seven U.S. crops in terms of harvested acres, covering over 2.6 million acres in 2013 to 2015,1 and sixth in terms of sales (cash receipts) with annual transactions of over 3.1 billion dollars. In addition, U.S. is a major rice exporter, accounting for about 8 to 9% of the annual volume of global rice trade (ERS, 2015). There are three types of rice grown in the U.S., classified according to the length of grain as long, medium and short grain. Moreover, several varieties of each type are produced each year. Long grain type of rice is almost entirely produced in the southern regions; i.e. Arkansas, Louisiana, Mississippi, Missouri and Texas; and long grain covers about 70% of the total U.S. rice production. Also, Arkansas produces about 65% of all long grain rice. The medium grain is mainly produced in California, while a much smaller production is located in the southern regions â&#x20AC;&#x201C; mostly in Arkansas and far behind in Louisiana. Medium grain production accounts for over 25% of the total U.S. rice production. Less than 2% of total rice production is short grain and nearly all U.S. short grain rice is grown in California (Childs, 2012). Figure 1 presents the time trend of rice production over types across regions between 1980 and 2015. More than 200 million cwt (centum or hundred weight equivalent to 100 pounds) of rice were annually produced on average during the 2000s decade. The blue and light blue areas together indicate long grain production from Arkansas-Missouri,

These years include periods of severe drought in California and Texas that resulted in reduced plantings; i.e. the size of industry may be quite larger than that of the data during these years.

AR_lng

LA_lng

TX_lng

AR_med

LA_med

CA_med

250 225 200

Millions cwt

175 150 125 100 75 50 25 0 1980

1985

1990

1995

2000

2005

2010

2015

Figure 1. U.S. rice production of different varieties and regions (USDA NASS; https://www.nass.usda. gov). AR_lng = AR-MO long grain; LA_lng = LA-MS long grain; TX_lng = Texas long grain; AR_med = AR-MO medium grain; LA_med = LA-MS medium grain; CA_med = CA medium grain. International Food and Agribusiness Management Review

624

Kim et al.

Volume 20, Issue 5, 2017

Louisiana-Mississippi, and Texas. The combined brown and light brown areas represent the medium grain production across Arkansas, Louisiana, and California. Global rice markets have very little substitution between types of long-grain and medium-grain rice. Japan, Korea and other Northeast Asian countries – the largest export markets for U.S. medium- and short-grain rice – consumes only ‘higher quality’ medium and short grain rice from California. Note that almost all U.S. rice purchased by Northeast Asia is from California. Conversely, Southern medium-grain export sales are much smaller and limited to North Africa and the Middle East. In the U.S. domestic market, substitution between grain types hardly occurs and only in case of processed food products, where mostly southern medium-grain is used.2 Price analysis of agricultural commodities in the U.S. has been studied extensively in the literature (e.g. Fackler and Goodwin, 2001; Serra and Goodwin, 2004; Stockton et al., 2010; Yu et al., 2007). However, it is surprising – given its economic value – that dynamic relationships among U.S. rice prices is still limited. Taylor et al. (1996) investigated the dynamic relationships between U.S. and Thai rice prices but price dynamics within U.S. rice markets was not addressed in their study. Djunaidi et al. (2001) study long-run relationships between long grain rice markets from Southern states and California, finding only existing for Southern states and not with the California rice market. In addition, McKenzie et al. (2002) study the efficiency of the U.S. long-grain rough rice futures market and find evidence in favor of it, concluding that rice futures market should be considered useful as a price risk management and forecasting tool. The analysis of price dynamics across regions and types is important to understand the U.S. rice market structure and, in turn, helpful for improving price transparency in the markets. Thus, the objective of this study is to investigate and identify the dynamic relationships of the prices of two U.S. rice types, i.e. medium and long grain, among major domestic producing markets. The analysis will provide insights in the price discovery process among separate U.S. rice markets. Findings on price discovery among rice markets and types can potentially mitigate marketing risk for producers and wholesales, also assist policy makers to stabilize the market through policy tools. We employ a structural multivariate Vector Autoregression model with an error correction term (VECM). Multivariate time series models such as the VECM have been commonly used in the literature of price analysis across regions. To formulate a structural VECM, a Directed Acyclic Graphs (DAG) from Pearl (1995, 2000) and Spirtes et al. (2000) is utilized to sort-out the instantaneous causal flows among the innovations from the VECM (Hoover, 2005) and used to construct the structural decomposition of the VECM residuals (Swanson and Granger, 1997).

2. Data We use average monthly f.o.b. (free on board)3 milled prices in $/cwt from major milling centers located in each specific region. We are interested in f.o.b. milled prices because the bulk of U.S. rice is sold as milled. In addition there is no prior study addressing price discovery of these milled rice markets. Based on product-weight, about 50% of rice exports is in the milled form (USDA, 2015: Tables 12-13); however, actually about two-thirds of U.S. rice exports are classified as milled rice (milled and brown rice on a roughbasis). In particular, we use price data for grain varieties of Arkansas long (AR_lng), Arkansas medium (AR_med), Louisiana long (LA_lng), Louisiana medium (LA_med), Texas long (TX_lng), and California medium (CA_med) obtained from the Agricultural Markets Service – USDA (USDA, 2017: Table 17). Prices considered are from August 1986 to December 2015. Figure 2 illustrates these prices for the study period. The spike of prices between 2008 and 2010 is related to global rice price movements over the same period (Childs and Kiawu, 2009). After 2010 long grains in Arkansas, Texas, and Louisiana (blue-light blue solid lines) are lower prices than the medium rice varieties (dotted lines). Descriptive statistics of the data are reported in Table 1. 2 3

We thank an anonymous referee for pointing this out. Shipping terminology that means price of product includes loading on top of truck, rail cargo, vessel, or other.

International Food and Agribusiness Management Review

625

Kim et al.

Volume 20, Issue 5, 2017 LA_lng

TX_lng

AR_lng

x_cme

AR_med

CA_med

LA_med

55 50 45 40

$/cwt

35 30 25 20 15 10 5 0

1987

1992

1997

2002

2007

2012

Figure 2. U.S. (milled) rice prices of grain size and milling states (Aug. 1986 ~ Dec. 2015) (adapted from USDA, 2017: Table 17). AR_lng = AR-MO long grain; LA_lng = LA-MS long grain; TX_lng = Texas long grain; AR_med = AR-MO medium grain; LA_med = LA-MS medium grain; CA_med = CA medium grain; x_cme = rough rice futures price from CME Group Inc. (Chicago Mercantile Exchange and Chicago Board of Trade). Table 1. Descriptive statistics of data ($/cwt), August 1986-December 2015 (353 observations) (adapted from USDA, 2017: Table 17).1 Arkansas long grain Louisiana long grain Texas long grain Arkansas medium grain Louisiana medium grain California medium grain CME rough rice futures 1

AR_lng LA_lng TX_lng AR_med LA_med CA_med x_cme

Mean

Std. Dev.

CV(%)

Min.

Max.

Autocorr.

18.99 18.57 20.18 20.98 20.86 24.54 9.44

6.34 6.18 6.43 8.30 8.23 10.30 3.68

33.40 33.27 31.89 39.57 39.43 41.95 38.99

8.56 9.13 10.50 10.06 10.00 11.50 3.51

42.50 43.25 44.00 46.25 43.25 52.25 21.48

0.984 0.983 0.984 0.990 0.993 0.993 0.971

Price data are not deflated.

Rough rice futures price is also included in the analysis. Incorporating rough rice futures price in the analysis can help us investigate whether it provides information in the price discovery of the U.S. milled rice markets since rough rice is a major input of milling facilities. Taylor et al. (1996) point out that rough rice futures markets provide market participants with information regarding local, national and international rice market as well as serving as a primary discovery mechanism. Rough rice futures prices are compiled from CME available from Quandl; http://tinyurl.com/ycy7s9u8.

International Food and Agribusiness Management Review

626

Kim et al.

Volume 20, Issue 5, 2017

3. Methods We employ the framework used by Bessler and Akleman (1998), Bessler and Yang (2003) and Stockton et al. (2010), which combines the DAG method and multivariate time series modeling, to explore the price dynamics of rice markets. Given the non-stationarity nature of the data, we specify a VECM of the U.S. rice market with the seven selected prices. After the VECM is estimated, the contemporaneous innovations (residuals) are obtained. The DAG analysis then identifies the contemporaneous causal relationships among these innovations. This enables to address our matter of interest, that is, the dynamics of the variables which are investigated by using innovation accounting (impulse response functions (IRF) and forecast error variance decompositions (FEVD)). First, the data series are tested for non-stationarity using the Augmented Dickey-Fuller (ADF) test considering a constant (Dickey and Fuller, 1979) and Kwiatkowski-Phillps-Schmidt-Shin (KPSS) test (Kwiatkowski et al., 1992). For the ADF test, the optimal lag length for the augmented terms was determined by minimizing the Schwarz-loss statistics (SL). A unit root was found in five out of seven price series based on the ADF test (Table 2). The KPSS test suggested all series are not stationary. The unit root test for the data in first difference is in the second half of Table 2 and suggests that those price series are integrated of order one, or I(1), given test results for the first difference is stationary. Table 2. Non-stationarity tests.1 Raw data

Long grain AR_lng

ADF test (non-zero mean) Test statistics -2.82 Lag using SL 1 5% critical value -2.86 NS Decision2 KPSS test (level stationary) Test statistics 2.15 3 Lags 5 5% critical value 0.463 Decision NS First difference

Medium grain LA_lng

TX_lng

x_cme

AR_med

CA_med

LA_med

-3.25 1 -2.86 S

-2.97 1 -2.86 S

-2.77 1 -2.86 NS

-2.50 1 -2.86 NS

-1.69 1 -2.86 NS

-2.40 1 -2.86 NS

2.26 5 0.463 NS

2.96 5 0.463 NS

2.34 5 0.463 NS

3.18 5 0.463 NS

3.98 5 0.463 NS

3.56 5 0.463 NS

ΔAR_lng ΔLA_lng ΔTX_lng Δx_cme

ADF test (non-zero mean) Test statistics -7.92 Lag using SL 0 5% critical value -2.86 Decision S KPSS test (level stationary) Test statistics 0.034 Lags 5 5% critical value 0.463 Decision S

ΔAR_med ΔCA_med ΔLA_med

-8.52 0 -2.86 S

-8.35 0 -2.86 S

-9.92 1 -2.86 S

-8.02 0 -2.86 S

-8.18 0 -2.86 S

-8.14 0 -2.86 S

0.064 5 0.463 S

0.025 5 0.463 S

0.050 5 0.463 S

0.042 5 0.463 S

0.034 5 0.463 S

0.045 5 0.463 S

AR_lng = AR-MO long grain; LA_lng = LA-MS long grain; TX_lng = Texas long grain; AR_med = AR-MO medium grain; LA_med = LA-MS medium grain; CA_med = CA medium grain; x_cme = CME rough rice futures; ADF = Augmented DickeyFuller test; SL = Schwarz-loss statistics; KPSS = Kwiatkowski-Phillps-Schmidt-Shin test. 2 NS = nonstationary; S = stationary. 3 The number of Newey-West lags, {4(T⁄100)29)}lags where T is the number of observations.

International Food and Agribusiness Management Review

627

Kim et al.

Volume 20, Issue 5, 2017

After confirming the presence of unit roots, the Johansen’s Trace test for co-integration (Johansen, 1991) was applied to determine the possible presence of any long-run stationary relationships among the prices. In addition, to determine the optimal lag of the VECM, we first determine the optimal lag of the corresponding level vector autoregression (VAR) since the optimal lag length of the VECM is one less than that of the corresponding level VAR. The optimal lag of the level VAR is determined based on the Schwarz Loss metric. The optimal number of lags in the series was determined as two. Thus, for the VECM, the optimal lag length is one. The Johansen trace test provides the information on the cointegrating vectors and results are reported in Table 3. Based on the trace-test statistics regarding the rank hypothesis, the number (r) of cointegrating vectors was determined to be five. Trace* and C* refer to the values of the trace statistic and the critical values at the 5% significance level considering an intercept, while Trace and C refer to the values of the trace statistic and the critical values at the 5% significance level considering a trend and intercept. Given these results, we apply the VECM to our series of prices based on the procedure described in Lütkepohl and Krätzig (2004). Let yt denote the vector of variables under consideration, yt'=[y1t, ...,y7t], where the subscript 1 represents the prices series of AR_lng, subscript 2 represents the price series of LA_lng and so on. The VECM model with one lags is represented as: ∆yt = Πyt–1 + Γ∆yt–1 + μ + et (t = 1,........T)

(1)

where Δ is the first difference operator (e.g. Δyt=yt–yt–1); yt is a (7×1) vector of prices; Π is a 7×7 coefficient matrix of rank r, i.e. number of co-integration vectors such that Π=αβ'. α is a 7×5 matrix of weights knows as the speed of adjustment parameters and β is the 5×7 matrix of cointegrating vectors. Γ is a 7×7 matrix of short-run dynamic coefficients; and et is a 7×1 vector of innovations. After estimating the VECM of 1 lag in Equation (1), we identify the contemporaneous structure of the innovations through the DAG analysis of the correlation matrix of residuals, êt. The DAG method, as described by Pearl (1995, 2000) and Spirtes et al. (2000), considers a non-time sequence asymmetry in causal relations among variables resulting in an illustration using arrows and vertices (variables) to represent the causal flow among a set of variables (Pearl, 2000). DAG represent a conditional independence relationship as shown by the recursive decomposition: n

Pr (v1,v2,... ... ...,vn) = Π i=1 Pr (vi│pri) (2) where Pr (.) is the joint probability of variables v1,v2,...,vn and pri represents ‘parents’ of vi , a minimal set of predecessors (variables that come before in a causal sense) that renders vi independent of all its other Table 3. Trace test on order of cointegration. Rank

Trace*1

C*1

Decision

Trace2

Decision

r=0 r≤1 r≤2 r≤3 r≤4 r≤5 r≤6

280.52 172.71 108.36 65.36 38.57 20.13 5.16

134.54 103.68 76.81 53.94 35.07 20.16 9.14

Reject Reject Reject Reject Reject Fail3 Fail3

324.00 216.00 142.51 93.50 52.25 27.37 8.97

150.35 117.45 88.55 63.66 42.77 25.73 12.45

Reject Reject Reject Reject Reject Reject Fail

Trace* and C* refer to the values of trace statistic and critical values at the 5% significance level with an intercept. and C refer to the values of trace statistic and critical values at the 5% significance level with a time trend and an intercept. 3 The first ‘fail to reject’ the null hypothesis occurs for r≤5. Thus, there are 5 cointegrating vectors. 2 Trace

International Food and Agribusiness Management Review

628

Kim et al.

Volume 20, Issue 5, 2017

predecessors (Pearl, 2000: 14-15). Geiger et al. (1990) have shown that there is a one-to-one correspondence between the set of conditional independencies among variables implied by Equation 2 and the graphical expression of variables in a directed acyclic graph. For example, consider four variables, v1, v2, v3, and v4. If there is causal relationship such as v1, v2 cause v3 and v3 causes v4, then the directed graphs that represents in this causal relationship can be represented in Figure 3. The directed graph is expressed as the probability distribution product by: Pr (v1,v2,v3,v4) = Pr (v1) Pr (v2) Pr (v3│v1,v2) Pr (v4|v3) (3) A Linear Non-Gaussian Acyclic Model (LiNGAM) algorithm, developed by Shimizu et al. (2006a), identifies the causal patterns in the case of non-normal data (Shimizu and Kano, 2008). Our study applies LiNGAM following a recent study by Lai and Bessler (2015). This algorithm determines the causal directionality based on functional composition (Pearl, 2009) by making use of independent component analysis (ICA). In effect, and as noted by Lai and Bessler (2015), the Central Limit Theorem (CLT) affirms that any mixture of independent variables generally has a distribution that is closer to a normal distribution than that of any of the original variables (Stone, 2004). Assuming we observe the mixtures, x=(x1,...,xn) from the independent (unobserved) variables v=(v1,...,vn), then x=Av where v are mutually independent components. From CLT, any of the v is less Gaussian than the (mixture) variables x. ICA’s main goal is to determine the ‘de-mixing’ matrix W, such that W maximizes the sum of the non-normal, mutually statistical independent components of ṽ where ṽ=W̃ x and x=A-1 (Shimizu et al., 2006b). In the case of LiNGAM (Lai and Bessler, 2015; Shimizu et al., 2006a), we assume existing causal relationships among the vector x=(x1,...,xn) characterized by the linear equation model: xi = ∑k(j)<k(i) bijxj + ei

(4)

where k(i) denotes a causal order of xi and xj is a direct cause of xi. The disturbances ei are mutually independent and non-Gaussian distributed with non-zero variances. Assuming each xi has a zero mean, we have: x=Bx+e (5) where B=[bij] is the coefficient matrix of the model. Solving for x, we obtain: x=(I–B)-1 e=Ae (6) From Equation 5 and non-Gaussian disturbances, we obtain the classical linear ICA model. Equation 5 may be rewritten as: e=(I–B)x=W̃ x (7) The LiNGAM algorithm operates by initially conducting ICA estimation to obtain the matrix A and the permuting and normalizing it before computing B. It is important to remark that some estimated edges between variables may be weak and are most likely zero in the generating model. The Wald test is appropriately used to determine if some remaining connection(s), such as these weak estimated edges, are to be ‘pruned’ as noted by Shimizu et al. (2006a) and Lai and Bessler (2015).

2 Figure 3. Example of directed acyclic graph.

International Food and Agribusiness Management Review

629

Kim et al.

Volume 20, Issue 5, 2017

After obtaining the DAG results, we estimate structural innovations directly from the reduced form residuals by applying the additional (obtained) contemporaneous restrictions (Lütkepohl, 2005: 362). We then use standard innovation accounting techniques to obtain inferences with respect to the dynamic adjustments in each of the variables from unexpected shocks in the series. The FEVD consists of when the variance of each variable’s forecasted error is decomposed, permitting the identification of the relative proportion of the movements in that sequence due to its own shocks, over shocks to the other variables. In the case that own shocks explain mostly all of the forecasted error’s variance of a specific series, this variable may be considered (weakly) exogenous with respect to the other variables in the system. Conversely, if a large proportion of the FEVD from a variable’s sequence can be explained by shocks to one or more of the other variables, then this variable is considered endogenous to the system. The FEVD approach likewise permits to draw inferences with respect to the magnitude and degree of influence on the sequence, among the variables in the system. In addition, IRFs are likewise determined through standard innovation accounting. IRFs permit to identify the dynamic adjustments, in terms of direction and magnitude, for each variable in the system in response to unit shocks in a particular system’s variable. The IRFs are generated by separately shocking innovations for each of the variables by one standard deviation.

4. Results 4.1 Contemporaneous causal structure Figure 4 displays the contemporaneous causal relationships among the variables, where each line is an edge indicating a relationship between the connected markets. As shown in Figure 4, milled rice markets are segregated by types and regions in the contemporaneous period. Arkansas, Louisiana, and Texas long grain prices and rough rice futures prices are connected. Median grain is segregated and only Arkansas and Louisiana markets are connected. California median grain price moves independently, perhaps because California medium grain is exported to East Asia, and Arkansas/Louisiana medium grain is exported mostly Mexico, Central America, the Caribbean, and the Middle East. Note that large amounts of U.S. southern medium grain rice is exported to Libya. Median grain rice in both markets are different in terms of quality and export destinations. Arkansas long grain appears to be the source of price discovery in the long grain markets, which is expected because of its size of production. Also, Arkansas medium grain is the source of information in the median grain markets. Based on Figure 4, we may conclude that Arkansas long grain and medium grain is the milled rice price reference in the southern regions and California medium market is exogenous in the contemporaneous period. It is relevant to re-emphasize that the DAG results in Figure 4 show only the contemporaneous (i.e. non-time) causal structure. The contemporaneous period here refers to the actual period in which a disturbance to the U.S. milled rice market may occur, e.g. a one-time-only shock to Louisiana long grain. It is also noted that

CA_med

AR_med

LA_med

AR_lng

TX_lng

x_cme

LA_lng

Figure 4. Directed acyclic graph of milled rice markets using LiNGAM algorithm. LiNGAM = Linear Non-Gaussian Acyclic Model. AR_lng = AR-MO long grain; LA_lng = LA-MS long grain; TX_lng = TX long grain; AR_med = AR-MO medium grain; LA_med = LA-MS medium grain; CA_med = CA medium grain; and x_cme = CME rough rice (long grain) futures price. International Food and Agribusiness Management Review

630

Kim et al.

Volume 20, Issue 5, 2017

the causal structure in Figure 4 shows the direction of instantaneous causal flows among the variables and does not suggest the magnitude or the sign (positive or negative) of the effect. 4.2 Innovation accounting Impulse responses shown in Figure 5 depict the responses of all variables to a one-time-shock in the innovation of one variable – when the other variables’ innovations remain constant. The (one-time) shock is positive and of a magnitude equal to one standard deviation of the innovation of the particular factor (variable), applied at a contemporaneous period (month zero), and leaving all other factor’s innovations constant throughout the period (Hamilton, 1994: 318). In the first column in Figure 5, Arkansas long grain market’s shock affects all the long milled rice markets, including futures market, and has stronger impacts on the long grain markets in the first few months. This significant positive impact is up to 5 to 6% on Louisiana and Texas long grain and then diminishes after six months. Similarly, in the second column, Louisiana long grain positively affects Arkansas long and Texas long grain markets initially, while it has a negative effect on these two markets in the long run, which is similar to the findings in Djunaidi et al. (2001). The impacts of shock in Texas long grain market on Arkansas and Louisiana long grains is very limited (third column). Thus, even though markets are co-integrated, Arkansas long grain has a sizably larger effect on the other long grain markets. This appears to indicate that grain markets are not segregated regionally but actually more set apart by types. In addition, Arkansas dominance among long grain markets may respond to its larger volume of production in comparison to the other producing states. The fifth column illustrates the effect from a shock to Arkansas medium grain. In this case, there is a significant positive effect of about 2-3% in both Louisiana and California’s medium grain markets. In addition, this effect appears to be rather permanent. It is anticipated that Arkansas medium affects Louisiana medium as both these grains are mainly used for processed foods (e.g. cereals); however, California medium is of high One-time-only-shock to AR_lng

LA_lng

TX_lng

x_cme

AR_med

CA_med

LA_med

AR_lng

Response

LA_lng TX_lng x_cme AR_med CA_med LA_med

Figure 5. Impulse response functions. Red boxes represent long grain and medium grain markets, respectively. AR_lng = AR-MO long grain; LA_lng = LA-MS long grain; TX_lng = TX long grain; and x_cme = CME rough rice (long grain) futures price; AR_med = AR-MO medium grain; LA_med = LA-MS medium grain; and CA_med = CA medium grain. International Food and Agribusiness Management Review

631

Kim et al.

Volume 20, Issue 5, 2017

quality serving mainly high-end markets and exports (Baldwin et al., 2011). Moreover, the production size of California medium grain is more than four times that of Arkansas. We conjecture that given that California medium holds a premium over Arkansas medium grain, it has to hold that premium when the Arkansas grain (of lower quality) increases its prices. Thus California medium grain ‘maintaining’ a premium price over the southern states’ less quality grain may be seen in the permanent effect from the positive shock coming from Arkansas medium. In addition to the impact of Arkansas medium grain on other medium grain markets, it also present positive impacts on the long grain markets for over one year. The substitution between long-, medium-, and short-grain rice is generally lacking, given that they are demanded by different markets – according to particular tastes and preferences (Childs and Burdett, 2000). However, the positive effect of Arkansas medium grain price may be due to land competition to long grain rice production in Arkansas, a major area of long grain rice production. As a result, price in domestic long grain rice markets increases in the long run given anticipated competition for land in the major supply region. A shock to California medium grain does not have any significant effect on the other markets (column six), responding to its ‘high quality’ characteristics. Thus the effect from shocks to California market seems to be segregated from other markets. Shocks on Louisiana medium grain (column seven) have no significant impacts on other medium grain markets. The FEVD series considering up to 18 months for the long, and medium grains are shown in Table 4, and grouped across varieties. The long grain’s variations of its prices’ forecasted errors, first three rows, are explained mostly by shocks from Arkansas long grain market as well as by shocks from its own regional long grain markets. Again may respond to milled rice major producer is Arkansas and a smaller later effect from Louisiana, the outbound shipping port. In the case of variation of Arkansas medium grain’s forecasted errors, fourth row, it is mostly explained by shocks to itself. For the case of unexpected changes to California’s forecasted errors (sixth row), the medium grain is affected by shocks to itself but mostly from shocks to Arkansas medium, corroborating findings from the IRFs. Regarding Louisiana’s variations in its forecasted errors, seventh row, it is mostly explained by shocks from Arkansas medium and a bit by its own shocks. As shown in the fourth column of Figure 5, futures price (x_cme) is the most important variable in the system, given that a shock to it produces a substantial significant positive effect in all markets. This is anticipated given that futures markets act as price discovery for cash markets (Leuthold et al., 1989). As shown in the fourth row of Table 4 the variation in forecasted errors of the futures markets seems that mainly shocks to futures market has effects on them, though also a bit of Arkansas long markets minimal initial effects and Louisiana long and Arkansas medium has effects after six months and a year, respectively. Conversely, variations in futures explains much of the variation in long grain prices after the second month, which is anticipated given that this is the primary type of grain market served by futures prices. In addition, CME futures has a smaller effect from variations in its prices on the medium grain markets in latter months, perhaps as a spillover information effect.

International Food and Agribusiness Management Review

632

Kim et al.

Volume 20, Issue 5, 2017

Table 4. Forecast error variance decompositions.1 Variation in AR_lng

LA_lng

TX_lng

x_cme

AR_med

CA_med

LA_med

Months 1 2 4 6 12 18 1 2 4 6 12 18 1 2 4 6 12 18 1 2 4 6 12 18 1 2 4 6 12 18 1 2 4 6 12 18 1 2 4 6 12 18

Accounted for by AR_lng

LA_lng

TX_lng

x_cme

AR_med

CA_med

LA_med

1.00 0.84 0.54 0.34 0.13 0.08 0.56 0.57 0.40 0.25 0.10 0.06 0.41 0.50 0.41 0.29 0.13 0.09 0.08 0.07 0.05 0.03 0.02 0.01 0.00 0.00 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00

0.00 0.01 0.01 0.02 0.12 0.16 0.44 0.24 0.09 0.06 0.12 0.16 0.12 0.10 0.06 0.04 0.08 0.12 0.00 0.00 0.02 0.05 0.13 0.17 0.00 0.00 0.01 0.04 0.10 0.13 0.00 0.00 0.00 0.01 0.03 0.04 0.00 0.00 0.01 0.03 0.09 0.12

0.00 0.00 0.00 0.01 0.04 0.04 0.00 0.00 0.00 0.01 0.03 0.04 0.47 0.30 0.16 0.10 0.05 0.04 0.00 0.00 0.01 0.02 0.04 0.03 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00

0.00 0.14 0.43 0.60 0.62 0.53 0.00 0.17 0.49 0.66 0.67 0.56 0.00 0.10 0.36 0.54 0.63 0.55 0.92 0.92 0.91 0.87 0.70 0.58 0.00 0.06 0.18 0.26 0.33 0.32 0.00 0.01 0.06 0.10 0.16 0.16 0.00 0.08 0.21 0.30 0.35 0.33

0.00 0.00 0.01 0.02 0.08 0.17 0.00 0.01 0.01 0.01 0.07 0.16 0.00 0.00 0.00 0.01 0.07 0.18 0.00 0.00 0.01 0.02 0.11 0.20 1.00 0.92 0.78 0.67 0.55 0.53 0.00 0.11 0.29 0.35 0.40 0.44 0.56 0.59 0.57 0.52 0.46 0.48

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.01 0.00 0.00 0.01 0.02 0.04 0.03 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.01 0.01 0.01 1.00 0.86 0.64 0.52 0.39 0.32 0.00 0.00 0.00 0.01 0.01 0.01

0.00 0.00 0.00 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.01 0.00 0.01 0.02 0.02 0.02 0.02 0.44 0.32 0.20 0.14 0.07 0.06

1 ar_lng = AR-MO long grain; la_lng = LA-MS long grain; tx_lng = TX long grain; x_cme = CME rough rice (long grain) futures price; ar_med = AR-MO medium grain; la_med = LA-MS medium grain; ca_med = CA medium grain.

International Food and Agribusiness Management Review

633

Kim et al.

Volume 20, Issue 5, 2017

5. Conclusions Rice is among the top seven U.S. crops in terms of harvested acres and sixth in terms of cash receipt. The U.S. exports about half of its rice production to international markets and accounts for over 10% of the annual volume of global rice trade. U.S. rice is produced in four distinct regions, i.e. Arkansas, Mississippi Delta, Texas and Southwest Louisiana, and California. There are different types of rice associated with a production region and there are different varieties produced within each type. California produces almost exclusively short and medium type of grain; while Arkansas, Texas, and Louisiana produce mostly long types of grain and also small amount of medium types of grain. Thus, understanding the dynamic integration of domestic markets is of significant relevance. In this study, we determine whether the U.S. rice markets are integrated across regions and whether these markets are integrated across its grain types. We applied the Directed Acyclic Graph approach to investigate the contemporaneous structure among seven U.S. f.o.b. milled rice prices and rough rice futures price in the context of multivariate time series modeling. Results are summarized as follows: ■■ Milled rice markets are segregated by types not by producing regions. ■■ In the long grain markets, Arkansas long is the leading reference price in the short term and rough rice futures price is the leading reference in the long term. ■■ In the medium grain markets, Arkansas is the leading reference prices in the short term and rough rice futures price is the limited impact on medium grain markets in the longer term. ■■ California medium grain market is weakly exogenous in the short run, but its price is affected by Arkansas-Missouri medium grain in the longer term. ■■ Rough rice futures prices are weakly exogenous in regards to milled rice prices. The U.S. rice is competing in the overseas markets with other exporting countries, such as Thai, and thus an obvious extension of the study is to include an international rice price – such as Thai price. Unfortunately, adding one more price series reduces the degrees of freedom of data significantly, which lessens the level of confidence of estimated parameters and its inferred results. It is important to note that despite our results not finding significant impact of California medium grain markets on the Southern U.S. medium grain markets, it is still well possible that this effect exists. This in light of southern medium grain production area increasing sharply during the recent California drought, and then sharply contracting once the California area returned to near-normal conditions. Thus it seems plausible that it may not necessarily be that California prices usually adjust to changes to southern medium grain prices, but the other way around, i.e. movements in California medium grain price (given its vast market size compared to that of the southern states) may affect southern medium-grain prices.

References Baldwin, K., E. Dohlman, N. Childs and L. Foreman. 2011. Consolidation and structural change in the U.S. rice sector. Outlook report No. RCS311D301. Available at: http://tinyurl.com/y9es2uue. Bessler, D.A. and D.G. Akleman. 1998. Farm prices, retail prices and directed graphs: results for pork and beef. American Journal of Agricultural Economics 80: 1145-1149. Bessler, D.A. and J. Yang. 2003. The structure of interdependence in international stock markets. Journal of International Money and Finance 22: 261-287. Childs, N. 2012. Rice, US news media resources. Economic Research Service, USDA, Washington, WA, USA. Available at: http://tinyurl.com/y7wq9t5j. Childs, N. and A. Burdett. 2000. Rice situation and outlook: the U.S. rice export market. U.S. Department of Agriculture, Economic Research Service, Washington, WA, USA, pp. 48-54. Childs, N. and J. Kiawu. 2009. Factors behind the rise in global rice prices in 2008. Economic Research Service, USDA, Washington, WA, USA.

International Food and Agribusiness Management Review

634

Kim et al.

Volume 20, Issue 5, 2017

Dickey, D.A. and W.A. Fuller. 1979. Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Association 74: 427-431. Djunaidi, H., K.B. Young, E.J. Wailes, L.A. Hoffman and N.W. Childs. 2001. Spatial pricing efficiency: the case of U.S. long grain rice. Available at: http://tinyurl.com/y7d9uyap. Fackler, P.L. and B.K. Goodwin. 2001. Spatial price analysis. In: Handbook of Agricultural Economics Volume 1, Part B, edited by B.L. Gardner and G.C. Rausser. North-Holland, Amsterdam, the Netherlands, pp. 971-1024. Geiger, D., T.S. Verma and J. Pearl. 1990 Identifying independence in bayesian networks. Networks 20(5): 507-534. Giraud, G. 2013, The world market of fragrant rice, main issues and perspectives. International Food and Agribusiness Management Review 16(2): 1-20. Hamilton, J.D. 1994. Time series analysis. Princeton University Press, Princeton, NJ, USA. Hoover, K. 2005. Automatic inference of the contemporaneous causal order of a system of equations. Econometric Theory 21: 69-77. Johansen, S. 1991. Estimation and hypothesis testing of cointegration vectors in gaussian vector autoregressive models. Econometrica 59: 1551-1580. Kwiatkowski, D., P.C.B. Phillips, P. Schmidt and Y. Shin. 1992. Testing the null hypothesis of stationarity against the alternative of a unit root. Journal of Econometrics 54: 159-178. Lai, P. and D.A. Bessler. 2015. Price discovery between carbonated soft drink manufactures and retailers: a disaggregate analysis with PC and LiNGAM algorithms. Journal of Applied Economics 18(1): 173-198. Leuthold, R.M., J.C. Jankus, J.E. Cordier. 1989. The theory and practice of futures markets. Lexington Books, Washington DC, WA, USA. Lütkepohl, H. 2005. New introduction to multiple time series analysis. Springer Verlag, Berlin, Germany. Lütkepohl, H. and M. Krätzig. 2004. Applied time series econometrics. Cambridge University Press, New York, NY, USA. McKenzie, A.M., B. Jiang, H. Djunaidi, L.A. Hoffman and E. Wailes. 2002. Unbiasedness and market efficiency tests of the U.S. Rice futures market. Review of Agricultural Economics 24(2): 474-493. Pearl, J. 1995. Causal diagrams for empirical research. Biometrika 82: 669-710. Pearl, J. 2000. Causality. Cambridge University Press, Cambridge, MA, USA. Serra, T. and B.K. Goodwin. 2004. Regional integration of nineteenth century U.S. egg markets. Journal of Agricultural Economics 55(1): 59-74. Shimizu, S., A. Hyvärinen, P.O. Hoyer and Y. Kano. 2006b. Finding a causal ordering via independent component analysis. Computational Statistics and Data Analysis 50: 3278-3293. Shimizu, S., P.O. Hoyer, A. Hyvärinen and A. Kerminen. 2006a. A linear non-gaussian acyclic model for causal discovery. Journal of Machine Learning Research 7: 2003-2030. Spirtes, P., C. Glymour and R. Scheines. 2000. Causation, prediction and search. MIT Press, Cambridge, MA, USA. Shimizu, S. and Y. Kano. 2008. Use of non-normality in structural equation modelling: application to direction of causation. Journal of Statistical Planning and Inference 138: 3483–3491. Stockton, M.C., D.A. Bessler and R.K. Wilson. 2010. Price discovery in Nebraska cattle markets. Journal of Agricultural and Applied Economics 42: 1-14. Stone, J.V. 2004. Independent component analysis: a tutorial introduction. MIT Press, Cambridge, MA, USA. Swanson, N.R. and C.W.J. Granger. 1997. Impulse response functions based on a causal approach to residual orthogonalization in vector autoregressions. Journal of the American Statistical Association 92: 357-367. Taylor, E.L., D.A. Bessler, M.L. Waller and M.E. Rister. 1996. Dynamic relationships between US and Thai rice prices. Agricultural Economics 14: 123-133. United States Department of Agriculture (USDA). 2015. Rice yearbook. Available at: http://tinyurl.com/ ya4hdhoa. United States Department of Agriculture (USDA). 2017. Rice yearbook. Available at: http://tinyurl.com/ y8nddkuk. Yu, T-H., D.A. Bessler and S.W. Fuller. 2007 Price dynamics in U.S. grain and freight markets. Canadian Journal of Agricultural Economics 55(3): 381-397. International Food and Agribusiness Management Review

635

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2016.0109 Received: 8 June 2016 / Accepted: 2 May 2017

The effects of price promotion on relative virtue and vice food products RESEARCH ARTICLE Josefa Parreño-Selva a, Francisco J. Mas-Ruizb, and Enar Ruiz-Condea aLecturer

and bProfessor and Department Chair, Department of Marketing, Faculty of Economics, University of Alicante, Ctra. de San Vicente, s/n. Campus de San Vicente del Raspeig, Alicante, Spain

Abstract Retailers use price promotion of light and regular products, but not all of these products are perceived as relative virtues and vices, respectively. This paper aims to identify whether consumers distinguish between the two product categories. Survey data is used to distinguish between each product category, and identifies low-fat milk as a light product that gives both immediate and delayed rewards. Daily scanner data from a hypermarket supports the effects of price promotions on sales within and between product categories, as expected. We expect that, (1) due to these light products representing more enduring involvement, demand is less price sensitive compared to demand for regular products; (2) as nonimpulse purchase products, price promotions of light products cannibalize the sales of other light products; and (3) the loss of light product benefits associated with switching means that price promotions of light products hurt regular product sales more than vice versa. Keywords: light product, regular product, relative virtues and vices, price promotions JEL code: M31 Corresponding author: pepi@ua.es

637

Parreño-Selva et al.

Volume 20, Issue 5, 2017

1. Introduction Verbal descriptors are useful in product labeling to indicate that foods and drinks are low in specific nutrients (Arambepola et al., 2008). These words include light, zero calorie, low fat, no fat and sugar free (hereafter, light product). They refer to diet food – that is, any food or drink whose recipe has been altered in some way to make it part of a body-modification diet (FDA, 2013). Light/regular classification has become widespread and is regulated by law, but there are doubts around whether consumers consider products with these labels as relative virtue or vice items, respectively. The virtue – vice distinction denotes which in a pair of goods is preferable when one considers the immediate and delayed consumption consequences (Wertenbroch, 1998). Relative virtues exchange small immediate costs (e.g. poorer taste of light cream cheese) for larger delayed rewards (e.g. health), whereas relative vices exchange small immediate rewards (e.g. good taste of regular-fat cream cheese) for larger delayed costs (e.g. future health problems) (Milkman et al., 2008; Okada, 2005; Van Doorn and Verhoef, 2011). Some authors, such as Wertenbroch (1998), have suggested that the vice-virtue distinction is not absolute; in fact, Wertenbroch’s study analyzed a total of 30 light and regular product categories, but only classified 21 of them as relative virtues and vices. Researchers have found a negative taste association (which equates to a short-term sacrifice) with healthy products (Raghunathan et al., 2006). Despite this, many people consume certain light products (e.g. light fruit drinks instead of regular fruit drinks) that give them a delayed reward (long-term social or health effects or any other long-term benefits) and do not consider these products as imposing an immediate, nonpecuniary cost – that is, a reason to believe that they are doing something they dislike in return for a future benefit. In fact, evidence has also pointed to a positive taste association with healthy products (Drewnowski, 1997; Smith, 2004). This implies that light products, which give a delayed reward, can be perceived as either having an immediate cost (i.e. relative virtue) or as not having an immediate cost. In the same way, people consume certain regular products (e.g. regular coffee instead of decaffeinated coffee) that give an immediate reward (pleasure, flavor, ease of use, fun, temptation, or anything else that makes the product appealing to the consumer in the short term), but do not regard these products as imposing a delayed, nonpecuniary cost – that is, a reason to believe that they are doing something they like to their future detriment. This implies that regular products, which give an immediate reward, can be perceived as either having a delayed cost (i.e. relative vice) or as not having a delayed cost. In this way, we can put light and regular products on a continuum along with relative virtue and vice products (Figure 1). In this paper, we empirically test this continuum of product forms based on the two extremes of immediate and delayed consequences. We build on several extant works (e.g. Dhar and Simonson, 1999; Hoch and Loewenstein, 1991) that have focused on whether consumers choose to forgo temptation to achieve a goal that lies on the opposite end of a single continuum, which ranges from a short-term desire for pleasure to a long-term goal that requires abstention from desired behaviors (e.g. indulging in tasty foods versus minimizing calories). This logic was

Immediate consequences

Relative vice (Regular product with immediate reward and delayed cost)

Regular product with immediate reward and no delayed cost

Light product with no immediate cost and delayed reward

Regular (not diet)

Relative virtue (Light product with immediate cost and delayed reward)

Light (diet)

Figure 1. Continuum of products. International Food and Agribusiness Management Review

638

Delayed consequences

Parreño-Selva et al.

Volume 20, Issue 5, 2017

discussed by Haws and Winterich (2013); in situations where consumers must navigate multiple goals, they tend to concentrate on one goal at a time, often at the expense of other goals, with affect-driven motivations often overriding cognitions. In a more general form, whichever goal is made more salient by the environment will take priority over other goals, and will inhibit the accessibility of competing goals through a shielding process. Accordingly, we consider a continuum between immediate and delayed consequences where it is possible to integrate a relative vice-virtue structure and a light-regular distinction. In addition, we build on the means-end chain model (Gutman, 1982), which emphasizes the role of values (defined as desirable end-states of existence) as a powerful force that governs individuals’ choices; that is, the way in which people cope with a wide diversity of products that can satisfy their own values, grouping them into classes to reduce the complexity of choice. This model considers situation influences; thus, in some cases the relation between attributes of products and consequences for consumers are easy to determine. For example, for beer drinkers, the ability to control weight is clearly tied to the caloric level in beer, resulting in the category of light beer as opposed to regular beer. However, in other cases such relations are not easy to determine, requiring intermediate distinctions between the desired consequences and the actual product attributes, which are part of the distinctions made at the grouping level. For example, a consumer who wants good health and sees a connection between diet and health could form a natural-artificial distinction that could apply to drinks by grouping them according to type in an ordered array of categories along the dimension defined by the natural-artificial distinction. Accordingly, we integrate the light-regular distinction into the relative vice-virtue structure in a continuum between immediate and delayed consequences. An interesting question that has emerged from research into relative virtues and vices is the effect of price promotions on purchase quantity (Wertenbroch, 1998). This is based on the idea of potential self-control, because consumers are tempted towards excess consumption of the relative vices they have in stock at home. The self-imposed constraint of rationing the quantity bought implies that consumers of relative vices will be less likely to buy large quantities (per shopping trip), compared to consumers of relative virtues, in response to price promotions. Nevertheless, some researchers have considered this effect as being conditioned by promotion type and cultural factors. Mishra and Mishra (2011) and Huyghe and Van Kerckhove (2013) found that consumers prefer a price discount to a bonus pack for relative vices (and vice versa for virtues) because the bonus pack (offering more of the product for the same price) leads to increased consumption of the vice. Similarly, Parreño-Selva et al., (2014) noted a bigger effect on sales for price promotions of regular beer than for alcohol-free beer. This may be because alcoholic beer is not considered a relative vice in Spain (according to the continuum of products in Figure 1, it would be a regular product with no delayed cost), meaning that consumers do not feel the need to use self-control when drinking it. Regular beer is perceived as a natural product of low alcoholic strength, which forms part of a healthy Mediterranean diet and is linked with moderate consumption at social gatherings (Cerveceros de España, 2001). This idea has also been supported by research in recent decades on health, epidemiology, and nutrition, which has found a link between moderate alcoholic beer consumption and health (Kaplan et al., 2000; Klatsky et al., 2003). Alternatively, we focus on the influence of price promotion on sales of products that are not characterized as relative virtues, and in particular on light products that do not impose an immediate cost and give a delayed reward (Figure 1). We expect that, due to these light products representing more enduring involvement1, the demand for promoted light products is less price sensitive compared to the demand for promoted regular products; furthermore, as nonimpulse purchase products, price promotions of light products cannibalize the sales of other products within the light category; finally, the loss of light product benefits associated with switching means that price promotions of light products hurt regular product sales more than vice versa. Our objective is twofold: first, we aim to find out whether consumers see light and regular products as relative virtues and vices, respectively; second, by focusing on one light product with no immediate costs 1

Enduring involvement is defined as an individual difference variable representing the arousal potential of a product or activity that causes personal relevance (Higie and Feick, 1989).

International Food and Agribusiness Management Review

639

Parreño-Selva et al.

Volume 20, Issue 5, 2017

and delayed rewards – low-fat milk – we calculate the net effect of price promotion of this light product on sales in comparison to the regular product (whole milk). Survey data on university students finds that light and regular products are seen as being different from relative virtues and vices. Daily scanner data from a hypermarket supports the own and cross effects within and across categories for milk, where low-fat milk is identified as a light product that is preferred from both perspectives – immediate and delayed.

2. Effect of price promotions on retail sales of regular products and light products with no immediate costs Traditionally, marketing researchers have examined whether price promotions are a useful way to increase sales and, to this end, have developed econometric models to estimate the price sensitivity of retail demand using scanner data (e.g. Blattberg et al., 1995; Pauwels et al., 2002; Van Heerde et al., 2002, 2004). The logic of these studies rests on the impact of promotions on the acceleration of purchases by consumers (buying earlier than usual and/or buying more than usual) and, therefore, on household stockpiling. Apart from these normative arguments about the effects of inventory size on consumption (lower cost per unit consumed), psychological elements also come into play. Certain product characteristics – for example, vice and virtue – related to inter-temporal distribution of the costs and benefits of consumption could also impact the effect of storage. This raises the possibility that some inventory effects on consumption are impulsive (Wertenbroch, 1998), and incites self-imposition of constraints on consumption. Alternatively, our study considers that relative virtues and vices are not fully equivalent to light and regular products sold by retailers, which will affect the own and cross effects of price promotions of these product categorizations. By focusing on light products, which do not impose an immediate cost and give a delayed reward, and on regular products, we distinguish the effects described in the following sections. 2.1 Own price promotion effect The own price promotion effect is the effect of price promotion on the sales of promoted product. Our study expects that sales are less sensitive to the price promotion of a light product that does not impose an immediate cost and gives a delayed reward than to the price promotion of a regular product. This own effect is argued through enduring involvement; an aspect that underlies both light products and the concept of organic (i.e. chemical-free) products. Basically, a light product that does not impose an immediate cost and gives a delayed reward represents more enduring involvement compared to a regular product (nondiet) because it is a virtuous2 choice due to its delayed reward (i.e. less negative long-term consequences). Similarly, organic products (natural yogurt, organic milk) represent more enduring involvement for consumers because these products are a means of achieving important life values (Bezawada and Pauwels, 2013). Consumers of organic products perceive these products to offer benefits related to taste, nutritional value, health, environmental protection, animal welfare, and ethics (Bourn and Prescott, 2002). However, despite there being some studies that contradict this (e.g. Gutman, 1982), we follow the common conclusion in the literature (e.g. Van Doorn and Verhoef, 2011) and believe that consumers consider taste and health concerns before more altruistic motivations, such as animal welfare and care for the environment, especially for food purchases, because consumers are characterized as ‘unashamedly selfish’. Taste and health benefits are also perceived by consumers of light products as benefits which do not impose an immediate cost and give a delayed reward; in addition, the

The characterization of involvement as the purchase of anything that is ‘virtuous’ (i.e. the purchase of any product with a delayed reward must be the act of someone with higher involvement) highlights the idea that people are more committed to ‘long-term thinking’. However, it does not correspond very closely with many explanations of consumer involvement in the literature (Laurent and Kapferer, 1985; Mittal and Lee, 1989).

International Food and Agribusiness Management Review

640

ParreĂąo-Selva et al.

Volume 20, Issue 5, 2017

choice of these light food products3 is not based on social goals either, because individual motives have a stronger influence on food choice compared to social ones (Van Doorn and Verhoef, 2011). Regarding the price promotion effect, despite the possibility of price promotions of organic products representing a strong buying incentive, with little potential for perceived quality erosion (Delvecchio et al., 2006), price promotions may be less effective for organic products than for conventional products. Organic products represent more enduring involvement (Makatouni, 2002). Thus, consumers not only consider benefits and costs at the moment of purchase when faced with a price promotion for organic products, but also consider the benefits of these products in the future with a reasonably low price (Bezawada and Pauwels, 2013). In consequence, a reduction in regular prices, but without increasing temporary promotions, would be more effective for organic than for conventional products. In terms of the idea that more enduring involvement underlies light products with no immediate cost, we expect that the own price promotion effect will be bigger for regular products than for light products that do not impose an immediate cost. 2.2 Cross effects within the light product (or regular product) category Cross effects within the category are the effects of price promotion on the sales of products that belong to the same product category as the promoted product (within-category effects). These cross effects can be complementary if the price promotion increases the sales of other products, or substitutive if the price promotion damages the sales of other products. Our study expects that complementary within-category effects of promoted regular products are greater than those of promoted light products, while the substitutive within-category effects of promoted regular products are smaller than those of promoted light products. These within category cross-effects are argued to rely on the degree to which the product counts as an impulse buy. In general terms, the extent to which a product category is impulse bought supports the expansion of the category produced by promotion (complementariness effect) (Narasimhan et al., 1996). The logic rests on the idea that promotion attracts consumers to the category and stimulates impulse buying. However, promotion does not generate a brand switch (nonsubstitution effect), given that a brand switch effect would imply that the consumer was planning to buy a specific brand but this intention was changed by the promotion, whereas impulse buying is, by nature, an unplanned purchase. Conversely, nonimpulse-bought products (i.e. relative necessities) are less elastic in terms of purchase incidence and stockpiling; therefore, promotions of these products will have a lesser sales effect within their own category (noncomplementariness effect) (Bell et al., 1999). Moreover, insofar as a product is a necessity (planned purchase), which implies that consumers have little flexibility to adjust the demand of the category, the only way for consumers to save money is via brand switching, which suggests a greater within-category cross-sales (substitution) effect. Basically, regular products are more likely to be bought on impulse (unplanned) compared to light products with no immediate cost. For example, Narasimhan et al. (1996) found that the following products show a drastic difference in their propensity for impulse buying: diet pills, which are perceived as a low-impulse product; and candy and mints, which are perceived as high-impulse products. Similarly, Wertenbroch (1998) considered candy and alcohol as impulse goods. That is, regular (nondiet) products are mainly characterized based on their immediate reward and, therefore, based on greater short-term temptation, and tend to be impulse bought. Conversely, light products with no immediate cost represent a virtuous choice as they are mainly characterized by less negative long-term consequences, and therefore tend to be bought less on impulse and more as a planned purchase. Thus, we can predict a complementary effect for the price promotions of regular products on the sales of other regular products within this category, and a substitution effect for price promotions of light products on sales of other light products within this category. That is, we can expect that 3 A difference regarding these light products is that organic product purchase is inhibited by low availability and distribution, price premium, and lack of consumer knowledge (Bezawada and Pauwels002C 2013); as credence goods (associated with a high degree of uncertainty), they require appropriate labeling (third-party certification) to mitigate uncertainty.

International Food and Agribusiness Management Review

641

Parreño-Selva et al.

Volume 20, Issue 5, 2017

the complementary within-category effects of promoted regular products are greater than those of promoted light products with no immediate costs. In addition, we can expect that the substitution within-category effects of promoted regular products are smaller than those of promoted light products with no immediate costs. 2.3 Cross effects between categories The cross effects between categories are the effects of price promotion on the sales of products that belong to different categories of promoted product (cross-category effects). These effects can also be complementary or substitutive. Our study expects a greater substitutive price promotion effect of a light product on sales of regular products (substitution cross-category effects) than vice versa, as well as a smaller complementary price promotion effect of a light product on sales of regular products (complementary cross-category effects) than vice versa. These cross effects between categories are argued to be based on the loss of light-product quality benefits associated with product switching; an aspect that also underlies organic products. This hinges on the idea that if light products with no immediate costs, such as with organic products (see above), offer intrinsic quality benefits to consumers (e.g. health benefits and taste), especially for food purchases, switching to conventional products represents a loss of these benefits, which consumers would seek to avoid (Bronnenberg and Wathieu, 1996). Thus, price promotions of organic products (and of light products) would hurt conventional product sales more than vice versa (Bezawada and Pauwels, 2013). In addition, we consider that the purchase of a promoted light product would reduce additional regular product purchases (substitution cross-category effect) more than vice versa, whereas purchase of a promoted regular product would increase additional purchases of light products (complementary cross-category effect) more than vice versa. That is, we can expect that the substitution cross-category effects of promoted light products – which do not impose an immediate cost – on regular product sales are greater than vice versa. In addition, we expect that the complementary cross-category effects of promoted light products – which do not impose an immediate cost – on regular product sales are smaller than vice versa.

3. Identifying light and regular product categories and relative virtue and vice product categories Analysis of the first objective of our study requires a research design suited to examining whether light and regular products are equivalent to relative virtues and vices for consumers. In order to identify light and regular product categories by distinguishing between relative virtue and vice product categories, we use Wertenbroch’s (1998) procedure, which uses a consumer survey to identify individuals’ temporal preference order for pairs of product categories. The study’s sample consists of 176 university students in the third year (an average age of 21 years) of a business and administration degree in Spain. The students were invited to dedicate five minutes to complete a questionnaire after class. The participant motivation was the possibility of receiving a reward, since they were told that €10 would be given to 10% (decided via raffle) of the students (randomly selected) who filled out the questionnaire. Finally, 52% of the 340 students who received the questionnaire completed it. We use seven pairs of product categories (butter vs margarine, regular vs decaffeinated coffee, regular vs light mayonnaise, regular vs diet soft drinks, white vs brown sugar, sugared vs low-sugar cereal, and whole vs low-fat milk). These pairs of product categories were selected by following the list of categories used by Wertenbroch (1998), and taking into account the criterion that surveyed students could understand them. For each pair, students rated (using a nine-point scale anchored at 1 and 9 by the two categories) the category that they would prefer to consume, considering first the immediate consequences (hereafter i) of consumption (e.g. taste, ease of use, fun, temptation, or any other short-term benefit) assuming identical delayed consequences (e.g. long-term social or health effects or any other long-term costs or benefits); and second only considering the delayed consequences (hereafter d) of consumption, assuming identical immediate consequences.

International Food and Agribusiness Management Review

642

Parreño-Selva et al.

Volume 20, Issue 5, 2017

We then rescaled the subjects’ i and d ratings from -4 to +4, such that a preference for the hypothesized vice (virtue) in a pair would always be indicated by a negative (positive) rating. Subsequently, we sorted responses (Table 1) depending on whether an individual preferred one category from one temporal perspective but expressed the reverse preference from the other temporal perspective (d>0 and i<0, or d<0 and i>0), or whether the individual at least weakly preferred the same category in a pair from both temporal perspectives (d≥0 and i≥0, or d≤0 and i≤0). Positive temporal reverse scores4 (d-i>0, because d>0 and i<0) mean that the category was preferred based on the delayed consequences but not on the immediate consequences, so it is a relative virtue, while the other category in the pair is a relative vice. Negative temporal reverse scores (d-i<0, because d<0 and i>0) mean that the category not preferred from the delayed consequences was preferred from the immediate consequences, so it is a relative vice, while the other category in the pair is a relative virtue. Finally, a pair in which a category was at least weakly preferred in both temporal perspectives, immediate and delayed (d≥0 and i≥0, or d≤0 and i≤0), received a temporal reversal score of zero (d-i=0), which indicates no inconsistent temporal preferences. This latter case would represent a light product with no immediate cost (d≥0 and i≥0), or a regular product with no delayed cost (d≤0 and i≤0), respectively. 3.1. Results of identification of light and regular product categories and relative virtue and vice product categories Our results (Table 1) allow us to distinguish the cases of relative virtue and vice, where individuals have temporal reversal preferences (d>0 and i<0), from the cases of light or regular products, where individuals prefer the same category in a pair from both temporal perspectives (d>0 and i≥0, and d=0 and i<0). The pairs of mayonnaise, cereal, and soft drinks products present mean values that are positive for d, negative for i, and significantly different from zero (d>0 and i<0), and positive mean temporal reverse scores that differ from zero at a 99% confidence level. This shows that the categories in these pairs create intertemporal conflict between preferences, so they can be characterized as relative vices and virtues. For example, with mayonnaise, the light-mayonnaise category gives people a delayed reward (d=0.94; P<0.0000) while the regular mayonnaise category gives an immediate reward (ī=-0.77; P<0.0000), with positive temporal reverse 4 For example (adapted from Wertenbroch, 1998), if an individual selected ‘1’ on the immediate scale (indicating a preference for one anchor; e.g. regular mayonnaise) and ‘8’ on the delayed scale (indicating a preference for the other anchor; e.g. light mayonnaise), these values were rescaled to -4 (immediate) and +3 (delayed). The temporal reversal score (delayed minus immediate) would be +7, so that regular mayonnaise would be classified as a relative vice and light mayonnaise as a relative virtue.

Table 1. Relative vice and virtue product categories and regular and light product categories.1 Product category

Immediate Delayed Product category consequences consequences (i) (d)

Mean n temporal reversal score

Relative vice Regular mayonnaise Sugared cereal Regular soft drinks Regular product with no delayed cost Sugar Regular coffee Regular product

i<0 and d>0 -0.772**** -1.153***** -0.676**** i<0 and d=0

1.35***** 1.26***** 1.05*****

176 176 176

1.26***** 0.58*****

176 176

0.95***** 0.09

176 176

Butter Whole milk 1*

0.943***** 0.454**

0.750****

-1.630***** -1.034***** i≥0 and d>0

0.125 0.079

-0.284 0.358*

0.914***** 1.534*****

Relative virtue Light mayonnaise Low-sugar cereal Diet soft drinks Light product Brown sugar Decaffeinated coffee Light product with no immediate cost Margarine Low-fat milk

P<0.10, ** P<0.05, *** P<0.01, **** P<0.001, ***** P<0.0001 in two-sided test. International Food and Agribusiness Management Review

643

Parre単o-Selva et al.

Volume 20, Issue 5, 2017

scores (d-i>0). This implies that the light-mayonnaise category preferred in terms of delayed consequences was not preferred in terms of immediate consequences, so it is a relative virtue while the other category is a relative vice. However, for the four remaining pairs of products (sugar, coffee, butter, and milk), consumers showed a preference for the same category from both temporal perspectives (immediate and delayed); thus, these regular and light pairs of products were deemed not to induce time-inconsistent preferences and cannot be characterized as relative vices or relative virtues. The butter and milk pairs show average values of d>0 that are significantly distinct from zero, being average values of i that are significantly distinct from zero (i>0) for milk and not statistically different from zero (i=0) for butter. Thus, for milk, the low-fat milk category gives a delayed reward (d=1.53; P<0.0000) and a weak immediate reward (朝=0.36; P<0.10). This preference for the same category suggests that this category represents a light product that does not impose an immediate cost. A similar result is evidenced in the margarine category. The margarine category gives people a delayed reward (d=0.914; P<0.0000) but the immediate reward is not significant (朝=-0.284; P>0.10). In consequence, there is no evidence that low-fat milk or margarine are relative virtues that impose an immediate cost. The product pairs of sugar and coffee have average values of d that are not statistically different from zero, while those of i are negative and significantly different from zero. The regular coffee category gives an immediate reward (朝=-1.03; P<0.0000), but its delayed reward is not statistically significant (d=0.07; P>0.10). That is, there is no evidence that regular coffee is a relative vice product that imposes a delayed cost. Hence, regular coffee can be identified as a regular product with no delayed cost. Similar results are found in the sugar category. In other words, the participants did not regard sugar and regular coffee as imposing a delayed cost. In brief, our results reveal that consumers distinguish light and regular products from relative virtues and vices, according to the continuum of Figure 1; thus, virtues and vices are not equivalent to light and regular products.

4. Estimating the net effect of price promotion of low-fat milk (light product) on sales in comparison to whole milk (regular product) Analysis of the second objective of our study required a research design suitable for estimating the total effect of price promotion. In this section, we focus on milk as an example of light and regular products, where whole milk is identified as a regular product while low-fat milk is identified as a light product that does not impose an immediate cost and gives a delayed reward. 4.1 Data Our study uses data from a hypermarket chain with 56 stores in the Alicante province (Spain). Insofar as the residents of the area have weekly incomes, live in apartments, and the parking in the hypermarket is restricted, we can expect that customers do not acquire many products during price promotions to store them and consume them later (no stockpiling effect). The database comprises daily data, for two years (20092010), at the item level of different products. Milk was chosen because, first, application of the Wertenbroch procedure (1998) to the data from the survey of Spanish university students (see previous section) classifies whole milk within the regular products category and low-fat milk within the category of light products that do not impose immediate costs and give delayed rewards. In particular, our analysis of price promotion groups two low-fat milk products, semi-skimmed and skimmed milk, into the light products category. Semiskimmed and skimmed milk have different nutrient levels; however, the delayed reward explains why they are included in this category of light products, since both are specifically indicated for diets that prohibit the consumption of full-fat milk due to the saturated fats, contained in its cream, that increase blood cholesterol levels. Second, the milk products family is ideal for the objectives of this study because the products of whole, semi-skimmed, and skimmed milk vary greatly in terms of sales and price promotions.

International Food and Agribusiness Management Review

644

ParreĂąo-Selva et al.

Volume 20, Issue 5, 2017

The descriptive analysis on milk products is shown in Table 2. From the items belonging to these products, we selected those that had been for sale for more than 450 days. A similar procedure was used by Leeflang et al. (2008). For each item, Table 2 shows the total sales income (regular price Ă&#x2014; product units) during the study period; the use of price changes; the average price; and the number of days that the product had a sales volume different from zero. At the price level, in Alteza 1 l there is no variability among whole, semi-skimmed, and skimmed versions, so these variables were removed from all sales models. Of the whole-milk products, Puleva calcio completa 1 l and Puleva 1 l have almost no price variability (6.03 and 6.20%, respectively); Table 2. Descriptive statistics at the item level.

Whole milk

Semi-skimmed milk

Skimmed milk

Item1

Sales income Number of (euros) days with sales

Average price

Percentage of days with change in daily price over 5%

Pascual Calcio 1 l Puleva Calcio 1 l Puleva Omega 3 1 l Puleva Calcio Completa 1 l Alteza Calcio 1 l Pascual 1 l Puleva 1 l Asturiana 1.5 l Asturiana 1 l Alteza 1 l Pascual Calcio 1 l Puleva Calcio 1 l Asturiana Naturfibra 1 l Alteza Calcio 1 l Puleva Calcio Soja 1 l Puleva A+D 1 l Flora 1 l Ram 1 l Pascual 1 l Asturiana 1.5 l Asturiana 1 l Asturiana Naturlinea 1 l Alteza 1 l Pascual Calcio 1 l Puleva A+D 1 l Puleva Calcio 1 l Alteza Calcio 1 l Puleva Calcio Soja 1 l Ram 1 l Pascual 1 l Asturiana 1.5 l Asturiana 1 l Asturiana Naturfibra 1 l Alteza 1 l

5,521.25 7,178.94 40,071.27 6,822.58 1,539.28 17,304.72 6,525.71 11,292.47 8,781.31 10,922.37 13,811.58 7,463.35 5,592.56 3,807.00 3,490.19 7,757.70 5,281.48 6,116.18 26,396.49 22,919.57 21,914.58 2,323.08 10,348.46 21,628.02 7,377.81 17,387.43 4,611.60 3,224.46 4,782.92 20,556.30 9,360.06 9,959.75 3,618.69 8,524.39

1.26 1.15 1.35 1.34 0.78 0.90 0.85 1.33 0.82 0.63 1.26 1.15 1.17 0.78 1.26 0.95 1.28 0.78 0.90 1.33 0.82 1.52 0.63 1.26 0.85 1.15 0.78 1.25 0.78 0.90 1.33 0.82 1.17 0.63

16.73 20.14 23.08 6.03 0.00 13.33 6.20 11.22 15.57 0.00 22.58 20.00 31.17 1.10 5.09 6.25 3.89 0.86 13.50 11.00 15.58 24.56 0.00 16.52 6.30 20.17 1.04 6.43 3.52 17.17 11.15 15.53 24.66 0.00

556 556 598 580 454 600 581 588 591 599 598 595 571 547 511 592 540 583 600 600 597 452 598 575 587 595 579 482 568 600 592 599 515 572

These items have a sell by date of 3 months.

International Food and Agribusiness Management Review

645

Parreño-Selva et al.

Volume 20, Issue 5, 2017

the same applies to the semi-skimmed (Puleva calcio soja 1 l, Puleva A+D 1 l, Flora 1 l and Ram 1 l) and skimmed products (Puleva A+D 1 l, Puleva calcio soja 1 l and Ram 1 l). This lack of variability led us to exclude these price index variables from the sales models. 4.2. Methodology The applied methodology is based on several models that estimate the total net effect of price promotions of light and regular milk products on sales, distinguishing the following effects: (1) own-item sales effect (effect of price promotion of light or regular product on own sales); (2) cross-period sales effect (cross effects between periods); (3) within-category sales effect (effect of price promotions of light products on sales of other light products, and effect of price promotions of regular products on sales of other regular products); and (4) between-category sales effect (effect of price promotions of light products on sales of regular products, and effect of price promotions of regular products on sales of light products). This entails modeling to enable decomposition of the impact of price promotions from previous effects. To this end, we adapt the decomposition model of the effect of price promotions on sales proposed by Leeflang et al. (2008), considering the possible differences between categories of light and regular products. Leeflang et al. (2008) assumed that the substitution effect between categories (the reduction in sales of some products of a category due to price promotion of products of another category) accounts for part of the category expansion effect. In this sense, the category expansion effect plus the between-categories substitution effect equals the net category expansion effect. This net effect measures the increase in the consumption ratio of the products in the category and the possible cross-store effects (sales increases resulting from attracting customers from competing stores) as a consequence of the price promotion. However, the profits gained from price promotions by retailers and manufacturers with products in different categories (c and c') is the sum of the net expansion effect of the category and the complementariness effect between categories, and is known as the total effect. Given that this study aims to analyze the existence of differences in the effects described above, at the level of light and regular product categories, incorporation of this differentiation results in the following relationships between the effects (we use the example of the promoted item belonging to the light category, c) LI = CE – OLWC – OLWCOP

(1)

OLWC = POLWC– NOLWC

(2)

NCE = CE + SBROC

(3)

TC = NCE + CBROC

(4)

Where LI is the effect on sales of the light promoted item; CE is the category expansion effect; OLWC is the effect on sales of other light products within the same category; OLWCOP is the effect on sales of light products within the same category in other periods; POLWC is the positive effect on sales of other light products within the same category; NOLWC is the negative effect on sales of other light products within the same category; NCE is the net category expansion effect; SBROC is the substitution effect between regular products in other categories; TC is the total effect of a combination of categories; and CBROC is the complementary effect between regular products in other categories. There is an exact relationship between total sales of a combination of categories (1, ..., c,... , c',... C) (TCc) in a certain period of time (t, t+w, where w refers the time period after the price promotion has occurred; w=1, 2, ...) and the following elements: sales of the promoted item in the category c (PIc); sales of other items in category c with the same brand and with different brands to the promoted item that substitute sales of this product (OISBSc/OIDBSc); sales of other items in category c with the same brand and with different International Food and Agribusiness Management Review

646

ParreĂąo-Selva et al.

Volume 20, Issue 5, 2017

brands to the promoted item that complement its sales (OISBCc/OIDBCc); sales of other items in category c' with the same brand and with different brands to the promoted item that substitute sales of this product (OISBSc/OIDBSc); sales of other items in category c with the same brand and with different brands to the promoted item that complement its sales (OISBCc/OIDBCc); and sales in other periods (OPc). Consequently, we obtain the following expression for decomposition of the effect of price promotions, which is similar to that used by Leeflang et al. (2008): 5) TCđ?&#x2018;?đ?&#x2018;? = PIđ?&#x2018;?đ?&#x2018;? + OISBSđ?&#x2018;?đ?&#x2018;? + OIDBSđ?&#x2018;?đ?&#x2018;? + OISBCđ?&#x2018;?đ?&#x2018;? + OIDBCđ?&#x2018;?đ?&#x2018;? + OPđ?&#x2018;?đ?&#x2018;?

đ??śđ??ś

+ â&#x2C6;&#x2018; (OISBSđ?&#x2018;?đ?&#x2018;?đ?&#x2018;? + OIDBSđ?&#x2018;?đ?&#x2018;?đ?&#x2018;? + OISBCđ?&#x2018;?đ?&#x2018;?đ?&#x2018;? + OIDBCđ?&#x2018;?đ?&#x2018;?đ?&#x2018;? )

(5)

đ?&#x2018;?đ?&#x2018;? â&#x20AC;˛ =1 đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?

Estimation of the demand (SALES) for a product, either regular or light, is conducted through the following regression model: đ??śđ??ś

đ??˝đ??˝đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?

6) đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;? = đ??śđ??śđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;? + â&#x2C6;&#x2018; â&#x2C6;&#x2018; (đ?&#x203A;žđ?&#x203A;žđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;â&#x2C6;&#x2014; đ?&#x2018;?đ?&#x2018;?đ?&#x2018;? đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;? đ??źđ??źđ??źđ??źđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;?đ?&#x2018;? đ?&#x2018;Ąđ?&#x2018;Ą ) + â&#x2C6;&#x2018; â&#x20AC;˛ =1 đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;? đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?

đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;?

đ?&#x2018;&#x;đ?&#x2018;&#x;đ?&#x2018;&#x;đ?&#x2018;&#x;

(đ?&#x203A;žđ?&#x203A;žđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;â&#x2C6;&#x2014;đ?&#x2018;?đ?&#x2018;?đ?&#x2018;? đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;?,đ?&#x2018;&#x;đ?&#x2018;&#x; â&#x2C6;&#x2018;

đ?&#x2018; đ?&#x2018; đ?&#x2018; đ?&#x2018;

đ??źđ??źđ??źđ??źđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;? ) + đ?&#x2018;Łđ?&#x2018;Łđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;? (6)

Where SALESjc,t are the sales of item jc; IPj'c',t is the price index of item j'c'; â&#x2C6;&#x2018;6s=1 IPjc,tâ&#x20AC;&#x201C;(s+6r) is the sum of the price index of item jc in the days of the week previous to the promotion (r=0 week previous; r=1 two weeks previous, etc.); and vjc,t is the error term. Furthermore, (Rjc+1) is the number of retards of item jc; Îł*j'c'jc is the parameter for the price index of item j'c' on product jc; and y*jc'jc,r captures the retarded effects of the aggregated price index. By using the same explanatory variables for all components of Equation 5, as per Van Heerde et al. (2004) and Leeflang et al. (2008), the identity in Equation 5 leads to the following expression:

7) đ?&#x203A;žđ?&#x203A;žđ?&#x2018;Ąđ?&#x2018;Ąđ?&#x2018;Ąđ?&#x2018;Ąđ?&#x2018;?đ?&#x2018;? = đ?&#x203A;žđ?&#x203A;žpiđ?&#x2018;?đ?&#x2018;? + đ?&#x203A;žđ?&#x203A;žoisbsđ?&#x2018;?đ?&#x2018;? + đ?&#x203A;žđ?&#x203A;žoidbsđ?&#x2018;?đ?&#x2018;? + đ?&#x203A;žđ?&#x203A;žoisbcđ?&#x2018;?đ?&#x2018;? + đ?&#x203A;žđ?&#x203A;žoidbcđ?&#x2018;?đ?&#x2018;? + đ?&#x203A;žđ?&#x203A;žopđ?&#x2018;?đ?&#x2018;? đ??śđ??ś

(7)

+ â&#x2C6;&#x2018; (đ?&#x203A;žđ?&#x203A;žoisbsđ?&#x2018;?đ?&#x2018;?đ?&#x2018;? + đ?&#x203A;žđ?&#x203A;žoidbsđ?&#x2018;?đ?&#x2018;?đ?&#x2018;? + đ?&#x203A;žđ?&#x203A;žoisbcđ?&#x2018;?đ?&#x2018;?đ?&#x2018;? + đ?&#x203A;žđ?&#x203A;žoidbcđ?&#x2018;?đ?&#x2018;?đ?&#x2018;? ) đ?&#x2018;?đ?&#x2018;? â&#x20AC;˛ =1 đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?

Where Îłtcc is the total effect of the combination of categories; Îłpic is the effect of the price index on sales of the promoted light (regular) item of category c; Îłoisbsc and Îłoidbsc refer to the within-substitution effects between the promoted item and other light (regular) items of category c, with the same brand and with other brands, respectively; Îłoisbcc and Îłoidbcc refer to the complementary effects between the promoted and other light (regular) items of category c, with the same brand and with other brands, respectively; and Îłopc is the cross effect between periods. Similarly, Îłoisbsc' and Îłoidbsc' (Îłoidbcc' and Îłoidbcc') are the substitution (complementary) effects of products with the same brand and with other brands in categories c' â&#x2030; c. Hence, the estimation for each individual item in category c (c=1,...,C) of the demand Equation 6, Ě&#x201A;Îł*j'c'jc and Ě&#x201A;Îł*jc'jc,r, gives us, by adding the corresponding parameters, the effects of Equation 7, and therefore the decomposition of the effect of price promotions, differentiating between light and regular products. Specifically, decomposition of the sales effect of price promotions entails seven modeling stages. Table 3 shows a summary of these stages. For more modeling details, see ParreĂąo-Selva et al. (2014).

International Food and Agribusiness Management Review

647

Parreño-Selva et al.

Volume 20, Issue 5, 2017

Table 3. Modeling steps (adapted from Leeflang et al., 2008). Step What

Why

1 2 3 4 5 6

To obtain comparable sales measures across categories To obtain comparable price measures across categories To eliminate day-of-the-week effects To eliminate trend and seasonality To obtain parameter estimates To obtain average daily effects in terms of sales

Multiply daily unit sales by average price Divide daily price by average price Standardize sales per day Filter sales, price Model estimation using transformed variables Multiply regression coefficients by average standard deviation per day Use significant coefficients to obtain decomposition effects

To obtain reliable decomposition results

4.3. Total effect of price promotion of the light product (low-fat milk; no immediate costs and delayed rewards) on sales in comparison to this effect for the regular product (whole milk) By focusing on milk as an example of light and regular products, we estimate the effects of the own and cross-filtered price index on filtered standardized sales for 34 item sales (10 regular item sales and 24 light item sales) and within- and cross-category variables. Hence, we estimate 34 sales equations. Our analysis of all series of residuals indicates no serious autocorrelation problems. We find serious correlation problems across three groups of covariates, because prices are frequently used simultaneously for these three groups of items with the same brand. As the prices are identical, we use the common index price. Table 4 shows the results of the decomposition of the three milk categories. ■■ Effect of price promotion of light and regular products on own sales In the regular-milk category, only one own significant effect was noted; this is for Puleva omega 3 1 l, and is equal to 146.63. For the light items, more products have an own significant effect; however, their sizes are lower – between 9.64 and 99.89, with an average of 52.53. Consequently, on average, price promotions increase own sales of regular products almost three times more than sales of light products. Given that prices are estimated in terms of €10, the coefficients in Table 4 represent the effect of a 10% reduction on a price of €10. For example, the estimated effect for semi-skimmed Puleva calcio 1 l reflects a sales increase (units × regular price) of 68.14 Euros after a 10% price reduction. Hence, Whole Puleva omega 3 1 l has the largest sales income over the study period. As we expected, the results show that the own price promotion effect is bigger for regular products than for light products that do not impose an immediate cost. It seems that due to the more enduring involvement implicit in these light products, consumers place limited importance on temporary price promotions at the moment of purchase. ■■ Cross effects between temporal periods We did not expect to find a stockpiling effect for light and regular products due to the characteristics of the local residents (with weekly wages and living in apartments) and of the hypermarket (with reduced parking). Indeed, we only found a significant effect for whole Puleva omega 3 1 l. Therefore, price promotions of regular products can be said to have a greater negative impact on sales in the periods following the promotion, compared to promotions of light products. That is, customers of this establishment do not buy light products during price promotions in order to store them for future consumption. In any case, the context of the study influences the stockpiling effect results and impedes adequate testing of the stockpiling effect.

International Food and Agribusiness Management Review

648

ParreĂąo-Selva et al.

Volume 20, Issue 5, 2017

Table 4. Decomposition of the effects of price promotions for regular and light products1. Net category expansion (%)

Total effect (%)

Same brand (%)

Other brand (%)

68.14

110

279

406

Semi Skimmed Pascual 1 l

84.36

122

246

155

Semi Skimmed Asturiana 1 l

47.86

166

183

Skimmed Puleva Calcio 1 l

43.18

100

9.64

129

242

113

213

345

132

Skimmed Pascual 1 l

99.89

130

Skimmed Asturiana 1.5 l

14.64

100

Average

52.53

158

154

216

Whole Puleva Omega 3 1 l

146.63 39

Average

146.63 39

Substitution

Complementary

Total

Substitution

Complementary Other brand (%)

Same brand (%)

Other brand (%)

Same brand (%)

Other brand (%)

Same brand (%)

Stockpiling effect (%)

Semi Skimmed Puleva Calcio 1 l

Own effect

Category expansion (%)

Cross effects between categories

Other brand (%)

Cross effects within the category

Light products

Skimmed Asturiana Fibra 1 l

Regular products

Whole-milk items: Pascual Calcio 1 l Puleva Calcio 1 l, Puleva Calcio, Pascual 1 l, Asturiana 1.5 l and Asturiana 1 l; semi-skimmed-milk items: Pascual Calcio 1 l and Asturiana Fibra l1, Asturiana Linea 1 l and Asturiana 1.5 l; skimmed-milk items: Pascual Calcio 1 l and Asturiana 1 l are not included because they show no significant own effects.

International Food and Agribusiness Management Review

649

Parreño-Selva et al.

Volume 20, Issue 5, 2017

■■ Cross effects within the category In regular products, complementary effects can be seen both with the same brand and with other brands (14% for items with the same brand and 6% for items with other brands). In addition, in light products there are complementary effects of price promotions on sales of other products within the same category, with an average of 54% of items with the same brand and 23% for items with other brands. These average results are not what we expected for the complementary effects. A possible explanation for the obtained result of greater complementary effects for light products than for regular products is that categories could be different with respect to other characteristics; for example, in market share (Narasinham et al., 1996). Thus, light labels may have been able to obtain high market shares because they are resistant to the effects of brand switching due to promotions of regular labels. With regard to substitution effects within the respective categories, in the case of skimmed Asturiana fibra 1 l, this effect is considerable (129%); for light products on average, the effect is 18%. However, for regular products the effect is null, and there is no cannibalization. That is, the substitution effects within the respective categories are bigger for light products than for regular products. These results are as expected. It seems that light products with delayed health benefits and no immediate costs are nonimpulse products (i.e. relative necessities), where purchase is planned, which implies that consumers have little flexibility to adjust the demand of the category, and the only way for consumers to save money is to switch brands (Bell et al., 1999). This suggests a substitution effect for price promotions of light products on sales of other light products within the category. Therefore, it implies that the effects of price promotions of light products on total sales of their own category could be null or negative; that is, the total sales of the category promoted could be reduced or, in the best case, unaltered. In summary, considering the total cross effects within the category, we can see that the complementary effect is bigger than the substitution effect, and is bigger for light products (77 and 18%, respectively) than for regular products (20 and 0%). In addition, the total cross effect within the category is bigger across items with the same brand than for items with other brands, for both light products (35 and 23%, respectively), and for regular products (14 and 6% respectively). This suggests that there is an umbrella effect and that this is stronger for the light category. ■■ Cross effects between categories In terms of the substitution effect between categories, unlike with regular products, with light products there is a substitution effect between brands (4%). In other words, buying promoted light products reduces additional purchases of regular products with other brands (substitution effect across categories). This result is as expected for the substitution effect, although in a smaller quantity. It seems that the consumer would lose the intrinsic quality benefits of light products5 (e.g. health benefits and better taste) when switching to regular products, which means that price promotions of light products hurt sales of regular products. This result suggests that if the managers want consumers to buy light products instead of regular products, price promotions could be an appropriated tool. We find a large number of complementary effects between categories. In fact, we evidence this effect in 63% (five out of eight) of the products. In addition, on average, the complementary effect is much higher in light than in regular products for both the same brand and for other brands (31 and 31; 11 and 6%, respectively). This result is not as expected for the complementary effect.

Specifically, the fact that consumers value highly the intrinsic quality benefits of light products would be opposite to that suggested by sustainable consumption research (Luchs et al., 2010).

International Food and Agribusiness Management Review

650

Parreño-Selva et al.

Volume 20, Issue 5, 2017

In brief, we can conclude that the cross effects between categories differ significantly between regular and light products. This result is supported by other studies’ suggestions that the cross effects differ greatly from one product category to another (Leeflang et al., 2008). Finally, if we jointly consider the estimated decomposition of the above effects of price promotions of regular and light products, we find that the total effect, in percentage terms, is much higher in light products than in regular products. This result is not as expected, given that we expected a bigger own effect, complementary within-category effect, and complementary between-category effect for regular products than for light products. Specifically, we find that the complementary cross effect between categories is much higher for light products, and there is no substitution effect for regular products (either within or between categories), nor is there a stockpiling effect for light products. If we only consider own effects price promotions of light products are less profitable than promotions of regular products. However, thanks to complementary cross effects within the category and cross effects between categories and to the lack of stockpiling effect, price promotions of light products are more profitable than promotions of regular products. This result is supported by other studies that have revealed the importance of considering the effect of price promotion across categories to identify the profitability of price promotion (Leeflang and Parreño-Selva, 2012).

5. Conclusions Analysis of the survey data reveals that light and regular products are not equivalent to relative virtues and vices. In addition, our results evidence that the own sales effect and the stockpiling effect of price promotions for milk is greater for the regular product than for the light product with no immediate cost. Similarly, we find a cannibalization effect of price promotions of light products on sales of other light products, as well as a substitution effect of promoted light products on additional purchases of regular products. Despite this, as there is no substitution effect for regular products within or between categories, and as the complementary effect is much higher for light products, the category expansion effect and the total effect is much higher for light products than for regular products. In this sense, although price promotions of light products are less profitable than promotions of regular products in terms of their own effect, the between-categories complementary effects and the lack of stockpiling effect allow price promotions of light products to be more effective. The management implications of these results are as follows: first, retailers and manufacturers should analyze consumers’ perceptions and the impact of price promotions on sales from the perspective of light and regular products on a continuum along with relative virtues and vices. The idea of this single continuum was suggested previously by Hoch and Loewenstein (1991) and Haws and Winterich (2013), and ranges from a short-term desire for pleasure to a long-term goal requiring abstention from desired behaviors; e.g. indulging in tasty foods versus minimizing calories. It is also supported by the logic of Gutman (1982), who stated that consumers who want good health and sees a connection between diet and health could form a natural-artificial distinction that could be applied to products by grouping them according to type in an array of categories along the dimension defined by the natural-artificial distinction. Furthermore, increasing consumer sensitivity to health issues in developed countries and the results obtained in this study support consumer interest in relation to light products with no immediate costs and delayed rewards, as well as to light products with immediate costs and delayed rewards (i.e. relative virtue). In addition, managers in certain countries must concentrate more on dairy products than on other products because dairy products comprise eight of the top 10 light products in countries such as Belgium (Viaene, 2015). Second, the larger own effect of virtue over vice price promotions, as detected by Wertenbroch (1998), suggests that consumption self-control of vice products (e.g. chocolate, alcoholic beer, and regular cigarettes) could lead marketing managers to segment and discriminate prices, offering a variety of packet sizes that in particular include small-sized vice products with price premiums (e.g. M&Ms sold in vending machines, alcoholic beer sold in packs of 25 cl bottles and in packs of 33 cl bottles, and cigarettes sold in packs of 10 and in packs of 20), as opposed to the discounts applied to virtue products. In this way, manufacturers of International Food and Agribusiness Management Review

651

Parreño-Selva et al.

Volume 20, Issue 5, 2017

vice products could meet the demands of consumers that practice rationing and those that do not in order to increase profits. However, the larger own effect of price promotions of regular products (whole milk) compared to that of light products (low-fat milk) that is found in our study would make price discrimination unnecessary in regular-product segmentation (e.g. whole milk in packs of 50 cl cartons and in school size cartons) because it seems that consumption self-control of this regular product is not present. Third, although the own effect of price promotions of the regular product, whole milk, is higher than for the light product, low-fat milk, the cross effects between categories allow price promotions of these light products to be more effective than those of regular products in terms of increasing the store’s general sales and profits. In particular, retailers may benefit from cross effects of price promotion between categories for low-fat milk, given that the complementary effects between categories are greater than the substitution effects between categories. From this perspective, for retailers it is more profitable to promote low-fat milk than whole milk. To summarize, retailers can use promotions of these regular products and light products (with no immediate cost) with little risk. Fourth, retailers should consider for milk that promotions of light products with no immediate cost hurt other light-product and regular-product sales. This cross effect within and between categories shows that a substitution effect (which becomes cannibalization when it affects the same brand) should be born in mind by retailers. This paper has certain limitations that represent potential areas for future research. First, the identification of light and regular product categories and relative virtue and vice product categories followed the procedure proposed by Wertenbroch (1998), which found that the category sets are not equivalent. However, Wertenbroch’s study is almost 20 years old, and during the last two decades the prevalence of light products has increased significantly, which has likely had an impact on individuals’ ratings of these products. Second, the focus on a student survey and on product categories of one hypermarket impedes generalization of the results regarding the continuum of products and better understanding of the behavior of promotions of light and regular products, respectively. Third, the data from a hypermarket with reduced parking space and situated in an area in which the residents have weekly incomes and live in apartments prevents analysis of the inventory effect, and therefore of the strategies used by establishments that want their customers to buy large quantities of products during price promotions to store for later use. Fourth, the reduced use of flyers by this hypermarket prevented us from studying this competitive instrument, which could have affected the results.

Acknowledgements The authors appreciate the financial support provided by Ministerio de Ciencia e Innovación of Spain (ECO2012-32001). The authors would like to thank our colleague Aurora Calderón-Martinez for her helpful comments and suggestions.

References Arambepola, C., P. Scarborough, A. Boxer and M. Rayner. 2008. Defining ‘low in fat’ and ‘high in fat’ when applied to a food. Public Health Nutrition 12(3): 341-350. Bell, D.R., J. Chiang and V. Padmanabhan. 1999. The decomposition of promotional response: an empirical generalization. Marketing Science 18(4): 504-526. Bezawada, R. and K. Pauwels. 2013. What is special about marketing organic products? How organic assortment, price, and promotions drive retailer performance. Journal of Marketing 77: 31-51. Blattberg, R.C., R. Briesch and E. Fox. 1995. How promotions works. Marketing Science 14(3): G122-G132. Bourn, D. and J. Prescott. 2002. A comparison of the nutritional value, sensory qualities and food safety of organically and conventionally produced foods. Critical Reviews in Food Science and Nutrition 42(1): 1-34.

International Food and Agribusiness Management Review

652

Parreño-Selva et al.

Volume 20, Issue 5, 2017

Bronnenberg, B. and L. Wathieu. 1996. Asymmetric promotion effects and brand positioning. Marketing Science 15(4): 379-394. Cerveceros de España. 2001. Libro Blanco de la Cerveza. Cerveceros de España, Madrid, Spain. Delvecchio, D., D. Henard and T. Freling. 2006. The effect of sales promotion on post-promotion brand preference: a meta-analysis. Journal of Retailing 82(3): 203-213. Dhar, R. and I. Simonson. 1999. Making complementary choices in consumption episodes: Highlighting versus balancing. Journal of Marketing Research 36: 29-44. Drewnowski, A. 1997. Taste preferences and food intake. Annual Review of Nutrition 17: 237-253. Food and Drug Administration (FDA). 2013. U.S. guidance for industry: a food labeling guide. Available at: http://tinyurl.com/yd4n4pga. Gutman, J. 1982. A means-end chain model based on consumer categorization processes. Journal of Marketing 46: 60-72. Haws, K. and K. Winterich. 2013. When value trumps health in a supersized world. Journal of Marketing 77: 48-64. Higie, R.A. and L.F. Feick.1989. Enduring involvement: conceptual and measurement issues. Advances in Consumer Research 16: 690-696. Hoch, S. and G. Loewenstein. 1991. Time-inconsistent preferences and consumer self-control. Journal of Consumer Research 17(4): 492-506. Huyghe, E. and A. Van Kerckhove. 2013. Can fat taxes and package size restrictions stimulate healthy food choices? International Journal of Research in Marketing 30(4): 421-423. Kaplan, N., B. Palmer and M. Denke. 2000. Nutritional and health benefits of beer. American Journal of the Medical Sciences 320(5): 320-326. Klatsky, A., G. Friedman, M. Armstrong and H. Kipp. 2003. Wine, liquor, beer, and mortality. American Journal of Epidemiology 158(6): 585-595. Laurent, G. and J.N. Kapferer. 1985. Measuring consumer involvement profiles. Journal of Marketing Research 22: 41-53. Leeflang, P.S.H. and J. Parreño-Selva. 2012. Cross-category demand effects of price promotions. Journal of the Academy of Marketing Science 40(4): 572-586. Leeflang, P.H., J. Parreño-Selva, A. Van Dijk and D.R. Wittink. 2008. Decomposing the sales promotion bump accounting for cross-category effects. International Journal of Research in Marketing 25: 201-214. Luchs, M.G., R.W. Naylor, J.R. Irwin and R. Raghunathan. 2010. The sustainability liability: potential negative effects of ethicality on product preference. Journal of Marketing 74(5): 18-31. Makatouni, A. 2002. What motivates consumers to buy organic food in the UK?: results from a qualitative study. British Food Journal 104: 345-352. Milkman, K.L., T. Rogers and M.H. Bazerman. 2008. Harnessing our inner angels and demons: what we have learned about want/should conflicts and how that knowledge can help us reduce short-sighted decision making. Perspectives on Psychological Science 3(4): 324-338. Mishra, A. and H. Mishra. 2011. The influence of price discount versus bonus pack on the preference for virtue and vice foods. Journal of Marketing Research 48(1): 196-206. Mittal, B. and M.S. Lee. 1989. A causal model of consumer involvement. Journal of Economic Psychology 10(3): 363-389. Narasimhan, C., S.A. Neslin and S.K. Sen. 1996. Promotional elasticities and category characteristics. Journal of Marketing 60: 17-30. Okada, E., 2005. Justification effects on consumer choice of hedonic and utilitarian goods. Journal of Marketing Research 42(1): 43-53. Parreño-Selva, J., F.J. Mas-Ruiz and E. Ruiz-Conde. 2014. Price promotions effects of virtue and vice products. European Journal of Marketing 48(7/8): 1296-1314. Pauwels, K., D.M. Hanssens and S. Siddarth. 2002. The long-term effects of price promotions on category incidence, brand choice, and purchase quantity. Journal of Marketing Research 39: 421-439. Raghunathan, R., R.W. Naylor and W.D. Hoyer. 2006. The unhealthy = tasty intuition and its effects on taste inferences, enjoyment, and choice of food products. Journal of Marketing 70(4): 170-184.

International Food and Agribusiness Management Review

653

ParreĂąo-Selva et al.

Volume 20, Issue 5, 2017

Smith, T. 2004. The McDonaldâ&#x20AC;&#x2122;s equilibrium: advertising, empty calories, and the endogenous determination of dietary preferences. Social Choice and Welfare 23(3): 383-413. Van Doorn, J. and P. Verhoef. 2011. Willingness to pay for organic products: differences between virtue and vice foods. International Journal of Research in Marketing 28: 167-180. Van Heerde, H.J., P.S.H. Leeflang and D.R. Wittink. 2002. How promotions work: SCAN*PRO-based evolutionary model building. Schmalenbach Business Review 54(3): 198-220. Van Heerde, H.J., P.S.H. Leeflang and D.R. Wittink. 2004. Decomposing the sales promotion bump with store data. Marketing Science 23: 317-334. Viaene, J. 2015. Consumer behavior towards light products in Belgium. British Food Journal 99(3): 105-113. Wertenbroch, K. 1998. Consumption self-control by rationing purchase quantities of virtue and vice. Marketing Science 17(4): 317-337.

International Food and Agribusiness Management Review

654

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2017.0006 Received: 10 January 2017 / Accepted: 3 June 2017

Zero-inflated ordered probit approach to modeling mushroom consumption in the United States RESEARCH ARTICLE Yuan Jiang a, Lisa A. Houseb, Hyeyoung Kimc, and Susan S. Percivald aPhD

Candidate and bProfessor, Food and Resource Economics Department, University of Florida, P.O. Box 110240, Gainesville, FL 32611, USA

cUI

Economist, Oregon Employment Department, 805 Union St NE, Salem, OR 97311, USA dProfessor,

Food Science and Human Nutrition Department, University of Florida, P.O. Box 110240, Gainesville, FL 32611, USA

Abstract This paper investigates the determinants of fresh and processed mushroom consumption in the United States by employing the zero-inflated ordered probit (ZIOP) model. The ZIOP model accounts for excessive zero observations and allows us to differentiate between genuine non-consumers and individuals who did not consume during the given period but might under different circumstances. The results indicate that the market for fresh mushrooms is larger than that for processed mushrooms. However, the market for processed mushrooms has a larger portion of potential consumers which might indicate more potential if appropriate marketing strategies are applied. The results also suggest that the decisions to participate in the market or not and the consumption frequency are driven by structurally different factors. A comparison of the ZIOP to other models is included to show the advantages of allowing for non-consumers and potential consumers to be analyzed separately. Keywords: fresh and processed mushrooms, zero-inflated ordered probit model, consumption behaviors, double hurdle model JEL code: D12, C35, C25 Corresponding author: jyspring@ufl.edu

ÂŠ 2017 Jiang et al.

655

Jiang et al.

Volume 20, Issue 5, 2017

1. Introduction The United States is the world’s second-largest producer of mushrooms, following China, with 16% of world output. Regarding the value of production, mushrooms are a leading U.S specialty crop, exceeded only by potatoes, tomatoes and lettuce. Consumption of mushrooms has been on the rise in the United States over the past several decades. Per capita consumption has quadrupled since 1965 (the first year for which reliable data are available). According to data compiled by the U.S. Department of Agriculture’s (USDA) Economic Research Service (ERS), per capita use of all mushrooms totaled about 4.02 pounds in 2015, compared with 0.69 pounds in 1965 (Figure 1). The mushroom market can be divided into two main categories: fresh and processed. Fresh mushrooms accounted for three-fourths of domestic consumption in 2015. Since 1990, per capita consumption of fresh mushrooms increased dramatically, while per capita consumption of processed mushrooms (mostly canned mushrooms) gradually declined. This deviation in the consumption trend happens for other fresh and processed fruits and vegetables as well. According to the USDA ERS ‘disappearance data’, annual per capita consumption of fruits and vegetables in both fresh and processed form has increased approximately 7.3% from 1979 to 2015, reaching 678 pounds. Over this time, the consumption of fresh fruits and vegetables has been increasing significantly faster than the consumption of processed fruits and vegetables. Between 1976 and 2015, fresh vegetable consumption went from being the smaller part of vegetable consumption (45%) to the majority of consumption (54%). Although there is a similar trend in fruit consumption, processed consumption still dominates the market. From 1979 to 2015, the market share of fresh fruit has increased from 36 to 45%. Most previous studies about mushrooms have focused on their nutritional and medical benefits. For example, Alam et al. (2008) and Chang and Buswell (1996) have shown that dietary mushrooms provide a wide variety of medicinal properties including anticancer, antibiotic, antiviral activities, immunity and blood lipid lowering effects. Anno et al. (2016) and Finimundy et al. (2014) reviewed the nutritional value of mushrooms and indicated that mushrooms were rich in protein with an important content of essential amino acids and fiber.

Processed mushroom 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0

Fresh mushroom

Total consumption

1970/71 1971/72 1972/73 1973/74 1974/75 1975/76 1976/77 1977/78 1978/79 1979/80 1980/81 1981/82 1982/83 1983/84 1984/85 1985/86 1986/87 1987/88 1988/89 1989/90 1990/91 1991/92 1992/93 1993/94 1994/95 1995/96 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15

U.S mushroom consumption/per capita

Aside from USDA disappearance data and retail sales information, few studies explore factors influencing mushroom consumption at the individual level. One exception is a study by Lucier et al. (2003). It indicates that compared to other consumers, Asian and non-Hispanic white consumers were the strongest consumers of mushrooms and per capita mushroom consumption was positively correlated with income. The study also found that men and women between 20 and 39 years old were the leading mushroom consumers, representing about 32% of the population, yet consuming 43% of all mushrooms.

Year

Figure 1. Mushroom consumption changes (data provided by U.S. Department of Agriculture’s Economic Research Service). International Food and Agribusiness Management Review

656

Jiang et al.

Volume 20, Issue 5, 2017

To help fill this gap and investigate the growing market for mushrooms, the purpose of this study is to investigate and compare the determinants of fresh and processed mushroom consumption. Additionally, we will examine prior models used to understand food demand based on survey data. When examining consumption of mushrooms, we can observe two main decisions: consumption (or nonconsumption) and consumption frequency. Although there are two main decisions, we can identify three categories of consumers: those that ‘never’ consume (non-consumers); those that indicate that they did not consume in the specific period (potential consumers); and those that consumed mushrooms in the specific time period (consumers). A challenge with the data is both non-consumers and potential consumers are typically represented by ‘zero’ consumption, hence creating a situation where two types of zeros may be driven by different behaviors (Harris and Zhao, 2007). As a result, in this paper, unlike other literature in the food consumption, we will use the zero-inflated ordered probit (ZIOP) model to examine fresh and processed mushroom consumption to allow the investigation of non-consumers, potential consumers, and consumers. To the best of our knowledge, it is the first time that the ZIOP model was employed to analyze food consumer’s behavior.

2. Literature review Although little research has been published concerning consumer behavior in the mushroom market, much empirical research has been conducted on general fruit and vegetable consumption, and on specific products (i.e. blueberries (Shi et al., 2011), lettuce (Hospido et al., 2009), fresh citrus (Gao et al., 2011), and tomato (Lucier et al., 2000)). In this section, we identify the factors that may influence consumer behavior in the mushroom market based on reviewing previous studies on fruit and vegetable consumptions. Previous research has shown a relationship between demographic factors and fruit and vegetable consumption. For example, studies found that there were large variations in fruit and vegetable consumption among regions, age groups, gender, and social classes. Consumers who have higher education, income, and social status would be more likely to have a higher consumption of fruits and vegetables (i.e. Ball et al., 2015; Konttinen et al., 2013). It indicates that education, ethnicity and household size also have been correlated with the frequency of vegetable consumption (Cook, 2011; McMahon et al., 2013). Culture, tradition, and familiarity are also found to play an important role in influencing consumption of fruits and vegetables. Culture and tradition were considered as the foundations on which all food choice decisions are built, and have a significant correlation with fruit and vegetable consumption (Pollard et al., 2002). Schroeter and House (2015) found that the impact of culture on food choice is large and varied. They found that family consumption of fruits was highly predictive of the individual’s consumption of fruits. McMahon et al. (2013) illustrated that familiarity links closely with consumers’ selections of vegetables and consumers are more likely to have a strong preference for familiar vegetables when they made food choices. In addition to effects of culture and food habits, a growing number of studies indicated that information about health and nutrition is important factors influencing fruit and vegetable consumption. Rekhy and McConchie (2014) and Cook (2011) found that a belief in the health benefits of fruit and vegetables would increase consumption, and consumers’ concerns about nutrition were positively related to the consumption behavior. Cost is another factor affecting the consumption of fruits and vegetables. Many studies found that price was one of the most influential factors on food choice, especially for those in lower socioeconomic groups, for example, students, the retired and the unemployed (Waterlander et al., 2013). Other studies suggest that price was one of the barriers to eating more fruits and vegetables for low-income families in the United States (Cassady et al., 2007). Other factors that have been identified as barriers to vegetable consumption include preparation time (Rekhy and McConchie, 2014), convenience (McMahon et al., 2013; Nijmeijer et al., 2004), sensory factors (taste International Food and Agribusiness Management Review

657

Jiang et al.

Volume 20, Issue 5, 2017

preferences) and ‘freshness’ (Cook, 2011; Coulthard and Blissett, 2009; Kaminski et al., 2000; Lucan et al., 2010).

3. Data An online consumer survey was designed to investigate consumption and knowledge of fresh and processed mushrooms. In September 2012, a random sample of 1,217 respondents in the United States was recruited through a national survey panel. The target sample included adults, aged 18 or older, living in the United States. A total of 1,217 respondents initiated the survey, and 674 respondents completed the survey, for a response rate of 55.4%. In order to eliminate redundancy and any perceived bias, the draft survey questionnaire was pretested by employing the cognitive interviewing. Pretesters were asked to provide feedback concerning the survey length, survey content, and question clarity, and the survey was revised based on their comments. Furthermore, to better assess the survey design and respondent characteristics, before the full launch of the survey, we performed a ‘soft launch’, in which 75 complete responses were collected and analyzed. The survey included three parts: attitudes and perceptions about health benefits of food in general, then mushrooms specifically; consumption habits related to fresh and processed mushrooms; and demographics. Due to the difficulty in collecting the price information across different purchase locations, purchase frequency information is used to represent the consumption amount for each household. In the survey, we first asked whether the respondent had ever purchased fresh or processed mushrooms, and for those that replied positively, whether they had purchased fresh or processed mushrooms in the past month. For those respondents who had purchased in the prior month, a second follow-up question asked how often they purchased during that period. One month was used to help with the accuracy of the data as it is difficult for people to recall purchases more than one month ago. The key dependent variables in this paper are two ordered variables: Fresh_Freq, and Prossed_Freq for the consumption of fresh mushrooms and processed mushrooms, respectively (Table 1). As described above, the variables of consumers’ consumption frequency are collected through the two questions in succession. At first, it was asked ‘Have you purchased fresh/processed mushrooms before?’, where a binary ‘Yes/No’ answer is required. Secondly, for those answered ‘Yes’ to the first question, another question how frequently mushrooms had been purchased in the past month with the answer options being ‘none/only 1 time/more than 1 time.’ Both the respondents who answered ‘No’ to the first question, and the respondents who answered ‘None’ to the second question would typically be treated as zero in the ordered variable of consumption, but the zeroes represent two decisions: whether to participate in the market and whether to purchase in the prior month. Thus the variable of Fresh_Freq/Prossed_Freq are three-level ordered variables, taking the form of the following choices: never participate or did not consume in the last month (y=0); consume one time in the last month (y=1); consume more than 1 times last month (y=2). The independent variables used in the analysis are shown in Table 1, along with their mean values and descriptions. Based on the literature review, the variables consist of three sets. The first set of variables includes sociodemographic characteristics: age of the respondents, gender, education level, race, household income, and weekly food budget, and vegetarian (or not). The second set of variables focuses on consumers’ health and nutrition knowledge regarding general food and then mushroom specifically. The last set of variables is a ranking of how important attributes of mushrooms, including price, taste, convenience, diversity, and safety are to consumers when making purchase decisions.

International Food and Agribusiness Management Review

658

Jiang et al.

Volume 20, Issue 5, 2017

Table 1. Variable descriptions. Variables

Description

Value (%)

Fresh_Freq

Consumption frequency of fresh mushrooms in the last month

37.2 38.9 23.9 55.7 29.1 15.2 45.1 77.0 22.5 18.9 21.4 31.2 5.3 27.0 10.7 15.3 21.5 10.6 14.9 7.4 11.1 7.7 75.5 2.7 11.0 18.6 18.5 12.2 36.3 32.7 10.8 3.5 4.5 6.5 82.7 44.1 59.7 39.7 48.6

Never/none 1 time More than 1 time Processed_Freq Consumption frequency of processed mushrooms in the Never/none last month 1 time More than 1 time Male % of sample male College % of sample with more than high school education Age Age in years (continuous in analysis) 18-29 years 30-39 years 40-49 years 40-69 years 70+ years Income Estimated household income $24,999 or less $25,000-$34,999 $35,000-$49,999 $50,000-$74,999 $75,000-$99,999 $100,000+ Hispanic % Hispanic Black % black/African American Asian % Asian White % white Otherrace % other races Knowledge_Immunity % who believe mushrooms boost immunity Preventative % who believe food helps relieve symptoms of illness Health_Aware % who are aware of specific health benefits of mushrooms Budget Food budget per week Less than $49 $50-99 $100-149 $150-199 $200-$249 $250+ Vegetarian % vegetarian Taste % who indicate taste as a reason for eating/not eating mushrooms Price % who indicate price as a reason for eating/not eating mushrooms Convenience % who indicate convenience as a reason for eating/not eating mushrooms Mushroom_health/safety % who indicate health as a reason for eating/not eating mushrooms Diversity % who indicate diversity as a reason for eating/not eating mushrooms

4. Method When using survey data to gather information on consumption behavior, two pieces of information are usually gathered: consumption (or not) and consumption frequency. However, there exists three types of consumers: non-consumers (those never consume), potential consumers (those who did not consume in the specific period, but who do consume at other points in time), and consumers (those who consume in the specific period). Although both the non-consumers and potential consumers report zero consumption, International Food and Agribusiness Management Review

659

Jiang et al.

Volume 20, Issue 5, 2017

they are driven by different factors. Non-consumers may choose not to consume mushrooms because of some stable reason, like eating habits, taste preferences, or allergies and they may be less likely to change their eating habits regarding mushrooms easily. At the same time, zero consumption might also be reported for potential mushroom consumers who did not consume during the prior month (or a specific time period studied). The potential mushroom consumer made their decision of â&#x20AC;&#x2DC;non-consumptionâ&#x20AC;&#x2122; as a corner solution to the standard demand problem. It can be expected that the underlying process driving the behaviors of these two types of zero-consumption consumers can be different. If we use a simple discrete choice model (i.e. an ordered probit/logit model) the two types of zeroes will not be differentiated, and the model may fail to capture the true reasons behind the zero observations. In previous research, logistic regression models were widely used to examine factors influencing fruit/ vegetable consumption frequency (i.e. Casagrande et al., 2007; Dehghan et al., 2011; Riediger et al., 2007). As developed, the logistic regression model has been combined with Heckman selection models in which the consumption frequency was considered as conditional on those choosing to participate, thus allowing factors influencing consumption frequency to be different from the factors influencing participation (Chern et al., 2003; Nayga, 1995). Double-hurdle models have also been employed to analyze the fruit and vegetable consumption (Reynolds, 1990; Shi et al., 2011). Another approach to modeling the data is suggested by Harris and Zhao (2007), who proposed the ZIOP. The ZIOP model consists of a split probit model and an ordered probit model with potentially different sets of covariates. The system of the probit model and the ordered probit is generated by two latent equations which allow for the differentiation between the two separate processes generating zero observations. Furthermore, the error terms of these two latent equations are allowed to be correlated. Although the use of ZIOP model is not entirely new to economic analysis1, to the best of our knowledge, this is the first time it is employed to analyze food consumerâ&#x20AC;&#x2122;s behavior. We define X reflecting individualsâ&#x20AC;&#x2122; characteristics including demographics, food habits (vegan or not), health and nutrition knowledge about mushrooms, and Z reflects the ratings of the importance mushrooms characteristics including taste, price, convenience, availability and diversity added to the daily diet. Since nonconsumers of mushrooms have not purchased mushrooms, we only include their individual characteristics in the participation stage, yet for consumption stage, we include both individualâ&#x20AC;&#x2122;s characteristics and mushroom characteristics. In the following model, we let the matrix W include both X and Z. The ZIOP model involves two latent equations: a probit selection equation and an ordered probit equation. The probit selection equation can be expressed as: đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2014; = đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;ź đ?&#x203A;ź đ?&#x203A;źđ?&#x203A;źđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; ; đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = {

1 đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x20AC;˛đ?&#x203A;źđ?&#x203A;ź đ?&#x203A;ź đ?&#x203A;źđ?&#x203A;źi > 0 0 đ?&#x2018;&#x153;đ?&#x2018;&#x153;đ?&#x2018;&#x153;đ?&#x2018;&#x153;â&#x201E;&#x17D;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;

(1)

Equation 1 works to analyze the binary decision to participate in the mushroom market or not. Where Ri is a dichotomous variable indicating whether or not consumers decided to participate, and Ri* is the latent variable measuring consumersâ&#x20AC;&#x2122; propensity for participation to purchase mushrooms. X is the vector of explanatory variables; u is the error term. Conditioning on participation (R=1), consumers need to further decide on how much to consume. The consumption frequency can be represented by a discrete variable Di which is generated by an ordered probit model through the second latent variable D*i. The ordered probit equation is expressed as Equation 2.

0 đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2014; â&#x2030;¤ 0 â&#x2C6;&#x2014; đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2014; = đ?&#x2018;&#x160;đ?&#x2018;&#x160;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x20AC;˛đ?&#x203A;˝đ?&#x203A;˝ đ?&#x203A;˝ đ?&#x203A;˝đ?&#x203A;˝đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; , if đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =1; đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = {j đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x203A;žđ?&#x203A;žđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; < đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x2030;¤ đ?&#x203A;žđ?&#x203A;žđ?&#x2018;&#x2014;đ?&#x2018;&#x2014; (j = 1, â&#x20AC;Ś J â&#x2C6;&#x2019; 1) J đ?&#x203A;žđ?&#x203A;žđ??˝đ??˝đ??˝đ??˝ < đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2014;

(2)

Downward et al. (2011) used the ZIOP model to analyze sports participation. A second study employed the ZIOP model was to analyze two types of peace in social science (Bagozzi et al., 2004).

International Food and Agribusiness Management Review

660

Jiang et al.

Volume 20, Issue 5, 2017

where W is the set of explanatory variables, including both X and Z, Îľ is the error term for the ordered probit equation, and Îł is the cutoff parameters. We include three consumption levels in the survey (0=no purchase in the past month, 1=purchased one time, and 2=purchased more than 1 time per month). Because the model assumes Îł0=0, we will estimate another cutoff from the model. The error terms of Equation 1 and 2 are allowed to be correlated, and the joint distribution function of (ui,Îľi) is assumed to be Gaussian, with zero means, unit variances, and correlation coefficient defined as Ď . Since in the ZIOP model, the decision on whether or not to participate and how much to consume are not separately determined, the indicators D and R are not individually observed. To observe the consumption frequency Y, it was given the following criteria: Y=R*D. According to this criteria, a positive Y was observed when R=1 and D>0; Y was observed as zero when R=0 or D=0. Thus, with this model specification, the probability of non-participation is: Pr(đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 0) = 1 â&#x2C6;&#x2019; ÎŚ(đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x20AC;˛đ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;ź

(3)

The full probabilities for observing different levels of Y are given by: Pr(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 0) = Pr(đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 0) + Pr(đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 1, đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 0) Pr(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = j) = {Pr(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = j) = Pr( đ?&#x2018;&#x2026;đ?&#x2018;&#x2026; đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026; đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;) (đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; đ?&#x2018;&#x2014; đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; đ?&#x2018;&#x2014; đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; Pr(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = J) = Pr( đ?&#x2018;&#x2026;đ?&#x2018;&#x2026; đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;)

Pr(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 0) = [1 â&#x2C6;&#x2019; ÎŚ(đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;ź)] + ÎŚ2 (đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝ Pr(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = j) = ÎŚ2 (đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;žđ?&#x203A;žđ?&#x2018;&#x2014;đ?&#x2018;&#x2014; â&#x2C6;&#x2019;đ?&#x2018;&#x160;đ?&#x2018;&#x160;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝ đ?&#x203A;˝đ?&#x203A;˝2 (đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;žđ?&#x203A;žđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; â&#x2C6;&#x2019;đ?&#x2018;&#x160;đ?&#x2018;&#x160;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝ = (đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; đ?&#x2018;&#x2014; đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; đ?&#x2018;&#x2014; đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; â&#x20AC;˛ â&#x20AC;˛ {Pr(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = J) = ÎŚ2 (đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x203A;źđ?&#x203A;źđ?&#x203A;ź đ?&#x203A;źđ?&#x203A;źđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x203A;˝đ?&#x203A;˝ đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;žđ?&#x203A;žđ??˝đ??˝đ??˝đ??˝ ; â&#x2C6;&#x2019;đ?&#x153;&#x152;đ?&#x153;&#x152;đ?&#x153;&#x152;

(4)

From Equation 4, we could indicate that the probability of observing zero level consumption includes two separate processes: the probability of non-participation (R=0) and the joint probability of the choice to participate, but choose to purchase zero. It also indicates the probability of observing a positive consumption level is the joint probability of the choice to participate and to consume at the j-level intensity. For almost all discrete choice models, marginal effects are useful to indicate the effectiveness of covariates on probability changes. For the ZIOP model, there are different sets of marginal effects which would be of interest to analyze. At first, it would be interesting to analyze the effectiveness of variables on the probability of â&#x20AC;&#x2DC;participationâ&#x20AC;&#x2122;. Then, it will be very interesting to calculate the marginal effects in the ordered model, and compare the effectiveness of the independent variables on the probability of different levels of consumption intensity conditional on participation. Whatâ&#x20AC;&#x2122;s more, based on the construct of ZIOP model, we can also observe the marginal effects of sets of explanatory variables on the overall probability for different levels of observed consumption. We calculate the marginal effects using the formulas shown in Harris and Zhao (2007: 1078). The standard errors of the marginal effects in this study can be calculated using the Delta method (Greene, 2003) or the simulated asymptotic sampling techniques. Like Harris and Zhao (2007), considering the complexity of the marginal effects, the sampling technique is used in this case. To be more specific, we randomly draw Î¸ (where Î¸ is the parameters in the ZIOP model) from multivariate normal distribution (Î¸Ě&#x201A;, Ě&#x201A; [Î¸]) 10,000 times, and for each draw, we calculate the marginal effects, and then calculate the standard var errors. These empirical standard deviations of the simulated marginal effect are the valid asymptotic estimates of the true marginal effectsâ&#x20AC;&#x2122; standard errors. Furthermore, from the ZIOP model, we also calculate the expected probability of observing different levels of consumption: the probability of observing a non-consumer is expressed in Equation 5; the probability of International Food and Agribusiness Management Review

661

Jiang et al.

Volume 20, Issue 5, 2017

observing a potential consumer (zero-consumption given participate) is expressed in Equation 6; and the probability of observing zero level of consumption is expressed in Equation 7; besides that, the probability of observing different levels of consumption given R=1 is given in Equation 8, and the probability of observing different levels of positive consumption is given in Equation 9.

E(đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =0) = Pr(đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =0|X) = 1 â&#x2C6;&#x2019; ÎŚ(đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x20AC;˛đ?&#x203A;źđ?&#x203A;źđ?&#x203A;ź

(5)

E(đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =0, đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =1)= Pr(đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =1, đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =0|X,W)= ÎŚ2 (đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;źđ?&#x203A;ź đ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝

(6)

E(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =0)= E(đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =0)+ E(đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =0, đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =1)= 1 â&#x2C6;&#x2019; ÎŚ(đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x20AC;˛đ?&#x203A;źđ?&#x203A;źđ?&#x203A;ź + ÎŚ2 (đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;źđ?&#x203A;ź đ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝ Pr(đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =d,đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =1|X,W)

E(đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =j|đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =1) =

Pr(đ?&#x2018;&#x2026;đ?&#x2018;&#x2026;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =1|X)

(7)

ÎŚ2 (đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x2018;&#x2014;đ?&#x2018;&#x2014; â&#x2C6;&#x2019;đ?&#x2018;&#x160;đ?&#x2018;&#x160;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝)â&#x2C6;&#x2019;ÎŚ2 (đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x203A;źđ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014;đ?&#x2018;&#x2014; â&#x2C6;&#x2019;đ?&#x2018;&#x160;đ?&#x2018;&#x160;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x20AC;˛ đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝đ?&#x203A;˝)

ÎŚ(đ?&#x2018;&#x2039;đ?&#x2018;&#x2039;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x20AC;˛đ?&#x203A;źđ?&#x203A;źđ?&#x203A;ź

(8)

E(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; =j) =Pr( R i = 1 , Di = j|X, W)= ÎŚ2 (Xiâ&#x20AC;˛ Îą, Îłj â&#x2C6;&#x2019; Wiâ&#x20AC;˛ Î˛; â&#x2C6;&#x2019;Ď ) â&#x2C6;&#x2019; ÎŚ2 (Xiâ&#x20AC;˛ Îą, Îłjâ&#x2C6;&#x2019;1 â&#x2C6;&#x2019; Wiâ&#x20AC;˛ Î˛; â&#x2C6;&#x2019;Ď ) (9)

5. Results

Results indicate that 37.2 and 55.7% of participants did not purchase fresh or processed mushrooms in the past month, respectively. Of those that had not purchased, 54.5 and 56.2% reported that they have never purchased fresh or processed mushrooms, respectively (Figure 2). Of the total respondents, approximately 18.5% indicated they were aware of health benefits of mushrooms, and approximately 21.6% reported that they believed that mushrooms would help with immunity. The estimated probabilities of different types of consumers for fresh and processed mushrooms from the ZIOP model are displayed in Table 2. Overall, the predicted probability of non-consumers of fresh mushroom is 18.0% (compared to the observed percentage of 20.3%), and the estimated predicted probability of potential consumers is 19.5% (compared to the observed percentage of 16.9%). For processed mushrooms, the estimated probability of non-consumers is 32.8% (compared to the observed percentage of 31.3%), and potential consumers is 24.0% (compared to the observed percentage of 24.4%). It indicates that the percentage of non-consumers of processed mushrooms is much higher than the percentage of non-consumers of fresh

Non-consumers Procesed mushroom Fresh mushroom 0%

Zero-consumption last month

31.3%

24.4%

20.3%

10%

16.9%

20%

30%

40%

50%

60%

Figure 2. Zero consumption for fresh and processed mushrooms. Table 2. Estimated probabilities for fresh and processed mushroom consumption. Binary probit

Ordered probit

ZIOP1 model

Fresh

Processed

0.18 0.20 0.39 0.23

0.33 0.24 0.40 0.03

Processed

Pr(Y=0) 0.35

0.57

Pr(Y=0) 0.36

0.60

Pr(Y>0) 0.65

0.43

Pr(Y=1) 0.50 Pr(Y=2) 0.14

0.30 0.10

Pr(R=0) Pr(R=1, D=0) Pr(R=1, D=1) Pr(R=1, D=2)

ZIOP = zero-inflated ordered probit. International Food and Agribusiness Management Review

662

Jiang et al.

Volume 20, Issue 5, 2017

mushrooms (18%), yet the predicted percentage of potential consumers of processed mushrooms is higher than that for fresh mushrooms (19.5%). Thus, we could see that the fresh mushroom is more popular now, yet the market of processed mushroom has strong potential with a larger portion of potential consumers. The OP model is conditional on Z, which treats all the observed zeros indifferently, and ZIOP model is conditional on both X and Z which allows zero observations to come from two different generating processes. Results for the likelihood ratio statistics for both fresh mushroom consumption and processed mushrooms consumption clearly reject the OP model. As for the information criteria, we can see that for processed mushroom consumption, the AIC clearly suggests the superiority of the ZIOP model over the OP model. For fresh mushroom consumption, the AIC suggests that the OP model is performing slightly better than the ZIOP model, although the difference is slight (Table 3). 5.1 Fresh mushroom consumption Marginal effects on Pr(y=0) using a ZIOP model, compared with the results from the probit and ordered probit models are shown in Table 42. For the ZIOP model, the overall marginal effect on Pr(y=0) was divided into two parts: the effect on non-participation (Pr(r=0)), and the effect on the participation with zero consumption Pr (r=1, d=0). In Table 6, we present marginal effects on the unconditional probabilities of positive levels of consumption (y=1, 2), using an ordered probit model versus the ZIOP model. Focusing on the demographic characteristics, age is significantly negative in the ordered probit (OP), however, when looking at the results from the ZIOP model, age is significantly negatively correlated with participation, but is not related to consumption frequency, indicating that younger people are more likely to consume fresh mushrooms, but not consume them more frequently. Males were more likely than females to purchase mushrooms according to the binary probit model, but no relationship was found in the OP or ZIOP models. The variables representing race and ethnicity also have different results depending on which model is used. In the binary probit model, results indicate that African Americans are less likely than Whites to purchase mushrooms and the ordered probit model suggests Asian, Hispanics, and people of other races are more likely to consume fresh mushrooms. In the ZIOP model, the same results are found for consumption frequency as found in the OP model, but not participation (in this case, no race or ethnicity variables significantly influenced participation). Income is positive and significant in the binary probit model and the participation stage of ZIOP model, which indicates that people with higher income are more willing to consume fresh mushrooms, but higher income donâ&#x20AC;&#x2122;t translate to more frequent consumption. Weekly food budget is significantly positive in the probit model and the OP model. In the ZIOP model, we find that budget is significantly correlated with higher fresh mushroom consumption frequency, but it does not significantly influence consumersâ&#x20AC;&#x2122; participation decisions. Thus, comparing the variable income and budget in the ZIOP model, we see that people with higher household income are more likely to try fresh mushrooms, and people with a higher weekly food budget are more likely to buy fresh mushrooms more 2

As marginal effects are more easily interpreted, regression results are not displayed in the text, however they are discussed and provided in Supplementary Tables S1 and S2.

Table 3. Mushroom consumption: summary statistics from ordered probit and ZIOP models1,2.

Log likelihood AIC LR versus OP

Fresh mushroom consumption

Processed mushroom consumption

ZIOP

-525.3 1,086.6 24.2*(df=13)

-513.2 1,088.4

-542.6 1,121.2 36.4**(df=13)

-524.4 1,110.8

1 ** 2

and * indicate statistical significance at 5 and 10% levels, respectively. ZIOP = zero-inflated ordered probit; AIC = Akaike information criterion; LR = Likelihood-ratio test; df = degrees of freedom. International Food and Agribusiness Management Review

663

Jiang et al.

Volume 20, Issue 5, 2017

Table 4. Fresh mushroom consumption: marginal effect for non-participation and zero.1 Binary probit

Male College Age Income Hispanic Black Asian Otherrace Budget Vegan Knowledge_Immunity Preventative Health_Aware Taste

Zero consumption given participation

Full zero consumption

Pr(y=0)

Pr(r=0)

Pr(r=1, d=0)

Pr(y=0)

-0.087** (0.041) -0.043 (0.049) -0.002 (0.013) -0.023** (0.011) -0.020 (0.096) 0.121* (0.069) -0.216*** (0.063) -0.176* (0.094) -0.053*** (0.019) 0.031 (0.088) -0.021 (0.066) -0.182*** (0.047) -0.202*** (0.046) –

-0.027 (0.037) 0.017 (0.046) 0.036** (0.012) -0.014 (0.010) -0.221*** (0.064) 0.023 (0.067) -0.130** (0.059) -0.197 (0.076) -0.058*** (0.017) 0.004 (0.078) 0.068 (0.061) -0.112** (0.046) -0.108** (0.047) -0.076*** (0.016) -0.017 (0.018) -0.024 (0.020) -0.046** (0.015) -0.008 (0.018)

-0.061 (0.044) 0.014 (0.052) 0.029** (0.014) -0.022* (0.012) -0.175 (0.198) 0.041 (0.076) -0.071* (0.042) 0.010 (0.104) -0.019 (0.022) -0.092 (0.112) 0.081 (0.117) -0.092** (0.050) -0.132*** (0.064) –

0.021 (0.024) 0.002 (0.030) -0.001 (0.009) 0.008 (0.007) -0.027 (0.076) -0.015 (0.043) -0.028 (0.047) -0.111 (0.071) -0.016* (0.010) 0.042 (0.049) -0.022 (0.065) 0.007 (0.031) 0.012 (0.034) -0.028*** (0.009) -0.000 (0.008) -0.018** (0.009) -0.018*** (0.007) -0.005 (0.008)

-0.040 (0.031) 0.016 (0.037) 0.028*** (0.010) -0.015** (0.008) -0.20 (0.142) 0.025 (0.053) -0.099* (0.065) -0.100 (0.082) -0.036*** (0.015) -0.050 (0.080) 0.059 (0.068) -0.085*** (0.038) -0.120*** (0.046) -0.028*** (0.009) -0.000 (0.008) -0.018** (0.009) -0.018*** (0.007) -0.005 (0.008)

–

Covenience

–

Mushroom_health

–

Diversity

–

Nonparticipation

Pr(y=0)

Price

Ordered probit ZIOP2 model

– – – –

Standard errors are in parentheses; ***, ** and * indicate statistical significance at 1, 5 and 10% levels, respectively. ZIOP = zero-inflated ordered probit.

International Food and Agribusiness Management Review

664

Jiang et al.

Volume 20, Issue 5, 2017

often. The variables for education was not related to mushroom consumption or frequency of consumption in any model. Considering consumers’ food habits, vegetarian was not found to be significantly correlated with either participation or consumption of mushroom. The estimated results indicated that consumers who were aware of the health benefits of mushrooms are significantly more likely to consume fresh mushrooms in the binary probit and ordered probit models. However, when looking at the results from the ZIOP model, we found that consumers’ awareness of health benefits only significantly influences the decision of participation, not consumption frequency. Similar results are found for consumers’ belief in certain foods ability to aid in recovery from illness. The coefficient of this variable is statistically significant in all three models, however, with the ZIOP model, it is only significant in the first stage (decision to participate). The belief that mushrooms boost immunity was not significant in any model. When looking at the mushroom characteristics in the ZIOP model, we find that taste, convenience and health benefits of mushrooms are the three significant factors which are positively correlated with fresh mushroom consumption frequency, while the price and diversity added to daily diet are not statistically significant in the case of fresh mushroom consumption. The marginal effects (shown in Tables 4 and 5) highlight some interesting results. As previously mentioned, one of the advantages of the ZIOP model is its capability to disentangle the total effect of a covariate on Pr(y=0) into those effects on the probabilities of the two types of zeros: Pr(r=0) and Pr(r=1, D=0). One example in the case of fresh mushroom consumption is the age effect. From the OP model, one would conclude that older people are more likely to be non-consumers (by 0.036, Table 4). When examining the age effect in the ZIOP model, we see that the dominant effect of age is on the probability to be a non-participant (by 0.029), yet it does not exert any influence on the probability to be a potential consumer. Thus, we conclude older people are more likely to be a non-participant for fresh mushroom consumption, but no more likely than other ages to be a potential consumer. Another example is consumers’ awareness of health benefits. In the OP model, when consumers were aware of the benefit of eating mushrooms, the probability of being non-consumer was lower (by -0.108, Table 4). However, when dividing the effect of health benefits into two categories, we could see that consumers’ awareness of health benefits only significantly correlated with consumers’ participation decision (by -0.132, Table 4), but did not significantly effect consumption frequency. Thus, we could conclude that people being aware of mushroom benefits would be more likely to try fresh mushrooms, yet it will not influence consumers’ decisions on how much to consume. Regarding the potential unobserved effect, the ZIOP model for fresh mushroom suggests that there is no significant correlation existing between the participation decision and consumption decision. This indicates that for the promotion of fresh mushrooms, marketing strategies to attract new consumers should be different than those targeting increased consumption frequency for existing consumers, and vice versa. 5.2 Processed mushroom consumption Marginal effect results for processed mushroom consumption are shown in Tables 6 and 73. Focusing on the demographic variables, gender and education were not significantly related to processed mushroom consumption for any model. Age is only significant in the binary probit model, but not significant in either stage of the ZIOP model. Income is significant in the ZIOP model, where it is negatively related to the consumption participation decision, yet positively related to consumption frequency. Weekly food budget is significantly positive in the probit model, OP model, and the participation-stage of ZIOP model, but is not correlated with consumers’ consumption frequency. This suggests that people with a higher food budget will be more willing to consume processed mushrooms. 3

Regression results for processed mushrooms are displayed in Supplementary Tables S1 and S2.

International Food and Agribusiness Management Review

665

Jiang et al.

Volume 20, Issue 5, 2017

Table 5. Fresh mushroom consumption: marginal effect for non-zero consumption levels.1

Male College Age Income Hispanic Black Asian Otherrace Budget Vegan Knowledge_Immunity Preventative Health_Aware Taste Price Covenience Mushroom_health Diversity 1 2

OP Pr(y=1)

ZIOP2 Pr(y=1)

OP Pr(y=2)

ZIOP Pr(y=2)

0.011 (0.016) -0.007 (0.018) -0.015*** (0.005) 0.006 (0.004) 0.016 (0.033) -0.010 (0.031) 0.035*** (0.010) 0.021 (0.030) 0.024*** (0.008) -0.002 (0.033) -0.024 (0.018) 0.052** (0.025) 0.035*** (0.012) 0.032*** (0.009) 0.007 (0.008) 0.010 (0.008) 0.019*** (0.007) 0.003 (0.008)

0.034 (0.026) -0.004 (0.032) -0.011 (0.010) 0.012* (0.008) 0.040 (0.112) -0.024 (0.047) 0.005 (0.055) -0.074 (0.085) -0.003 (0.014) 0.056 (0.061) -0.042 (0.074) 0.037 (0.032) 0.054* (0.034) -0.017 (0.013) 0.001 (0.006) -0.012 (0.011) -0.011 (0.009) -0.003 (0.006)

0.016 (0.022) -0.010 (0.028) -0.021*** (0.007) 0.008 (0.006) 0.206** (0.093) -0.013 (0.036) 0.096* (0.055) 0.176* (0.103) 0.034*** (0.009) -0.002 (0.011) -0.044 (0.044) 0.059*** (0.023) 0.072* (0.036) 0.044*** (0.009) 0.010 (0.011) 0.014 (0.011) 0.027*** (0.009) 0.005 (0.011)

0.005 (0.025) -0.013 (0.031) -0.018* (0.008) 0.003 (0.007) 0.162** (0.068) -0.001 (0.044) 0.094* (0.046) 0.174* (0.085) 0.039*** (0.012) -0.006 (0.049) -0.016 (0.046) 0.048 (0.033) 0.066* (0.034) 0.045*** (0.012) -0.001 (0.012) 0.030* (0.015) 0.029** (0.011) 0.009 (0.011)

Standard errors are in parentheses; ***, ** and * indicate statistical significance at 1, 5 and 10% levels, respectively. ZIOP = zero-inflated ordered probit.

Regression results indicate that consumers’ awareness of health benefits of mushrooms is not significant in any model. Consumers’ belief that foods can help when sick is not statistically significant except for the binary probit model. However, consumers’ knowledge of the effectiveness of mushroom enhancing immunity is significant in both the binary probit model and the ZIOP model. This variable is significant in both participation stage and the consumption stage, but with opposite directions. According to the ZIOP model, consumers who know mushrooms enhance immunity are more likely to try the processed mushrooms but less likely to purchase more frequently.

International Food and Agribusiness Management Review

666

Jiang et al.

Volume 20, Issue 5, 2017

Table 6. Processed mushroom consumption: marginal effect for non-participation and zero.1 Binary probit

Male College Age Income Hispanic Black Asian Otherrace Budget Vegan Knowledge_Immunity Preventative Health_Aware Taste

Zero consumption given participation

Full zero consumption

Pr(y=0)

Pr(r=0)

Pr(r=1, d=0)

Pr(y=0)

-0.011 (0.041) 0.020 (0.049) -0.028** (0.013) 0.002 (0.011) 0.063 (0.094) 0.122* (0.063) 0.069 (0.077) 0.014 (0.122) -0.069*** (0.019) -0.133 (0.086) -0.111* (0.064) -0.134** (0.046) -0.060 (0.054) –

0.023 (0.039) 0.066 (0.049) -0.004 (0.013) 0.004 (0.011) -0.070 (0.101) 0.070 (0.066) 0.083 (0.069) 0.124 (0.103) -0.059*** (0.018) -0.134* (0.080) -0.036 (0.068) -0.075 (0.049) 0.031 (0.052) -0.033** (0.017) -0.031* (0.020) -0.044** (0.021) -0.019 (0.016) 0.002 (0.020)

-0.037 (0.058) 0.006 (0.067) 0.025 (0.019) 0.026* (0.015) -0.488 (0.552) -0.027 (0.157) 0.059 (0.091) -1.747*** (0.205) -0.060*** (0.024) -1.699 (4.482) -0.241*** (0.070) -0.027 (0.066) -0.040 (0.078) –

0.051 (0.039) 0.021 (0.043) -0.016 (0.014) -0.017* (0.010) 0.231 (0.198) 0.051 (0.106) -0.003 (0.063) 0.637*** (0.110) 0.017 (0.016) 0.480 (4.918) 0.192*** (0.067) 0.005 (0.046) 0.049 (0.052) -0.009* (0.005) -0.015** (0.008) -0.012 (0.100) -0.008 (0.007) -0.008 (0.007)

0.013 (0.031) 0.027 (0.036) -0.009 (0.009) -0.009* (0.005) -0.257 (0.368) 0.024 (0.067) 0.056 (0.048) -1.111*** (0.197) -0.042*** (0.014) -1.218 (3.55) -0.048 (0.050) -0.022 (0.032) 0.008 (0.040) -0.009* (0.005) -0.015** (0.008) -0.012 (0.100) -0.008 (0.007) -0.008 (0.007)

–

Covenience

–

Mushroom_health

–

Diversity

–

Nonparticipation

Pr(y=0)

Price

Ordered probit ZIOP2 model

– – – –

Standard errors are in parentheses; ***, ** and * indicate statistical significance at 1, 5 and 10% levels, respectively. ZIOP = zero-inflated ordered probit.

International Food and Agribusiness Management Review

667

Jiang et al.

Volume 20, Issue 5, 2017

Table 7. Processed mushroom consumption: marginal effect for non-zero consumption levels.1

Male College Age Income Hispanic Black Asian Otherrace Budget Vegan Knowledge_Immunity Preventative Health_Aware Taste Price Taste Covenience Mushroom_health Diversity 1 2

OP Pr(y=1)

ZIOP2 Pr(y=1)

OP Pr(y=2)

ZIOP Pr(y=2)

-0.012 (0.021) -0.035 (0.024) 0.002 (0.007) -0.003 (0.006) 0.036 (0.048) -0.041 (0.041) -0.050 (0.044) -0.077 (0.071) 0.032*** (0.011) 0.063** (0.032) 0.021 (0.04) 0.043 (0.027) -0.017 (0.030) 0.018* (0.010) 0.017* (0.011) 0.018* (0.010) 0.024** (0.012) 0.011 (0.009) -0.001 (0.011)

0.018 (0.034) -0.006 (0.037) 0.013 (0.010) -0.014 (0.010) 0.268 (0.308) 0.007 (0.084) -0.035 (0.050) 0.977* (0.136) 0.034** (0.013) 0.934 (5.696) 0.125* (0.045) 0.016 (0.036) 0.020 (0.043) 0.000 (0.003) 0.001 (0.004) 0.000 (0.003) 0.001 (0.004) 0.000 (0.003) 0.000 (0.003)

-0.010 (0.017) -0.031 (0.024) 0.002 (0.006) -0.002 (0.005) 0.034 (0.053) -0.029 (0.025) -0.034 (0.025) -0.047 (0.033) 0.026*** (0.008) 0.070 (0.049) 0.015 (0.028) 0.032* (0.019) -0.013 (0.022) 0.015** (0.008) 0.014* (0.009) 0.015** (0.008) 0.019** (0.009) 0.009 (0.007) -0.001 (0.009)

-0.032 (0.016) -0.021 (0.018) -0.004 (0.005) 0.004 (0.005) -0.011 (0.071) -0.031 (0.035) -0.021 (0.031) 0.133* (0.081) 0.008 (0.009) 0.283 (8.864) -0.077* (0.028) 0.006 (0.020) -0.028 (0.021) 0.008* (0.005) 0.014** (0.006) 0.008* (0.005) 0.012 (0.009) 0.008 (0.006) 0.008 (0.007)

Standard errors are in parentheses; ***, ** and * indicate statistical significance at 1, 5 and 10% levels, respectively. ZIOP = zero-inflated ordered probit.

Considering mushroom characteristics, results of the regression model indicate that taste and price are the two significant factors influencing the consumption frequency of processed mushrooms. Processed mushrooms with better taste and more reasonable price will significantly increase consumersâ&#x20AC;&#x2122; consumption frequency. Different from fresh mushrooms, convenience and health benefits are no longer significant in the case of processed mushroom consumption.

International Food and Agribusiness Management Review

668

Jiang et al.

Volume 20, Issue 5, 2017

The marginal effects again highlight some interesting differences from the ZIOP model to the OP model for some explanatory factors, such as income and consumers’ knowledge of mushroom benefits on enhancing immunity. Compared to the OP model where income is not significant, the ZIOP model reveals reserve effects of income on consumption participation, and consumption frequency. As shown in Table 6, an increase in the household income results in a 0.026 increase in the probability of non-participation of processed mushrooms, and a 0.017 decrease in the probability of participation with zero consumption. The prior effect indicates that people with higher income might not prefer eating processed mushrooms, and the latter effect indicates that processed mushrooms are a normal good for the participant. Overall, there is a -0.009 net negative effect on the probability of observing zero consumption for the increase in the household income. Thus, the effect of income on the overall zero consumption is approximately zero as the two impacts counter effect each other in the binary and OP equations. However, more information is obtained by using the ZIOP model. Another difference was found in the effect of consumers’ knowledge about the effectiveness of mushroom enhancing immunity. With a single latent equation, like the OP model and Binary Probit model, we assume that there is a homogenous ‘benefit-awareness’ effect that affects an individual moving from non-consumers to consumers of processed mushrooms. However, when employing the ZIOP model, we see that consumers’ knowledge of the effectiveness of mushroom enhancing immunity will significantly decrease the probability of non-participation (Pr(r=0)) by 0.241, while significantly increasing the probability of zero-consumption (Pr(r=1, y=0)) by 0.192. This appears to indicate that although people’s knowledge of mushroom benefits might attract consumers to try processed mushrooms, it does not impact frequency of consumption and might lead to consumers trying the product, but at very infrequent times (hence showing up as potential consumers, but not influence frequency of consumption of frequent consumers). Regarding the potential unobserved effect, the ZIOP model of the processed mushrooms suggests that there exists a significant negative correlation between the two-stage decisions. Using the ZIOP model, results indicate that actions that might attract people to try processed mushrooms might be successful in getting consumers to sample, but might not be successful in creating regular consumers. This relationship suggests that for the promotion of processed mushrooms, attracting new consumers and then increasing their consumption frequency are entirely different challenges. It also reveals the advantage of using the ZIOP model. Ignoring the possible two different zero generating processes, the correlation between the unobserved factors influencing participation and consumption stages might lead to difficulties in correctly making marketing recommendations.

6. Conclusions This study contributes to the literature by conducting an in-depth analysis of mushroom consumption, as well as demonstrating the effectiveness of the ZIOP model in food consumption research. In this paper, a ZIOP model was employed to analyze the significant factors influencing consumers’ behavior in both the fresh and processed mushroom markets accounting simultaneously for the probability of consumption participation and frequency. The ZIOP model allows us to distinguish between zero observations in the survey data which might come from two different sources: genuine non-consumers and zero-consumption participants (potential consumers). The latter were considered as potential consumers because these zero-consumption participants choose not to consume at the current time, but would potentially consume under different circumstances. The market for fresh mushroom is larger than that for processed. In our study, the portion of non-consumers of processed mushrooms is 31.3%, compared to 20.3% for fresh mushrooms. However, the percentage of potential consumers of processed mushrooms is larger than that of fresh mushrooms, indicating that the processed mushroom market may have potential if appropriate marketing strategies are applied. Considering the factors influencing mushroom consumption, our study indicated that the reasons driving non-consumers and potential consumers are different, thus emphasizing the contribution of using a model such as the ZIOP, which specifically allows for this distinction. The reasons behind non-consumers are International Food and Agribusiness Management Review

669

Jiang et al.

Volume 20, Issue 5, 2017

mostly stable demographic attributes like age, income level, consumersâ&#x20AC;&#x2122; perceptions towards mushrooms and ethnicity â&#x20AC;&#x201C; factors that do not change (or do not change quickly). The reasons behind potential consumers are more related to economic reasons like food expenditure. Consumersâ&#x20AC;&#x2122; knowledge and awareness of the health benefits of mushroom are significant for participation and consumption for both fresh and processed mushrooms. This suggests that a policy of advertising the health benefits of mushrooms would be a good method to encourage mushroom consumption. What is more, we also find that once consumers make their determination to participate, sensory factors, like taste are important factors influencing consumption frequency. In this case, improvement in taste in the product will likely lead to increased consumption among consumers, but not an increased quantity of consumers in the market. Comparing strategies for fresh and processed mushrooms: taste, convenience, and health should be the key points promoting fresh mushrooms; while lower prices and better taste would help processed mushroom attract more consumption. The ZIOP model allows us to look into the interesting difference of some explanatory factors on the participation stage and consumption stage. A key example is the effect of household income on processed mushroom consumption. In this case, an increase in household income will cause an increase in the probability of non-participation, but a fall in the probability of participation with zero consumption. This latter effect indicates that the processed mushroom is a normal good for participants, although there is a net negative effect on the probability of observing zero consumption for an increase in the household income. However, basing policy advice on the ordered probit model, one would incorrectly conclude that the processed mushroom is an inferior good, and income is negatively related to both participation and higher consumption. From this study, we could see that the factors influencing potential consumers are different from those influencing non-consumers, and some variables even exert reverse effects on the decisions to participate and the consumption frequency. Thus, there might be structurally different reasons driving non-participants and potential consumers. In the survey design of food consumption studies, to better analyze the market structure and behaviors of different types of consumers, it would be important to not only collect information on consumption and non-consumption, but also the information that differentiates the potential consumers from non-consumers.

Acknowledgements We would like to thank Christian Tejero (University of Florida, Food Science and Human Nutrition Undergraduate Research Intern) for his efforts in data collection.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2017.0006. Table S1. Fresh mushroom consumption: regression results. Table S2. Processed mushroom consumption: regression results.

References Alam, N., R Amin, A. Khan, I. Ara, M.J. Shim, M.W. Lee and T.S Lee. 2008. Nutritional analysis of cultivated mushrooms in Bangladesh-Pleurotus ostreatus, Pleurotus sajor-caju, Pleurotus florida and Calocybe indica. Mycobiology 36(4): 228-232.

International Food and Agribusiness Management Review

670

Jiang et al.

Volume 20, Issue 5, 2017

Anno, A.H., H.K. Konan, J.P.E. Kouadio, E.A. Dué and L.P. Kouamé. 2016. Chemical composition and nutritional value of two edible mushrooms from three regions of Côte d’Ivoire. Journal of Basic and Applied Research 2: 119-125. Bagozzi, B.E., D.W. Hill Jr, W.H. Moore and B. Mukherjee. 2015. Modeling two types of peace: the zeroinflated ordered Probit (ZiOP) model in conflict research. Journal of Conflict Resolution 59(4): 728-752. Ball, K., K. E. Lamb, C. Costa, N. Cutumisu, A. Ellaway, C.B.M. Kamphuis, G. Mentz, J. Pearce, P. Santana, R. Santos, A.J. Schulz, J.C. Spence, L.E. Thornton, F.J. van Lenthe and S.N. Zenk. 2015. Neighbourhood socioeconomic disadvantage and fruit and vegetable consumption: a seven countries comparison. International Journal of Behavioral Nutrition and Physical Activity 12(1): 68. Casagrande, S.S., Y. Wang, C. Anderson and T.L. Gary. 2007. Have Americans increased their fruit and vegetable intake?: The trends between 1988 and 2002. American Journal of Preventive Medicine 32(4): 257-263. Cassady, D., K.M. Jetter and J. Culp. 2007. Is price a barrier to eating more fruits and vegetables for lowincome families? Journal of the American Dietetic Association 107(11): 1909-1915. Chang, S.T. and J.A. Buswell. 1996. Mushroom nutriceuticals. World Journal of Microbiology and Biotechnology 12(5): 473-476. Chern, W.S., K. Ishibashi, K. Taniguchi and Y. Yokoyama. 2003. Analysis of food consumption behavior by Japanese households. FAO Economic and Social Development Paper. FAO, Rome, Italy, pp. 152: 1-21 Cook, R. 2011. Tracking demographics and US fruit and vegetable consumption patterns. Department of Agricultural and Resource Economics, University of California, Davis, CA, USA. Coulthard, H. and J. Blissett. 2009. Fruit and vegetable consumption in children and their mothers. Moderating effects of child sensory sensitivity. Appetite 52(2): 410-415. Dehghan, M., N. Akhtar-Danesh and A.T. Merchant. 2011. Factors associated with fruit and vegetable consumption among adults. Journal of Human Nutrition and Dietetics 24(2): 128-134. Downward, P., F. Lera-Lopez and S. Rasciute. 2011. The zero-inflated ordered probit approach to modelling sports participation. Economic Modelling 28(6): 2469-2477. Finimundy, T.C., A.J.P. Dillon, J.A.P. Henriques and M.R. Ely. 2014. A review on general nutritional compounds and pharmacological properties of the Lentinula edodes mushroom. Available at: http:// tinyurl.com/y77o3mww. Gao, Z., L.O. House, F.G. Gmitter Jr, M.F. Valim, A. Plotto and E.A. Baldwin. 2011. Consumer preferences for fresh citrus: impacts of demographic and behavioral characteristics. International Food and Agribusiness Management Review, 14(1): 23-40. Greene, W.H. 2003. Econometric Analysis. Prentice Hall, New Jersey, NJ, USA. Harris, M.N. and X. Zhao. 2007. A zero-inflated ordered probit model, with an application to modeling tobacco consumption. Journal of Econometrics 141(2): 1073-1099. Hospido, A., L.M. Canals, S. McLaren, M. Truninger, G. Edwards-Jones and R. Clift. 2009. The role of seasonality in lettuce consumption: a case study of environmental and social aspects. The International Journal of Life Cycle Assessment 14(5): 381-391. Kaminski, L.C., S.A. Henderson and A. Drewnowski. 2000. Young women’s food preferences and taste responsiveness to 6-n-propylthiouracil (PROP). Physiology and behavior 68(5): 691-697. Konttinen, H., S. Sarlio-Lähteenkorva, K. Silventoinen, S. Männistö and A. Haukkala. 2013. Socio-economic disparities in the consumption of vegetables, fruit and energy-dense foods: the role of motive priorities. Public Health Nutrition 16(05): 873-882. Lucan, S.C., F.K. Barg and J.A. Long. 2010. Promoters and barriers to fruit, vegetable, and fast-food consumption among urban, low-income African Americans-a qualitative approach. American Journal of Public Health 100(4): 631-635. Lucier, G., J.E. Allshouse and B.H. Lin. 2003. Factors affecting US mushroom consumption. US Department of Agriculture, Washington, WA, USA. Lucier, G., B.H. Lin, J. Allshouse and L.S. Kantor. 2000. Factors affecting tomato consumption in the United States. Vegetables and Specialties/VGS 282: 26-32.

International Food and Agribusiness Management Review

671

Jiang et al.

Volume 20, Issue 5, 2017

McMahon, A.T., L. Tapsell, P. Williams and J. Jobling. 2013. Baby leafy green vegetables: providing insight into an old problem? An exploratory qualitative study examining influences on their consumption. Health Promotion Journal of Australia 24(1): 68-71. Nayga, R.M. 1995. Determinants of US household expenditures on fruit and vegetables: a note and update. Journal of Agricultural and Applied Economics 27(02): 588-594. Nijmeijer, M., A. Worsley and B. Astill. 2004. An exploration of the relationships between food lifestyle and vegetable consumption. British Food Journal 106(7): 520-533. Pollard, J., S.L. Kirk and J.E. Cade. 2002. Factors affecting food choice in relation to fruit and vegetable intake: a review. Nutrition Research Reviews 15(02): 373-387. Rekhy, R. and R. McConchie. 2014. Promoting consumption of fruit and vegetables for better health. Have campaigns delivered on the goals? Appetite 79: 113-123. Reynolds, A. 1990. Analyzing fresh vegetable consumption from household survey data. Journal of Agricultural and Applied Economics 22(2): 31. Riediger, N.D., S. Shooshtari and M.H. Moghadasian. 2007. The influence of sociodemographic factors on patterns of fruit and vegetable consumption in Canadian adolescents. Journal of the American Dietetic Association 107(9): 1511-1518. Schroeter, C. and L.A. House. 2015. Fruit and vegetable consumption of college students: what is the role of food culture? Journal of Food Distribution Research 46(3): 131-152. Shi, L., L. House and Z. Gao. 2011. Consumer structure of the blueberry market: a double hurdle model approach. In: 2011 American Agricultural Economics Association Annual Meeting, Pittsburgh, 2426 July 2011. AAEA, Milwaukee, WI, USA. Waterlander, W.E., M.R. de Boer, A.J. Schuit, J.C. Seidell and I.H. Steenhuis. 2013. Price discounts significantly enhance fruit and vegetable purchases when combined with nutrition education: a randomized controlled supermarket trial. The American Journal of Clinical Nutrition 97(4): 886-895.

International Food and Agribusiness Management Review

672

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2014.0181 Received: 12 February 2014 / Accepted: 25 May 2017

Cooperation among Ugandan farmers: cultivating social capital RESEARCH ARTICLE J.L. Morrow, Jr. a, Richard Patrick Joyce IIIb, William J. McMahonb, Antonio M. DeMaiab, S. Caleb McVickerb, Ashley E. Parsonsb, and Kristin Wilcoxc aAssociate

Professor of Business Administration, and bResearch Assistant, Department of Business and Accounting, Birmingham-Southern College, 900 Arkadelphia Road, P.O. Box 549023, Birmingham, AL 35254, USA cTechnical

Specialist, Food Security and Cooperatives, Global Communities, 8601 Georgia Avenue, Suite 300, Silver Spring, MD 20910-3440, USA

Abstract A survey was administered to 183 Ugandan farmers in August 2014 to assess the factors that influence their willingness to become members of a proposed new agricultural cooperative. In particular, we were interested in a better understanding of how farmers viewed the social benefits associated with cooperation. These social benefits have the potential to become valuable sources of social capital. Four valid and reliable measures of social benefits were identified. Social benefits that farmers may use to get by (bonding networks) had two dimensions: emotional support and social support. Social benefits that farmers may use to get ahead (bridging networks) also had two dimensions: tangible and intangible resource sharing. The desire to gain these social benefits from cooperation (except for emotional support) emerged as strong predictors of farmersâ&#x20AC;&#x2122; willingness to cooperate in a proposed new agricultural cooperative. Surprisingly, the expected economic benefits of cooperation did not have a significant effect on willingness to cooperate. Implications and suggestions for future research and cooperative development and management are also discussed. Keywords: farmer cooperative formation, Uganda, social capital, social benefits JEL code: P13, Q13 Corresponding author: bmorrow@bsc.edu

ÂŠ 2017 Morrow et al.

673

Morrow et al.

Volume 20, Issue 5, 2017

1. Introduction A cooperative alliance may be defined as an inter-firm arrangement that involves the utilization of resources from autonomous organizations for the joint accomplishment of individual goals (Parkhe, 1993). The alliance form of organization is widely used throughout the food and agribusiness industry and the agricultural cooperative may be one of the oldest, most well established forms of strategic alliance in the world (Oregon State University, 2004). Farmers may decide to join co-ops for several reasons. The most obvious reason farmers join cooperatives is to satisfy their economic goals, or the desire to become better off financially. This may occur when their co-op membership enables the farmer to reduce costs and/or increase revenues. Both supply (input) and marketing (output) co-ops may be used to accomplish these objectives. One could certainly argue that at its core, a farmer owned cooperative is a business organization used as a tool to accomplish economic objectives. However, there is also no doubt that these same organizations have a large social component such that an agribusiness firm does not operate solely as a production function ‘stripped of all human identity’ (Wilson and Kennedy, 1999: 191). Indeed, cooperatives have been characterized as having dual attributes – an economic attribute and a social attribute (Nilsson et al., 2012). Goler von Ravensburg (2010) referred to the economic motivations to join a cooperative as direct effects but also suggested that farmers may be motivated to join cooperatives in hopes of realizing indirect effects that accrue to members based on their social interactions within the cooperative. Similarly, Liu and Sumelius (2010), in their study of members of a Finnish dairy cooperative, concluded that farmers did not join and participate in a cooperative solely to accomplish economic objectives and that future research should consider sociological perspectives as a basis for understanding cooperative members’ motivations and behaviors. A cooperative must have the support of its members in order to survive and a pure economic motivation for member’s support and participation appears incomplete (Liu and Sumelius, 2010). The aim of a cooperative is to create benefits for its members (Nilsson et al., 2012). Thus, before a cooperative is formed, it is important to understand what benefits members desire to achieve. Nilsson et al. (2012) have even suggested that some large complex cooperatives have been losing a portion of their stocks of social capital leading to a decrease among members in their willingness to engage in cooperative behaviors. They attribute this decline in part due to a dissatisfaction among members that their expectations were not being met, which leads to members ultimately abandoning their membership. While it seems clear that farmers may enjoy economic benefits from joining agricultural cooperatives, it remains unclear what social benefits farmers may be seeking when making a decision to join and actively participate in a new agricultural cooperative. Thus, our research addresses the following question: What social benefits do Ugandan farmers hope to gain from participating in a proposed agricultural cooperative and how do these anticipated benefits affect their willingness to cooperate (WTC)? We contend that these desired social benefits act as antecedents in a decision to join a cooperative and that they also have the potential to later develop into valuable sources of social capital within the cooperative.

2. Farmer cooperatives in Africa Connolly (2014) noted that much of sub-Saharan Africa has ample land and water resources to potentially meet the food needs of a growing population. However he also notes that small farmers in Africa have difficulty accessing both supply and output markets. The formation of new cooperatives has the potential to provide this market access, which of course is an economic benefit. Understanding the social benefits that farmers in Africa hope to achieve by joining a new agricultural cooperative is particularly important in countries where there is a mixed history of cooperation among farmers. In Uganda, cooperatives were established as early as 1900 (Kyazze, 2010) and reached the peak of their economic influence in the early 1970s (Kyazze, 2010). However, for a variety of reasons, participation in cooperatives began to decline. As recently as 1992, cooperatives still accounted for 22% of marketed agricultural produce, but by 2001, this

International Food and Agribusiness Management Review

674

Morrow et al.

Volume 20, Issue 5, 2017

percentage had fallen to 2% and then to 1% in 2006 (Kyazze, 2010). Today, agricultural cooperatives in Uganda are experiencing a resurgence, due in part to government support. In nearby Ethiopia, a coffee cooperative was found to provide both economic and social benefits for their members in such endeavors as coffee production and marketing, but these social benefits were not explicitly identified (Meskela and Teshome, 2014). In a study that examined a dairy goat cooperative in neighboring Tanzania, Lie et al. (2012) suggested that farmer cooperatives in many African countries have a long and rather difficult history for reasons that range from poor management and governance structures to lack of working capital and supporting organizations. This suggests that starting new cooperatives in these areas may be more difficult than typically expected and that it is important to consider the resources that the proposed cooperative would control (Lie et al., 2012). Finally, social capital, which is a productive resource built from social interactions (Uphoff and Wijayaratna, 2000), has been a widely studied construct but its precise antecedents are not well understood (Liang et al., 2015). In a study that examined a highly successful 30 year old cooperative in Zambia, the organization’s current stock of social capital was found to be related to the initial conditions that were present prior to joining the cooperative and that these conditions contributed to the cooperative’s eventual success (Mungandi et al., 2012). Among these initial conditions are prospective members’ expectations about their future cooperative activities. Mungandi et al. (2012) found that the initial conditions faced by a cooperative positively influenced long term success. Thus, when forming new cooperatives in sub-Saharan Africa one key to success is gaining an understanding of potential members’ expectations about the social benefits that may be available to them.

3. Willingness to cooperate The willingness of individuals to cooperate has been examined in such settings as a willingness to reduce automobile use in order to reduce pollution (Nordlund and Garvill, 2003), racial differences in WTC with police (Viki et al., 2006) and the willingness of private agricultural firms to participate in research studies (Al-Rimawi and Al-Karablieh, 2002). A common theme throughout all of these studies is that cooperation is the act of working with others to achieve a common purpose and that one’s ‘willingness to cooperate’ is an intention to exhibit cooperative behavior. Thus, we define WTC as the extent to which an individual farmer is agreeable to joining and participating in an agricultural cooperative. In their study of entrepreneurial behavior, Steensma et al. (2000) argued that an individual’s attitude toward cooperation reflected the extent to which he/she is receptive toward cooperative strategies. Further, this attitude was indicative of a belief that cooperation was necessary for success. Thus, a farmer’s attitude toward cooperation reflects the degree to which he/she believes that there are benefits from cooperation and therefore may be predisposed toward exhibiting cooperative behavior (Zhang et al., 2006). By their nature, individuals seek to fulfill their self-interests (Williamson, 1979). Indeed, classical economic theory is founded on the premise that we are collectively ‘better off’ if we individually pursue our self-interests (Smith, 1776). Just since the mid twentieth century have scholars begun to develop an understanding of how cooperation may be expected to occur in an economic exchange (Axelrod, 1984; Nash, 1950). Much of the theoretical literature on the formation of strategic alliances (e.g. cooperatives) is grounded in the strategic choice paradigm (Child, 1972), which argues that individual decisions are based on cognitive processes that result in rational choices, such as maximizing revenues and minimizing costs (Williamson, 1979). However, most of these economic views of cooperation ignore the value that one may gain from social interactions within cooperative organizations. Thus, prior to joining a cooperative prospective members are expected to make decisions about the likely consequences of their actions including an assessment of the possible social benefits that other members may be able to provide. The decision to join a cooperative will be made by calculating the expected risks and benefits in hopes of maximizing the potential gains and these expected gains from cooperative membership are likely to be both economic and social (Barraud-Didier et al., 2012). Put simply, a WTC will arise if cooperation enables an individual to accomplish goals that could not be International Food and Agribusiness Management Review

675

Morrow et al.

Volume 20, Issue 5, 2017

accomplished through competition, and while perhaps not immediately apparent, it is reasonable to suggest that in some cases, these desired goals may be both economic and social in nature.

4. Social capital in cooperative organizations Social capital refers to the idea that there is value in our social fabric. Like our economic capital enables us to buy resources, our social capital enables us to access resources, providing access to information through the members of our networks and facilitating the achievement of common goals (Seferiadis et al., 2015: 171). Social capital has been studied extensively and there is ample theoretical and empirical research that suggests social capital is a valuable socio-economic resource (Aguilar and Sen, 2009; Grootaert et al., 2004; Putnam, 1995, 2000; Woolcock and Narayan, 2000). Social capital has also been linked to cooperation and an increase in individual farm performance (Uphoff and Wijayaratna, 2000). While there are many different perspectives of what social capital actually is, we suggest that social capital consists of the value that is available to individuals and groups from multiple dimensions of their various social relationships (Adler and Kwon, 2002; Paldam, 2000; Sporleder and Wu, 2006). This may include social attributes such as counsel and advice, sympathy, trust, forgiveness and the like that are offered to individuals by friends and acquaintances (Adler, 2001; Dore, 1983). Moreover, social ties of one kind (e.g. friendship) may be used for different purposes, such as moral and material support or perhaps work and nonwork advice (Adler and Kwon, 2002). Finally the pursuit of oneâ&#x20AC;&#x2122;s social goals fosters the development of social capital, which may serve as motivation for individuals to seek membership and active participation in cooperative organizations (Collier, 2002). This suggests that for social capital to have an opportunity to develop and flourish, it must first have a place â&#x20AC;&#x201C; a place where individuals may come together in a social context, or what Cilliers and Wepener termed a structural opportunity to meet (Cilliers and Wepener, 2007; Seferiadis et al., 2015). Cooperatives have been referred to as people centered businesses that offer social benefits for members by providing opportunities for increased social interactions and by providing a structure to enhance economic activities (Kwapong and Hanisch, 2013; Majee and Hoyt, 2010). Thus, while an agricultural cooperative is typically thought of as an economic organization, it clearly may also serve as a mechanism for the development of social capital (Wilson, 2000). Elder and her colleagues (Elder et al., 2012) found that coffee farmers in Rwanda, who were members of a cooperative, developed social capital by interacting with their neighbors. Farmers in their study felt the cooperative provided the place where individual farmers could gain advantages by listening to the ideas of other farmers, although the authors attributed the development of social capital to these increased social interactions, and not to the cooperative itself. Despite this, it seems clear that these social interactions were facilitated by the group structure provided by the cooperative they had joined. Gomez-Limon et al. (2014) argued that social capital may lead to what they called bonding and bridging networks. Bonding networks are horizontal in nature, in that they occur among those who are on the same level. These are people we perceive to be similar to ourselves, such as our friends and family. Bridging networks on the other hand are vertical and serve to link us with those who we perceive to be different, such as those who have resources that we do not have. This suggests that from our social interactions, we identify other people who may be able to provide us with resources and support that we value but do not have (Elder et al., 2012; Gomez-Limon et al., 2014).

5. Conceptual framework 5.1 Bonding networks: social capital useful for getting by A bonding network is created from social capital that is useful for getting by, because it provides support for people in managing issues that arise during their day-to-day activities (de Souza Briggs 1997; GomezLimon et al., 2014). These benefits flow from people within a community who depend on one another in International Food and Agribusiness Management Review

676

Morrow et al.

Volume 20, Issue 5, 2017

order to cope with life’s challenges that may arise unexpectedly every day (Woolcock and Narayan, 2000). In western societies, these actions may include such things as offering a neighbor a ride to work, picking up a friend’s child from school, helping to care for an ailing parent or sharing a cup of coffee with a friend while discussing life’s problems or concerns. Further, the social capital developed from these getting by social activities is likely to come from people who are similar to us, such as our friends, family and neighbors (Lickerman, 2013). This type of social capital has been referred to as bonding networks, because it acts to bring people closer together within their communities in ways that may give them advantages over others who do not share these social benefits (de Souza Briggs, 1997; Gomez-Limon et al., 2014). We suggest social capital that is present in these bonding networks is likely to develop in two distinct ways. First, is what we term social support, which is the extent to which farmers know others with whom they can socialize during their leisure time. Farmers in the study by Elder et al. (2012) talked specifically about the opportunities involved in such simple social interactions as ‘meeting people to share a bottle of beer and ideas’ and that through discussions with other farmers ‘I don’t feel isolated’ (Elder et al., 2012: 2365). Similarly, Putnam (1995, 2000) notes that such simple social interactions as bowling in a league or having coffee with a friend leads to the creation of social capital. He goes further to suggest that social capital declines as individuals begin to engage in fewer social interactions. The second type of social capital that is present in bonding networks is emotional support, which we define as knowing others who can provide help and comfort in time of need. Life in much of Africa can be difficult, and many farmers likely live in rural areas, often with low levels of connectivity to local or wider networks. During difficult and challenging times, these people would naturally seek emotional support from a friend or someone they know (Lickerman, 2013). Emotional support has been conceptualized as expressions of care, concern, value, love and interest, especially during life’s difficult times (Burleson, 2003; Cutrona and Russell, 1990). Hypothesis 1a: bonding networks will consist of two dimensions: social support and emotional support. Hypothesis 1b: social support will have a positive effect on WTC. Hypothesis 1c: emotional support will have a positive effect on WTC. 5.2 Bridging networks: social capital useful for getting ahead In addition to getting by, social capital may also be used to help individuals get ahead (de Souza Briggs, 1997; Gomez-Limon et al., 2014). This means that individuals can use the capital developed from their social relationships in an effort to improve their life’s circumstances in hopes of becoming ‘better off.’ This type of social capital is likely to accrue from those within our social network who have both tangible (e.g. tools and equipment) and intangible (e.g. knowledge and skills) resources that we do not have (Galunic et al., 2012). In other words, this social capital is likely to come from people who are different from us (GomezLimon et al., 2014; Han et al., 2014), in the sense that these people possess resources that we need but do not have. In western societies, examples may include seeking legal advice from an acquaintance who is an attorney, medical advice from a casual friend who is a physician or borrowing a truck from a co-worker. Social capital that develops from this type of sharing has been termed ‘bridging networks’, because it serves to move people into a new, stronger position in life relative to those who do not share this social benefit (de Souza Briggs, 1997; Gomez-Limon et al., 2014). Majee and Hoyt (2010) found that frequent interaction among cooperative members enabled network building among members (horizontal) and between members and others with whom they came in contact (vertical). These vertical interactions were also found to provide cooperative members access to new resources. Aguilar and Sen (2009) noted that social capital may enable people to solve problems and get ahead in life. This means that a better understanding of social capital within cooperative organizations may lead to better strategies for International Food and Agribusiness Management Review

677

Morrow et al.

Volume 20, Issue 5, 2017

upward mobility (e.g. getting ahead). Accessing skills and know-how (e.g. intangible resources) from other farmers collectively led to the development of social capital among potato farmers in Uganda that allowed them to access new markets for their crops (Kaganzi et al., 2009). Hypothesis 2a: bridging networks will consist of two dimensions: tangible and intangible resource sharing. Hypothesis 2b: sharing tangible resources will have a positive effect on WTC. Hypothesis 2c: sharing intangible resources will have a positive effect on WTC. 5.3 Economic benefits of cooperation Goler von Ravensburg (2010) wrote extensively about the economic benefits associated with the cooperative form of organization. These benefits are essentially grounded in classical economic theory which seeks to identify the means of maximizing the benefits while minimizing the costs associated with utilizing an organization to achieve economic objectives. If the farmer owned cooperative is viewed as a business firm pursuing profit maximization, Barney (2011) succinctly notes that value creation within a firm is anything that enables a firm to increase revenues and decrease costs. Historically, farmers have formed supply coops (which provide inputs such as seed, fertilizers, herbicides, pesticides, tools and equipment) with the goal of minimizing costs while marketing co-ops, which assist the farmer in selling or marketing his/her output, may be used for both cost efficiencies and maximizing revenues. Consistent with these arguments (Goler von Ravensburg, 2010), we have conceptualized the economic benefits of cooperation as the extent to which cooperation will enable an individual farmer to decrease costs (e.g. efficiency actions) and increase revenues (e.g. effectiveness actions). Hypothesis 3a: the economic benefits of cooperation will consist of two dimensions: efficiency (cost reductions) and effectiveness (revenue increases). Hypothesis 3b: the economic benefits of cooperation, both efficiency and effectiveness, will have positive effects on WTC. A summary of our proposed model and its hypothesized effects on WTC are presented in Figure 1.

Social support Bonding networks

Economic benefits of cooperation

Bridging networks

H 1a

Emotional support

H 1b H 1c

Cost reductions H 3a

H 2a

Revenue increases Sharing tangible resources Sharing intangible resources

H 3b

Willingness to cooperate

H 2b H 2c

(WTC) = β0 + β1 (Social support) + β2 (Emotional support) + β3 (Cost reductions) + β4 (Revenue increases) + β5 (Sharing tangible resources) + β6 (Sharing intangible resources) + ε

Figure 1. The effects of economic and social benefits (bonding and bridging networks) on willingness to cooperate among potential members of a new farmer cooperative. International Food and Agribusiness Management Review

678

Morrow et al.

Volume 20, Issue 5, 2017

6. Methodology A survey methodology was used to collect data over a three day period from 183 farmers representing six different parishes within the Bushenyi District of western Uganda where there were no existing agricultural cooperatives at that time. We contacted the district commercial officer who works in cooperative development within the district and he identified a local mobilizer who arranged for a group of farmers to meet with us at various locations around the district (typically health centers, schools or churches). These farmers were informed that the purpose of our research was to gain an understanding of the factors that might facilitate cooperation within their community in hopes of developing better approaches to cooperative development. The participating farmers were not compensated except that we provided soft drinks and snacks while they waited to be interviewed. We surveyed approximately 60 farmers each day with these numbers divided about evenly between a morning and an afternoon session. Each session was held at a different parish within the Bushenyi District and we surveyed all farmers who were present at each session. Because many of these farmers were illiterate and/or had poor eyesight, the survey was administered orally by a group of five facilitators. Prior to data collection, each facilitator underwent a day long training exercise where we reviewed each question in the survey to make certain that everyone understood the questions and how they were to collect the responses. There were 74 men and 100 women who participated in the study (gender was missing from nine surveys). Among the survey participants, the average age was 46.76 years, the farmersâ&#x20AC;&#x2122; average household size was 6.94 people and the average farm size was 3.32 acres.

7. Measures 7.1 Control variables The dependent variable in our study was WTC in a proposed new agricultural cooperative and our explanatory variables were the anticipated economic and social benefits that might accrue to the farmers from their cooperation. Thus, we wanted to control for additional variables that might impact oneâ&#x20AC;&#x2122;s WTC. Farm size (measured in acres), age of the farmer (measured in years), household size (including the farmer) and gender were chosen as control variables. Gender was dummy coded 1=male and 0=female. 7.2 Development of measurement scales In developing our measurement scales, we began with our theoretical definition of each construct and then wrote a series of questions that appeared to correspond to this definition. When measuring latent variables in the social sciences, one should avoid single-item scales because their psychometric properties are likely to be poor or unknown (Furr, 2011). We set of minimum requirement of identifying at least a three-item scale and evaluated the psychometric properties of each scale using factor analysis to assess construct validity, and where appropriate, discriminant validity. Factor analysis is a data analytic technique used to determine which questions in a survey are associated with a single construct, as known as a factor (Bagozzi et al., 1991). For questionnaire data, factor analysis is used to determine the construct validity of the survey instrument with the resulting factors expected to correspond to the underlying constructs (Shmueli, 2010). In other words, if a group of questions that were written to measure a single construct all load together on a single factor, this provides evidence that we appear to be measuring what we expected to measure, which validates our measurement scales. Once a scale has been validated, the scores from each item in the scale are averaged to yield the measure for that construct. Discriminant validity is used to assess whether survey respondents appear to be differentiating between two distinct but similar constructs and provides evidence that we successfully identified valid and unique measures for each construct (Bagozzi et al., 1991). Finally, Cronbachâ&#x20AC;&#x2122;s alpha was used to determine the reliability of each scale by assessing the internal consistency or average correlation of the items within International Food and Agribusiness Management Review

679

Morrow et al.

Volume 20, Issue 5, 2017

each scale (Shrout and Fleiss, 1979). We describe each of our measurement scales below and the questions that were used in each scale are presented in Supplementary Methods S1. 7.3 Willingness to cooperate Earlier we defined WTC as the extent to which an individual farmer is agreeable to joining and participating in a proposed new agricultural cooperative. WTC, the dependent variable, was measured using a four-item scale that was developed based on this definition. 7.4 Economic benefits of cooperation We sought to develop survey questions that would measure two dimensions of the anticipated economic benefits of cooperation: cost minimization (efficiency) and revenue maximization (effectiveness). For example, for the efficiency dimension we asked such questions as ‘being a member of an agricultural oriented group might help me lower the cost of my supplies and/or reduce the time it takes to buy supplies’. Among the questions that we developed to measure the effectiveness dimension were ‘being a member of an agricultural oriented group might help me find new markets for my crops and/or find more people to trade with’. 7.5 Bonding networks We defined bonding networks as flowing from social interactions that are built around personal relationships and provide support for people to help them with the day-to-day issues of their lives. We further suggested that these networks are likely to have two distinct dimensions. Social support is the extent to which farmers hope to meet others with whom they may socialize during their leisure time and emotional support, which reflects whether individual farmers hope to meet others who may provide them with help and comfort in time of need. We developed a total of nine questions to capture these two dimensions of bonding networks. 7.6 Bridging networks Bridging networks are believed to be based on developing new social contacts who may be able to provide access to resources that a farmer may need but currently does not have. Access to these resources may allow the farmer to improve their personal or work lives. We have suggested that these bridging networks would consist of two dimensions that include the desire to meet others who have tangible resources (e.g. tools and equipment) and intangible resources (e.g. knowledge and skills) that one may be able to borrow or utilize. Six questions were developed to measure these two distinct dimensions of bridging networks. 7.7 Testing the hypotheses It is possible that endogeneity may exist if one of our independent variables is correlated with the error term. This may occur because of simultaneity or omitted variables. If ordinary least squares regression is used to estimate models where endogeneity is present, the effort will be inefficient and yield biased coefficients (Heckman, 1979). To remedy model misspecification due to unobserved factors, we employed the Twostage Least Squares (TSLS) procedure (Newey, 1987). After testing the hypotheses, we conducted a post hoc analysis using hierarchical regression in order to understand how much the social benefit variables contributed to the variation in WTC relative to the variance explained by the economic and control variables.

International Food and Agribusiness Management Review

680

Morrow et al.

Volume 20, Issue 5, 2017

8 Results 8.1 Tests for reliability and validity of the measurement scales The factor analyses and Cronbach’s alpha results are presented in Table 1. The four-item WTC scale was factor analyzed and each item loaded on a single factor with factor loading coefficients that ranged from 0.672-0.746. This provides evidence of construct validity using this scale to measure WTC. Cronbach’s alpha for this four-item scale was 0.68, which provides evidence of reliability. Also using factor analysis, we identified a four-item scale for emotional support and a five-item scale for social support. More importantly, we were able to demonstrate discriminant validity between these two similar constructs because they each loaded on a separate factor. This provides support for Hypothesis1a and suggests that we are measuring two different dimensions of social capital that exist within an individual’s Table 1. Results of factor analyses tests for construct and discriminant validity and Cronbach’s alpha test for reliability.1 Willingness to cooperate (WTC) α=0.68

Economic benefits of cooperation α=0.82

Question WTC 1 WTC 2 WTC 3 WTC 4

Question Economic benefits 1 Economic benefits 2 Economic benefits 3 Economic benefits 4 Economic benefits 5 Economic benefits 6 Economic benefits 7 Economic benefits 8

Factor loading 0.746 0.739 0.718 0.672

Factor loading 0.761 0.717 0.700 0.694 0.692 0.686 0.658 0.656

Bonding networks Emotional support α=0.73 and Social support α=0.67 Question Emotional 1 Emotional 2 Emotional 3 Emotional 4 Social 1 Social 2 Social 3 Social 4 Social 5

Factor loading 1 0.801 0.774 0.653 0.616 0.019 0.170 0.236 0.093 0.191

Factor loading 2 0.196 0.013 0.171 0.349 0.744 0.637 0.599 0.597 0.487

Bridging networks Sharing of tangible α=0.69 and intangible resources α=0.68 Question Tangible 1 Tangible 2 Tangible 3 Intangible 1 Intangible 2 Intangible 3 1

Factor loading 1 0.843 0.752 0.682 0.062 0.122 0.423

Factor loading 2 -0.028 0.215 0.246 0.864 0.801 0.591

α = Cronbach’s alpha International Food and Agribusiness Management Review

681

Morrow et al.

Volume 20, Issue 5, 2017

bonding network. Cronbach’s alpha for the emotional support and social support dimensions were 0.73 and 0.67 respectively while the nine items loaded on two distinct factors ranging from 0.616-0.801 for the emotional support scales and 0.487-0.744 for the social support scale. When the six items that were developed to measure bridging networks were factor analyzed, we identified a three-item scale for the sharing of tangible resources and another three-item scale for the sharing of intangible resources. As with the bonding network, these two distinct factor loadings demonstrate discriminant validity and supports Hypothesis 2a, which hypothesized two different dimensions of social capital that exists within an individual’s bridging network. The factor loading coefficients ranged from 0.682-0.843 for the tangible dimension and 0.591-0.864 for the intangible dimension. The tests for reliability using Cronbach’s alpha were 0.69 and 0.68, respectively. When the eight questions that were developed to measure two dimensions of the anticipated economic benefits of cooperation (efficiency and effectiveness) were factor analyzed, we were only able to identify a single dimension for this construct. In other words, we do not have empirical support for our conceptualization of a two dimensional construct and Hypothesis 3a is not supported. Instead, all of the scale items loaded on a single factor with the factor loading coefficients ranging from 0.656-0.761. This means that the farmers in our study do not appear to differentiate between the dual economic benefits of increasing revenues and decreasing costs. Thus, we were left with an eight-item scale that measures a single dimension of the economic benefits of cooperation that is a mix between efficiency and effectiveness benefits. Cronbach’s alpha was 0.82. All scaled measures were on a 7-point scale (1=strongly disagree; 7=strongly agree). The means, ranges, standard deviations and Pearson correlation coefficients for all of the variables used in this study and the results of the TSLS regression procedure used to test the hypotheses are presented in Tables 2 and 3. Variance inflation factors (VIF) were used to test the data for multicollinearity and the highest VIF coefficient was 1.758 indicating that multicollinearity was not a problem in the dataset. The overall TSLS model had an F value of 5.213, which was statistically significant (P<0.001) and the R2 was 0.23. None of the control variables was statistically significant. Joining a cooperative in hopes of meeting others who may provide social support had a statistically significant (P<0.10) and positive effect on WTC. This finding supports Hypothesis 1b. However, the desire to meet others who could provide emotional support did not have a statistically significant effect on WTC. Thus, Hypothesis 1c was not supported. Together, these findings provide partial support for the theoretical argument that Ugandan farmers are motivated to cooperate in hopes of meeting other farmers who may provide social and emotional support. The desire to join a cooperative in order to meet other farmers with whom one might share tangible resources had a statistically significant (P<0.01) and positive effect on WTC. This finding supports Hypothesis 2b. The desire to cooperate in hopes of meeting other farmers who are able to provide one with intangible resources also had a statistically significant (P<0.05) and positive effect on WTC, which supports Hypothesis 2c. Together, these findings support the proposition that Ugandan farmers are motivated to cooperate, at least in part, in hopes of meeting other farmers who may be able to help them get ahead by providing both tangible and intangible resources to enhance their farming operation. Finally, joining a cooperative in order to achieve economic benefits did not have a statistically significant on WTC, which fails to support Hypothesis 3b. The results of the post hoc analysis using hierarchical regression are presented in Table 4. In step one, we added the control variables. This model had an F value of 0.579, which was not statistically significant. The R2 was 0.02. In step two, we added the economic benefits variable which increased the R2 by 0.12, which was a statistically significant increase (P<0.001). In the third step all of the social benefits variables were added, which increased R2 by 0.09, which was a statistically significant increase (P<0.01). We elected to add the economic benefits variable in the second step, even though we knew it would be statistically insignificant in the final analysis in order to test its contribution to explaining WTC in the absence of the social benefits International Food and Agribusiness Management Review

682

Morrow et al.

Volume 20, Issue 5, 2017

Table 2a. Minimum, maximum, mean and standard deviation. Variable

Min

Max

Mean

s.d.

Farm size (measured in acres) Age of the farmer Household size Willingness to cooperate Economic benefits of cooperation Emotional support Social support Sharing of tangible resources Sharing of intangible resources

1 14 2 5.00 4.67 5.00 5.20 4.33 5.33

20 88 18 7.00 7.00 7.00 7.00 7.00 7.00

3.32 46.76 6.94 6.67 6.45 6.42 6.43 6.30 6.58

2.98 15.62 2.51 0.28 0.47 0.44 0.39 0.59 0.44

Table 2b. Pearson product moment correlation coefficients.1 1 2 3 4 5 6 7 8 9 10

Farm size Age of the farmer Gender Household size Economic benefits WTC2 Emotional support Intangible resources Tangible resources Social support

1.00 0.25*** 0.16* 0.02 -0.04 -0.01 0.03 0.02 0.03 0.03

0.28*** -0.06 -0.06 -0.11 0.01 -0.03 -0.08 -0.02

-0.06 0.01 0.03 0.01 0.12 0.07 0.09

0.13† 0.13† 0.31*** -0.15* 0.02 -0.16*

0.45*** 0.25*** 0.54*** 0.52*** 0.46***

0.25*** 0.38*** 0.35*** 0.46*** 0.34*** 0.39*** 0.39*** 0.43*** 0.54*** 0.56***

1† 2

= P<0.10; * = P<0.05; ** = P<0.01; *** = P<0.001. WTC = Willingness to cooperate.

Table 3. Results of two-stage least squares regression analysis predicting willingness to cooperate.1 Independent variables

β estimate

Constant Farm size Age of the farmer Gender Household size Economic benefits Emotional support Social Support Sharing tangible resources Sharing intangible resources F R2

3.598*** -0.001 -0.002 0.010 0.015 0.081 0.021 0.086† 0.152** 0.135* 5.213*** 0.23

1†

= P<0.10; * = P<0.05; ** = P<0.01; ***= P<0.001.

International Food and Agribusiness Management Review

683

Morrow et al.

Volume 20, Issue 5, 2017

Table 4. Results of hierarchical regression analyses testing the effects of social capital and economic benefits on willingness to cooperate.1 Step 1

Step 2

Step 3

1†

Independent variables

β estimate

Constant Farm size Age of the farmer Gender Household size F R2 Constant Economic benefits F R2 change Total R2 Constant Economic benefits Emotional support Social support Sharing tangible resources Sharing intangible resources F R2 change Total R2

6.685*** 0.000 -0.002 0.038 0.014 0.579 0.02 4.890*** 0.281*** 5.191*** 0.12*** 0.14 3.598*** 0.081 0.021 0.086† 0.152** 0.135* 5.213*** 0.09** 0.23

= P<0.10; * = P<0.05; ** = P<0.001; *** = P<0.001.

variables. If the social benefits variables had been added in the second step, they would have increased the R2 by 0.21 and the economic benefits (added in the third step) would have had no additional explanatory power. In other words, the anticipated economic benefits of cooperation alone explain about 12% of the variation in WTC but in the presence of the expected social benefits of cooperation the anticipated economic benefits explained zero. All of this suggests that the expected social benefits from participating in a proposed new agricultural cooperative account for almost all of the variation in Ugandan farmers’ WTC and in the presence of these social benefits the economic benefits are not statistically significant.

9. Conclusions and recommendations This research suggests that the expected social benefits associated with participating in a proposed new agricultural cooperative play a significant role in determining Ugandan farmers’ WTC. It appears farmers may be motivated to join an agricultural cooperative, as least in part, in an effort to build social capital that they currently do not have by interacting with other farmers. Further, among the farmers in our sample, the potential economic benefits of cooperation appear insignificant in the presence of the potential social benefits. From these social interactions, famers in our study hope to create bonding networks that have two dimensions: emotional support, which is the desire to meet others who can provide help and comfort in time of need and social support, which is the desire to meet others with whom one can socialize during leisure time. Even though the emotional support dimension did not have a statistically significant effect on WTC, finding support for the social support dimension and construct and discriminant validity for the two measurement scales are important new findings. Regarding the lack of the support for the emotional support dimension, one reasonable explanation is that the farmers in our study are meeting their emotional support needs in other group activities. International Food and Agribusiness Management Review

684

Morrow et al.

Volume 20, Issue 5, 2017

The social interactions that farmers in our study hoped to achieve by joining a proposed cooperative also led to the creation of bridging networks, which had two dimensions: the desire to meet other people who have needed tangible resources, such as tools and equipment, and intangible resources, such as knowledge and skills (Galunic et al., 2012). Again, we found evidence of construct and discriminant validity of the measurement scales used to capture these two dimensions of bridging networks. Our discovery of these four distinct dimensions of social capital has important managerial implications for the operation of cooperatives in Uganda. Emotional and social support will likely develop over time as cooperative members are placed in social settings with each other. While these social settings may include gatherings in which members simply relax, enjoy the company of each other and perhaps share a meal together, they may also include gatherings that provide formal seminars on topics of interest to members that are unrelated to farming activities (such as health and wellness seminars). Ugandan farmers also appear interested in meeting others with whom they can share both tangible and intangible farm-related resources. This suggests that cooperative managers should consider organizing formal seminars in which people who cooperative members perceive as experts offer advice, guidance and share know-how on a variety of farm-related topics. Managers should also consider developing a formal system that enables farmers to share tangible resources with each other. Perhaps a ‘resource bank’ of some sort could be established where tools, equipment and other resources could be ‘deposited’ and then ‘borrowed’ by cooperative members. A better understanding of Ugandan farmers’ views regarding the economic benefits that should accrue to them from cooperation presents an interesting topic for future research. Consistent with classical economic theory, we attempted to develop measures of two separate dimensions of the economic benefits of cooperation: lowering costs (efficiency) and maximizing revenues (effectiveness). However, results from our study suggest that farmers view the potential economic benefits of cooperation as only having one dimension, which appears to represent a mix of both efficiency and effectiveness goals. Perhaps these farmers do not view their farming activities as a typical business model in which managers attempt to maximize revenues and minimize costs. Despite their many years of farming experience, discussions suggest that farmers do not see farming as a business activity where the goal is higher income for the household. Discussions further suggest they may be engaged largely in subsistence farming because they are not employed elsewhere. There were also indications that any surplus was sometimes used to barter for other goods or services. Future research should continue to identify measures that will fully capture the theorized economic benefits of maximizing revenues (effectiveness) and minimizing costs (efficiency) to better understand how this may impact motivations to cooperate.

Acknowledgements This study was made possible by the generous support of the American people through the United States Agency for International Development (USAID). The contents are the responsibility of Global Communities and Dr. J.L. Morrow, Jr. and do not necessarily reflect the views of USAID or the United States Government. The Joseph S. Bruno Endowed Fund at Birmingham-Southern College also contributed funds for this research. The authors would like to thank Cara Bidwell, Chris Ibyisintabyo, Richard Mumuni and Annet Tumwesige for their assistance with this research study.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2014.0181. Methods S1. Measurements scales.

International Food and Agribusiness Management Review

685

Morrow et al.

Volume 20, Issue 5, 2017

References Adler, P.S. 2001. Market, hierarchy, and trust: the knowledge economy and the future of capitalism. Organization Science 12: 215-234. Adler, P.S. and S. Kwon. 2002. Social capital: prospects for a new concept. Academy of Management Review 27: 17-40. Aguilar, J.P. and S. Sen. 2009. Comparing conceptualizations of social capital. Journal of Community Practice 17: 424-443. Al-Rimawi, A.S. and E.K. Al-Karablieh. 2002. Agricultural private firms’ willingness to cooperate with public research and extension in Jordan. Journal of International Agricultural and Extension Education 9(3): 5-13. Axelrod, R. 1984. The evolution of cooperation. Basic Books, New York, NY, USA. Bagozzi, R.P., Y. Yi and L.W. Phillip. 1991. Assessing construct validity in organizational research. Administrative Science Quarterly 36: 421-458 Barney, J. 2011. Gaining and Sustaining Competitive Advantage, 4th ed. Prentice Hall, Saddle River, NJ, USA. Barraud-Didier, V., M. Henninger and A. Akremi. 2012. The relationship between members’ trust and participation in the governance of cooperatives: the role of organizational commitment. International Food and Agribusiness Management Review 15(1): 1-24. Burleson, B. 2003. The experience and effects of emotional support: what the study of cultural and gender differences can tell us about close relationships, emotion, and interpersonal communication. Personal Relationships 10: 1-23. Child, J. 1972. Organization structure, environment and performance: the role of strategic choice. Sociology 6: 2-21. Cilliers, J. and C. Wepener. 2007. Rituals and the generation of social capital in contexts of poverty: a South African exploration. International Journal of Practical Theology 11: 39-55. Collier, P. 2002. Social capital and poverty: a microeconomic perspective. In: The role of social capital in development: an empirical assessment, edited by C. Grootaert and T. van Bastelaer. Cambridge, New York, NY, USA, pp. 19-41. Connolly, A.J. 2014. A glimpse into the future: a lens through which to consider ‘Africa’s rising’. International Food and Agribusiness Management Review 17(B): 9-18. Cutrona, C. and D. Russell. 1990. Type of social support and specific stress: toward a theory of optimal matching. In: Social support: an interactional view, Wiley series on personality processes. edited by B. Sarason, I. Sarason and G. Pierce. Wiley and Sons, New York, NY, USA, pp. 319-366. de Souza Briggs, X. 1997. Social capital and the cities: advice to change agents. National Civic Review 86: 111-117. Dore, R. 1983. Goodwill and the spirit of market capitalism. British Journal of Sociology 34: 459-482. Elder, S., H. Zeffiffi and F. Le Billon. 2012. Effects of fair trade certification on social capital: the case of Rwandan coffee producers. World Development 40: 2355-2367. Furr, M. 2011. Scale construction and psychometrics for social and personality psychology. SAGE Publications Ltd., Thousand Oaks, CA, USA. Galunic, C., G. Ertug and M. Gargiulo. 2012. The positive externalities of social capital: benefiting from senior brokers. Academy of Management Journal 55: 1213-1231. Goler von Ravensburg, N. 2010. Economic and other benefits of the Entrepreneurs’ Cooperative as a specific form of Enterprise Cluster. International Labour Organization, Geneva, Switzerland. Gomez-Limon, J., E. Vera-Toscano and F. Garrido-Fernández. 2014. Farmers’ contribution to agricultural social capital: evidence from southern Spain. Rural Sociology 79(3): 380-410. Grootaert, C., D. Narayan, V.N. Jones and M. Woolcock. 2004. Measuring social capital: an integrated questionnaire. World Bank, Washington, WA, USA. Han, J., J. Han and D. Brass. 2014. Human capital diversity in the creation of social capital for team creativity. Journal of Organizational Behavior 35: 54-71. Heckman, J. 1979. Sample selection bias as a specification error. Econometrica 47(1): 153-161.

International Food and Agribusiness Management Review

686

Morrow et al.

Volume 20, Issue 5, 2017

Kaganzi, E., S. Ferris, J. Barnham, A. Abenakyo, P. Sanginga and J. Njuki. 2009. Sustaining linkages to high value markets through collective action in Uganda. Food Policy 34: 23-30. Kwapong, N.A. and M. Hanisch. 2013. Cooperatives and poverty reduction: a literature review. Journal of Rural Cooperation 41(2): 114-146. Kyazze, L.M. 2010. Cooperatives: the sleeping economic and social giants in Uganda. Coop Africa Working Paper No. 15. International Labour Organization, Geneva, Switzerland. Liang, Q., Z. Huang, H. Luc and X. Wang. 2015. Social capital, member participation, and cooperative performance: evidence from Chinaâ&#x20AC;&#x2122;s Zhejiang. International Food and Agribusiness Management Review 18(1): 49-78. Lickerman, A. 2013. The true meaning of friendship: what is it that makes a true friend? Psychology Today. Available at: http://tinyurl.com/ppeq2ke. Lie, H., K. Rich, L. Kurwijilac and A. Jervell. 2012. Improving smallholder livelihoods through local value chain development: a case study of goat milk yogurt in Tanzania. International Food and Agribusiness Management Review 15(3): 55-85. Liu, Y.X. and J. Sumelius. 2010. Analysis of the factors of the participation of farmers in the management of cooperatives in Finland. Journal of Rural Cooperation 38(2): 134-155. Majee, W. and A. Hoyt. 2010. Are worker-owned cooperatives the brewing pots for social capital? Community Development 41: 417-430. Meskela, T. and Y. Teshome. 2014. From economic vulnerability to sustainable livelihoods: the case of the Oromia Coffee Farmers Cooperatives Union (OCFCU) in Ethiopia. International Food and Agribusiness Management Review 17(B): 103-108. Mungandi, S., D. Conforte and N.M Shadbolt. 2012. Integration of smallholders in modern agri-food chains: lessons from the KASCOL model in Zambia. International Food and Agribusiness Management Review, 15(3): 155-176. Nash, J.F. 1950. The bargaining problem. Econometrica 18: 155-162. Newey, W.K. 1987. Efficient estimation of limited dependent variable models with endogenous explanatory variables. Journal of Econometrics 36: 231-250. Nilsson, J., G.L.H. Svendsen and G.T. Svendsen. 2012. Are large and complex agricultural cooperatives losing their social capital? Agribusiness 28(2): 187-204. Nordlund, A.M. and J. Garvill. 2003. Effects of values, problem awareness, and personal norm on willingness to reduce personal car use. Journal of Environmental Psychology 23: 339-347. Oregon State University. 2004. A brief introduction to agricultural cooperatives. Oregon State University, Corvallis, OR, USA. Paldam, M. 2000. Social capital: one or many? Definition and measurement. Journal of Economic Surveys 14(5): 629-653. Parkhe, A. 1993. Strategic co-op structuring: a game theoretic and transaction cost examination of interfirm co-operation. Academy of Management Journal 36: 794-829. Putnam, R.D. 1995. Tuning in, tuning out: the strange disappearance of social capital in America. Political Science and Politics 28: 664-683. Putnam, R.D. 2000. Bowling alone: the collapse and revival of American community. Simon and Schuster, New York, NY, USA. Shmueli, G. 2010. To explain or to predict? Statistical Science 25(3): 289-310. Seferiadis, A., S. Cummings, M. Zweekhorst and J. Bunder. 2015. Producing social capital as a development strategy: implications at the micro-level. Progress in Development Studies 15: 170-185. Shrout, P.E. and J.L. Fleiss. 1979. Intraclass correlations: uses in assessing rater reliability. Psychological Bulletin 86: 420-428. Smith, A. 1776. The wealth of nations. W. Strahan and T. Cadell, London, UK. Sporleder, T., and S. Wu. 2000. Social capital and vertical ties in agrifood supply chains. Journal on Chain and Network Science 6(1): 1-7. Steensma, H.K., L. Marino and K.M. Weaver. 2000. Cooperative strategies: a cross-cultural analysis of entrepreneurs. Journal of International Business Studies 31: 591-609.

International Food and Agribusiness Management Review

687

Morrow et al.

Volume 20, Issue 5, 2017

Uphoff, N. and C.M. Wijayaratna. 2000. Demonstrated benefits from social capital: the productivity of farmer organizations in Gal Oya, Sri Lanka. World Development 28: 1875-1890. Viki, G.T., M.J. Clumer, A. Eller and D. Abrams. 2006. Race and willingness to cooperate with the police: the roles of quality of contact, attitudes towards the behavior and subjective norms. British Journal of Social Psychology 45: 285-302. Williamson, O.E. 1979. Transaction cost economics: the governance of contractual relations. Journal of Law and Economics 22: 233-261. Wilson, P.N. 2000. Social capital, trust, and the agribusiness of economics. Journal of Agricultural and Resource Economics 25: 1-13. Wilson, P.N. and A.M. Kennedy. 1999. Trustworthiness as an economic asset. The International Food and Agribusiness Management Review 2(2): 179-193. Woolcock, M. and D. Narayan. 2000. Social capital: implications for development theory, research, and policy. World Bank Research Observer 15: 225-249. Zhang, L., H. Wang, L. Wang and W. Hsiao. 2006. Social capital and farmerâ&#x20AC;&#x2122;s willingness-to-join a newly established community-based health insurance in rural China. Health Policy 76(2): 233-242.

International Food and Agribusiness Management Review

688

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2017.0021 Received: 28 March 2017 / Accepted: 7 August 2017

The mediated partnership model for sustainable coffee production: experiences from Indonesia RESEARCH ARTICLE Atika Wijaya

a,b,

Pieter Glasbergenc, and Surip Mawardid

aPhD

Researcher and cProfessor, International Centre for Integrated assessment and Sustainable development (ICIS), Maastricht University, P.O. Box 616, 6200 MD Maastricht, the Netherlands bLecturer,

Department of Sociology and Anthropology, Faculty of Social Sciences, Semarang State University, Kampus Unnes Semarang 50229, Indonesia dMember

of Scientific Board, Indonesian Coffee and Cocoa Research Institute (ICCRI), Jember 68118, Indonesia

Abstract This study demonstrates how a bottom-up agricultural development project, rooted in the practices of the smallholders and their (local) networks, might relate to global sustainability standards and certification schemes. Such an initiative starts with the economic interests of the farmers and may result in sustainability certification afterwards. An approach was investigated to implement more sustainable coffee production at the local level in Indonesia â&#x20AC;&#x201C; the Mediated Partnership Model (Motramed). We conceptualize this model, initiated by an Indonesian research institute, as a form of collaborative governance to create a new, more sustainable management practice. We particularly focus on four critical mechanisms in the partnering process: defining a common problem, building trust, exploring collaborative advantages, and establishing a leadership role. Empirical research was conducted in three regions: Bali, Flores, and Java. In conclusion, we define several critical factors that need to be addressed to further develop similar bottom-up partnerships for more varied agricultural commodities. Keywords: mediated partnership, sustainable coffee, collaborative governance, production network, Indonesia JEL code: O130, Q110, Q130, Q180 Corresponding author: atika.wijaya@maastrichtuniversity.nl

ÂŠ 2017 Wijaya et al.

689

Wijaya et al.

Volume 20, Issue 5, 2017

1. Introduction A more sustainable production of agricultural commodities is generally seen as being in the interest of the livelihood of farmers in the South (Bitzer et al., 2008; Giovannucci and Koekoek, 2003). This interest is particularly reflected in the steady growth of sustainability standards and certification schemes initiated by Northern-based business and Non-Governmental Organizations (NGOs) or partnerships between them (Bartley, 2010; Bitzer and Glasbergen; 2015; Fuchs et al., 2011). As external initiatives from the top of the value chain, these schemes are channeled downwards to the bottom of the chain as new conditions for production (Giovannucci and Ponte; 2005). At the bottom of the chain, the producers need to make sense of the standards, decide on how to respond, and handle them given local socioeconomic circumstances and preferences (see also Bitzer et al., 2013; Hatanaka, 2010; Marin-Burgos et al., 2015; Schouten et al., 2016; Vellema et al., 2013). Participation in these market-based initiatives is voluntary, but it has gradually become a prerequisite for access to parts of the global market. However, it is still doubtful whether Northernbased certification is the most effective way to improve farmersâ&#x20AC;&#x2122; livelihood (Abdulsamad et al., 2015; Van Rijsbergen et al., 2016). This study shows how bottom-up agricultural development projects, rooted in the practices of the smallholders and their (local) networks, might relate to these external certification inputs from the North. In this study, we analyze such a project: Motramed (in English: mediated partnership model), which is different from the former initiatives on coffee production, as the certification standards are not a starting point for a change in the practices of coffee producers, but they may result from practices that are changed with the intention to improve the living conditions of the farmers first. Motramed was initiated in 2001 by Indonesian Coffee and Cocoa Research Institute (ICCRI). ICCRI is a research center that specializes in the cultivation and processing techniques of coffee and cocoa1. Motramed is based on the following considerations (Hartatri and De Rosari, 2011; Mawardi et al., 2006; Soemarno et al., 2009; Virgiano, 2012). The first consideration is the idea that the farmersâ&#x20AC;&#x2122; income should be increased by improving coffee quality and shortening the marketing channel. The problems of low quality and a long supply chain are seen as interconnected, and therefore, improving both is regarded as an inseparable formula for improving the livelihood of farmers. Second, coffee quality should be improved through technology transfer. ICCRI has the responsibility of disseminating technology to the farmers. In Motramed, ICCRI introduces the procedure of wet processing (WP) of coffee cherries, which meets international standards. By following WP, farmers will gain added value from processed coffee that they do not gain by directly selling coffee without advanced processing (Mawardi et al., 2006). Third, the marketing channel should be simplified by building partnerships between farmers and exporters. The long supply chain between farmers and exporters has resulted in an inefficient marketing channel. Because of this long chain, there is only small value for each player, of which farmers get the least value due to their weak bargaining position (Astuti et al., 2015; Hartatri, 2016; Neilson, 2008). Therefore, in Motramed, directly connecting farmers and exporters results in a shorter coffee supply chain. Thus, farmers will receive a higher price than they would when they sell to local traders or middlemen. Although these considerations focus on economic improvement, Motramed also addresses social and environmental aspects in order to comply with sustainability principles. Motramed encourages farmers to work in collectivity by establishing farmer organizations. Through these organizations, it is easier to coordinate the activities of coffee processing, and collective marketing will create a stronger bargaining 1

ICCRI was established in 1911 and is located in Jember, East Java. Structurally, ICCRI is still part of the Indonesian government, under coordination by the Ministry of Agriculture, the Ministry of Research, Technology and Higher Education, and the Ministry of State-Owned Enterprises. However, in terms of its operational budget, ICCRI is a self-funding institution. ICCRI has a wide network, which was established from numerous collaborations and research projects with stakeholders, ranging from international to local organizations, research centers, companies and governments.

International Food and Agribusiness Management Review

690

Wijaya et al.

Volume 20, Issue 5, 2017

position for farmers (Mawardi et al., 2006). Motramed promotes shade tree applications and applies organic farming (Mawardi, 2009a). However, in the partnering process, economic aspects receive priority, as they are considered an incentive to encourage farmers and other actors to participate. This study aims to contribute to our knowledge of the intricacies of implementing such a bottom-up partnership model to induce more sustainable production at the local level. We are particularly interested in the factors that are conducive and limit change. From our analysis, we aim to draw on lessons from similar initiatives to improve the economic position of farmers in the sustainable production of agricultural commodities at the local level. We first introduce an analytical framework. Next, we explain our research methods. In the subsequent section, we analyze a pilot project in Bali that developed into a model of change, followed by an analysis of two other Motramed cases. In the conclusion, we reflect on these findings.

2. Analytical framework Herein, we conceptualize Motramed as a form of collaborative governance to create a new, more sustainable management practice. The term ‘collaborative governance’ refers to the interactive process through which the changes in the production network will be realized. In the Motramed case, the actors involved in coffee production and trade need to create a new mutually beneficial relationship. In this process, they need to develop a common definition of the problems, a shared vision of the future, and commitments to activities that need to be employed. Van Tulder and Pfisterer (2013) refer to this kind of process as the creation of a ‘partnering space.’ A partnering space is a framework within which the positions and roles of actors from different societal sectors are considered in a collaborative arrangement. Through their interactions, the actors are assumed to be able to identify opportunities for aligned actions and build a synergistic relationship. In such a relationship, the actors realize more through cooperation than they are able to realize on their own. The concept of partnering space resonates with the view of Austin and Seitanidi (2012), who declared that creating value is the central justification for multi-actor partnering. They define collaborative value as transitory, and the enduring benefits are relative to the costs that are generated due to the interactions of the collaborators and those accrued by organizations, individuals, and society. Based on these studies, we can analytically clarify the process and mechanisms that are instrumental and critical to Motramed as a partnering process. Most generally, we assume that this process includes generating the ability, desirability, and willingness to develop a new production network. First, the actors involved in the production network must recognize a common problem. As Schruijer (1999: 2) observed, collaboration can only be created when a shared problem definition is developed. Regarding Motramed, there must be recognition of inefficient coffee production that is caused by existing production methods and connections to the market. A critical mechanism to create an accepted problem definition would be the transfer of knowledge and expertise regarding more promising production networks. According to Loconto (2016), mobilizing knowledge is fundamental to better linking small farmers with markets in developing countries and creating transitions to a more sustainable production network. Such knowledge transfer includes information on what to do, how to do it, when to do it, and why. Second, the actors need to recognize their interdependent relations and the need for collaboration. Regarding Motramed, a critical mechanism is fostering trust among actors and commitment to cooperate in the change process. Trust particularly encapsulates an emotional argument, which is the reduction of feelings of risk and vulnerability in the partnering process. Trust is regarded as a significant component of collaboration in networks (Glasbergen, 2011). Trust diminishes the risk of opportunistic behavior in a relationship, increases the probability that actors will invest their resources and is supposed to stimulate learning and the exchange of information and knowledge (Klijn et al., 2010). In a trader-supplier relationship, trust includes recognition of the competence of the supplier to deliver the required quality and quantity of the product (Saad et al., 1999: 43). Everlasting trust for all is not instantaneously created. Building trust is a social process that needs

International Food and Agribusiness Management Review

691

Wijaya et al.

Volume 20, Issue 5, 2017

to be managed, maintained and supported by positive experiences (Barroso-Méndez et al., 2016; Brewer and Haylarr, 2005; Glasbergen, 2011). Third, the actors in the partnership need to recognize the advantages of cooperation, and these advantages should be regarded as long-lasting. The central mechanism should be the creation of collaborative advantage. Regarding Motramed, potential actors need to find advantages that they cannot achieve without the partnership (Brinkerhoff, 2002; Emerson et al., 2011). According to Glasbergen (2011), collaborative advantages must relate to resources, skills, and a balance between benefits and costs among partners. In this study, collaborative advantages will be analyzed by looking at the actors’ interests in coffee production and how they distribute responsibilities among themselves in accordance to their capacity and resources. Fourth, the functioning of an intermediary entity that fulfills a leadership role in the network is important. The leadership role denotes the capacity to bring the different actors together around the issue and to facilitate communication to take the partnership forward. Leadership is not something that comes naturally. As Huxham and Vangen (2000) observed, leadership is often thwarted by dilemmas and difficulties. Therefore, it requires abundant energy, skill, and personal attention to chaperone the cause and drive a partnering process forward. Leadership intervention tasks include appreciation or envisioning, convening, problem structuring, designing the process, reflective intervention, conflict handling, brokering and institutional entrepreneurship (Gray, 2007). Because participation in Motramed is voluntary, the leadership role occurs in a non-hierarchical context, which implies that it is mainly based on specific competences that are valued by the other actors. The process of employing these mechanisms should be seen as an iterative and cyclical process rather than as a linear process. It is an empirical question as to how these mechanisms unfold in practice. Here, we take them as a ‘lens’ to study Motramed and the lessons that can be learned.

3. Research methods The current study applies a qualitative research approach to understand the evolvement of Motramed as a local partnership to induce more sustainable production of Arabica coffee at the producer level. We employed three main research methods. First, semi-structured interviews were used to gain a comprehensive view of the actors in the process of collaboration to develop a new management practice through Motramed. The informants were chosen through purposive sampling. We first interviewed the initiator of Motramed, i.e. ICCRI’s researchers. They led us to the key informants in each region, who were interviewed by snowball sampling in accordance with the need for data in the field (Table 1). The semi-structured interviews were helpful to detect the informants’ framing of coffee production problems. We also crosschecked the information that we had previously obtained from the informants. We sometimes Table 1. Composition of informants from the three locations. No.

Institutions

Number of informants

1. 2. 3. 4. 5. 6. 7.

Indonesian Coffee and Cocoa Research Institute (ICCRI) Farmer groups and cooperatives The Estate Crops Agency (Dinas Perkebunan) Exporter Bank NGO and community organizations related to coffee Perhutani (State-owned enterprise on state forestry) Total

2 12 5 3 2 4 1 29

International Food and Agribusiness Management Review

692

Wijaya et al.

Volume 20, Issue 5, 2017

conducted interviews more than once so that the informants could talk freely without interruption from other informants. Second, we attended three meetings related to Motramed: two annual member meetings of cooperatives in Bajawa and Bondowoso and a coffee stakeholders meeting in Bondowoso. We collected information on the interaction patterns among the actors in the partnering process, particularly focusing on trust and commitment building and the role of leadership. Third, we analyzed documents related to Motramed both from governmental bodies and ICCRI, such as the Memorandum of Understanding (MoU) in some regions, published reports of the implementation of Motramed in some regions, and the requirements books of Geographical Indications. We also analyzed statistical data for coffee production, productivity, land area and coffee prices to gather data regarding the changes that resulted from Motramed. From 7 to 10 November 2015, we went to the Bajawa and Manggarai regions in Flores to conduct preliminary research with the aim of gaining a clearer picture of Motramed before finalizing the research framework. Then, from the end of December 2015 to March 2016, we completed our research in three different regions in Indonesia. We selected these regions based on specific features and the progress of Motramed. The first region is Kintamani, Bali, which was chosen as a pilot project that began in 2001. The second region is Bajawa, Flores, which applied Motramed in 2004. The third region is Bondowoso, East Java, which applied Motramed in 2011 with more stakeholders involved since the initiation of Motramed.

4. Case 1: the pilot project of the partnering process in Kintamani, Bali In this section, we will analyze the partnering process based on the four mechanisms introduced in the previous section. The period of analysis covers the beginning of the partnership in 2001 until 2015. Originally, Motramed was not a model with a fixed design, but it developed over time. The origin was closely related to the application of a Geographical Indication (GI) protection system2. A pilot project was initiated by ICCRI in cooperation with CIRAD (a French agricultural research organization) and was supported by the Provincial Estate Crops Agency of Bali (Disbun Bali). Kintamani, Bali, was chosen as the pilot project location because of its popularity as an Arabica coffee producer. 4.1 Common problem definition Before the pilot project started, ICCRI had conducted a baseline study regarding the potential of Arabica coffee in Kintamani. This study showed that the price of Arabica coffee in Kintamani was lower than that of Robusta coffee, i.e. 5,200 IDR/kg for Arabica and 5,400 IDR/kg (USD 0.48) for Robusta (interview with the senior researcher of ICCRI), although Arabica coffee is generally regarded as being of higher quality. Consequently, farmers had lost their motivation to maintain their coffee plantations and preferred to change to tangerine trees, which offered more profit. This massive conversion became an environmental threat for Bali because Kintamani is a highland area and coffee plants have a function in preserving groundwater. The study concluded that the low prices were caused by the low quality and long supply chain at the production level. These basic problems were not only problems for farmers but also a concern of the local government. Therefore, it was necessary to attract more farmers and improve their coffee plantations through the creation of a price incentive. ICCRI concluded that this goal could be realized by building a partnership among the actors involved in coffee production. ICCRI played a role as a provider of knowledge regarding the improvement of coffee quality and disseminated this knowledge to the staff of Disbun Bali and farmers.

This system is part of Law No. 15, year 2001, regarding trademarks. GI is a sign that indicates the origin of products. GI is important because geographical factors, including human and natural factors, result in specific characteristics and qualities. GI was created as a response against the misuse of geographical names by Japanese and Dutch companies (Mawardi, 2009a).

International Food and Agribusiness Management Review

693

Wijaya et al.

Volume 20, Issue 5, 2017

4.2 Building trust It was not easy to convince farmers to participate in this partnership. At first, ICCRI and the staff of Disbun Bali purposively selected four potential Subak Abian (SA)3 – Kertawaringin, Triguna Karya, Bhaktiyasa and Ulian Murn – that showed a strong willingness to try the new technique of WP introduced by ICCRI. The staff of Disbun Bali explained how they tried to convince farmers to participate in the partnership: Together with ICCRI, we worked on improving the understanding of farmers to change their habits from strip picking, which dominates green and yellow cherries, to fully red picking. We presented economic calculations to them showing that processing coffee will result in higher prices than directly selling coffee to local traders. ICCRI also showed farmers how to better manage their coffee plantations. Farmers were not only persuaded by the economic advantages; the health aspects of the production process, e.g. a more hygienic process, also played a role. With more advanced knowledge, the farmers valued coffee more highly than they did before, and they even started to consume this high quality coffee themselves. Meanwhile, the government also supported farmers by establishing a processing unit for each SA and providing machines to these units. During the harvest period, ICCRI researchers and the staff of Disbun Bali intensively assisted farmers to ensure that they would comply with the standard operating procedures (SOP). The next step was connecting farmers to exporters. Previously, farmers and exporters did not interact because exporters preferred to buy coffee beans from their suppliers. Many exporters were reluctant to directly interact with farmers. In their view, it was risky for their business because of the low quality of coffee and because farmers were reportedly unable to keep their commitment to an agreement (according to interviews with the exporter). Farmers were also anxious about trading with exporters because exporters might be deceitful in their pricing. Thus, their relationship was undermined by doubt and misunderstanding. However, the presence of ICCRI and the staff of Disbun Bali sufficiently assured both farmers and exporters of the value of initiating interaction between farmers and exporters. ICCRI succeeded in inviting an exporter from France in 2002. However, this exporter did not buy coffee from the farmers because the quality did not yet meet the standards, as the farmers were still in the early process of learning WP. Later, in 2003, with CIRAD’s network, PT Indo CafCo under the ECOM Group was willing to participate. Despite concerns about dealing with farmers, the Director of PT Indo CafCo was very interested in the Kintamani Coffee brand. ICCRI’s intensive monitoring of farmers convinced him to take part in this partnership. Thus, PT Indo CafCo’s trust was strongly influenced by the credibility of ICCRI as the mediator and the active role of Disbun Bali in supporting the farmers. 4.3 Exploring collaborative advantage In partnering processes, all actors need to find long-lasting advantages that they cannot achieve without the partnership. These collaborative advantages can be analyzed by examining their interests and the distribution of responsibilities. ICCRI, as the mediator in this partnership, had an interest in disseminating its research results to the community as part of their obligations as a research center. Their research results consisted of not only knowledge and skills regarding coffee but also agricultural products, ranging from seedlings to machineries. ICCRI observed an opportunity to expand its network in the coffee sector.

Subak Abian (SA) is a traditional farmer organization that adheres to the Hindu philosophy of ‘Tri Hita Karana’, which translates as three causes of happiness that can be obtained when one has a good relationship with God, other people and the environment. SA is involved in not only agricultural activities but also religion and social activities.

International Food and Agribusiness Management Review

694

Wijaya et al.

Volume 20, Issue 5, 2017

The local government’s interest was to improve the regional economy through coffee farming, which was regarded as a potential area that could be developed. With this interest, the local government also expected that the increase in farmers’ welfare would lead to the growth of other sectors, such as education and health. However, the agency did not have the ability to achieve this alone. By cooperating with ICCRI, the local government could achieve its goals. The local government also helped ICCRI disseminate its knowledge and agricultural products. The support of the government consisted of providing machines (some of which were bought from ICCRI), tools, technical assistance, and loans for farmer groups. Farmers had an interest in earning a higher income. They were highly motivated to participate in the partnership if there was a price incentive for their work. Previously, farmers did not have much price-setting power because of the low quality of their coffee. After applying the WP technique, farmers could produce green bean coffee grade 1 with a good taste profile, as tested in the CIRAD and ICCRI laboratories4. Better quality led to a better price at the farmer level (Figure 1). Figure 1 shows that in 2001, before the quality improvement program of Motramed had begun, the coffee price in Bali at the farmer gate was three times lower than the world coffee price. However, in the next few years, the Bali coffee price gradually increased and had a small margin compared to the world coffee price because of its improved quality. In 2006, the Bali coffee price was higher than the world coffee price by approximately 7%, and farmers received a price premium. Although the discounted Bali coffee price fluctuated between -0.8% at the lowest and -30% at the highest from the world coffee price in 2002-20155, it still showed a large improvement since the initiation of Motramed. The price fluctuation was strongly influenced by the world coffee price and the currency exchange rate. During the partnership with PT Indo CafCo as an exporter, farmers received almost the same price as the international price6. The increase in coffee price attracted more SA to participate in the program. They believed that by complying with the SOP of WP, farmers could earn a higher income. In the first two years, ICCRI and Disbun Bali only focused on four SA until farmers could produce better quality coffee. Afterwards, four SA were added each year for training and support in coffee quality improvement (CGIP, 2007). Although many more SA showed interest in the partnership, they could not all receive this aid due to the limited budget of the government. 4

Additional results from the laboratory test of Kintamani Arabica coffee can be found in Mawardi (2009b). A more elaborate explanation of this chart is presented in Supplementary Table S1. 6 The world coffee price in the chart uses the grade 1 price of the ‘Other Milds’ coffee category from International Coffee Organization (in which Indonesian Arabica coffee is part of Other Milds category), whereas the Bali coffee price is for unsorted beans at the farmer gate. If the farmer gate price is converted to grade 1, 10% needs to be added. Thus, the farmers were receiving a good price from PT Indo CafCo because they produced grade 1 Arabica coffee. 5

World coffee price

Bali coffee price

60,000

Price in IDR

50,000

38,719 38,854

39,219

40,000

30,066

32,840

20,000 14,095 12,561 10,000 5,200

2001

15,941 12,131

12,000

2002

12,000

2003

24,500 24,806

28,600 23,671

27,400

53,833

56,833 47,452

32,492 37,871

36,375

24,690 23,055 24,960

30,000

52,503

52,429

28,083

14,800

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

Year

Figure 1. The price of Bali coffee compared with the world coffee price (adapted from Mawardi et al., 2006; The Estate Crops Agency of Bali Province, 2015). International Food and Agribusiness Management Review

695

Wijaya et al.

Volume 20, Issue 5, 2017

The exporterâ&#x20AC;&#x2122;s interest was to obtain a sustainable coffee supply with good quality and quantity. With this partnership, PT Indo CafCo obtained an agreed quantity that met its quality standards and the expectations from their overseas buyers. During the partnership, PT Indo CafCo contributed by allocating an assistance fee to ICCRI, offering their staff to monitor the processing in the field, and providing processing machines. In addition, PT Indo CafCo made an advanced payment if farmers had difficulties obtaining capital to start coffee processing. These contributions showed that the exporter understood the farmersâ&#x20AC;&#x2122; interest in obtaining a price incentive for their work. Without easy access to capital and machines, it would be difficult for farmers to improve their coffee quality. In 2008, Kintamani Bali Arabica Coffee was awarded a Geographical Indication certificate, which was the first GI certificate in Indonesia, based on the Community of Geographical Indication Protection (CGIP)7. Although the GI certificate is not a commercial certificate for trading, it ensured that Kintamani Arabica coffee is a regional asset protected by law. The GI certificate regulates the procedure of coffee production, processing, quality control, and marketing. This certificate added value to the Bali Kintamani Arabica coffee brand, as it became better known on national and international markets for its quality and unique taste. This certificate is an additional advantage of the partnership for all involved actors. 4.4 On the leadership role Although there was no assigned leadership in this partnership, this role was taken by ICCRI. With its renowned reputation as an expert in coffee production and its understanding of the social, political and economic context within which production occurs, ICCRI could link the farmers, government and exporters and could facilitate communication among them. Leadership thus implied facilitating a partnering process based on the creation of consensus on a preferable future for all stakeholders in the coffee production process. During the partnering process in Kintamani, several obstacles influenced the progress of the partnership. First, PT Indo CafCo decided to end its coffee trading in Bali after six years (2003-2009). Although their main reason was a change in internal policy caused by a change in director, it was also clear that the exporter no longer observed collaborative advantages that made it worthy to continue the partnership, as explained by a former staff member: We closed our coffee trade in Bali in 2009 because the director resigned. The new director did not want to continue the partnership because it was complicated, and it did not give much profit or quantity. Another reason was that the company wanted to focus on its main area in Medan, Sumatera, which could produce a larger quantity. We may get a lot of profit through such a partnership, but it takes a long time to reach that level. Second, after PT Indo CafCo left Bali, the local market became overly dynamic and competitive among buyers. Farmers preferred to sell their coffee to buyers with the highest offer. Consequently, farmers started to sell smaller quantities individually and this put the spirit of collectivity at risk. Meanwhile, ICCRI could not continue its work because there was no financial support from PT Indo CafCo or the local government 8. As commented by a farmer representative:

CGIP, established in 2005, assisted in the process of obtaining a GI certificate, and it later played a role as supervisor of the quality of Bali Kintamani Arabica coffee. The members of CGIP include 61 SA (representing 3,201 households) and 6 local coffee processors (with WP technique), and the local governments, ICCRI, CIRAD and PT Indo CafCo, which are members of the advisory board (CGIP, 2007: 4).

Although ICCRI formally ended its leadership role in the partnership, it still maintains an informal relationship with farmers and local governments. For example, ICCRI still consults regarding coffee quality and the coffee market through mobile phone contacts with the farmers and Disbun Bali. 8

International Food and Agribusiness Management Review

696

Wijaya et al.

Volume 20, Issue 5, 2017

When PT Indo CafCo was still here, Kintamani coffee was a leading commodity. We won a national contest in coffee taste. Unfortunately, a few farmers now no longer follow the SOP because there is no guarantee from the buyers. We think that there is no company like PT Indo CafCo, which gave us advantages as long as we produced good quality coffee. Third, farmers hesitated to continue the spirit of Motramed through the strategy of â&#x20AC;&#x2DC;one-door marketingâ&#x20AC;&#x2122; developed by the provincial government of Bali. This strategy aimed to stabilize coffee prices, regulate supply and demand, maintain commitment to quality and control its distribution so that unfair competition could be avoided (according to an interview with the director of CGIP Cooperative). The government initiated the cooperative of CGIP, which was expected to bring back the spirit of collectivity among SA and become a representative of coffee farmers to sell Bali Kintamani Arabica Coffee. However, not all SA were willing to participate in collective marketing through the CGIP Cooperative, with one reason being that farmers were reluctant to pay fees to the cooperative, even though the members would ultimately benefit from the fees. 4.5 Evaluation Previously, the farmers in this region were not well organized and were dependent on local collectors, and they sold unsorted, low-quality beans. The local government, i.e. Disbun Bali, did not have the knowledge to intervene. Motramed changed the old production network to a new one and created a fixed relationship between farmers and the exporter. The local government received advanced knowledge regarding coffee processing from ICCRI, and the governmental staff was able to train farmers. This partnering process was possible due to the leadership role of ICCRI and its close cooperation with the local government. After a decade of Motramed, the market in Bali is now more developed, produces better-quality coffee and is more dynamic. The positive side is that many buyers come to Bali to purchase good-quality coffee in smaller quantities than the exporter did. This creates competition among buyers, and farmers have more options to sell their coffee to the preferred buyer. However, this also has a negative side, as farmers tend to go back to an individualized system of trading. Farmers have gradually become more reluctant to participate in collective marketing through the CGIP Cooperative. A positive side effect is that the local government of Bali adopted this model and applied it to other commodities, such as cashew nuts and Robusta coffee, in other districts.

5. Intermezzo: the philosophy of Motramed Based on the successful implementation of the pilot project in Kintamani, ICCRI conceptualized the Motramed partnering process into four stages, as shown in Figure 2. ICCRI also presented the model at the national level at the Ministry of Agriculture and at the national coffee symposium in August 2006. Figure 2 shows three main actors in Motramed: farmer groups, coffee exporters and the mediator. In the first stage or pre-partnership, there are only relationships between the mediator and farmer groups and the mediator and exporters. Farmers and exporters do not have any relationship yet. The mediator explores the possibilities of developing a partnership. In the second stage, the mediator begins to connect farmer groups to exporters. The mediator transfers knowledge to farmers about quality improvement, while exporters can explore their advantages. In the third stage, the relation between farmer groups and the exporter is stronger. Here, the mediator plays a role as a consultant in the process. In the fourth stage, the independence stage, the mediator reduces its involvement and plays a role as a consultative actor. The mediator plays a crucial role as a mobilizer of the other actors. The mediator should be a neutral actor who has credibility and is trusted by all actors involved in coffee production. The mediator should have expert knowledge on coffee and needs to be competent to create liaisons among the main actors in coffee production. In this model, the mediator is ICCRI, because the ICRRI has advanced agricultural technology related to coffee production and is a respected actor in the sector. Moreover, the mediator needs to fulfill several roles: conducting a baseline study to explore the possibilities of a partnership, finding exporters as a partner, building trust between farmers and exporters, providing training for farmers, monitoring the process International Food and Agribusiness Management Review

697

Wijaya et al.

Volume 20, Issue 5, 2017

Farmer groups

Exporters

Mediator

Farmer groups

Exporters

Mediator Stage 2

Stage 1

Farmer groups

Exporters Mediator

Exporters

Mediator Stage 4

Stage 3

Figure 2. The implementation stages of Motramed (adapted from Mawardi et al., 2006: 84). of knowledge transfer, controlling coffee quality, and sustaining negotiations to achieve mutual agreement between farmers and exporters. In carrying out these roles, the mediator always coordinates with relevant governmental agencies. Although it is not stated clearly in the model, ICCRI cannot perform this work without an active supporting role of local governments through their budget allocations, human resources, and facilities. Local governments, at either the provincial or district level, can be considered a client who, due to their limited capability, asks ICCRI to be their consultant in coffee development. Their involvement in Motramed is supposed to help them develop their regional economy. However, the level and form of involvement of local governments in Motramed may vary from one region to another.

6. Case 2: the operationalization of Motramed in Bajawa, Flores Valuable lessons obtained from the Kintamani case improved the implementation of Motramed in Bajawa, Flores. Initially, the Provincial Estate Crops Agency of East Nusa Tenggara (ENT) in 2004 asked ICCRI to explore the possibility of implementing the model in some Arabica coffee regions on Flores Island, including Bajawa and Manggarai. In its feasibility study, ICCRI found similar problems to those in Kintamani. The old management practice in Bajawa was worse than that in Kintamani, as farmers did not have any knowledge of post-harvest and coffee processing, and this lack of knowledge resulted in a much lower coffee price than that in Bali. In Bajawa, District of Ngada, farmers only received approximately 1,100-1,500 IDR, or 0.15 USD/kg for Arabica coffee cherries9, which is four times less than the coffee price in Bali before the partnership was implemented. Farmers did not have any bargaining power because the quality was very low and the prices were set by local traders or collectors on the local market. Only the District Estate Crops Agency of Ngada (Disbun Ngada) actively supported Motramed and officially started the partnership with ICCRI in 2005. The objective of Disbun Ngada was to improve coffee farmersâ&#x20AC;&#x2122; welfare and to protect Bajawa coffee as their regional asset10. Disbun Ngada, which was supported by the provincial government of ENT, provided two budget allocations: assistance for the operational costs of ICCRI and facilities for farmers, including assisting farmers in the process.

Based on interviews with the senior researcher of ICCRI and the staff of the District Estate Crops Agency of Ngada. Previously, coffee from Bajawa was widely known as Ruteng coffee (a sub-district in Manggarai) because all the post-harvest processing was performed in Ruteng, and Bajawa was only a supplier of coffee cherries. Consequently, there was no economic advantage to process coffee in Bajawa (interview with the staff of Disbun Ngada). 10

International Food and Agribusiness Management Review

698

Wijaya et al.

Volume 20, Issue 5, 2017

We observed that the new management practice in Bajawa was much better organized than the one in Kintamani. Although the same technique of coffee processing used in Kintamani was applied in Bajawa, the network of farmer groups was stronger. Farmers received training regarding coffee quality, group dynamics and the economic system of togetherness to improve the organizational capacity of the farmer groups. In Kintamani, although ICCRI did not recognize the need for this training, improving farmers’ organizational capacity resulted in the acquisition of a crucial asset. Bajawa farmers were culturally different because they were very active, open to change and eager to develop Arabica coffee. As a start, two farmer groups that were willing to receive assistance and try something new were selected: Fa Masa and Suka Maju. Earlier, many farmers refused to participate because they did not want to become the subject of an experiment, as expressed by the staff of Disbun Ngada: At the beginning of our work, we often heard from farmers that they did not need us, and they asked us to leave. However, we are grateful that we have indomitable leaders (i.e. ICCRI researchers and the head of Disbun Ngada) who encouraged us to continue to help farmers because of the farmers’ ignorance on how to produce better coffee. With the support and promise of ICCRI and the government, farmers were encouraged to continue the process. In fact, the farmer groups quickly improved the quality of their production. Disbun Ngada and ICCRI developed a processing unit11 in each farmer group. The training courses and the processing unit were new for Motramed, and they were developed based on the evaluation of Kintamani’s experience. As a next step, ICCRI tried to invite exporters to join Motramed. Previously, there were no exporters in this region because yields were low and transporting and shipping costs were high. The first large exporter to enter Flores Island was PT Indokom Citra Persada, which ICCRI successfully invited. Involvement with Motramed was a big decision for this company, as they used to buy coffee from a supplier, and building a partnership with farmers was something new. However, the reputation of ICCRI in the coffee sector was seen as a guarantee of success, and PT Indokom felt confident with the support of the local government in Motramed. In addition to creating trust, all respective actors needed to explore the advantages of participating in Motramed. Applying Motramed in the Bajawa region was a challenge because of the lack of resources, such as bad infrastructure and inadequate human resources, which were common in the Eastern part of Indonesia. ICCRI’s interest was to make Motramed as successful as it was in Bali despite the unfavorable conditions. This interest matched well with the interest of the local government to promote their region, improve regional income by increasing farmers’ income and develop Arabica coffee. The farmers also obtained many advantages from participating in Motramed, such as new knowledge and skills in coffee processing, economic calculations, and negotiations with buyers, and they became more aware of coffee quality. One of the farmer groups expressed this awareness of quality as follows: When farmers process coffee based on the SOP, the prices never decline. Farmers do not want to go back to the traditional methods. ‘Quality is a must!’ We really feel the advantages of producing good quality coffee. PT Indokom understood the effort made by farmers to improve coffee quality. Therefore, after the farmers succeeded at producing green beans with good quality, PT Indokom gave them a better price than the local market. Table 2 below shows some remarkable coffee prices in Bajawa since Motramed was implemented.

In Bahasa, this is called Unit Pengolahan Hasil (UPH). This UPH functions as a business unit to support collective processing and trade with the exporter.

International Food and Agribusiness Management Review

699

Wijaya et al.

Volume 20, Issue 5, 2017

Table 2. The price history of Flores Bajawa Arabica Coffee (based on interviews with Bajawa actors 2015 and adapted from Soemarno et al., 2009; The Estate Crops Agency of Ngada District, 2015). Year

Price

Notes

2004

1,100 IDR per 1 kg of red cherries (equal to 6,600 IDR per kg green bean) 17,500 IDR per 1 kg of green beans

Farmers directly sold red cherries at the local market, as they did not yet know how to process coffee.

2005

2007 2013 2014

20,500 IDR per 1 kg 27,500 IDR per 1 kg 11,750 IDR per 1 liter of wet parchment bean (equivalent to approximately 45,000 IDR per 1 kg of green beans)

Processed coffee with a better quality. While the price of green beans at the local market was 12,000 IDR, Bajawa farmers received a 30% higher price from PT Indokom. Green beans (unsorted). This was a good price, as the world coffee price fell this year. Starting this year, farmers sold wet coffee parchment to PT Indokom. Wet parchment is processed coffee that has 30% moisture content. This solution (i.e. wet HS) was decided upon because of (1) high rainfall that did not support the drying process; and (2) farmers’ limited capital, whereas aid from the government was insufficient for all farmer groups. Farmers still benefited from this price because they did not have to do all the processing.

PT Indokom obtained advantages from Motramed as well. First, the supply chain was shorter because the company purchased directly from farmers. Second, PT Indokom acquired coffee with assured quality, which suited their consumers’ standards. PT Indokom encouraged farmers by giving aid, such as machines and tools to support coffee processing. At this stage, the relationship between farmers and the exporter became closer over time. Frequent meetings to discuss problems and solutions were possible. For example, because of the excellent quality, PT Indokom asked for a larger quantity; however, the farmers encountered difficulties related to labor, capital and natural factors. The win-win solution consisted of selling dried parchment coffee (dried HS)12. By selling dried HS, farmers saved time in processing but still received a good price. The district government showed a strong commitment and was very supportive in controlling the processes in each farmer group. In the meantime, the local government started to transfer the Motramed model to other commodities. Moreover, other coffee actors in Bajawa made greater efforts to develop the Arabica coffee commodity by obtaining sustainability certifications and the GI certificate, thus institutionalizing a more sustainable production process. PT Indokom helped Bajawa farmers become organic certified and sent its Internal Control System team to assist farmers in preparing their farms. In 2007, Bajawa coffee farms obtained organic certification from the USDA and next obtained Rainforest Alliance certification in 2008. According to the farmer groups, the sustainability certification not only positively impacted health and the environment but also farmers’ income13. For example, when the world coffee prices fell in 2013, Bajawa coffee farmers still received a good price from PT Indokom because of its quality and the growing demand by international consumers for a sustainability guarantee. Additionally, the Bajawa coffee stakeholders established the CGIP for Bajawa Flores Arabica Coffee and prepared the requirements book to apply for the GI certificate.14 In 2013, Bajawa Arabica coffee was awarded

Dried HS is processed coffee beans that still contain parchment (shell horn). If the shell horn is removed with huller machines, the product is called green beans (CGIP, 2011: 89). 13 Based on interviews with the director of PT Indokom and the Internal Control System staff. 14 The requirements book contains the history of Bajawa coffee, coffee processing, community, etc. By 2011, CGIP members consisted of 25 farmer groups that produced red cherries (representing 600 household farmers), 14 UPH, and 1 roaster (CGIP Flores Bajawa, 2011: 14).

International Food and Agribusiness Management Review

700

Wijaya et al.

Volume 20, Issue 5, 2017

the GI certificate by the Indonesian government. With this certificate, the Flores Bajawa Arabica coffee brand is now better prepared to preserve the coffee tradition in Bajawa. 6.1 The follow-up of Motramed After the ICCRI reduced its role as a mediator, the district government, together with other stakeholders, such as PT Indokom, the CGIP Bajawa, and an NGO that became involved, further developed the coffee sector. First, the CGIP Cooperative, which was more structured than the one in Kintamani, began to function as a one-door-marketing system of Flores Bajawa Arabica Coffee. In price negotiations, PT Indokom and the CGIP Cooperative sat together to discuss the prices and tonnages. Collective marketing through the CGIP Cooperative strengthened the farmersâ&#x20AC;&#x2122; position in negotiating the price with the company. The cooperative became a symbol of a shared commitment of the coffee stakeholders to develop Bajawa. For every kg of coffee sold through this cooperative, farmers or Unit Pengolahan Hasil (UPH) must contribute a marketing fee of approximately 500 IDR (or 0.5 cents in USD). With this marketing fee, the CGIP Cooperative has collected 66 million IDR that will be used to prepare the Fair Trade accreditation. Second, a new MoU was concluded in 2015 between the district government of Ngada with the CGIP cooperative, PT Indokom, Bank of East Nusa Tenggara, VECO Indonesia and ICCRI. More actors became interested in joining this MoU due to the achievements of Motramed in Bajawa. VECO Indonesia, as an NGO, will for example play a facilitation role for farmers in preparing for the Fair Trade certificate through capacity improvement. The Bank of East Nusa Tenggara will help farmers by providing access to loans for coffee processing. With this MoU, the program was more focused on on-farm management training to increase the production and expansion of coffee land area than the previous MoU. Motramedâ&#x20AC;&#x2122;s main focus was on post-harvest quality improvement; there was not much attention to on-farm improvements. Consequently, land area and production tended to stagnate over the past few years (Table 3). Third, the District government of Ngada issued local regulation number 6, year 2016, regarding the protection of Bajawa coffee as a regional asset. This regulation aimed to further protect Bajawa coffee and to confer more benefits to the community. It covers issues such as land protection to avoid chemical farming inputs and diversion of land use, collective marketing and labor during the harvest period. Table 3. Growing area, production and productivity of Bajawa Arabica coffee, 2010-2014 (adapted from The Estate Crops Agency of Ngada District (Disbun Ngada), 2015). No.

Year

Growing area (ha)

Production (ton)

Productivity

1 2 3 4 5

2010 2011 2012 2013 2014

5,677 5,651 5,627 5,510 5,720

1,996 2,584 3,047 3,298 3,448

739 766 811 867 825

International Food and Agribusiness Management Review

701

Wijaya et al.

Volume 20, Issue 5, 2017

7. Case 3: the further development of Motramed in Bondowoso, East Java Bondowoso, East Java, applied to Motramed in 2011. The partnering process in this case differs from that of the previous regions. The partnership model was introduced to the Provincial Estate Crops Agency in East Java in approximately 2006, immediately after the development in Flores. ICCRI attempted to contact officials at the provincial level, assuming that it might be easier to cover many districts at once. However, this approach was never continued. Similar to what occurred in Bali and Manggarai, the district governments were not willing to support a program from the provincial government. Although this was not a major constraint in the implementation of Motramed, it influenced the progress and continuation of the partnership. In this section, we particularly address what is different from the previous regions in the partnering process. The first difference concerns the 2011 MoU of the partnership, which was signed by seven actors and is valid for five years and operationalized in annual work plans. The new actors had different interests. The Bank of Indonesia was interested in maintaining the stability of the Indonesian currency. One method to accomplish this was to increase the export of agricultural commodities, such as Arabica coffee. The Bank of East Java would profit from providing loans to farmer groups. Perhutani is a public company engaged in the protection of forests. Although Perhutani was not directly involved with coffee processing, it became involved because 99% of the coffee plantations in Bondowoso are farmed in forest areas protected under the management of Perhutani. Perhutani’s interest is to ensure that the coffee plantations are managed in accordance with the principles of protected forest management. Following each actor’s interests, the roles of the respective actors were clearly stated in the MoU (Table 4). Based on the MoU and in contrast to the previous regions, the leadership role in Bondowoso was not solely handled by ICCRI; it was shared with the district government and Bank of Indonesia Jember. The second difference compared to the other regions is that the quality improvement in Bondowoso showed rapid progress. This progress was made possible by the improved conditions of the old management practice in Bondowoso compared with those in Bajawa. The farmers already had some knowledge regarding postharvest quality and the market. Unlike previous regions, where it took 1-2 years for the coffee to be exported, it only took approximately 8 months for the five selected farmer groups in Bondowoso to produce coffee with improved quality and export it through PT Indokom. These five farmer groups also had the opportunity Table 4. The roles of actors in Motramed Bondowoso (adapted from Memorandum of Understanding, 2011). No.

Actors

District Government of Bondowoso Bank of Indonesia, Jember

3. 4.

6. 7.

Roles

• Provides facilities and infrastructure and facilitates training and mentoring. • Facilitates technical aid, such as research, training and the provision of information, in order to improve farmer competency. • Facilitates access to financing. Indonesian Coffee and Cocoa • Provides coaching in all stages of coffee production, including Research Institute (ICCRI) cultivation, processing and marketing. PT. Regional Development Bank • Provides financing to assist with the development of Arabica of East Java (Bank Jatim) coffee farmers as long as doing so is in accordance with the bank’s regulations. ‘Perhutani’ public company • Provides forest area that could be used to develop Arabica coffee as long as this is in accordance with principles of sustainable forest management. PT. Indokom • Is a marketing partner, i.e. buyer. Association of Bondowoso coffee • Organizes farmers and farmer groups in the area of Bondowoso. farmers • Participates in a program to empower coffee farmers. International Food and Agribusiness Management Review

702

Wijaya et al.

Volume 20, Issue 5, 2017

to visit ICCRI’s coffee plantation in Jember so that they could learn faster. The following was expressed by a member of a farmer cooperative: After the first harvest period, when our coffee was exported to the Swiss and was recognized by the Mayor, PT Indokom, the Provincial government’s representative and other actors, I was full of emotion. Now, I understand the benefits of good processing for the farmer and that the responsibility is in the hands of farmers; the government and other actors can only help farmers improve their production. For farmers, the collaboration was a great advantage, as they had previously only applied traditional processing (drying in the street without a proper technique) and had only received 7,000 IDR per kg. After they succeeded in improved coffee processing and exporting, the coffee price increased to 22,000 IDR per kg of green beans in 2011. For PT Indokom, this partnership was advantageous, as this region is located near its office headquarters in East Java, which made distribution easier and more efficient. Despite its rapid progress, we found a few threats at the farmer level that were also discussed at the annual cooperative and coffee stakeholder meetings. First, there was an indication of dishonest behavior by a few farmers regarding profit sharing to Perhutani. According to the cooperation agreement between Perhutani and the forest village community, farmers are required to share 30% of the coffee production (i.e. red cherries) with Perhutani. However, there were a few unscrupulous individuals who misused the agreement for their personal profit. Second, there were a few farmer groups that did not return the loan to Bank Jatim at the agreed time. The Head of Disbun Bondowoso often reminded farmers of the importance of Bank Jatim’s trust in the partnership for the continuation of their coffee business. Third, keeping farmers in the collective commitment was problematic. Coffee farmers in Bondowoso could achieve positive results because they applied the SOP for coffee quality introduced by ICCRI. However, we observed that a few farmers were tempted to obtain more short-term profit by inconsistently following the SOP. Some of the farmers who had participated since the beginning of Motramed became concerned about this condition (as stated in interviews with farmer groups). These problems might not have occurred if the farmer organizations had sufficiently intervened. In Bondowoso, there are three important farmer organizations; however, their roles overlap, and these organizations failed to maintain the farmers’ commitment to develop Arabica coffee. This is similar to the situation in Kintamani, Bali. Keeping farmers in a collective is a social problem that is difficult to solve, and it can become one of the constraining factors in the partnership. The following was stated by the senior researcher at ICCRI: I affirm to farmers that in developing agriculture, the capital problem only plays a small part. If the problems of developing agriculture are 100%, then 60% of it is a social problem, such as selfishness, internal conflicts, dishonest behavior and more. The technological problem covers approximately 30% of the problem, and the rest is a capital problem. Farmers always complain about the lack of capital as their main problem, but this is not always the case. If the farmers have good organization and are solid and united, then banks, local governments or other actors will not hesitate to give them aid and financial facilitation. Although the new production practices are better organized, as shown by the MoU, the issue of farmer commitment needs to be noted. Nevertheless, all actors plan to continue the MoU for the second term and further address the capacity improvement of farmer organizations.

8. Conclusions This study examined a bottom-up development initiative to create a new partnership for more sustainable coffee production at the local level in Indonesia. We conceptualized the partnering process of the Motramed as a form of collaborative governance to create a new management practice.

International Food and Agribusiness Management Review

703

Wijaya et al.

Volume 20, Issue 5, 2017

Motramed aims to change the practices common in coffee production throughout Indonesia to improve farmersâ&#x20AC;&#x2122; welfare. Traditional production is characterized by farmers that possess inadequate knowledge regarding quality and price setting. They directly sell their unsorted coffee beans to local traders without further processing. These farmers do not have much capital and face difficulty obtaining funding to develop their coffee farms. Moreover, coffee farmers usually work individually. Although they are sometimes members of farmer groups, these farmer groups are often incapable of assisting farmers. Furthermore, there is no close relationship between farmers, exporters and governments in the old management practice. The new production methods that Motramed aims to bring about are quite different. The most important changes are that farmers (1) have obtained new knowledge and skills regarding coffee processing, which improves the quality and quantity of their production; (2) have applied collective marketing of coffee in better-organized farmer groups that are part of higher-level regional trading organizations; (3) have direct access to the international market through a permanent relationship with an exporter; (4) are sufficiently educated regarding price-setting, have acquired an improved bargaining position and can more easily obtain funding from banks; (5) have become a member in organizations that trade in sustainability certified coffee; and (6) are sustained by local governments that now recognize that coffee production can expand regional economic development and have acquired knowledge to further sustain the farmers. In this study, we defined four mechanisms: defining a common problem, building trust, creating collaborative advantages and establishing leadership, which we used as a lens to study the changes brought about by Motramed. We analyzed the implementation of the partnerships in three different regions and particularly focused on the constraining and conducive factors that are critical to the success of Motramed. Our main conclusion is that it is particularly difficult for the process of sustainable change to continue after the initial improvement in coffee production. As we have seen, there is a great risk that tensions between individual interests and collective interests will manifest themselves. This indicates that Motramed still creates rather fragile new production practices. A first critical factor is the scope of the problem. Motramed started with a small and localized problem definition in coffee production, i.e. low quality and inefficient marketing. This restricted definition ensured a practical approach involving only specialized actors to address the problems. Motramed thus primarily focused on the economic aspects of coffee production while the ecological and social aspects were mainly seen as a possible side effect. Only during the partnering process did the relevance of sustainability certification become clear as an important factor in new production practices. It is suggested that a more elaborate model might better balance its focus among the economic, social and environmental aspects of the production process, including the prospect of acquiring certification at the beginning. A second critical factor is the significant role of district governments. ICCRI would not be able to implement the partnership without the support of local governments in organizing farmers, allocating operational budgets and providing staff. For the local governments, Motramed is a way to promote their regions and improve the regional economy. In the most successful cases, the district government, together with ICCRIâ&#x20AC;&#x2122;s researchers, is continuously and intensively involved in daily practices at the farmer level. The Kintamani and Bondowoso cases in the early years show that a lack of responsiveness from the district governments towards Motramed led to the stagnation of its implementation. Therefore, the implementation of Motramed also depends on the willingness of the district governments to consider coffee production a priority. A third critical factor is the development of a shared responsibility between the provincial and district governments for the partnership approach. As we have seen, the intention to expand more regions might not be in accordance with the political situation. Since the implementation of regional autonomy in Indonesia, each of the districts has its own priorities, which may differ from higher level government policies. The cases in our study show that the Motramed model works well with the support of the district governments. Nevertheless, regarding budgeting and coordination, it is suggested that the support of provincial governments

International Food and Agribusiness Management Review

704

Wijaya et al.

Volume 20, Issue 5, 2017

is essential. Initiators of a partnership model such as Motramed should be aware of the political tensions that they might create in a multilevel governance situation. A fourth critical factor relates to the organization of the farmers. Among the three regions, Bajawa farmers developed the strongest farmer groups and showed the strongest commitment to keep farmers in collective action. In other regions, the organizations faced the difficulty of convincing farmers that they should abide by the rules of the CGIP organization. After farmers increased their knowledge regarding quality and marketing, the market grew and more buyers became interested, and therefore, farmers had increasing options for selling their coffee. However, this also had a negative side, as some farmers tended to show opportunistic behavior and focused on their own short-term economic profits. These farmers preferred to sell in smaller quantities to many buyers rather than to only to one exporter under the CGIP organization. This may result in a decline of trust between farmers and exporters. Therefore, strengthening the capacity of farmer organizations and raising awareness regarding the virtues of collective action is an essential asset of the partnership model. A fifth critical factor is regarding the question of how many and what type of actors to involve in the partnership. Inducting more actors may result in additional risks because of different interests and perceived problems. In the third region, we observed that Motramed involved more actors from various backgrounds as partners. The positive side of this is that with additional actors and different resources, more aspects of the coffee sector can be addressed, which might hasten progress. If there are additional participating actors, then there are additional interests that require consideration and these interests may be competing or conflicting. However, our research shows that the participation of actors who are able to secure funding, such as banks, positively influences the results of Motramed. Recently, ICCRI has started to expand the Motramed model to many regions in Indonesia. Together with local governments, Motramed has been initiated not only for Arabica coffee but also for Robusta coffee and cocoa. Discussion is also continuing regarding the adoption of the Motramed model in a national program of the Ministry of Agriculture. This will create additional opportunities to develop a nationwide bottom-up development process to improve smallholder production of agricultural commodities, but it may also lead to bureaucratization of the approach with much red tape. Further research on the conditions of an effective national program should be the next step in advancing this type of bottom-up sustainable development of the smallholder coffee sector with a connection to global certification schemes.

Acknowledgement This research was conducted as part of the joint research project on Social and Economic Effects of Partnering for Sustainable Change in Agricultural Commodity Chains in Indonesia. The project involves a bilateral cooperation between Maastricht University and Lampung University, with the financial support from the Royal Netherlands Academy of Arts and Sciences (KNAW) and the Ministry of Research, Technology and Higher Education (MENRISTEKDIKTI) of Republic Indonesia. The authors are very grateful to Bustanul Arifin, Purwo Santoso, Pieter Leroy, and Astrid Offermans for their valuable comments and input to (earlier versions of) this paper.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2017.0021. Table S1. The price history of Bali coffee compared with the world coffee price.

International Food and Agribusiness Management Review

705

Wijaya et al.

Volume 20, Issue 5, 2017

References Abdulsamad, A., S. Stokes and G. Gereffi. 2015. Public-private partnerships in global value chains: can they actually benefit the poor? USAID Leverage Economic Opportunities (LEO) Report 8. Duke University, Durham, NC, USA. Available at: http://tinyurl.com/y7qggszx. Astuti, E.S., A. Offermans, R. Kemp and R. CĂśrvers. 2015. The impact of coffee certification on the economic performance of Indonesian actors. Asian Journal of Agriculture and Development 12. Austin, J.E and M.M. Seitanidi. 2012. Collaborative value creation: a review of partnering between nonprofits and businesses: part I. Value creation spectrum and collaboration stages. Nonprofit and Voluntary Sector Quarterly 41(5): 726-758. Barroso-MĂ¨ndez, M.J., C. Galera-Casquet, M.M. Seitanidi, and V. Valero-Amaro. 2016. Cross-sector social partnership success: a process perspective on the role of relational factors. European Management Journal 34: 674-685. Bartley, T. 2010. Certification as a mode of social regulation. Jerusalem Papers in Regulation and Governance Working Paper No. 8 May 2010. The Hebrew University, Jerusalem, Israel. Available at: http:// tinyurl.com/yckn49ya. Bitzer, V., M. Francken and P. Glasbergen. 2008. Intersectoral partnerships for a sustainable coffee chain: really addressing sustainability or just picking (coffee) cherries? Global Environmental Change 18: 271-284. Bitzer, V. and P. Glasbergen. 2015. Business-NGO partnerships in global value chains: part of the solution or part of the problem of sustainable change? Current Opinion in Environmental Sustainability 12: 35-40. Bitzer, V., P. Glasbergen and B. Arts. 2013. Exploring the potential of intersectoral partnerships to improve the position of farmers in global agrifood chains: findings from the coffee sector in Peru. Agriculture and Human Values 30: 5-20. Brewer, B. and M.R. Haylarr. 2005. Building public trust through public-private partnerships. International Review of Administrative Sciences 71: 475-492. Brinkerhoff, J.M. 2002. Government-nonprofit partnership: a defining framework. Public Administration and Development 22: 19-30. Central Bank of Indonesia. 2016. Selected foreign exchange middle rates against rupiah at Bank Indonesia year 2001-2015. Central Bank of Indonesia (BI), Jakarta, Indonesia. Community of Geographical Indication Protection of Bajawa Flores Arabica Coffee (CGIP). 2011. The requirements book of geographical indication. CGIP, Flores, Indonesia. Community of Geographical Indication Protection of Bali Kintamani Coffee (CGIP). 2007. The requirements book of geographical indication. CGIP, Bali, Indonesia. Emerson, K., T. Nabatchi and S. Balogh. 2011. An integrative framework for collaborative governance. Journal of Public Administration Research and Theory 22: 1-29. Fuchs, D., A. Kalfagianni and T. Havinga. 2011. Actors in private food governance: the legitimacy of retail standards and multistakeholder initiatives with civil society participation. Agriculture and Human Values 28: 353-367. Giovannucci, D. and F.J. Koekoek. 2003. The state of sustainable coffee: a study of twelve major markets. International Institute for Sustainable Development (IISD), the International Coffee Organization (ICO) and the United Nations Conference on Trade and Development (UNCTAD) with the support of the World Bank and the International Development Research Centre. Available at: http://tinyurl. com/ya57aes5. Giovannucci, D. and S. Ponte. 2005. Standards as a new form of social contract? Sustainability initiatives in the coffee industry. Food Policy 30: 284-301. Glasbergen, P. 2011. Understanding partnerships for sustainable development analytically: the ladder of partnership activity as a methodological tool. Environmental Policy and Governance 21: 1-13. Gray, B. 2007. The process of partnership construction: anticipating obstacles and enhancing the likelihood of successful partnerships for sustainable development. In: Partnerships, governance and sustainable development. Reflections on theory and practice, edited by P. Glasbergen, F. Biermann and A.P.J. Mol. Edward Elgar, Cheltenham, UK. International Food and Agribusiness Management Review

706

Wijaya et al.

Volume 20, Issue 5, 2017

Hartatri, D.F.S. 2016. Influence of quality improvement activities and direct selling through mediated partnership model on supply chain, farm-gate price and Indonesian households specialty coffee farmers’ income. Pelita Perkebunan (a Coffee and Cocoa Research Journal) 32: 52-66. Hartatri, D.F.S. and B. de Rosari. 2011. Arabica coffee farming and marketing chain analysis in Manggarai and East Manggarai districts. Pelita Perkebunan (a Coffee and Cocoa Research Journal) 27: 55-67. Hatanaka, M. 2010. Certification, partnership, and morality in an organic shrimp network: rethinking transnational alternative agrifood networks. World Development 38: 706-716. Huxham, C. and S. Vangen. 2000. Leadership in the shaping and implementation of collaborative agendas: how things happen in a (not quite) joined-up world. The Academy of Management Journal 43: 1159-1175. International Coffee Organization (ICO). 2016. World coffee price from 2001-2015 (annual and monthly averages). International Coffee Organization (ICO), London, UK. Available at: http://www.ico.org/ new_historical.asp. Klijn, E.-H., J. Edelenbos and B. Steijn. 2010. Trust in governance networks. Administration and Society 42: 193-221. Loconto, A. 2016. The role of knowledge in transitions to sustainable food systems: examples from institutional innovations. Knowledge and Information for Sustainable Food Systems 203. Marin-Burgos, V., J.S. Clancy, J.C. Lovett. 2015. Contesting legitimacy of voluntary sustainability certification schemes: valuation languages and power asymmetries in the roundtable on sustainable palm oil in Colombia. Ecological economics 117: 303-313. Mawardi, S. 2009a. Advantages, constraints and key success factors in establishing origin-and traditionlinked quality signs: the case of Kintamani Bali Arabica coffee geographical indication,. Available at: Retrieved from http://tinyurl.com/ya8trskl. Mawardi, S. 2009b. Establishment of geographical indication protection system in Indonesia: a case in coffee. In: Report of Worldwide Symposium on Geographical Indications, Sofia, 10-12 June 2009. Available at: http://tinyurl.com/yae4a6bo. Mawardi, S., C. Ismayadi, A. Wibawa, Sulistyowati and Yusianto. 2006. Model kemitraan bermediasi (Motramed) untuk pengembangan agribisnis kopi melalui perbaikan mutu dan sistem pemasaran di tingkat kelompok tani. In: Proceeding of Simposium Kopi, Surabaya, 2-3 October 2010. Surabaya, Indonesia. Memorandum of Understanding. 2011. Memorandum of Understanding (MoU) of Bondowoso coffee partnership. The District of Bondowoso, East Java, Indonesia. Neilson, J. 2008. Global private regulation and value-chain restructuring in Indonesian smallholder coffee systems. World Development 36: 1607-1622. Saad, M., K. Rowe and P. James. 1999. Developing and sustaining effective partnerships through a high level of trust. In: Public and private sector partnerships: furthering development, edited by L. Montanheiro, B. Haigh, D. Morris, and M. Linehan. Sheffield Hallam University Press, Sheffield, UK. Schouten G., S. Vellema and J.V. Wijk. 2016. Diffusion of global sustainability standards: the institutional fit of the ASC-Shrimp standard in Indonesia. Revista de Administração de Empresas 56: 411-423. Schruijer, S.G. 1999. Leadership and organizational change: an introduction. European Journal of Work and Organizational Psychology 8: 1-8. Soemarno, D., S. Mawardi, M. Maspur and H. Prayuginingsih. 2009. Added value improvement on arabica coffee wet process method using model kemitraan bermediasi (Motramed) on unit Pengolahan Hasil at Ngada Residence-NTT. Pelita Perkebunan (a Coffee and Cocoa Research Journal) 25: 38-55. The Estate Crops Agency of Bali Province. 2015. The price of Bali coffee 2001-2015. The Estate Crops Agency of Bali Province, Bali, Indonesia. Estate Crops Agency of Ngada District. 2015. The price of Ngada coffee 2004-2014. The Estate Crops Agency of Ngada District, Ngada, East Nusa Tenggara, Indonesia. Van Rijsbergen, B., W. Elbers, R. Ruben, S.N. Njuguna. 2016. The ambivalent impact of coffee certification on farmers’ welfare: a matched panel approach for cooperatives in Central Kenya. World Development 77: 277-292.

International Food and Agribusiness Management Review

707

Wijaya et al.

Volume 20, Issue 5, 2017

Van Tulder, R. and S. Pfisterer. 2009. Creating partnering space: exploring the right fit for sustainable development partnerships. In: Social partnerships and responsible business: a research handbook, edited by M. M. Seitanidi and A. Crane. Routledge, Third Avenue, NY, USA, pp. 105-124. Vellema, S., G. Ton, N.D. Roo and J.V. Wijk. 2013. Value chains, partnerships and development: using case studies to refine programme theories. Evaluation 19: 304-320. Virgiano, L. 2012. A retrospective study of Motramed (in English: Mediated Partnership Model) program for arabica coffee farmers in Bondowoso, Indonesia. Master thesis, Malmรถ University, Malmรถ, Sweden.

International Food and Agribusiness Management Review

708

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2016.0171 Received: 8 November 2016 / Accepted: 21 May 2017

Exploring the applicability of a sustainable smallholder sourcing model in the black soybean case in Java RESEARCH ARTICLE August R. Sjauw-Koen-Fa a, Vincent Blokb, and Onno Omtac aPhD

Candidate, bAssociate Professor and cProfessor, Management Studies Group, School of Social Sciences, Wageningen University, Hollandseweg 1, P.O. Box 8130, 6700 EW Wageningen, the Netherlands

Abstract Food and agribusiness multinational enterprises are redesigning their sourcing strategies to tap into the underused food production potential of small-scale farms in a way that improve farmersâ&#x20AC;&#x2122; livelihood. The problem is that current widely applied sourcing models do not include improvement of livelihood of the producers/farmers. The present article explores the applicability of a sustainable smallholder sourcing model with a list of critical success factors, in which business objectives and corporate social responsibility perspectives are combined. To this end, the black soybean supply chain in Java/Indonesia is studied. It was found that the black soybean case can be conceptualized by the sourcing model. Most of the critical success factors were present, but also some differences were identified. The differences enable to fine-tune some critical success factors. The sustainable sourcing model can help in (re-)designing sourcing strategies to secure sustainable and more equitable supply from small-scale farmers from a business perspective. Keywords: agribusiness, CSR, smallholder inclusion, sustainable sourcing, value chain JEL code: O13, Q13, Q01, Q02 Corresponding author: august.sjauw-koen-fa@wur.nl

ÂŠ 2017 Sjauw-Koen-Fa et al.

709

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

1. Introduction The challenge in meeting the increasing global food demand (FAO, 2009) is that food supply must be achieved in an era that is characterized by resource constraints imposed by planetary resource boundaries and climate change, while dietary patterns are shifting and consumers concerns of food production are intensifying. In addition, corporate actors are increasingly called upon to play a proactive role in solving pressing global challenges such as food security, climate change and poverty reduction (e.g. UN Sustainable Development Goals 2015-2030). In response to the challenging global food demand, food and agribusiness multinational enterprises (F&A MNEs), such as Unilever, Hershey, Mars, Mondelez, Cargill, Ferrero and Nestlé, are redesigning their sourcing strategies to tap into small-scale farms (smallholders) in developing and emerging economies. This relates largely to commodities such as cocoa, coffee, bananas, tea, and cotton. The F&A MNEs’ aim is to secure sustainable supply in a way that includes improvement of livelihood/economic welfare of smallholders. Smallholders in developing and emerging economies, however, face productivity, product quality and transactional constraints in their effort to access high value-adding food markets (e.g. London et al., 2010; Rivera-Santos and Rufin, 2010). To overcome these constraints, smallholders need to become more advanced. F&A MNEs can potentially help smallholders with these challenges, because of their dominant position in agricultural value chains (e.g. Reardon et al., 2009; Rossignoli and Moruzzo, 2014; Sjauw-Koen-Fa, 2010). The problem is that the conventional sourcing strategies by F&A MNEs mainly focus on complying with consumers’ concerns regarding environmental and societal issues of production (e.g. Bush and Bain, 2004; Henson and Humphrey, 2009; Trienekens et al., 2012), rather than on improving farmers’ livelihood/economic welfare. This raises the question how conventional sourcing strategies by F&A MNEs from smallholders can address smallholders’ livelihood improvement simultaneously with other business objectives. We have found several studies assess the impact of conventional sourcing models on farmers’ livelihood (e.g. Blackman and Rivera, 2011; Dragusanu et al., 2014; Jaffee and Henson, 2004; Ruben and Fort, 2011; Rueda and Lambin, 2013) and some empirical case studies scrutinize particular aspects of smallholder interest by F&A MNEs (Alvarez et al., 2010; Perez-Aleman and Sandilands, 2008). The question remains how F&A MNE’s can best include smallholders’ interest in their sourcing practices to secure a sustainable and more equitable supply, while maintaining competitive advantage. This has not been studied extensively yet. In order to find an answer to this sourcing question, Sjauw-Koen-Fa et al. (2016) have developed a Sustainable Smallholder Sourcing model (3S-model) with a list of critical success factors, based on a literature review, in which the delivery of value for improvement of livelihood/economic welfare to smallholders is integrated. The objective of this research is to explore the applicability of this smallholder sourcing model by using primary data from a value chain analysis of the black soybean supply chain of an F&A MNE in Java. The value chain analysis was conducted in November-December 2013 aimed at learning how the F&A MNE can do business with smallholders in a way that improves the farmers’ livelihood (Tait, 2015). The black soybean supply chain in Java was scaled, providing longitudinal data (2007-2013), proved to work and had produced progressive results. As such, this case is considered as a best practice example for exploring the applicability of the developed Sustainable Smallholder Sourcing model. In the next section, the framework of the developed Sustainable Smallholder Sourcing model with the list of critical success factors (CFSs) will be introduced, followed by the materials and methods section. Finally, the case findings of the study are presented and the lessons learned are discussed and concluded.

International Food and Agribusiness Management Review

710

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

2. Introduction to the Sustainable Smallholder Sourcing model As we want to explore the applicability and the dynamics of the Sustainable Smallholder Sourcing model (3S-model) with the list of CFSs by studying a best practices empirical case in the present article, we first explain the framework of the Model in this section. The conceptual elements of the 3S-model are: 1. ‘Upgrading’ (Humphrey, 2004; Humphrey and Schmitz, 2002), which is defined as a transition of firms/ smallholders to higher value added activities or interventions in production to improve technology, knowledge and skills in order to increase the benefits or profits deriving from the participation in regional or global production networks. 2. ‘Supplier development program’ (Hahn et al., 1990; Watts et al., 1995), which is defined as long-term cooperative efforts between a lead food firm and its suppliers to upgrade smallholder farming systems to secure sustainable supply from a business perspective and improving smallholder livelihood at the same time. 3. ‘Global value chain governance structure’ (Gereffi et al., 2005), which is defined as non-market coordination. We used their framework to determine the emerging coordination combination with the characteristics of smallholder supply chains from Riisgaard et al. (2010), in order to determine the best governance structure to coordinate smallholder supply chains for supplier developed by F&A MNEs. This is the Captive governance type in which supply chain actors are locked in for upgrading smallholder farming system by the lead firm. The CSFs are based on Supply Chain Management literature on the domains of sourcing/strategic purchasing and buyer-supplier relationships, and the literature on Subsistence Market and Bottom of the Pyramid on the domains of competitive and institutional environment, networks, and farmer business models. The aim was to identify leverage points/synergistic connections between MNE (the buyer), sourcing strategies (top-down approach of the supply chain), and small-scale farmer business models (bottom-up approach of the supply chain). Enabling us to define the six sub-questions based on the synergistic connections that were found. Frameworks that provide a complete overview of the challenges of supply chain management and the bottom of the economic pyramid BOP that were used to identify leverage points/synergistic connections for smallholder inclusion in high value-adding supply chains were: ■■ For the supply chain management domains: the research framework of Chen and Paulraj (2004) and Carter and Rogers (2008). ■■ For the BOP domains: The BOP producers constraints frameworks of London et al. (2010) and Rivera-Santos and Rufin (2010). Figure 1 represents the framework of the 3S-model. The 3S-model consists of 6 building blocks of which F&A MNE, intermediaries, smallholders and Partnership are the core, while the other two building blocks are control variables. The sourcing process of sustainable smallholder supply consists of two activities: (1) the buying process in the supply chain (the axis ‘F&A MNE – Intermediaries – Smallholders’); and (2) the upgrading process in the supply chain (the partnership consisting of F&A MNE, intermediaries and input suppliers). In this model, government, donor, public bodies, Non-governmental organizations (NGOs) and private foundations supporting the upgrading of the smallholder farming system can join in the partnership. The building block ‘livelihood improvement smallholders’ represents the impact of the 3S-model. The CSFs which should leverage external and internal organizational challenges of sustainable smallholder inclusion by the F&A MNE, are located at the conjunctions of the chain ‘Partnership – Smallholder farming system’ and ‘Partnership – F&A MNE’, respectively. There are 5 external (outside the F&A MNE) and 2 internal organizational (within the F&A MNE) critical success factors: International Food and Agribusiness Management Review

711

Sjauw-Koen-Fa et al.

Internal organizational CSFs

Volume 20, Issue 5, 2017

Government, NGOs, public bodies, private foundations, social investors

Secured supply

External CSFs

Upgrading

F&A MNE

Partnership for sustainable smallholder supplier development

Procurement & Operation & CSR

Intermediaries

Market pressure and incentives

Smallholder farming system

Livelihood improvement of smallholders

inputflow outputflow

Figure 1. Sustainable Smallholder Sourcing model (3S-model) (Sjauw-Koen-Fa et al., 2016). NGO = Nongovernmental organization; CFS = critical success factor; F&A MNE = food and agribusiness multinational enterprises; CSR = corporate social responsibility. ■■

External organization CSFs: ◆ CSF 1: selected smallholders are commercial/market-oriented Smallholders that are potentially able to adopt upgrading interventions to meet F&A MNE’s supply requirements from a business perspective successful are commercial/market oriented small-scale farmers. These are according the segmentation of smallholder households of Christen and Anderson (2013) they grow cash crops that are sold usually in local, regional or export markets and have limited access to inputs, financial services and market information. ◆ CSF 2: building partnership for upgrading Building a sustainable smallholder supply chain is a long term process that needs to be a cooperative one, which is based on partnership-like and long-term contracts to achieve mutual interest, rather than transactional ‘buyer-seller’ relationships that are driven by bargaining power and shortterm contracts to achieve quick-wins at low cost by the buyer (Monczka et al., 1998; Trent and Monczka, 2002). Therefore, F&A MNEs need to collaborate with non-traditional actors, such as NGOs’, development and governmental agencies, and social investors, as partner and facilitator in the companies’ smallholder supply chains. The aim is to bring public and private resources and capabilities together which are needed to upgrade smallholder farming systems effectively (e.g. Hahn and Gold, 2013; London et al., 2010) and to generate relational rents spread over the supply chain network (Dyer and Singh, 1998). ◆ CSF 3: building a captive governance structure based on a cooperative ‘buyer-seller’ relationship To include smallholders effective in high value-adding supply chains, F&A MNEs need to get closer to smallholders in collaboration with supply chain partners, that are willing to co-invest resources and time, share risks and rewards and maintain relationships over a longer period of time (Krause and Ellram, 1997; Landros and Monczka, 1989). The 3S-model proposes that as lead firm, F&A MNEs can best apply a captive governance structure (Gereffi et al., 2005) to lock-in all supply chain actors for upgrading smallholder farming system, based on a cooperative rather than a transactional ‘buyer-seller’ relationship (Hahn et al., 1990; Landros and Monczka, 1989; Mohr and Spekman, 1994).

International Food and Agribusiness Management Review

712

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

CSF 4: building effective producers’ organizations A producers’ organization (PO) is based on the principle that acting collectively improves the position of its members, such as smallholders, and creates growth opportunities in farm productivity and income. They can fit together activities of sellers (farmers) and buyers (traders and processors) to more effectively meet market requirements than smallholders can achieve individually. Areas in which POs can play a role in strengthening the coordination in smallholder supply chains are: reducing transaction costs and marketing risks, enabling collective action and facilitating access to finance (e.g. Chambo, 2009; Onumah et al., 2007; Torero, 2011). ◆ CSF 5: providing access to finance to smallholders In many developing countries financial markets are imperfect because of high transaction costs and information asymmetries. These imperfections are likely to be binding on smallholders that lack collateral credit history and connections. However, access to affordable financial services is essential in order for smallholders to meet investment and working capital requirements, and other financial services such as insurance to cover risk and savings, to unlock their potential (Chalmers et al., 2006; IFC and GPFI, 2011; Miller and Jones, 2010; Sjauw-Koen-Fa, 2012). Internal organizational CFSs (within the F&A MNE) are: ◆ CSF 6: presence of a clear smallholder sourcing strategy by the F&A MNEs Firms have different responses to social responsibility and social issues such as smallholder inclusion. This refers to a firm’s corporate commitment and capacity, such as mechanisms, procedures, arrangements, behavioral patterns, sustainability codes and standards to anticipate social issues (e.g. Gold et al., 2013). Social responsiveness of firms can range from ‘doing nothing’ to ‘doing much’ regarding CSR (Caroll, 1979; Maignan et al., 2002; Van Tilburg et al., 2012). In the supply chain literature, the involvement and commitment of the top management has been emphasized, because they understand best the needs of supply chain management as they have the most knowledge of the firm’s strategic imperatives to remain competitive in the market place (Hahn et al., 1990; Monczka et al., 1998). The presence of a proactive CSR strategy, including a responsible supplier guide and sustainable agricultural codes, which is supported by the top management, is a critical precondition for long-term investments in sustainable smallholder supply (Gold et al., 2013; Monczka et al., 1998). ◆ CSF7: use of Cross-functional sourcing teams The challenge is that within large firms profit driven short term business activities and policy/ strategic driven long term CSR activities are operating in different structures. However, the sustainability and CSR strategy should be integrated into core business operations, activities and products and services of the firms (e.g. Blok et al., 2013; Spence and Bourlakis, 2009). The use of Cross-functional teams, led by Procurement and Operation and including CSR of the F&A MNE, may help to harmonize organizational values, routines and resources, and to interact effectively with supply chain counterparts (Driedonks et al., 2014; Olsen and Boxenbaum, 2009; Trent and Monczka, 1994). ◆

■■

3. Materials and methods 3.1 Research setting ■■ Soybeans and sweet soy sauce in Indonesia by Unilever Indonesia In Indonesia close to 90% of the soybeans are processed into food products ‘tahu’ and ‘tempe’, which are the main source for proteins, vitamins and fat for low-income households in Indonesia. A smaller amount is used for production of soy milk, sauces and other traditional Javanese food products. Per capita consumption of soybeans of Indonesia has increased significantly, while domestic soybean production has been declining in the past decades, due to large import of cheaper (yellow) soybeans most from the USA. This is possible because Indonesia applied a liberal trade policy (e.g. Dartanto et al., 2011). The lack of infrastructure to support government initiatives and farmers’ lack of access to better farming technology are seen as major International Food and Agribusiness Management Review

713

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

causes for the souring of domestic production. Moreover, import soybeans were preferred by the local food processing industry, because they had a better and more constant quality than domestic grown soybeans. Consequences of the liberal trade and ‘cheap’ basic food security policies of the Government was that the involvement of the government in local agricultural supply chains development, such as black soybean production, remains at arm’s length. They provide mainly general agricultural support, such as extension services, empowerment of cooperatives and facilitating sustainable agriculture platforms. Unilever Indonesia is a subsidiary of Unilever that produces and sells a number of brands locally. It had implemented the fundamentals of corporate social responsibility (CSR) from the early 1970s onwards, which is basically focused on market and community development support (Urip, 2010). In 2000, Unilever Care Foundation Indonesia (Unilever CSR) was established in Indonesia to strengthen the Unilever Indonesia’s commitment to sustainable development of Small and Medium Enterprises in Indonesia, while maintaining a competitive edge, i.e. CSR became an integral part of Unilever Indonesia ‘s business strategy for Indonesia. Unilever Indonesia is helping small-scale paddy famers in Java to produce black soybeans of a high and constant quality used as key ingredient of sweet soy sauce. The specific taste of the black soybeans, the traditional recipe and the use of local small-scale paddy farmers makes it possible to advertise and sell the sweet soy sauce to local supermarkets. ■■ Historical background of the black soybean supply chain In 2001, the Unilever Indonesia acquired a majority stake in a Javanese company producing a regional brand of sweet soy sauce for the Indonesia market. This brand has maintained its classic taste due to the consistent high-quality taste of locally produced black soybeans. It was recognized that the supply of black soybean would not be enough to meet the growing demand of the brand sweet soy sauce, because many smallholders in Indonesia switched to other crops due to the low price of imported yellow soybeans. To solve this problem, the Unilever Indonesia chosen to develop its own black soybean supply chain of smallscale paddy farmers in Java, in addition to still purchasing additional black soybean from selected/qualified regional commodity traders. However, the small-scale paddy farmers needed to be trained in cultivating black soybean according to good agricultural practices and access to inputs. In 2002 a pilot upgrading farm program with two cooperatives including 12 small-scale paddy farmers has started, and since 2003 the University of Gaja Mada (seed-supplier) has enter into a strategic partnership with Unilever Indonesia providing guidance on how to grow black soybean by paddy farmers and for breeding an improved black soybean variety. They had also selected a high yield variety of black soybean (Malika) that became a cornerstone of the upgrading program. A partnership consisting of Seed-suppliers and selected cooperatives from East and Central Java (Cooperatives) was formed and a program to upgrade small-scale paddy farmers to grow black soybean on contract base was set up. The period 2002-2007 is seen as the pilot phase of the black soybean supply chain. Around 2007 about 5,000 farmers from eight cooperatives were participating in the planting of black soybean, covering an area of about 1,200 hectares on Java. They contributed 10-20% of the black soybean demand in the period, while the remaining quantity was purchased from regional commodity traders operating in other areas in Indonesia (Hasibuan-Sedyono, 2010). By 2007 the black soybean supply chain was scaled to a full commercial level. Unilever Indonesia became full owner of the soy sauce company and Malika was certified by the Indonesian authorities and became a cornerstone of the black soybean supply chain. The present study focused on the period 2008-2013 which is the upscaling phase of the black soybean supply chain.

International Food and Agribusiness Management Review

714

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

■■ Case selection criteria The black soybean supply chain in Java of Unilever Indonesia was selected to study the applicability of the 3S-model because: 1. This case study, which includes a value chain analysis, was part of a broader joint research program Sunrise 2.0, commissioned by Unilever (one of the largest consumer goods company of the world with a clear proactive CSR strategy (Van Tilburg et al., 2012), and Oxfam (an international confederation of a number of NGOs working together with partners and local communities). The aim was to learn how Unilever can do business with smallholders in a way that improves the smallholders’ livelihood. The research approach included that Unilever viewed smallholder supply chains top-down (from F&A MNE’s perspectives), while Oxfam viewed them bottom-up (from farmer’s perspectives). The combined perspectives were also integrated in the used methods, tools and interview questionnaires. The signed Memorandum of Understanding of the Sunrise research project 2010-2015 (see final report of the Sunrise project, available at http://tinyurl.com/y8h4tsqf) was led down that Unilever and Oxfam were funders of the joint research program. Both organizations might use the outcome for their own purposes and interests. These research settings ensured more balanced results of conclusions, rather than in case of dominance of one of the two research partners. Accordingly, the black soybean supply chain in Java provided a unique case for empirical studying the integration of business and CSR perspectives for smallholder inclusion in high value-adding supply chains: it combines the business and CSR perspective which is the basis of the developed Smallholder Sourcing model with the list of critical success factors. 2. As the black soybean supply chain in Java is a scaled supply chain that provided sufficient historical data (2007-2013) and opportunities to review the evolution of the black soybean supplier development program and the governance conditions over a period of time. Moreover, it provided also opportunities to identify and interview all relevant supply chain actors including procurement, operation and CSR managers, farmers, intermediaries, input suppliers, government, field workers and public stakeholders, and to do field observations to verify the accuracy of understanding and consistency of the collected data and information. 3.2 Methods ■■ Research approach To explore the applicability of the sourcing model with the list of CSFs, we used the found black soybean supply chain map from the value chain analysis to learn about the partnership model for supplier development and the buying and the upgrading sourcing process including the role of the different actors and trading relationship in the supply chain. We also used the outcome of the black soybean farmer business model to assess the impact of the sourcing model on farmers livelihood. Finally, we match the information and lessons of the black soybean supply chain with the elements of the smallholder sourcing model to explore similarities and difference to draw conclusion about the applicability. ■■ Back ground field study The value chain analysis of the black soybean supply chain was conducted by the lead author in the period June 2013 to April 2014. The desk research consisted of an evaluation of the sourcing strategy and the CSR policy of the Unilever Indonesia, reviewing publications regarding the black soybean sector in Indonesia, and collection of relevant information and data about the supply chain. The field research in Java was conducted in Jakarta and Jogjakarta from 24 November to 4 December 2013. The practical toolkits of the LINK methodology of the International Center for Tropical Agriculture were applied to map the black soybean supply chain and to explore the farmers’ business model and to get an indication of the impact of the applied black soybean sourcing model on smallholders’ livelihood.

International Food and Agribusiness Management Review

715

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

Before the field research in Java was conducted, the black soybean supply chain were preliminary mapped in cooperation with Unilever Indonesia. The aim was to select and invite participants of all categories of supply chain actors (input suppliers, farmers, cooperatives, region commodity traders, managers of the Unilever Indonesia, government representatives, field workers, NGOs) for a multi-stakeholder workshops, to select interviewees for semi-structured interviews and for a field observations. The number of smallholders (out of about 8,200 distributed across nine areas in Java) that could be invited for the workshop and interviews was limited (in total 17), because they had to come (fly) to Jogjakarta. We selected smallholders that had experience with the black soybean upgrading program for several years, because we primary focus on why smallholders stay in the black soybean program. The assumption we made was that profitability and income security of producing black soybean according to the conditions of the upgrading program are foremost the key consideration to join and to stay in the program. Moreover, farmers were free to join the black soybean supply program, because they can grow other crops such as corn, chili peppers and ground nuts for the market. We cross-checked the impact of the applied black soybean supply program on smallholder livelihoods during the workshops and field visits and by means of personal communication with supply chain actors. The program of the field research conducted in Java consisted of the following elements: ■■ A multi-stakeholder workshop with all key stakeholders of the black soybean supply chain including Unilever Indonesia Procurement and CSR managers, Cooperative executives, farmers, seed supplier, field workers, NGO, local government servant of the black soybean supply chain (n=22) was held to explore the (trade) relations and the flow of products, services and payments between stakeholders in order to map the black soybean supply chain. Topics discussed during the multi-stakeholder workshop were: what are the core processes in the supply chain? How is the supply chain organized? Who are the key partners? How do products, payments, services and information flow through the supply chain? What are the external influences that affect the performance of the supply chain? ■■ A farmers’ workshop (n=17) to map the farmers’ business model. We used the business model canvas exercise to get an indication of cost-revenue structure. In addition, we used the standard cost price calculation of black soybeans that is used to determine the contract price in order to calculate the break-even price of black soybean at farm gate level (Indonesian Rupiah/kg). ■■ 23 semi-structured interviews with representatives of different categories of stakeholders of the partnership and regional commodity black soybean suppliers were conducted: executives of cooperatives (n=3), Unilever Indonesia-procurement manager (n=1), Unilever Indonesia Operation manager (n=1), Unilever Indonesia-Supplier development manager (n=1), Unilever Indonesia CSR managers (n=2), representatives of the seed supplier (n=1), field assistants (n=2), government extension agents (n=1), women groups (n=2), NGO (n=1), regional commodity traders (n=3) and farmers (n=5). The aim was to explore each profile and each relationship in contract terms, and all barriers, drivers and success factors, performance indicators, and future perspectives. For each stakeholder category a semi-structured questionnaire was developed (Supplementary Methods S1). All interviews were recorded and transcribed in English. A Bahasa interpreter was hired in case interviewees and participants of the workshops were not able to communicate in English. Reports of the workshops were drawn up and the interviews were transcript. ■■ A field observation to a cooperative nearby Jogjakarta was paid to get an in-depth view of the organization and the practice such as storage and sorting facilities they provide to farmers. Two executives of a farmers’ women group were interviewed to explore the role of farmers’ women in the black soybean supply chain. ■■ A meeting with a delegation of Indonesian Human Rights committee for Social Justice about the black soybean supply program arranged by Oxfam Indonesia in Jakarta. Practical toolkits from the Sunrise 2.0 research program that were used to explore the impact of the upgrading program of black soybean on smallholder livelihood, were from of the LINK (acronym from ‘LINKing’ smallholder to markets) methodology was developed by the International Center of Tropical Agriculture (available at: http://tinyurl.com/yb6gaatd):

International Food and Agribusiness Management Review

716

Sjauw-Koen-Fa et al.

■■ ■■

Volume 20, Issue 5, 2017

the business model canvas exercise to draw the famer business model during the farmers’ workshop; the score cart within the New Business Model Principles was filled out in the multi-stakeholder workshop to examine the inclusiveness of the black soybean program.

The aim of the Methodology is to foster inclusive trading relation between farmer organization and formal markets. It has been conducting action research on inclusive business model since 2007.

4. Case findings 4.1 Design of the black soybean supply chain map Figure 2 shows the black soybean supply chain map that resulted from the multi-stakeholder workshop and information collected from the interviewees. There were two supply chains of black soybeans to Unilever Indonesia: (1) the traditional supply from smallholders of other areas to regional commodity traders to Unilever Indonesia; and (2) the new developed supply chain from smallholders (member farms) to Cooperatives to Unilever Indonesia. The latest supply chain has become increasingly the supply source, while the traditional supply chain was used as a leverage to meet total black soybean demand of Unilever Indonesia. The farmer’s price and delivery conditions of the black soybeans of both supply chain are equivalent. In the present study we focused on the development black soybean supply chain of Unilever Indonesia. However, both supply chains are interrelated by the use of a similar price, product quality and delivery conditions. The developed black soybean supply chain consisted of two activities: (1) the buying processes (the axis Unilever Procurement – Cooperatives – smallholders) led by Procurement; and (2) the upgrading

Unilever Indonesia

Buying commitment, price guaranty, logistics, payments

Financial access, capacity building, operational support

CSR (foundation)

Cooperatives Breeders’ seeds

Support facilities & breeding research

Procurement Soybeans

Input package

Soybeans

Communication, TA, SAC, monitoring

Seed-supplier

Training, advisor

Partnership

Soybeans

Commodity traders

Smallholders (members)

Contract farming

Inputs & services

Local SMEs

Buying commitment, price guaranty, payments

Soybeans

Meso support

Government Plant area selection, TA, support cooperative development Input flow, support services

Output flow

Smallholders (of other areas) Meso support

Contract farming

Figure 2. Black soybean supply chain map in Java. SMEs = Small and medium-sized enterprises; TA = Technical assistance; SAC = Sustainable Agriculture Code. International Food and Agribusiness Management Review

717

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

processes (the partnership consisting of Unilever CSR – Cooperatives – Seed supplier (circle)) led by Unilever CSR, because of their mission to support Unilever Indonesia corporate sustainable development in Indonesia (Urip, 2010: 99-122). The different arrows represent the flow of products, payments and upgrading interventions provided by supply actors during the planting season. The sourcing process of black soybean started with the calculation of the required beans needed for the next season to produce sweet soy sauce. Buying conditions were: the price was guaranteed before planting and there was a commitment to buy all black soybeans that was produced and had the required quality. With this information, Unilever-CSR consulted Cooperatives and Seed-suppliers to explore how many of the demanded black soybeans could be produced by the smallholders that are small-scale paddy farmers. The outcomes of the assessment and the agreed terms were written down in a Memorandum of Understanding signed by Cooperatives and Seed supplier. Side selling by farmers was not permitted but was not penalized either. Remaining soybeans that the MNE needed for the next season came from selected regional commodity traders that operated in other areas than those operated in by the Cooperatives. These traders had a long standing supply relationship with Unilever Indonesia. During the planting seasons, field assistants of Unilever Indonesia-CSR were frequently in touch with farmers, and each month a meeting was held with field assistants and cooperatives. One month before the harvest, Unilever Indonesia paid 80% in advance on sales that allowed them to provide a cash loan to farmers before the harvest and the remaining part was paid shortly after delivery. The total estimated harvest to be delivered was determined by Cooperatives along with Unilever-CSR. After the harvest, the Cooperatives facilitated the collection of the beans, the sorting (by hired women farmers), storage and payment to farmers or farmer groups. Finally, the beans were collected by Unilever Indonesia-Procurement and transported to the soy sauce factory near Jakarta. The involvement of the government in the black soybean supply chain was on ‘arm’s length’, i.e. they were not directly involved in the partnership. This was due to the liberal economic development and international trade policy of Indonesia that allowed large imports of cheaper yellow soybeans. The critical performance indicators of the black soybean supply chain in the period 2008-2013 are shown in Table 1. The aggregated performance figures disguise, however, variances and differences between cooperatives/ regions and farmers, because the agronomical conditions and soil quality vary widely between regions. This is an important point of concern for the Unilever Indonesia to minimize supply risks, also for the smallholders, such as in 2012. Production was exceptional higher (more one third) than forecast due to favorable weather conditions. Due to the fact that Unilever Indonesia is committed to buy all the produced black soybeans. In 2013 Unilever Indonesia started with certification of black soybean farmers according companies Sustainable Agriculture Code (SAC) for black soybean production. This consisted of minimum sustainability standards regarding soil management, crop and animal husbandry, working conditions and environmental resources, which are applied to their suppliers and farmers who supply them. It was reported that in 2014 about 65% of the farmers were certified as SAC farmer. Table 1. Performance indicators of the black soybean supply chain (2007-2012) (data provided by Unilever Foundation Indonesia). Number of smallholders Yield increase on average Supply of total soybean demand Return/costs-ratio per unit (pre-calculation 2013)

5,000-8,300 360-700 kg per hectare From 20 to 60% 1.8

International Food and Agribusiness Management Review

718

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

The growth of number of smallholders participating the black soybean supply program, the yield increase on average and using pre-calculated return/cost-ratio (R/C-ratio) much larger than one for the pricing in the period 2007-2012, indicated that the black soybean case is a best practice empirical case in which Unilever Indonesia source effectively from smallholders. 4.2 Indicators for smallholder livelihood improvement The impact of the applied black soybean upgrading program on smallholder livelihood was measured in two ways (paragraph 3.2). First, Figure 3 shows the findings of the business model canvas exercise on the business model of black soybean farmers in Java as it was applied in a multi-stakeholders workshop. The found business model canvas of black soybean smallholders clearly shows that the balance of the costrevenue structure of the farmers’ business model was positive, which is an indicator of improved livelihood of the farmers. This was also confirmed by exploring the standard cost price calculation per unit of black soybean production used for the upgrading program. The outcome of the farmers’ business model was that the calculated total production costs per kg of black soybean was more than 60% of the contract price (farm gate price) of the upgrading program (2013). This means that planting black soybeans was profitable for farmers according to the farmers’ business model canvas exercise. Also during the farmers’ workshop it was confirmed that the value of the black soybean as intercrop is that it is a secure and reliable source of income. Second, Figure 4 shows that the average scores on the six inclusive business principles of the New Business Model Principles were positive. The conclusion is that the black soybean supply program was inclusive according to the New Business Model Principles. We conclude that based on the positive outcomes of the two approaches the applied black soybean sourcing model indicates a positive impact on smallholder livelihood. Partners Partners provide: • Improved seeds • knowledge and advice • extension services • financial services

Key activities On-farm: • Paddy • Black soybean • Corn • Peanuts • Chili

Off-farm labor (30-50% of income)

Value proposition • Can provide needed volumes • Reliable delivery • Quality – dry, clean, no brokens, no insects

Key resources

Relationships with cooperatives are contract based and involve service provision. Those who side-sell are blacklisted. Relationships with traders are formed around prices negotiated based on quality at harvest. Channels

• Land • Labor • Fertilizer • Pesticides • Social capital (groups) • Water • Information/knowledge • Access to finance • Equipment (need) Cost structure

Customer relationships

Farmers group: motivator picks up harvest, they pay fuel costs (or by bike where close enough) Market: bring it themselves on public transportation

Revenue

structure

Figure 3. Business model of black soybean farmers in Java. International Food and Agribusiness Management Review

719

Customers

• Unilever (for cooperatives) • Cooperatives • Cattle farmers (for rejected beans, used for cattle feed) • Commodity traders/brokers at the local market

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017 Chain wide collaboration 5 4 Measurements of outcomes

Effective market linkage

2 1 0

Fair and transparent governance

Inclusive innovation

Equitable access to services

Figure 4. Inclusiveness of the black soybean business model in Java. 4.3 Critical Success Factors of the black soybean supply chain The found CSFs of the black soybean supply chain in Java were: ■■ CSF 1: the selected farmers to grow black soybean are commercial/market-oriented small-scale paddy farmers. We found that the characteristics of paddy farmers participating in the black soybean supply program meet the characteristics of commercial farms in tight value chains according Christen and Anderson (2013). The farm size of these farmers was 0.3 hectare on average. These farmers were likely to carry out a variety of other activities. They sold almost all of the paddy on the local market and used the dry season to plant an intercrop such as chili peppers, corn and groundnuts. They were members of a cooperative and participated in farmers’ groups to cultivate black soybean as an intercrop. ■■ CSF 2: a long-term partnership was formed and a supplier development program was set up for upgrading. We found that Unilever Indonesia has set up a partnership with selected farmers’ cooperatives and seed suppliers and created a supplier development program to upgrade paddy farmers to cultivate black soybean. Unilever-Procurement led the buying processes while Unilever-CSR led the upgrading processes of the upgrading program, because upgrading small-scale paddy farmers was considered also a community development activity. This is consistent with the mission of Unilever-CSR, namely to strengthen the Unilever Indonesia’s commitment to sustainable development to maintain a competitive edge. ■■ CSF 3: the governance structure of the black soybean supply chain is of a captive type based on a cooperative ‘buyer-seller’ relationship for black soybean supplier development. The found governance structure of the black soybean supply chain can be classified according the classification of Gereffi et al. (2005) a captive type of governance, because the supply chain were locked in by Unilever Indonesia. The relationship between partners is based on a cooperative instead of a short-term transactional buyer-seller relationship, because of the long-term business perspective of the upgrading program. The communication within the black soybean partnership is a two-way and open system. There are regular meetings, visits to farmers during the planting period, and standard cost price calculations are used during the negotiations. The local government did not participate in the black soybean partnership consisting of Unilever Indonesia, the cooperatives and the University of Gadja Mada (seed supplier).

International Food and Agribusiness Management Review

720

Sjauw-Koen-Fa et al.

■■

Volume 20, Issue 5, 2017

SF 4: cooperatives were empowered in order to strengthen the vertical coordination of the black C soybean supply chain. We found that the Cooperatives participating in the upgrading program of the black soybean supply chain were key contracting partners representing the member farmers. They play a central role in facilitating the flow of information, inputs, provision of upgrading support, and collection, sorting, storage and delivery of the contracted black soybeans. Farmers are aggregated in groups in order to communicate effectively and to lower transactional costs. Unilever Indonesia-CSR supports the improvement of management capabilities and the financial access of the cooperatives to strengthen vertical coordination in the black soybean supply chain. CSF 5: Unilever Indonesia offers a prepaid system before the harvest and a buying commitment of all produced soybeans. The Unilever Indonesia offered price guarantee before planting and a buying commitment of all black soybeans harvested. Hence that Unilever Indonesia is not a financial institution. However, these buying conditions eased the credit demand and lowered the risks of the black soybean smallholder farmers. Farmers received 80% in advance on sales one month before the harvest, while the remaining 20% was paid within two weeks of delivery to the factory. This payment system combined with the buying commitment of the black soybeans produced eased the credit demand, lowered costs and reduced the risks of the black soybean smallholders. In addition, it was noticed in the farmers’ workshop and during the interviews that this finance system was of great value especially to small-scale farmers because it is a secured source of income. CSF 6: presence of a clear smallholder sourcing strategy and commitment to secured sustainable black soybean supply combined with a proactive CSR strategy from a business perspective. The CSR strategy was established to strengthen Unilever Indonesia’s commitment to sustainable development in Indonesia. The final aim of this CSR strategy is to maintain competitive advantage of Unilever Indonesia, while continuously ensuring the business commitments to community building, creation of employment and wealth, as well as caring for the environment. In practice this means that Unilever Indonesia-CSR supports the sustainability performance of the projects of the business units of Unilever Indonesia. The development of an alternative black soybean supply chain fits into the sustainable business development approach of the Unilever Indonesia. CSF 7: use of cross-functional sourcing teams consisting of Procurement and CSR with clear division of tasks, resources and incentives for effective black soybean supplier development, but both are focus on the same Unilever Indonesia inclusive goal. Within the Unilever Indonesia, close coordination between Unilever Indonesia-CSR and Unilever Indonesia-Procurement staff enabled them to run the upgrading program in a smooth way. Unilever Indonesia-CSR played a program management role in the upgrading processes from input supply to the sorting out stage of the soybeans harvested. Procurement steered buying processes in the collection and logistics of the soybeans from the cooperatives to the soy sauce factory, and the payment of the delivered black soybeans. Unilever Indonesia-Procurement as well as the Unilever Indonesia-CSR were focused on the same strategic corporate goal, i.e. the development of a sustainable smallholder supply chain to strengthen Unilever Indonesia market position in Indonesia and contributing to smallholders’ livelihood. There was a clear division of tasks between the Unilever Indonesia-CSR, which is a companies’ foundation, and the Unilever Indonesia-Procurement, which is a department of Unilever Indonesia. However, both had compatible tasks, complementary competences, resources/ funds and incentives, and there was an open communication and understanding between them.

The overall finding of the case study is that the CSFs in the black soybean case were in line with the CSFs of the 3S-model. However, differences found were the role of Unilever Indonesia in providing affordable farm financing (CSF 5) and the business form Unilever Indonesia-CSR.

International Food and Agribusiness Management Review

721

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

4.4 Matching of the black soybean map with the 3S-model frame To illustrate the applicability and the dynamics of the 3S-model we have matched the empirical black soybean supply chain map (Figure 2) with the of the 3S-model (Figure 1). The result is shown in Table 2. We conclude that the black soybean supply map is in general in line with the framework (building blocks) of the 3S-model with related CSFs.

5. Discussion and conclusions The purpose of present article was to explore the applicability and the dynamics of the sustainable sourcing model for sustainable smallholder supply (3S-model) based on a best practice case as empirical background. In this model the business perspective (to secure stable access to sustainable smallholders’ commodity supply) and CSR perspective (improvement of smallholder livelihood) are integrated. The overall finding of the study is that the dynamics of the Unilever case can be understood with the help of the 3S-model. Similarities include the use of a partnerships model for upgrading, a captive governance structure, and the existence of a clear proactive and committed corporate sustainable smallholder sourcing strategy. At the same time we also found differences regarding the role of the ULI in farm financing, the business form of suppliers and the cross functional sourcing team that influence the concept of the 3S-model. The following lessons are learned from these differences. First, regarding building partnerships for upgrading (CSF 2): although the 3S-model for sustainable smallholder sourcing has a business perspective, input suppliers can also be public organizations instead of just private companies alone. We think that in the pilot and start-up phase of a smallholder supply development program, this isn’t a constraint, because of the supportive character of the program. However, in the scale up phase of the supply chain, the limits of a public organization in a business setting could be more pressing, because of the non-commercial orientation of the public organization to grow together with the business in a competitive global market environment (e.g. provision of long term investment capital and profit making). Table 2. Black soybean supply chain map according to the 3S-model.1 Building blocks of the 3S-model

Black soybean supply chain map

F&A MNE

ULI represented by: • Procurement (buying black soybeans from cooperatives). • ULI-CSR (a company’s foundation, leading upgrading processes). Intermediary Cooperatives: representing farmers in the partnership, organizing member farmers for production, facilitating upgrading and delivering processes of black soybean. Smallholders Commercially oriented small-scale paddy farmers growing black soybeans on a contractual base. Partnership model Consisting of MNE, cooperatives and seed supplier (a university), using a farmer development program for upgrading. Government, NGOs, public bodies, private • Government has been involved on arm’s length. foundation, social investors: (control variables) • NGO empowered women group. • Local SME’s provided farm services. Smallholder livelihood improvement Positive indication (e.g. R/C-ratio>one) 1

F&A MNE = food and agribusiness multinational enterprises; ULI = Unilever Indonesia; CSR = corporate social responsibility; NGO = Non-governmental organization; SME = Small and medium-sized enterprises; R/C-ratio = return/cost-ratio.

International Food and Agribusiness Management Review

722

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

Second, regarding providing access to finance to smallholders (CSF 5): F&A MNEs can play an important role to lower financing cost and risks for smallholders by offering buying commitments and price guarantees and down payments before planting, although they are no credit institutions themselves. These facilities can attract rural banks to provide farm financing to smallholders, because such buying guarantees lower financing risks for bank credits (smallholders in developing regions often miss reliable collaterals, land titles and professional book keeping over a long period of time). Moreover, this type of producers’ contractual financial relationship between MNE – Intermediary – Smallholders in the value chain could also provide opportunities for value chain financing by banks. F&A MNEs are then taking the lead (as lead contractor) in the value chain in the farm financing. The concept of value chain financing can be defined as financial services and products flowing to and/or through value chain participants in order to address and alleviate driving constraints to growth and competiveness of that value chain (e.g. Miller and Jones, 2010). Third, regarding the sourcing organization of the F&A MNE for governing long-term smallholder supplier development programs effectively (CSF 7): because there are process-related, cognitive, structural and incentive-related challenges to overcome, one of the most critical points is the refining of the traditional role and capabilities of Unilever Indonesias’ CSR and Procurement departments and their activities within the F&A MNE. The use of cross-functional teams consisting of Procurement and CSR staff and resources (CSF 7 of the 3S-model) as we found in the black soybean case. In this case, Unilever Indonesia-CSR related activities are employed via the company’s foundation, i.e. an independent non-profit organization with a corporate sustainability mission and resources, that works at the Bottom of the economic Pyramid to strengthens a company’s license to operate, rather than a department of the core business organization of F&A MNEs. The advantage of this internal organization within F&A MNEs is that CSR goals can be combined with core business sourcing goals of the F&A MNE, i.e. to make a company (more) business inclusive. Based on the lessons learned from the black soybean case study the following CSFs have been sharpened: ■■ CSF 5: ‘Providing access to finance to smallholder’ has been change into ‘Lowering financing costs and risks of smallholders, because MNEs are no credit institutions. However, they can offer to buy the produce at a guaranteed price and do down payment before planting to farmers to ease smallholders’ financing needs. ■■ CSF 7: ‘Use of cross-functional sourcing teams’ has been changed into ‘Use of cross-functional sourcing teams that combine corporate sourcing and CSR goals’, because a common focus on the corporate sourcing goals of the team is more important than the organization a team member. However, there are questions to be discussed left. First, despite a positive indication of the contribution to the smallholder sourcing model livelihood, the question remains whether smallholders do get an equitable piece of the cake, i.e. do they get the real price for their produce that covers at least all costs and risks? A clear answer to this question can hardly be given because of several reasons, for instance because of the business development and learning character of supplier development programs. Primary sources for raising farm income are improvement of the productivity (higher yields per hectare) and product quality (higher price). Moreover, smallholders in developing economies are mostly not familiar with cost price calculations and bookkeeping, and lack price and market information or are surrendered to the practices of a middlemen (e.g. London et al., 2010). Second, the question is what Unilever would do in case there is an excessive supply of black soybeans? The consideration of Unilever to develop an own smallholder black soybean supply chain has been driven by the expectation that demand of black soybean will exceed supply on the one hand. On the other hand, it was an opportunity to express Unilever’s CSR commitments in community development building, creating employment and livelihood improvement of small-scale farmers in Indonesia (Urip, 2010). Accordingly, Unilever offered price guaranty before planting and a buying commitment of all black soybeans harvested to

International Food and Agribusiness Management Review

723

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

farmers. The agreed terms were written down in a Memorandum of Understanding. These two considerations compelled Unilever to fulfill the obligations avoiding reputational risks. However, in 2012 the production of the contracted black soybeans was 35% higher than the forecast, due to favourable weather conditions. Due to the fact that Unilever Indonesia was committed to buy all the harvested black soybeans the storage capacity proved insufficient and question raised about the accuracy of the production forecasting system. The third questions is, what will happen with the black soybean sourcing model in case of severe downturns of Unilever or in case of crop failure because severe drought or extreme soya price falls? On the one hand, there is no guaranty that Unilever will never change it strategy, such as in case of a downturn, because firms need to be profitable in order to survive and grow in a challenging global food system. On the other hand, the best guaranty that F&A MNEâ&#x20AC;&#x2122;s take their responsibility serious is the level of their CSR commitment. In case of Unilever, the company has a proactive CSR strategy, implying that is has the capacity, procedures, arrangements, behavior patterns, sustainable codes and standards to anticipate on social issues, and a committed top management (Van Tilburg et al., 2012). This is a precondition for long-term investment in smallholder supplier development programs in order to secure a (long-term) sustainable and more equitable commodity supply from a business perspective (CSF 6). Although we have illustrated in this study that from a business perspective MNEs can include smallholders in a sustainable and more equitable way in high value-adding supply chains from a business perspective, the overall effect of F&A MNEs in solving global food security and sustainable development challenges must not be overestimated. Constraints are among other things their short-term commercial and business model orientation, and their relatively small scale in the global food system compared to the estimated 200 million small-scale commercially oriented farmers operating at the Bottom of the Pyramid (Christen and Anderson, 2013), i.e. F&A MNEs cannot do it alone (e.g. World Economic Forum, 2011). Nonetheless, we believe that the fact that the value chain analysis of the black soybean supply chain is commissioned by both an international NGO (OXFAM) and a corporate actor (Unilever), provides a unique case for studying integration of business and CSR perspectives in smallholder inclusion in high value-adding supply chains, which is the underlying basic principle of the 3S-model with CSFs and drivers of supply chain dynamics. The added value of this research is, especially to management scholars, that the black soybean case illustrates that MNEs can include smallholders in a sustainable and more equitable way in high value adding supply chains. It can help in (re-)designing (conventional) sustainable smallholder sourcing strategies. However, there are limitations to this study because the findings were based on a single case. Food sectors, geographical conditions, the political context and sourcing strategies of F&A MNEs may differ. Accordingly, we recommend further case study research in order to further confirm, modify, or fine-tune the 3S-model.

Acknowledgements Authors are indebted to Unilever Penduli Indonesia and Procurement PT Unilever Indonesia for facilitating the field research. They want to specially acknowledge the interviewees, Andre Setiawan, Syaiful Hakim, Sinta Kaniawati, Theresa Wuryanti and Hanah Schiff (epven, Ltd.) for their kind cooperation. They are also grateful to the Sunrise 2.0 project team, especially to Justin Tait (Sunrise), Ximing Hu (Unilever) and Juni Sul (Oxfam) that commissioned the value chain analysis.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2016.0171. Methods S1. Topics of the semi structured interviews. International Food and Agribusiness Management Review

724

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

References Alvarez, G., C. Pilbeam and R. Wilding. 2010. Nestlé Nespresso AAA sustainable quality program: an investigation into the governance dynamins in a multi-stakeholder supply chain network. Supply Chain Management 15(2): 165-182. Blackman, A. and J. Rivera. 2011. Producer-level benefits of sustainable certification. Conservation Biology 25: 1176-1185. Blok, V., A.R. Sjauw-Koen-Fa and S.W.F. Omta. 2013. Effective stakeholder involvement at the base at the pyramid: the case of Rabobank. International Food and Agribusiness Management Review 16(A): 39-44. Busch, L. and C. Bain. 2004. New! Improved? The transformation of the global agrifood system. Rural Sociology 69(3): 321-346. Caroll, A.B. 1979. A three-dimensional conceptual model of corporate performance. Academy of Management Review 4(4): 497-505. Carter, C.R. and D.S. Rogers. 2008. A framework of sustainable supply chain management: moving toward new theory. International Journal of Physical Distribution and Logistics Management 38(5): 360-387. Chalmers, G., M.D. Wenner, P. Tiffen and E. Gálvez. 2006. Lessons learned in agricultural value chain financing. Paper presented on the seminar ‘Agricultural value chain finance’. FAO, and Academia de Centroamérica, San José, Costa Rica. Chambo, S.A. 2009. Agricultural co-operatives: role in food security and rural development. UN expert group meeting on co-operatives April 28-30 2009, UN, New York, USA. Available at: http://tinyurl. com/yanfdnud. Chen, I.J. and A. Paulraj. 2004. Towards a theory of supply chain management: the constructs and measurements. Journal of Operations and Management 22: 119-150. Christen, R.P. and J. Anderson. 2013. Segmentation of smallholders households: meeting the range of financial needs in agricultural families. Forum note 85, Consultative Group to Assist the Poor (CGAP), Washington, WA, USA. Dartanto, T. and M. Usman. 2011. Volatility of world soybean prices, import tariffs and poverty in Indonesia: a CGE-Microsimulation analysis. The Journal of Applied Economic Research 5: 139-165. Dragusanu, R., D. Giovannucci and N. Nunn. 2014. The economics of fair trade. Journal of Economic Perspectives 28(3): 217-236. Driedonks, B.A., J.M.P. Gevers and A. van Weele. 2014. Success factors for sourcing teams: how to foster sourcing team effectiveness. European Management Journal 32(2): 288-304. Dyer, J.H. and H. Singh. 1998. The relational view: cooperative strategy and sources of inter organizational competitive advantage. Academy of Management Review 23(4): 660-679. FAO. 2009. Global agriculture towards 2050. High-level expert forum 12-13 October. FAO, Rome, Italy. Gereffi, G., J. Humphrey and T. Sturgeon. 2005. The governance of global value chains. Review of International Political Economy 12(1): 78-104. Gold, S., R. Hahn and S. Seuring, 2013. Sustainable supply chain management in the ‘Base of the Pyramid’ food projects – A path to triple bottom line approaches for multinationals? International Business Review 22: 784-799. Hahn, R. and S. Gold. 2013. Resources and governance in ‘base of the pyramid’-partnerships: assessing collaborations between business and non-business actors. Journal of Business Research 67: 1321-1333. Hahn, C.K., C.A. Watts and K.Y. Kim. 1990. The supplier development program: a conceptual model. Journal of Purchasing and Materials Management 26: 2-7. Hasibuan-Sedyono, C. 2010. Growing together in partnership: the case of Unilever and the black soybean farmers in Indonesia, Binus Business School, Global Compact Network, Jakarta, Indonesia. Henson, S. and J. Humphrey. 2009. The impacts of private food safety standards on the food chain and on public standards-setting processes. Paper prepared for FAO/WHO 2009. Available at: http://tinyurl. com/yceh5wtz. Humphrey, J. 2004. Upgrading in global value chains. Working Paper No 28. International Labour Office, Geneva, Switzerland. International Food and Agribusiness Management Review

725

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

Humphrey, J. and H. Schmitz. 2002. How does insertion in global value chains affect upgrading in industrial clusters. Regional Studies 36(9): 1017-1027. IFC and GPFI. 2011. Scaling up access to finance for agricultural SMEs policy review and recommendations. IFC, Washington, WA, USA. Jaffee, S. and S. Henson. 2004. Standards and Agro-food exports from developing countries: rebalancing the debate. World Bank Policy research working paper 3348. Available at: http://tinyurl.com/ybyorlcf. Krause, D.R. and L.M. Ellram. 1997. Critical elements of supplier development: the buying perspective. European Journal of Purchasing and Supply Chain management 3(1): 21-31. Landros, R. and R.M. Monczka. 1989. Cooperative buyer/seller relationships and a firmâ&#x20AC;&#x2122;s competitive posture. Journal of Purchasing and Material management 25(3): 9-18. London, T., R. Anupindi and S. Sheth. 2010. Creating mutual value: lessons learned from ventures serving base of the pyramid producers. Journal of Business Research 63: 582-594. Maignan, I., B. Hillebrand and D. McAlister. 2002. Managing socially-responsible buying: how to integrate non-economic criteria into the purchasing process. European Management Journal 20(6): 641-648. Miller, C. and L. Jones. 2010. Agricultural value chain finance: tools and lessons. Food and Agriculture Organization of the United Nations and Practical Action Publishing, Rome, Italy and Warwickshire, UK. Mohr, J. and R. Spekman. 1994. Characteristics of partnership success: partnership attributes, communication behavior, and conflict resolution techniques. Strategic Management Journal 15: 135-152. Monczka, R.M., K.J. Petersen, R.B. Handfield and G.L. Ragatz. 1998. Success factors in strategic supplier alliances: the buying company perspectives. Decision Sciences 29(3): 553-577. Olsen, M. and E. Boxenbaum. 2009. Bottom-of-the-Pyramid: organizational barriers to implementation. California Management Review 51(4): 100-125. Onumah, G., E. Junior, R. Davis, U. Kleih and F.J. Proctor. 2007. Empowering smallholder farmers in markets: changing agricultural marketing systems and innovative responses by producers organizations. ESFIM Working Paper 2. IFAD, CTA and AgriCord, Rome, Italy. Perez-Aleman, P. and M. Sandilands. 2008. Building value at the top and bottom of the global supply chain: MNC-NGO partnership and sustainability. California Management Review 51(1): 24-49. Reardon, T., C.B. Barrett, J.A. BerdeguĂŠ and J.F.M. Swinnen. 2009. Agrifood industry transformation and small farmers in developing countries. World Development 37(11): 1717-1727. Riisgaard, L., S. Bolwig, F. Matose, S. Ponte, A. du Toit and N. Halberg. 2010. Integrating poverty and environmental concerns into value-chain analysis: a strategic framework and practical guide. Development Policy Review 28(2): 195-216. Rivera-Santos, M. and C. Rufin. 2010. Global village vs small town: understanding networks at the base of the pyramid. International Business Review 19: 126-139. Rossignoli, C.M. and R. Moruzzo. 2014. Retail power and private standards in the agri-food chain. Agroecology and Sustainability Food Systems 38: 1108-1124. Ruben, R. and R. Fort. 2011. The impact of fair trade certification for coffee farmers in Peru. World Development 40(3): 570-582. Rueda, X. and E.F. Lambin. 2013. Responding to globalization: impact of certification on Colombia smallscale coffee growers. Ecology and Society 18(3): 21. Sjauw-Koen-Fa, A.R. 2010. Sustainability and security of the global food supply chain. Economic Research Department of the Rabobank Group, the Netherlands. Available at: https://www.academia. edu/16331687. Sjauw-Koen-Fa, A.R. 2012. Framework for an inclusive food strategy. Co-operatives key for smallholder inclusion into value chains, Economic Research department, Rabobank Group. Available at: http:// tinyurl.com/ycbgve74. Sjauw-Koen-Fa, A.R., V. Blok and S.W.F. Omta. 2016. Critical success factors for smallholder inclusion in high value-adding supply chains by Food and Agribusiness Multinational Enterprises. International Food and Agribusiness Management Review 19(1): 83-112. Spence, L. and M. Bourlakis. 2009. The evolution from corporate social responsibility to supply chain responsibility: the case of Waitrose. Supply Chain Management 14(4): 291-302. International Food and Agribusiness Management Review

726

Sjauw-Koen-Fa et al.

Volume 20, Issue 5, 2017

Tait, J. 2015. Project Sunrise: final report. Unilever and Oxfam UK. Available at: http://tinyurl.com/y8h4tsqf. Torero, M. 2011. A framework for linking small farmers to markets. Conference on â&#x20AC;&#x2DC;New directions for smallholder agricultureâ&#x20AC;&#x2122; 24-25 January 2011, IFAD, Rome, Italy. Available at: http://tinyurl.com/ yarnfgzz. Trent, R.J. and R.M. Monczka. 1994. Effective cross-functional sourcing teams: critical success factors. International Journal of Purchasing and Material Management 30: 2-11. Trent, R.J. and R.M. Monczka. 2002. Pursuing competitive advantage through integrated global sourcing. Academy of management executive 16(2): 66-80. Trienekens, J.H., P.M. Wognum, A.J.M. Beulens and J.G.A.J. van de Vorst. 2012. Transparency in complex dynamic food supply chains. Advanced Engineering Informatics 26: 55-65. Urip, S. 2010. CSR strategies: corporate social responsibility for competitive edge in emerging markets. John Wiley (Asia) Pte. Ltd., Singapore. Van Tilburg, R., R. van Tulder, M. Francken and A. da Rosa. 2012. Duurzaam ondernemen waarmaken, Stichting Management Studies, van Gorcum, Assen, the Netherlands. Watts, C.A., K.Y. Kim and C.K. Hahn. 1995. Linking purchasing to corporate competitive strategy. Journal of Supply Chain Management: 31: 2-8. World Economic Forum. 2011. Realizing a new vision for agriculture: a road map for stakeholders. WEF, Geneva, Switzerland.

International Food and Agribusiness Management Review

727

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2016.0093 Received: 1 May 2016 / Accepted: 14 April 2017

Understanding the determinants of adoption of enterprise resource planning (ERP) technology within the agrifood context: the case of the Midwest of Brazil RESEARCH ARTICLE Caetano Haberli Jr a, Tiago Oliveirab, and Mitsuru Yanazec aProfessor,

Superior School of Advertising and Marketing (ESPM), Universidade de São Paulo (USP), Butantã, São Paulo 03178-200, Brazil; PhD Candidate at IMS, Universidade Nova de Lisboa, Campus de Campolide, 1070-312 Lisbon, Portugal and School of Communication and Arts, Universidade de São Paulo (USP), São Paulo, Brazil; Owner, Agroipes, Institute of Research and Sectorial Studies, Rod. Sc 405 4850, Florianopolis, Santa Catarina 88065-000, Brazil bAssistant

Professor and Coordinator, NOVA Information Management School (NOVA IMS) Universidade Nova de Lisboa, 1070-312 Lisbon, Portugal

cProfessor

of Marketing & Communication and Coordinator, School of Communications and Arts, Universidade de São Paulo (ECA/USP), Butantã, São Paulo 03178-200, Brazil

Abstract The object of this study is to investigate the determinants of adoption of Enterprise Resource Planning (ERP) technology in agricultural farms located in the Central-West region of Brazil. The data was collected from 200 in-depth interviews with soy, corn and cotton farmers from the State of Mato Grosso, Brazil. Structural Equations methodology was used to analyze the data and hypothesis. The conceptual model was proposed by combining Diffusion of Innovations and Technology-Organization-Environment theories. The results provide information to agribusiness owners, managers and administrators to promote and incentivize the use of ERP. Politicians and farmers can evaluate each scenario and support their political and administrative decisions through the evaluation of socioeconomic and environmental performances of agricultural exploration as a result of technological innovation. This leads to a need for an analytical tool for the farmers, with the objective of supporting the adoption of optimized ERP for agri-food activities. Keywords: enterprise resources planning, ERP technology, management models, agribusiness JEL code: Q0 Corresponding author: caetanohaberlijr@me.com

729

Haberli et al.

Volume 20, Issue 5, 2017

1. Introduction Although the Brazilian agricultural production represents a significant share of total world food production, Brazilian farms do not have a well-organized business structure, neither do they have adequate control of their production process to reach a new level of efficiency and effectiveness. This is due to a lack of Enterprise Resource Planning (ERP) in the farms. As a result, it can cause considerable production loss (Orsi et al., 2017). The decision makers in the agricultural sector deal with volatile and risk variables such as management of physical storage, controlling transportation costs, exposure to climate issues, vulnerability to weeds, pests and diseases. The ERP system can minimize the risks on decisions taken on this environment. The purpose of this study is to understand the determinants of the adoption of ERP as a management model compatible with the farm’s needs, and also to evaluate the benefits of this model to provide improvement in the competitiveness among farms. The only few research found on the topic explores daily operational routines focusing essentially on production and productivity. It is difficult to find studies regarding the applicability of ERP through an organizational and processual point of view, specifically regarding the direct effects of the use of this technology and some aspects of this business. Brazil is a major world food producer (Table 1). This scientific study evaluates the best practices for ERP in Brazilian farms to uphold the country’s position among the main world producers of protein, fiber and energy. Studies about the conceptual model of future farm management information system debates how farmer’s paradigms are changing the management tasks in order to achieve economic sustainability and interaction to the environment (Sørensen et al., 2010). In 2011, Sørensen et al. developed a study to support and guide the functional requirements for a future management information system. The paper concerning ERP in agriculture, ‘Lessons learned from the Dutch horticulture’, evaluates and explores the experiences of the applicability of ERP in agri-food companies (Verdouw et al., 2015). In a paper about farm management information system called, ‘Current situation and future perspectives’, the authors acknowledge that information systems in the farms evolved from only keeping records to more complex systems supporting production management (Fountas et al., 2015). In order to fulfill the increased demands from partners, consumers, government organizations and food processing companies, farms need to develop a knowledge-based economy which shares information and organized data (Wolfert et al., 2010). These needs inspired us to establish an integrated research model by gathering the determinants of adoption of an adequate ERP and combining the Technology Organization Environment (TOE) (Tornatzky and Fleischer, 1991) and Diffusion of Innovation Theory (DOI) framework (Rogers, 1993). The purpose of this model is to provide information to decision-makers (i.e. politicians and farmers) and to encourage the evaluation of the farm’s results based on its resource planning choices. This motivation is the result that can be seen on Table 1. Brazil position in the world ranking of food producing (adapted from USDA and Agri-Business Sector Value, 2015, 2016; https://www.fas.usda.gov/data).

Brazil export market share (%) Brazil production market share (%) World export ranking – Brazil World production ranking – Brazil

Orange Sugar Coffee Soy Juice

Beef

Chicken Corn

Pork Cotton

77 54 1st 1st

21 16 1st 2nd

36 11 1st 3rd

8 3 4th 4th

47 22 1st 1st

27 32 1st 1st

42 31 1st 2nd

International Food and Agribusiness Management Review

730

24 9 2nd 3rd

13 6 3rd 5th

Haberli et al.

Volume 20, Issue 5, 2017

the performances from innovations on the socioeconomical, environmental and agricultural exploration. (Janssen and Van Ittersum, 2007). To evaluate the research model and investigate the determinants, we collected the data of 200 soy, corn and cotton producers in the Mato Grosso state. Therefore, this study presents a holistic evaluation of the determinants to make a theoretical contribution to the adoption of ERP for agricultural farms.

2. Agri-business in Brazil and background about Enterprise Resource Planning Brazilian agri-business has been improving in the last decade. Brazilian farmers excelled in production techniques and overcame technological issues to reach high productivity levels comparable to the larger world food producers. This progress can be certified by looking at Brazil’s position on the world ranking of food production and food exports in 2013/2014 (Table 1). ERP Systems require simultaneous changes in the business process, information sharing and the use of complex data (Amoako-Gyampah and Salam, 2004). They process information from different functional areas and integrate them to identify and incorporate the best business practices (Kumar and Van Hillegersberg, 2000). An understanding of the processed and integrated information from different functional areas (Madapusi and D’Souza, 2012) can support ERP development for agribusiness companies. Finding the critical elements of the simultaneous changes that are going on and identifying the success drivers can define a different approach to implement ERP (Zhang et al., 2005). According to Ruivo et al. (2012), the implementation of ERP allows companies to increase its value, achieve trade efficiency, enhance internal collaboration and improve business analysis which are important determinants in this process. Therefore, we will analyze how the ERP systems can contribute to agricultural production organizations. Agribusiness is part of a globalized economic environment. Universal operations are indispensable to the integration of providers, partners and customers (Yusuf et al., 2004). The increasing necessity for food production has led to new research in order to optimize its productivity. This has been achieved by the increased use of technology which incorporates the ideal combination of specific software and hardware (Orlovski et al., 2012). Efforts to adopt information technology (IT) and systems such as ERP support business integration and decision-making (Yusuf et al., 2004). Management principles and techniques, sustainability and evaluation of the farm project, management support, process reengineering, consulting and budget services are crucial elements for ERP’s implementation (Ehie and Madsen, 2005). Considering the innovation process has two main stages of adoption and implementation (Damanpour and Schneider, 2009), the overall results can be significantly increased by a combination of organizational factors as well as the use of technology and innovation (Karimi et al., 2007). Benefits obtained from the automation of business processes and the use of the ERP system improve decison-making in all organizational levels (Velcu, 2010) which applies to agribusiness companies. However, it is necessary to face and accept the issue that managers still do not have the knowledge and technical skills to handle the system and processes which can produce inaccurate data gathering and some mistrust regarding the use of this technology (Hakim and Hakim, 2010). ERP implementation is a slow process and demands resources (Tsai et al., 2011). An understanding of the process and information packages among functional areas is necessary. Although ERP implementation represents a significant investment (Madapusi and D’Souza, 2012; Zeng and Skibniewski, 2013), it can have an important impact in the organization’s operational and process performances (Madapusi and D’Souza, 2012). Besides the high investment level, implementation risks are also high. Countless complex elements in the organization can interfere in the implementation such as

International Food and Agribusiness Management Review

731

Haberli et al.

Volume 20, Issue 5, 2017

user’s low-level of acceptance of the technology, changes in the information environment, instability in the management environment and the complexity of the ERP system (Hung et al., 2012). On the other hand, after a successful ERP implementation, it is possible to observe relevant effects in the social capital which is the (1) learning opportunity; (2) desire to learn; and (3) increase in abilities. Those results can be attributed to the complexity of the ERP system which compels users to solve challenges, acquire knowledge and develop new abilities to run tasks and make decisions (Ruivo et al., 2012). It is important for the managers to set the priorities and goals to be reached in each ERP implementation phase which will contribute to the maximization of the whole process (Ram et al., 2013). The improvements in process efficiency achieved by the ERP system can deliver the competitive advantage needed by organizations in a global market context where their strategies are affected by many different elements especially the competing companies (Rouyendegh et al., 2014). The ERP implementation process and the achievements reached by it are distinct for each organization (Rouyendegh et al., 2014). According to studies of Small and Medium Enterprises (SMEs), improvements in strategic planning for Information Systems (IS) helps the companies to understand the benefits that the ERP system can offer (Zach et al., 2014). It is also important to state that ERP experts are not easy to find in the market which can deliver an extra challenge to SMEs as they need to train and capacitate their employees on the use of this tool (Esteves, 2014). A particular study of SMEs in Portugal shows that ERP implementation was a determinant for the company’s performance in management, finance and tax accounting as well as the company’s management control (Ruivo et al., 2014). Although it is possible to find good results in the field, a considerable number of companies do not reach the expected goals after implementing the ERP system, These failures can be accounted for by the improper use of the system and its full resources (Chou et al., 2014; Ruivo et al., 2012). In many cases, the use of ERP does not achieve business process control, costs reductions, increase in profits and an influence on the companies key’s performance indicators (Gajic et al., 2014). Therefore, organizations must find ways to simplify the use of the system because once the system reaches its efficiency, it will provide the direct learning ability and desire to the users (Chou et al., 2014). It is important to understand that any progress on the use of the strategic assets will also contribute directly to business development (Wood et al., 2014).

3. Methodology The study is focused on the innovation adoption phase. Two theories are usually used to explore similar cases in organizations of distinct nature; DOI (Rogers, 1993) and TOE Framework (Tornatzky and Fleischer, 1991). TOE Framework identifies the process used by a company to adopt and implement innovations by considering the technological, organizational and environmental context (Tornatzky and Fleischer, 1991). The technological context embraces relevant internal and external technology as tools and processes while the organizational context is related to the company’s features and its assets such as company size, hierarchy, process procedures, administrative structure, human resources, extra resources and employee connections. The environmental context is influenced by market elements such as the size and structure of the industry, company’s competitors, macroeconomics and the regulatory environment. All three contexts can present opportunities and threats which influence how a company sees, searches and adopts new technologies. On the other hand, DOI Theory studies the spread of innovations and how it is communicated through channels over time and inside a particular social environment (Rogers, 1993). Each individual is deemed to hold different levels of innovation acceptance. This paradigm of diffusion was spread during the fifties and sixties among sociology researchers of rural areas (Valente and Rogers, 1995). International Food and Agribusiness Management Review

732

Haberli et al.

Volume 20, Issue 5, 2017

DOI and TOE Theories have been widely used in studies concerning the adoption of innovative technology and they have consistent empiric support (Oliveira et al., 2014). The benefits of merging TOE concepts to reinforce the DOI theory are already well recognized (Hsu et al., 2006). Using DOI and TOE together helps to provide a more comprehensive perspective about technology adoption including the technological context aspects, organizations and external environment (Zhu et al., 2006a). DOI and TOE Theories (Figure 1) complement each other successfully (Park et al., 2012). This paper presents a conceptual model combining DOI and TOE and also includes nine constructs. Based on the DOI theory, the constructsâ&#x20AC;&#x2122; Relative Advantage (RA), Complexity (Cx), Compatibility (Cp) and Cost Savings (CS) were selected. The first three (RA, Cx and Cp) are attributes of innovation. Technology Readiness (TR), Top Management Support (TMS), Farm Size (FS), Competitive Pressure (CP) and Regulatory Support (RS) are constructs used from the TOE theory. The TR construct is related to the technological context, TMS and FS are related to the organizational context and CP and RS are related to the environmental context. Rogers (1993) addresses 5 adoption factors, i.e. relative advantage, compatibility, complexity, trialability and observability. The trialability and observability are not widely used in IT innovation studies (Chong et al., 2009). Following the general orientation of the research IS, we disconsidered these two attributes because they are not relevant to the ERP technology (Oliveira et al., 2014). In many IT innovation studies, trialability and observability are excluded because they are not consistently related to the diffusion process of innovation (Martins et al., 2016). Rogers (1993) defines Innovation Relative Advantage as the degree in which innovation is understood as a better option than the idea it is replacing at that moment. Studies confirmed that Relative Advantage

Technology context Innovations characteristics

Technology readiness (TR)

Relative advantage (RA) Complexity (CX)

Organizational context

ERP adoption

Top management support (TMS)

Compatibility (Cp)

Farm size (FS)

Cost savings (CS)

Environmental context

DOI

Competitive pressure (CP) Regulatory support (RS)

TOE

Figure 1. Research Model combining Technology, Organization and Environment (TOE) Framework and Diffusion of Innovation Theory (DOI). ERP = Enterprise Resource Planning. International Food and Agribusiness Management Review

733

Haberli et al.

Volume 20, Issue 5, 2017

is a significant variable and is positively related to innovation adoption (Premkumar and Roberts, 1999). Innovations that present clear benefits on creating strategic efficiency (i.e. the increase in the number of prizes received for harvest or credit; the anticipation of business) and operational efficiency (i.e. the reduction on expenses funding) have greater chances of adoption. If the benefits of ERP technology exceed the benefits of current practices and procedures, the adoption of ERP technology would be positively influenced. H1(+):Relative Advantage has a positive influence on ERP adoption. Complexity is the degree that an innovation is considered to be relatively hard to understand and use (Rogers, 1995). There is a better chance of approval if new technology is integrated and assimilated in business operations during the implementation phase. ERP can also gather real-time information to support main decisions in complex operations. However, its Complexity creates some doubts about its implementation and for this reason, it decreases the chances of approval. Therefore, Complexity is negatively associated to adoption: H2(-): Complexity has a negative influence on ERP adoption. Compatibility is the perception and degree of alignment with previously established values (Valente and Rogers, 1995). This is an important determinant of innovation adoption (Tornatzky and Klein, 1982). ERP adoption can support high risk decisions by anticipating the purchase of an input material such as seeds and fertilizers, the optimum timing to acquire defense products for crops considering historical weather data and forecasts which have an impact on pests and crop diseases, as well as the sale and production of crops considering macroeconomical forecast. â&#x20AC;&#x2DC;The main motivation for bringing this hypothesis is because the industry trend on operating in a volatile and high risk environment (Xouridas, 2015). H3(+):Compatibility has a positive influence on ERP adoption. Innovation adoption which leads to Cost Savings are considered good for the company. If the cost can be controlled and accounted for, there is a higher likelihood of companies adopting the technology (Tornatzky and Klein, 1982). Cost Savings is verified as a relevant variable for innovation adoption. H4(+): Cost Savings has a positive influence on ERP adoption. Expertise is an important factor which is positively related to new technology adoption (Nordin et al., 2014). Companies which are not familiar with IT are probably not aware of innovations, or are more resistant to adopting new technologies. Organization features including structural elements and specialized human resources affect the technological context concerning innovation adoption. H5(+): Technology Readiness has a positive influence on ERP adoption. Many studies show that the top management is creating a supportive environment with adequate resources for new technology adoption (Premkumar and Roberts, 1999). The support from the top management plays a relevant role in innovation adoption because it guides budget relocation, integration of services and reengineering of processes (Chou et al., 2014). The top management is one of the determinants of the organizationâ&#x20AC;&#x2122;s culture. The adjustment of the organizationâ&#x20AC;&#x2122;s culture because of information systems is indispensable for ERP implementation success. Therefore, the success of ERP implementation increases as the top management promotes and supports it within the organization culture (Ke and Wei, 2008). H6(+): Top Management Support has a positive influence on ERP adoption. Studies indicate that the organization size is related to the impact of new technologies adoption (Zhu et al., 2006a). Large farms should have larger budgets for improvements, and they are capable of experiencing innovations faster them small properties. Moreover, large farms are more capable of raising funds from banks and investors. Going against the odds, some small farms are capable of taking risks in new technology (Martins et al., 2014). H7(+): Farm Size has a positive influence on ERP adoption. Competitive Pressure is the force a company experiences from similar competitors in the same industry (Gatignon and Robertson, 1989). One characteristic of the agribusiness is a commoditized market, which strives towards a perfect competition environment. This scenario makes the adoption of new technologies like International Food and Agribusiness Management Review

734

Haberli et al.

Volume 20, Issue 5, 2017

ERP a non-essential tool for the competitive strategy in the market, because it only delivers process innovation instead of product innovation. Otherwise, considering the global competition, this external pressure from producer countries can become relevant and can be strategic for the company. H8(+): Competitive Pressure has a positive influence on ERP adoption. In Brazil, the government regulation to support the adoption of new technologies is not clearly defined yet. This hypothesis is concerned about the legal protection of farm activities. H9(-): Regulatory Support has a negative influence on ERP adoption. A proper tool was developed and adapted to collect data from the companies in this study. To evaluate others studies concerned about the same constructs in the Table 2, we examined papers found related to researches in the agribusiness. Some of the constructs were not found in the same study field and to fulfill the gap, we extended the research to other commoditized markets. Interviews were realized between August 1st and 21st of 2014 in Mato Grosso (Central-West region of Brazil). The tool applied to collect data was a structured survey focused on measuring the variables/determinants of ERP adoption described in the Table 2. The questions proposed were based on the DOI and TOE theories and were validated by applying in-depth interviews (Boyce and Neale, 2006) with ten agribusiness consultants. A quantitative method was used in this study and personal interviews with owners or farms managers were conducted on site (farms). Experts and researchers in the agriculture sector collected data through personal interviews. Consistency was maintained by using a 5 point ranking system varying from ‘strongly disagree’ to ‘strongly agree’ for the evaluation of DOI variables such as Rrelative Advantage, Complexity, Compatibility and Cost Savings as well as TOE variables like Technology Readiness, Top Management Performance, Farm Size, Competitive Pressure and Regulatory Support. We have found some qualitative evidences during the interviews: (1) how can an ERP help me to solve my management problems? (Lambim, 2000); (2) does the ERP can solve my long term problems? (Bloch and Richins, 1983). Therefore, we have noticed that the perception of the ERP process and ERP image can have a strong influence on the farm management and also on the individual’s behavior. (De Toni and Schuler, 2007). The interviews were conducted with the necessary care considering these aspects. We used ‘simple random sampling’ as the criteria to select the sample for this study. In addition to the inquiries on Table 2, farmers were asked about their job position in the farm, level of education, how they managed the farm, whether there were other professionals working in the farm administration, precision agriculture. These questions gave us qualitative information about the interviewees. The sample is composed of 200 valid interviews with soy, corn and cotton producers with medium (2,471 to 9,884 acre) and large farm sizes (above 9,885 acre). This can be seen in Table 3. Small producers were not included in this study because the ERP system can be easily found in medium and large farms. The survey sample is composed of interviewees with the following profiles. Interviewees with college degrees made up 19%, 14.5% had incomplete college degrees while 38.5% had high school certificates. The interviewees’ average age was 38 years. 56.5% responded that there were other specialized professionals in management positions in their farms while 30.5% were the only ones in charge. 75.5% of the interviewees performed at least one precision agriculture operation in the crops waiting to achieve: (1) 86.1% reduced costs of planting, caring for and harvesting crops; (2) 76.2% reduced losses related to pests; (3) 28.5% prevented weather conditions; and (4) 12,6% obtained sustainability participation credits (low carbon rates agriculture).

International Food and Agribusiness Management Review

735

Haberli et al.

Volume 20, Issue 5, 2017

Table 2. Data collection tool: quantitative variables. Constructs

Items

Reference

Relative Advantage RA1 ERP1 allows more efficiency in managing business operations RA2 An adequate ERP use improves operations quality RA3 An adequate ERP use allows a faster execution of specific assignments RA4 Using ERP – Enterprise Resource Planning enables new opportunities RA5 ERP allows increment of business productivity Complexity CX1 ERP use requires high mental effort CX2 It is frustrating to use ERP. CX3 It is too complex to use ERP on commercial operations CX4 It is too complex to use ERP on production operations CX5 Adoption of ERP requires complex skills from farm’s employees Compatibility Cp1 I can’t find an ERP that fits this farm’s work structure Cp2 I can’t find a perfectly compatible ERP for my business operation Cp3 I can’t find an ERP compatible with the culture and corporate values of my farm Cp4 I can’t find an EPR compatible with computers and programs (hardware and software) in my farm Cost Savings CS1 ERP benefits outweigh its adoption cost CS2 ERP adoption reduces overall and environmental costs CS3 ERP adoption costs are low Technology Readiness TR1 There is enough knowledge in the farm to use ERP to support its operations TR2 There are required skills in the farm to implement a more effective ERP Top Manager Support Farm’s management supports ERP implementation TMS1 Farm’s top management plays a strong leadership role and gets involved in the TMS2 ERP process TMS3 Farm’s Top management is inclined to take risks (economic and organizational) to adopt an ERP Farm Size - from 2.471 to 4.942 acre - from 4.943 to 7.413 acre - from 7.415 to 9.884 acre - above 9.885 acre Competitive Pressure The farm believes its own ERP influences other businesses in the same region The Farm is under external market pressure to adopt an ERP Some farmers from the same region use ERP Regulatory Support RS1 There is no legal protection for agriculture activities RS2 Existing laws and regulations are enough to protect agriculture activities

International Food and Agribusiness Management Review

736

Gul et al., 2014; Helitzer et al., 2014; Sarker and Ratnasena, 2014

Batz et al., 1999; Montedo, 2012; Peshin, 2013

Fu et al., 2007; Gerber et al., 1996

Ghadim et al., 2005; Pannell et al., 2014; Sangle, 2011 Nordin et al., 2014

Chou et al., 2014

Premkumar and Roberts, 1999

Zhu et al., 2003, 2004

Zhu et al., 2006b

Haberli et al.

Volume 20, Issue 5, 2017

Table 2. Continued. Constructs

Items

Reference

ERP Adoption (ERPA) ERPA1 At this moment, what would you say about the possibility of adopting ERP? - I have never considered it. - there is a pilot project running - I’ve already considered the possibility and I will not adopt it - I want to adopt it in the future - I’ve already adopted it (less than 1 year) ERPA2 How long will it take to adopt ERP in your farm? - less than 1 year - 1 to 2 years - 2 to 5 years - more than 5 years - I don’t know. 1

Hong et al., 2008

ERP = Enterprise Resource Planning.

4. Results Structural Equation Modeling (SEM), was applied in this study. SEM combines statistical data and qualitative causal assumption for testing and estimating causal relations. Researchers recognized the possibility of distinguishing between measuring models and structures and have started to consider the measurement error (Henseler et al., 2009). It is possible to find two different divisions of SEM techniques. They are the covariance technique and the technique based on variance. Based on the variance technique, it is possible to use the Partial Least Squares (PLS) in cases where not all items in the data are normally distributed (P<0.01, based on Kolmogorov-Smirnov test) or the research model was not tested in the concerning literature or if the research model is considered complex. In this case, we used de SMART PLS 2.0 M3 software (SmartPLS GmbH, Hamburg, Germany) (Ringle et al., 2005) to analyze relations defined by the theoretical model. The model was evaluated in two steps; first, the variables were analyzed to determine their capability to measure each one of the constructs. Second, the structural relations were analyzed among the constructs (Table 4). Measurement model validation was evaluated based on three criteria: construct reliability, convergent validity and discriminant validity. The reliability of each construct is a measurement of internal consistency of its indicators and presents the adequacy of measurement scale. To evaluate its reliability, we adopted a composite reliability indicator which is considered better compared to Cronbach’s Alpha which can underestimate results (Hock and Ringle, 2010). Following the reliability indicator, values for reliability composite above 0.700 are adequate. Based on Table 5, it is possible to observe values above 0.700 for reliability composite which indicates an adequate model. Table 3. Sample distribution. Culture

Total (%)

Base

Soy Corn Cotton Total

43.5 41.0 15.5 100.0

87 82 31 200

International Food and Agribusiness Management Review

737

Haberli et al.

Volume 20, Issue 5, 2017

Convergent validity evaluates the extension in which the indicator is capable of measuring a latent variable (construct). It can be verified by observing patterns of factorial loading and average variance extracted (AVE). Factorial loading above 0.700 (Im et al., 1998) and variances above 0.5 (Hair et al., 1995) were accepted as high and significant. At Table 5, it is possible to find only 4 variables (Cx1, Cp4, TMP3 and RS2) which have factorial loading below 0.700. Since items with factorial loading below 0.400 should be removed from the analysis, we kept all the constructs in the study because all of them presented variances above 0.500. These results are an assurance that the indicators are legitimate representatives of the analyzed constructs. The constructâ&#x20AC;&#x2122;s discriminant validity was evaluated using two criteria: Fornell-Larcker and Cross-loadings. Fornell-Larcker advocates that the square root of AVE needs to be greater than the correlation of the construct (Fornell and Larcker, 1981). Cross-loadings requires loading of each indicator to be greater than the crossTable 4. Measurement model combining Technology Organization Environment and Diffusion of Innovation Theory.1 Latent variables

RA1 RA2 RA3 Cx1 Cx2 Cx3 Cx4 Cp1 Cp2 Cp3 Cp4 CS2 CS3 TR1 TR2 TMP1 TMP2 TMP3 FS CP1 CP3 RS1 RS2 ERPA1 ERPA2 AVE Reliability composition

TMS

ERP adoption

0.775 0.775 0.765 -0.130 -0.190 -0.140 -0.090 0.356 0.323 0.245 0.233 0.236 0.152 0.426 0.446 0.458 0.355 0.173 0.065 0.423 0.396 -0.164 0.004 0.587 0.546 0.595 0.815

-0.243 -0.070 -0.123 0.636 0.766 0.780 0.736 -0.396 -0.420 -0.319 -0.317 0.240 0.318 -0.124 -0.151 0.014 0.061 0.169 -0.181 0.007 0.067 0.453 0.286 -0.247 -0.259 0.535 0.821

0.302 0.222 0.334 -0.283 -0.357 -0.346 -0.350 0.838 0.882 0.790 0.662 -0.125 -0.223 0.156 0.212 0.084 -0.026 -0.248 0.079 0.168 0.047 -0.372 -0.293 0.376 0.354 0.635 0.873

0.101 0.294 0.165 0.395 0.166 0.270 0.244 -0.112 -0.168 -0.239 -0.209 0.818 0.799 0.139 0.166 0.226 0.235 0.177 -0.076 0.135 0.147 0.178 0.201 0.237 0.168 0.654 0.790

0.389 0.350 0.471 -0.083 -0.214 -0.044 -0.106 0.198 0.254 0.158 0.044 0.201 0.093 0.819 0.847 0.384 0.160 0.174 0.067 0.435 0.252 -0.228 0.095 0.497 0.416 0.694 0.819

0.354 0.331 0.429 0.097 -0.024 0.085 0.112 0.002 -0.031 -0.032 -0.046 0.256 0.213 0.237 0.336 0.826 0.774 0.568 0.003 0.305 0.271 0.023 0.131 0.414 0.378 0.535 0.771

0.130 -0.025 0.044 -0.134 -0.141 -0.097 -0.153 0.029 0.086 0.040 0.114 -0.077 -0.046 0.087 0.027 0.050 -0.048 -0.016 1.000 -0.064 -0.086 -0.154 -0.100 0.153 0.178 1.000 1.000

0.398 0.414 0.354 -0.055 -0.026 0.103 0.091 0.167 0.138 0.051 0.056 0.146 0.132 0.310 0.409 0.325 0.202 0.262 -0.090 0.859 0.762 -0.022 0.087 0.357 0.324 0.659 0.794

-0.107 -0.108 -0.113 0.348 0.439 0.277 0.332 -0.326 -0.411 -0.307 -0.264 0.119 0.236 -0.124 -0.155 -0.039 0.089 0.158 -0.165 -0.041 0.068 0.952 0.544 -0.220 -0.262 0.601 0.737

0.456 0.444 0.450 -0.136 -0.229 -0.159 -0.210 0.351 0.356 0.247 0.202 0.173 0.166 0.377 0.407 0.375 0.303 0.171 0.170 0.315 0.249 -0.251 -0.092 0.975 0.971 0.947 0.973

1 ERP = Enterprise Resource Planning; RA = Relative Advantage; Cx = Complexity; Cp = Compatibility; CS = Cost Savings; TR =

Technologic Readiness; TMS = Top Management Support; FS = Farm Size; CP = Competitive Pressure; RS = Regulatory Support; ERPA = Enterprise Resource Planning Adoption; AVE = average variance extracted.

International Food and Agribusiness Management Review

738

Haberli et al.

Volume 20, Issue 5, 2017

Table 5. Correlation between constructs and median variance extracted from square root (diagonal).1 RA CX Cp CS TR TMS FS CP RS Adoption

0.771 -0.190 0.371 0.241 0.523 0.481 0.065 0.504 -0.141 0.583

0.732 -0.460 0.344 -0.166 0.082 -0.181 0.041 0.484 -0.259

0.797 -0.214 0.222 -0.029 0.079 0.140 -0.416 0.376

0.808 0.183 0.290 -0.076 0.172 0.218 0.210

0.833 0.346 0.067 0.433 -0.168 0.471

TMS

0.731 0.003 0.355 0.061 0.408

1.000 -0.090 -0.165 0.170

0.812 0.009 0.350

0.775 -0.247

ERPA

0.973

RA = Relative Advantage; Cx = Complexity; Cp = Compatibility; CS = Cost Savings; TR = Technologic Readiness; TMS = Top Management Support; FS = Farm Size; CP = Competitive Pressure; RS = Regulatory Support; ERPA = Enterprise Resource Planning Adoption.

loadings (Chin, 1998; Götz et al., 2010; Grégoire and Fisher, 2006). As presented in Table 5, the square roots of AVE (diagonal elements) are greater than the correlation between each structure’s pairs (elements outside of diagonal). Table 5 also presents loading patterns higher than cross-loadings. In conclusion, both criteria were satisfied. According to the results, we can conclude that the measuring model presents construct reliability, convergent validity and discriminant validity. For this reason, it is adequate to test the structural model. To perform the analysis of the structural relations model, the statistical significance of the independent variables was evaluated to explain the ERP adoption. In addition, the R2 related was also evaluated. The results showed that the proposed model could explain 48.1% of variation in ERP adoption (Figure 2). The construct indicators are represented by rectangles and constructs are represented by circles. On Table 6, it is possible to confirm the hypothesis presented in this work. To reach those results, the signal and significance of the structural model coefficients were evaluated. This was the signal between the explanatory variables (independents) and ERP adoption (dependent variable). The significance levels of factorial loads were estimated using a bootstrap of 5,000 samples. The following results have indicated statistical significance: Relative Advantage (=0.227; P<0.01), Top Management Performance (=0.198; P<0.01), Compatibility (=0.194; P<0.01), Cost Savings (=0.171; P<0.01), Technology Readiness (=0.140; P<0.05), Complexity (=-0.120; P<0.10) and Farm Size (=0.113; P<0.10). To sum up, hypotheses H1, H2, H3, H4, H5, H6 and H7 were validated. On the other hand hypotheses H8 and H9 could not be validated in this study.

5. Discussions and conclusions Relative Advantage is the most important variable to explain ERP adoption. When the Relative Advantage increases a standard unit, ERP adoption increases 0.227 standard units subsequently. On the other hand, Regulatory Support and Competitive Pressure were not significant to ERP adoption. ERP is a process innovation and not a product innovation. However, agribusiness is about the production of agricultural commodities and competitive pressure does not apply among producers in this business. Soy, corn and cotton producers compete with producers worldwide but they do not necessarily compete with local producers. Agribusiness is an industry with perfect competition which is characterized by the lack of product differentiation and the similarities regarding the structure of cost among the farm. Future studies can explore in-depth the International Food and Agribusiness Management Review

739

Haberli et al.

Volume 20, Issue 5, 2017

RA1 0,775 0.775 RA2

0,775 0.775

RA RA

0,765 0.765

0,819 0.819 0,847 0.847

TR1 TR2

RA3 CX1 CX2 CX3

ERPA1 ERPA1

0,636 0.636 0,766 0.766 0,780 0.780

CX CX

0,975 0.975

Cp1

Cp3

0,836 0.836 0,881 0.881 0,790 0.790

Cp Cp

0,662 0.662

CS3

0,818 0.818 0,799 0.799

TMS TMS

0,774 0.774

TMS2 TMS2

0,568 0.568 TMS3 TMS3 0,198*** 0.198***

ERP ADOPTION 2 R = 48,1%

0,113* 0.113*

HA FS

0,064 0.064 CP

0,171*** 0.171***

Cp4 CS2

0,194*** 0.194***

0,826 0.826

0,971 0.971

-0,120* -0.120*

CX4

Cp2

0,140** 0.140**

0,227*** 0.227***

0,736 0.736

TMS1 TMS1

ERPA2 ERPA2

1.000 1,000

0,859 0.859 0,762 0.762

-0,084 -0.084

CS CS

DOI DOI

0,952 0.952 0,544 0.544

CP1 CP2

RS1 RS2

TOE

Figure 2. Measurement model combining Technology Organization Environment (TOE) and Diffusion of Innovation Theory (DOI). ERP = Enterprise Resource Planning; RA = Relative Advantage; Cx = Complexity; Cp = Compatibility; CS = Cost Savings; TR = Technologic Readiness; TMS = Top Management Support; FS = Farm Size; CP = Competitive Pressure; RS = Regulatory Support; ERPA = Enterprise Resource Planning Adoption. competitive pressure among farmers. Relative Advantage allows an anticipation of the harvest during the harvest period which gives farmers a negotiation advantage with the buyers. Since regulatory support for the sector is relatively recent, farmers need more time to understand in-depth and adapt to its rules. Besides this, the more the tools fit, the greater the possibility of ERP adoption. On the other hand, when it is less complex, the possibility of adoption is higher. To increase the chances of ERP system adoption, the providers will have to have a better understanding of its tools and technological processes, which is the hardware and software, used in the farms. However, without Top Management Support ERP adoption cannot succeed. The gateway to ERP adoption is in the farm ownerâ&#x20AC;&#x2122;s hands. We believe this study can contribute to the development of processes and tools with indicators related to this market. This paper can also help consultants who want to develop ERP systems for farms, by bringing features which are related to ERP adoption in the rural segment. Nonetheless, this paper can also motivate new International Food and Agribusiness Management Review

740

Haberli et al.

Volume 20, Issue 5, 2017

Table 6. Hypotheses analysis.1 Hypotheses

Results

H1(+) Relative Advantage has a positive influence on ERP2 adoption H2(-) Complexity has a negative influence on ERP adoption H3(+) Compatibility has a positive influence on ERP adoption H4(+) Cost Savings has a positive influence on ERP adoption H5(+) Technology Readiness has a positive influence on ERP adoption H6(+) Top Management Support has a positive influence on ERP adoption H7(+) Farm Size has a positive influence on ERP adoption H8(+) Competitive Pressure has a positive influence on ERP adoption H9(-) Regulatory Support has a negative influence on ERP adoption

Validated (= 0.227***) Validated (= -0.120*) Validated (= 0.194***) Validated (= 0.171***) Validated (= 0.140**) Validated (= 0.198***) Validated (= 0.113*) Not validated (= 0.064) Not validated (= -0.084)

1 *, ** 2

and *** indicate significant differences at P<0.10, <0.05 and <0.01, respectively. ERP = Enterprise Resource Planning.

research about the adoption of technology related to the farm organizations’ resource planning particularly for universities connected to the rural sector. The study’s contribution is directly related to the determinants of ERP adoption for farmers. This paper did not discuss the ERP’s specificities for the rural sector. This sector presents some particularities as there are high levels of uncertainty in production due to weather or there is a high number of small or medium companies. These characteristics interfere negatively to ERP adoption compared to others sectors such as industrial sector. The following studies can contribute to the determination of the sector’s specific features. We have also considered relevant studies about diffusion states (i.e. intention, adoption and use) and explored if the determinants of intention, adoption and use are the same. The next step is to consider the study of innovation adoption including the theory of individual behavior on farmers and the value that this adoption can deliver to the sales process, production process, procurement process and contracts. Because what we have so far is limited to research about Innovation Diffusion which explores innovation adoption from the individual decision maker’s point of view, as farmers, doctors and consumers (Fliegel and Kivlin, 1966a, 1966b). The only innovation factors measured are the ones observed by an individual adopter (Damanpour and Schneider, 2009). In addition, this paper does not discuss ERP development or the impact of Cloud Computing, the Internet of Things or analytic insight platform on the future of the ERP system. For next studies, we are considering to research the trends of ERP with the same target audience, including Platform of Analytical Insights, Internet of Things and Cloud Computing. Share knowledge and experiences can provide a healthy competitive environment in the agricultural sector for all countries. Usually, farmers are scattered, disorganized, deficient in resources and also exposed to natural disasters, market uncertainties and pricing failures (Ahmad et al., 2016). Future researches can be based on gathering ‘the wealth of scientific knowledge and agricultural domains in a cloud-based ERP to develop an e-agriculture platform of resources planning. It can contribute to strength the agriculture activity of a region or a country. The major questions will be: (1) can we feed 8.5 billion people in 2030? (www. unric.org); (2) can we get more of our land to control losses?; (3) can we better protect the environment while sharing more sustainable decisions? The answers for these questions must come from result of new production process applied in the field and in the crops, also from the controls of the production processes and from the management sharing.

International Food and Agribusiness Management Review

741

Haberli et al.

Volume 20, Issue 5, 2017

References Ahmad, T., S. Ahmad, and M. Jamshed. 2016. A knowledge based Indian agriculture: with cloud ERP arrangement. In: Proceedings of the 2015 International Conference on Green Computing and Internet of Things, ICGCIoT 2015 pp. 333-340. Available at: http://tinyurl.com/ya6e7o9p. Amoako-Gyampah, K. and A.F Salam. 2004. An extension of the technology acceptance model in an ERP implementation environment. Information and Management 41: 731-745. Batz, F.J., K.J. Peters and W. Janssen. 1999. The influence of technology characteristics on the rate and speed of adoption. Agricultural Economics 21: 121-130. Bloch, P.H. and M.L. Richins. 1983. A Theoretical model for the study product of importance perceptions. The Journal of Marketing 47(3): 69-81. Boyce, C. and P. Neale. 2006. Conducting in-depth interviews: a guide for designing and conducting indepth interviews. Evaluation 2: 1-16. Chin, W.W. 1998. The partial least squares approach for structural equation modeling. Modern methods for business research. In: Modern methods for business research: methodology for business and management. Lawrence Erlbaum Associates, Mahwah, NJ, USA, pp. 295-336. Chong, A.Y.-L., K-B. Ooi, B. Lin and M. Raman. 2009. Factors affecting the adoption level of c-commerce: an empirical study. The Journal of Computer Information Systems 50(2): 13-22. Chou, H.W., Y.H. Lin, H.S. Lu, H.H. Chang and S.B. Chou. 2014. Knowledge sharing and ERP system usage in post-implementation stage. Computers in Human Behavior 33: 16-22. Damanpour, F. and M. Schneider. 2009. Characteristics of Innovation and innovation adoption in public organizations: assessing the role of managers. Journal of Public Administration Research and Theory 19: 495-522. De Toni, D. and M. Schuler. 2007. Gestão da imagem: desenvolvendo um instrumento para a configuração da imagem de produto. Revista de Administração Contemporânea 11(4): 1-16. Ehie, I.C. and M. Madsen. 2005. Identifying critical issues in enterprise resource planning (ERP) implementation. Computers in Industry 56: 545-557. Esteves, J.M. 2014. An empirical identification and categorisation of training best practices for ERP implementation projects. Enterprise Information Systems 8: 665-683. Fliegel, F.C. and J.E. Kivlin. 1966a. ATtributes of innovations as factors in diffusion. American Journal of Sociology 72: 235-248. Fliegel, F.C., J.E. Kivlin. 1966b. Farmers perceptions of farm practice attributes. Rural Sociology 31: 197-206. Fornell, C. and D.F. Larcker. 1981. Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research 18(1): 39-50. Fountas, S., G. Carli, C.G. Sørensen, Z. Tsiropoulos, C. Cavalaris, Vatsanidou, B. Liakos, M. Canavari, J. Wiebensohn and B. Tisserye. 2015. Farm management information systems: current situation and future perspectives. Computers and Electronics in Agriculture 115: 40-50. Fu, Z., J. Yue, D. Li, X. Zhang, L. Zhang and Y. Gao. 2007. Evaluation of learner adoption intention of e-learning in China: a methodology based on perceived innovative attributes. New Zealand Journal of Agricultural Research 50: 609-615. Gajic, G., S. Stankovski, G. Ostojic, Z. Tesic and L. Miladinovic. 2014. Method of evaluating the impact of ERP implementation critical success factors a case study in oil and gas industries. Enterprise Information Systems 8: 84-106. Gatignon, H. and T.S. Robertson. 1989. Technology diffusion – an empirical-test of competitive effects. Journal of Marketing 53: 35-49. Gerber, A., V. Hoffmann and M. Kugler. 1996. The diffusion of eco-farming in Germany and the concomitant development of the knowledge system supporting it. Berichte Uber Landwirtschaft 74: 591-627. Ghadim, A.K.A., D.J. Pannell and M.P. Burton. 2005. Risk, uncertainty, and learning in adoption of a crop innovation. Agricultural Economics 33: 1-9. Götz, O., K. Liehr-Gobbers and M. Krafft. 2010. Evaluation of structural equation models using the partial least squares (PLS) approach. In: Handbook of Partial Least Squares, Springer, Heidelberg, Germany, pp. 691-711. International Food and Agribusiness Management Review

742

Haberli et al.

Volume 20, Issue 5, 2017

Grégoire, Y. and R.J. Fisher. 2006. The effects of relationship quality on customer retaliation. Marketing Letters 17(1): 31-46. Gul, M., G. Akpinar, V. Demircan, H. Yilmaz, T. Bal, S.E. Arici and S.G. Yilmaz. 2014. Economic analysis of early-warning system in apple cultivation: a Turkish case study. Custos E Agronegocio on Line 10: 165-179. Hair, J.F., R.E. Anderson, R.L. Tatham and W.C. Black. 1995. Multivariate data analysis, 4th edition. PrenticeHall, Upper Saddle River, NJ, USA. Hakim, A. and H. Hakim. 2010. A practical model on controlling the ERP implementation risks. Information Systems 35: 204-214. Helitzer, D.L., G. Hathorn, J. Benally and C. Ortega. 2014. Culturally relevant model program to prevent and reduce agricultural injuries. Journal of Agricultural Safety and Health 20: 175-198. Henseler, J., C.M. Ringleand and R.R. Sinkovics. 2009. The use of partial least squares path modelling in international marketing. New challenges to international marketing. In: New challenges to international marketing (advances in international marketing, volume 20), edited by R.R. Sinkovics and P.N. Ghauri. Emerald Group Publishing Limited, Bingley, UK, pp. 277-319. Hock, M. and C.M. Ringle. 2010. Local strategic networks in the software industry: an empirical analysis of the value continuum. International Journal of Knowledge Management Studies 4(2): 132. Hong, H.Y, J.H. Moon, C.W. Yoo and Y. Choe. 2008. An analysis of social-psychological factors that influence the intention to use the agricultural information system ‘LFcenter System’. Journal of Agricultural Extension and Community Development 15: 659-681. Hsu, P.-F., K.L. Kraemer and D. Dunkle. 2006. Determinants of e-business use in US firms. International Journal of Electronic Commerce 10: 9-45. Hung, W.H., C.F. Ho, J.J. Jou and K.H. Kung. 2012. Relationship bonding for a better knowledge transfer climate: an ERP implementation research. Decision Support Systems 52: 406-414. Im, S.K., V. Grover and S. Sharma. 1998. The use of structural equation modelling in research. University of South Carolina, Columbia, SC, USA. Janssen, S. and M.K. van Ittersum. 2007. Assessing farm innovations and responses to policies: a review of bio-economic farm models. Agricultural Systems 94: 622-636. Karimi, J., T.M. Somers and A. Bhattacherjee. 2007. The impact of ERP implementation on business process outcomes: a factor-based study. Journal of Management Information Systems 24: 101-134. Ke, W. and K.K. Wei. 2008. Organizational culture and leadership in ERP implementation. Decision Support Systems 45: 208-218. Kumar, K. and J. Van Hillegersberg. 2000. ERP – Experiences and evolution. Communications of the Acm 43: 22-26. Lambin, J.J. 2000. Marketing estratégico. 4a. ed. Portugal. McGraw-Hill, Tradução Domingos Silva, Portugal. Madapusi, A. and D. D’Souza. 2012. The influence of ERP system implementation on the operational performance of an organization. International Journal of Information Management 32: 24-34. Martins, C., T. Oliveira and A. Popovic. 2014. Understanding the Internet banking adoption: a unified theory of acceptance and use of technology and perceived risk application. International Journal of Information Management 34: 1-13. Martins, R., T. Oliveira and M.A. Thomas. 2016. An empirical analysis to assess the determinants of SaaS diffusion in firms. Computers in Human Behavior 62: 19-33. Montedo, U.B. 2012. The process of constructing new competencies in the family agricultural production unit. Work-a Journal of Prevention Assessment and Rehabilitation 41: 76-80. Nordin, S.M., S.M. Noor, M.S. Saad and M. Bin. 2014. Innovation Diffusion of new technologies in the malaysian paddy fertilizer industry. 2nd World Conference on Business, Economics and Management 109: 768-778. Oliveira, T., M. Thomas and M. Espadanal. 2014. Assessing the determinants of cloud computing adoption: an analysis of the manufacturing and services sectors. Information and Management 51: 497-510. Orlovski, R., L.A. Kisielewicz and S.R.A. Ribeiro 2012. Segurança de informação em sistemas agrícolas: estudo de caso fundação abc. Revista de Engenharia e Tecnologia 4(3).

International Food and Agribusiness Management Review

743

Haberli et al.

Volume 20, Issue 5, 2017

Orsi, L., I. De Noni, S. Corsi and L.V. Marchisio. 2017. The role of collective action in leveraging farmers’ performances: lessons from sesame seed farmers’ collaboration in eastern Chad. Journal of Rural Studies 51: 93-104. Pannell, D.J., R.S. Llewellyn and M. Corbeels. 2014. The farm-level economics of conservation agriculture for resource-poor farmers. Agriculture Ecosystems and Environment 187: 52-64. Park, S.-H., J. Eo and J.I. Lee. 2012. Assessing and managing an organization’s green IT maturity. Mis Quarterly Executive 11: 127-140. Peshin, R. 2013. Farmers’ adoptability of integrated pest management of cotton revealed by a new methodology. Agronomy for Sustainable Development 33: 563-572. Premkumar, G. and M. Roberts. 1999. Adoption of new information technologies in rural small businesses. Omega-International Journal of Management Science 27: 467-484. Ram, J., D. Corkindale and M.L. Wu. 2013. Implementation critical success factors (CSFs) for ERP: do they contribute to implementation success and post-implementation performance? International Journal of Production Economics 144: 157-174. Ringle, C.M., S. Wende and A. Will. 2005. SmartPLS 2.0 M3. Available at: http://www.smartpls.de. Rogers, E.M. 1993. The diffusion of innovations model. Diffusion and Use of Geographic Information Technologies 70: 9-24. Rogers, E.M. 1995. Lessons for guidelines from the diffusion of innovations. Joint Commission Journal on Quality Improvement 21: 324-328. Rouyendegh, B.D., U. Bac and T.E. Erkan. 2014. Sector selection for erp implementation to achieve most impact on supply chain performance by using ahp-topsis hybrid methoD. Tehnicki Vjesnik-Technical Gazette 21: 933-937. Ruivo, P., T. Oliveira and M. Neto. 2012. ERP post-adoption: value impact on firm performance. In: 7th Iberian Conference on Information Systems and Technologies (CISTI), edited by I. Tecnologias and L.P. Isegi Unl. ISEGI UNL, Lisbon, Portugal. Ruivo, P., T. Oliveira and M. Neto. 2014. Examine ERP post-implementation stages of use and value: empirical evidence from Portuguese SMEs. International Journal of Accounting Information Systems 15:166-184. Sangle, S. 2011. Adoption of cleaner technology for climate proactivity: a technology-firm-stakeholder framework. Business Strategy and the Environment 20: 365-378. Sarker, R. and S. Ratnasena. 2014. Revealed comparative advantage and half-a-century competitiveness of Canadian agriculture: a case study of wheat, beef, and pork sectors. Canadian Journal of Agricultural Economics-Revue Canadienne D Agroeconomie 62: 519-544. Sørensen, C.G., S. Fountas, E. Nash, L. Pesonen, D. Bochtis, S.M. Pedersen, B. Basso and S.B. Blackmore. 2010. Conceptual model of a future farm management information system. Computers and Electronics in Agriculture 72(1): 37-47. Sørensen, C.G., L. Pesonen, D.D. Bochtis, S.G. Vougioukas and P. Suomi. 2011. Functional requirements for a future farm management information system. Computers and Electronics in Agriculture 76(2): 266-276. Tornatzky, L.G. and M. Fleischer. 1991. The processes of technological innovation. The Journal of Technology Transfer 16: 45-46. Tornatzky, L.G. and J.K. Klein. 1982. Innovation characteristics and innovation adoption-implementation – a meta-analysis of findings. Ieee Transactions on Engineering Management 29: 28-45. Tsai, M.T., E.Y. Li, K.W. Lee and W.H. Tung. 2011. Beyond ERP implementation: the moderating effect of knowledge management on business performance. Total Quality Management and Business Excellence 22: 131-144. Valente, T.W. and E.M Rogers. 1995. The origins and development of the diffusion of innovations paradigm as an example of scientific growth. Science Communication 16: 242-273. Velcu, O. 2010. Strategic alignment of ERP implementation stages: an empirical investigation. Information and Management 47: 158-166. Verdouw, C.N. R.M. Robbemond and J. Wolfert. 2015. ERP in agriculture: lessons learned from the Dutch horticulture. Computers and Electronics in Agriculture 114: 125-133. International Food and Agribusiness Management Review

744

Haberli et al.

Volume 20, Issue 5, 2017

Wolfert, J., C.N. Verdouw, C.M. Verloop and A.J.M. Beulens. M. 2010. Organizing information integration in agri-food â&#x20AC;&#x201C; A method based on a service-oriented architecture and living lab approach. Computers and Electronics in Agriculture 70(2): 389-405. Wood, B.A., H.T. Blair, D.I. Gray, P.D. Kemp, P.R. Kenyon, S.T. Morris and A.M. Sewell. 2014. Agricultural science in the wild: a social network analysis of farmer knowledge exchange. Plos One 9. Xouridas, S. 2015. Agricultural financial risks resulting from extreme events. Journal of Agricultural Economics 66(1): 192-220. Yusuf, Y., A. Gunasekaran and M. Abthorpe. 2004. Enterprise information systems project implementation: a case study of ERP in Rolls-Royce. International Journal of Production 87: 251-166. Zach, O., B.E. Munkvold and D.H. Olsen. 2014. ERP system implementation in SMEs: exploring the influences of the SME context. Enterprise Information Systems 8: 309-335. Zeng, Y.J. and M.J. Skibniewski. 2013. Risk assessment for enterprise resource planning (ERP) system implementations: a fault tree analysis approach. Enterprise Information Systems 7: 332-353. Zhang, Z., M.K.O. Lee, P. Huang, L. Zhang and X.Y. Huang. 2005. A framework of ERP systems implementation success in China: an empirical study. International Journal of Production Economics 98: 56-80. Zhu, K., S. Dong, S.X. Xu and K.L. Kraemer. 2006b. Innovation diffusion in global contexts: determinants of post-adoption digital transformation of European companies. European Journal of Information Systems 15: 601-616. Zhu, K., K.L. Kraemer and S. Xu. 2006a. The process of innovation assimilation by firms in different countries: a technology diffusion perspective on e-business. Management Science 52: 1557-1576. Zhu, K., K. Kraemer and S. Xu. 2003. Electronic business adoption by European firms: a cross-country assessment of the facilitators and inhibitors. European Journal of Information Systems 12: 251-268. Zhu, K., K.L. Kraemer, S. Xu and J. Dedrick. 2004. Information technology payoff in e-business environments: an international perspective on value creation of e-business in the financial services industry. Journal of Management Information Systems 21: 17-54.

International Food and Agribusiness Management Review

745

Haberli et al.

Volume 20, Issue 5, 2017

International Food and Agribusiness Management Review

746

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2017.0001 Received: 2 January 2017 / Accepted: 13 April 2017

An empirical investigation of patent and trademark ownership propensity and intensity in the U.S. food and drink industry RESEARCH ARTICLE Jasper Grashuis

and Stanley Kojo Daryb

aPostdoctoral

Fellow, and bPhD Graduate, Department of Agricultural and Applied Economics, University of Missouri-Columbia, Columbia, 65211 MO, USA

Abstract We use patent and trademark ownership data to study product, process, and marketing innovation by 157 manufacturers in the U.S. public food and drink industry. For the 2000-2014 period, most patented innovations relate to processes for manufacturing and designs for marketing, whereas patented innovations in food and drink products or compositions are relatively few. Meanwhile, intellectual property in general is more often protected by trademark ownership. Empirically, we specify a panel logistic model and a panel negative binomial model to study the relationship of firm characteristics to the propensity and intensity of patent and trademark ownership, respectively. In each model, firm size exhibits a significant and positive relationship to the propensity and intensity of patented innovations in products, processes, and marketing. Past innovation, past income, and firm age also have a positive relationship to patent and trademark ownership in most models, whereas leverage is only estimated to negatively relate to the propensity and intensity of trademark ownership. We use our main findings and conclusions to inform research, management, and policy implications. Keywords: food and drink industry, intellectual property, patents, trademarks, panel analysis JEL code: L66, O34, Q13 Corresponding author: grashuisj@missouri.edu

ÂŠ 2017 Grashuis and Dary

747

Grashuis and Dary

Volume 20, Issue 5, 2017

1. Introduction While the food and drink industry is traditionally not characterized by relatively intense research and development (R&D) (Allred and Park, 2007; Galizzi and Venturini, 1996), recent developments in the overall agri-food industry have forced or motivated food and drink manufacturers to increasingly engage in product, process, and marketing innovation for economic value creation (Johnson et al., 2009). At the forefront is product differentiation as consumers in the developed world are increasingly critical and fragmented, which implies consumer satisfaction is in part dependent on the improvement of old products and the introduction of new products (Avermaete et al., 2004; Grunert, 2005). A related development is the heavy emphasis on food safety and quality, which is manifested by the proliferation of public and private standards and regulations for producers and manufacturers (Aung and Chang, 2014). Furthermore, the increasing degree of consolidation and concentration in the food retail sector implies food and drink manufacturers face stiffer competition for scarce marketing opportunities (Adjemian et al., 2016). We place primary emphasis on intellectual property, which provides the legal and economic framework to connect innovation to value creation (O’Donnell et al., 2008). To be specific, innovation is the commercial manifestation of ex ante investment in R&D, which implies value generation in tangible and intangible assets and resources, intellectual property in particular. If left unprotected, other individuals and organizations can appropriate the economic value. In fact, analogous to asset ownership in general, ex ante investment in R&D and innovation is irrational if ex post rent protection is suboptimal. In the food and drink industry, intellectual property is in practice often protected by means of patents, trademarks, trade secrets, and copyrights (O’Donnell et al., 2008). While there is good reason to assume food and drink manufacturers use trade secrets,1 we limit our study to patents and trademarks as its ownership is a matter of public record. As explained by O’Donnell et al. (2008) and Hall et al. (2014), a patent grants a limited monopoly to exclude other individuals and organizations from making, using, or selling an invention for 20 years. By comparison, a trademark is the exclusive right to use words, names, symbols, or any combination thereof to identify and distinguish a good from other goods (Hall et al., 2014). Patents and trademarks offer different protections, and trademark ownership is often pursued in case of unpatented innovations (Flikkema et al., 2015), which implies consideration of both patents and trademarks may provide a richer conceptualization of innovation by food and drink manufacturers than either patents or trademarks alone. Although we do not claim patent and trademark data capture or indicate the full extent of innovation in the food and drink industry, the recent literature has used patent and trademark data extensively as standard measures of innovation (Flikkema et al., 2015; Hall et al., 2014; Moser, 2013; Schautschick and Greenhalgh, 2016). While we use patent and trademark data to infer innovation, it is important to adopt a multi-dimensional conceptualization of both ‘soft’ and ‘hard’ innovation in the food and drink industry, as demonstrated in the recent literature (Baregheh et al., 2012; Capitanio et al., 2010; Ciliberti et al., 2016; Gehlhar et al., 2009; Minarelli et al., 2015; Triguero et al., 2013; Vancauteren, 2016). Therefore, we study product, process, and marketing innovation, where ‘[a] product innovation is the introduction of a good or service that is new or significantly improved with respect to its characteristics or intended uses’, ‘[a] process innovation is the implementation of a new or significantly improved production or delivery method’, and ‘[a] marketing innovation is the implementation of a new marketing method involving significant changes in product design or packaging, product placement, product promotion or pricing’ (Organization for Economic Co-operation and Development, 2005). In the U.S food and drink industry, understanding of patent and trademark ownership and its determinants is limited. In general, applied research on innovation in the U.S. food and drink industry is not extensive. Recent empirical literature on food and drink innovation for the most part emphasizes small European enterprises and uses survey data. Because of differences in social, economic, and political environments, it is uncertain 1

In perhaps the only study of its kind, Cohen et al. (2000) determined food manufacturers find trade secrets to be more effective as compared to patents in terms of appropriating rent from product and process innovations.

International Food and Agribusiness Management Review

748

Grashuis and Dary

Volume 20, Issue 5, 2017

if findings and conclusions from such studies are generalizable to the U.S. food and drink industry, which implies a considerable gap in the literature. We therefore formulate the following research question: what is the relationship of firm characteristics to product, process, and marketing innovation propensity and intensity?2 Specifically, we emphasize firm size, firm age, past income, past innovation, and leverage as the firm characteristics of interest. We approach the research question by means of panel analysis on 157 U.S. public food and drink manufacturers for the 2000-2014 period. The article proceeds as follows. Section 2 provides the background of our study, and Section 3 contains a brief overview of relevant literature. In Section 4 we discuss our methodology, including the data, the summary statistics, the empirical model specifications, and the model variables. We present and discuss the results of our empirical analysis in Section 5. Section 6 contains the summary and conclusions, including a discussion of implications for academics, practitioners, and policy makers.

2. Background Following the OECD (2005), we interpret innovation as the discovery of a new or the improvement of an existing and useful product, process, or marketing method. Economically, there exist different perspectives of innovation. For example, innovation is the catalyst in the entrepreneurial theory of the firm, where Schumpeter (1942) described innovation as creative destruction. According to Schumpeter (1942), constant disequilibrium is the natural state of the overall economy as bold, creative individuals and organizations make constant innovations in products and processes. As first movers, these innovators have temporary power in input supply or output demand markets and can thus appropriate profit. Kirzner (1997), however, placed emphasis on the reaction to the creative destruction. Through alertness, individuals and organizations react to profit possibilities in the spot market. To appropriate profit, the entrepreneur may make innovations in products and processes (Kirzner, 1997, 2009). In doing so, the primary effect of innovation is market equilibration. While different in conceptualization, the practical interpretations of innovation by Schumpeter and Kirzner are not too dissimilar as value generation and profit appropriation are the main processes (Bostaph, 2013). However, these theories do not consider explicitly the legal or economic implications of knowledge or intellectual property and its protection. To start, Arrow (1962) argued the acquisition of technological knowledge may explain variability in per capita income across countries. By extension, it is possible knowledge may inform heterogeneity in firm competitiveness. Posner and Landes (2003) defined intellectual property as an idea, invention, discovery, or any human product of potential value separable from a unique physical embodiment. Without protection of such intellectual property, the inventor or innovator is susceptible to ex post rent appropriation by other individuals and organizations. In fact, ex post rent protection must be guaranteed in order to secure ex ante investment (Posner and Landes, 2003), which corresponds to the main theme of property rights theory. In practice, intellectual property is protected by patents, trademarks, trade secrets, copyrights, and sui generis rights (Oâ&#x20AC;&#x2122;Donnell et al., 2008). Although the legal definition and execution varies, each mechanism is an indicator of intellectual property and economic value protection. Therefore, as noted by Flikkema et al. (2015), Hall et al. (2014), Moser (2013), Nagaoka et al. (2010), and Schautschick and Greenhalgh (2016), increasingly more studies use patent and trademark ownership data to inform innovation.

3. Literature review Generally, studies of patent data in relation to the food and drink industry use two types of approaches: (1) within-country firm comparisons, and (2) cross-country firm comparisons. Examples of the former approach are Gopinath and Vasavada (1999), Martinez and Briz (2000), and the Government of Australia (2014), who studied such topics as market structure, sales performance, and innovation type in relation to patent ownership. 2 As we explain in the data section, propensity is a binary measurement of innovation (1 if patent or trademark ownership is non-zero, 0 otherwise) and intensity is a continuous measurement of innovation (total owned patents or trademarks).

International Food and Agribusiness Management Review

749

Grashuis and Dary

Volume 20, Issue 5, 2017

Examples of the second approach are Alfranca et al. (2003), Allred and Park (2007), Martinez and Rama (2012), and Van Galen et al. (2013), who emphasized R&D expenditure and new product introductions. While patent ownership is the common denominator in the cited studies, there is much diversity in the samples and methodologies, which is characteristic of the general literature on innovation in the food and drink industry. For example, innovation is studied in terms of core competences (Traill and Meulenberg, 2002), the causal impact of retailer concentration (Weiss and Wittkopp, 2005) and vertical integration (Karantininis et al., 2010), innovation probability determinants (Capitanio et al., 2010), local and regional networking (Gellynck and Kuhne, 2010; Gellynck et al., 2007), product and process innovation complementarity (Triguero et al., 2013), market liberalization (Ghazalian and Fakih, in press), and external knowledge acquisition (Ciliberti et al., 2016). Meanwhile, our study is most relatable to Avermaete et al. (2004), who analyzed the causal impact of firm size on innovation, as well as Bhaskaran (2006) and Giovannetti et al. (2011), who studied the different characteristics of food and drink innovators and non-innovators, and Baregheh et al. (2012) and Minarelli et al. (2015), who emphasized the difference in food product, process, and marketing innovation. As stated in the introduction, our objective is to combine the three themes by studying the relationship of firm characteristics to product, process, and marketing innovation by U.S. food and drink manufacturers.

4. Methodology 4.1 Demographic and financial data First, we collected secondary data for food and drink manufacturers listed on U.S. stock exchanges during the 2000-2014 period.3 We extracted the full population of public food and drink manufacturers from Compustat4, which yielded a total of 180 firms. We deleted 23 firms for various reasons: (1) the firm is listed on the Canadian stock exchange; (2) the firm is not primarily active in the U.S.; (3) the firm is not primarily active in the food and drink industry; (4) the firm has under $1 million in revenue; and (5) missing information.5 The 110 observations with less than $1 million in revenue are deleted because of the large disproportionate impact on the sample. These observations are primarily of firms in the development stage with low revenue, negative income, substantial debt, low equity, and no patent or trademark ownership. As startups are not our primary interest, the exclusion of these observations likely contributes to better empirical estimation. Altogether, the final sample is comprised of 157 firms and 1,355 firm-year observations. As illustrated in Table 1A (Panel A), the geographical distribution of firm headquarters is rather even, although most (54 of the 157) are in California, Illinois, and New York. In terms of sectoral distribution, we use the standard industrial classification (SIC) system in which each firm is classified based on the business activity which generates the most revenue. As such, food and drink manufacturers may have activities in multiple categories, which implies caution is necessary when interpreting the data. As reported in Table 1B (Panel B), drink manufacturers form approximately 27% of the sample, whereas there are relatively few fats and oils manufacturers and bakery manufacturers. Table 2A (Panel A) reports the basic summary statistics for the demographic and financial data. The age of the mean firm is approximately 59 years, which reflects the year of incorporation to the present.6 As companies with revenue of $1 million or less have been deleted, there are few companies below the age of 3

The following three-digit SIC codes are included: 20 (food manufacturing), 201 (food manufacturing – meat products), 202 (food manufacturing – dairy), 203 (food manufacturing – fruits and vegetables), 204 (food manufacturing – grain), 205 (food manufacturing – bakery), 206 (food manufacturing – confectionery), 207 (food manufacturing – fats and oils), 208 (food manufacturing – beverages), and 209 (food manufacturing – miscellaneous). 4 https://wrds-web.wharton.upenn.edu/wrds. 5 Although the food and drink industry is assumed to be well-integrated, in particular the American-Canadian market, non-U.S. observations are deleted in order to facilitate robust as well as parsimonious analysis. Otherwise, inclusion of such observations raises the chance variability in performance is explained by cross-country differences in income, population, and other macro-economic indicators. 6 The date or year of the initial public offering (IPO) is not used to calculate firm age because few companies start on the public market. Using the year of incorporation allows consideration of resource and knowledge acquisition prior to the IPO.

International Food and Agribusiness Management Review

750

Grashuis and Dary

Volume 20, Issue 5, 2017

Table 1. Distribution of U.S. public food and drink manufacturers by region and by sector. Panel A: distribution by region

Panel B: distribution by sector

State

Census region

Total

% of total

California Illinois New York Missouri Colorado New Jersey Other Total

West Midwest Northeast Midwest West Northeast

23 18 13 11 10 9 73 157

18.70 14.63 10.57 8.94 8.13 7.32 46.50 100.00

SIC1

Description

Total

% of total

20 201 202 203 204 205 206 207 208 209

Food Manufacturing Food Manufacturing – Meat Products Food Manufacturing – Dairy Food Manufacturing – Fruits and Vegetables Food Manufacturing – Grain Food Manufacturing – Bakery Food Manufacturing – Confectionery Food Manufacturing – Fats and Oils Food Manufacturing – Beverages Food Manufacturing – Miscellaneous Total

8 16 13 21 15 8 12 4 42 18 157

5.10 10.19 8.28 13.38 9.55 5.10 7.64 2.55 26.75 11.46 100.00

SIC = standard industrial classification.

Table 2. Demographic, financial, patent and trademark ownership data statistics. Panel A: summary statistics of demographic and financial data

Panel B: summary statistics of patent and trademark ownership data

Variable

Mean

Median

Std. dev.

Firm age Employees Total assets (million) Total liabilities (million) Total equity (million) Total revenue (million) Net income (million)

58.95 13,561.04 4,736.05 2,973.59 1,734.31 4,788.44 314.01

48.00 1,870.00 485.07 178.85 201.01 569.20 15.03

45.13 31,297.95 11,178.71 7,027.81 4,392.59 10,583.05 980.75

Variable

Mean

Median

Std. dev.

Product innovations Process innovations Marketing innovations Total owned patents Patenter (>0 patents) Total owned trademarks Trademarker (>0 trademarks)

0.55 1.02 1.05 2.29 0.20 6.06 0.58

0.00 0.00 0.00 0.00 0.00 1.00 1.00

2.49 4.04 4.95 8.76 0.40 11.77 0.49

ten. Also, some companies in the panel are the products of mergers or acquisitions, which implies the mean firm age is in fact underestimated. Based on the means, medians, and standard deviations for the balance sheet and income statement data, there is great heterogeneity in the sample. While the median firm has $490 million in total assets and $571 million in total sales, both figures rise to $4.8 billion for the mean firm. As International Food and Agribusiness Management Review

751

Grashuis and Dary

Volume 20, Issue 5, 2017

such, the sample is characterized by large influential observations in the right tail, which we address in the empirical model by means of log transformation of most variables in order to obtain normal distributions. 4.2 Patent and trademark data Second, our patent and trademark data source is the U.S. Patent and Trademark Office (USPTO), which maintains an online database of all patents and trademarks, both granted and rejected as well as pending.7,8 We categorize each patent as either product, process, or marketing innovation. Correspondingly, we filed a patent as product innovation if its main claim relates to food products or food compositions. We inferred process innovation by the terms ‘method’, ‘system’, or ‘process’, and marketing innovation by the terms ‘design’ or ‘package’. We make no similar distinction for trademarks, in part because the quality of the data is not high enough to allow definitive categorization. The innovation is assumed to take place in the year the patent or trademark is filed, not granted, as there is no good reason to assume the invention is not used during the application process. Table 2B (Panel B) presents the summary statistics for the full panel and the full period. Based on patent ownership data, public food and drink manufacturers for the most part engaged in process and marketing innovation. However, the mean is not the most informative number as manufacturers with no patent ownership impose a downward bias. Of the 157 firms in the sample, only 37 (24%) patented one or more innovations during the 2000-2014 period. For the sub-sample of 37 firms, the mean number of patented innovations per year increases to approximately 6.41, which in turn is affected by seven large outliers. Specifically, Kraft Foods (786), Coca-Cola Company (480), PepsiCo (404), General Mills (342), Archer Daniels Midland (225), ConAgra (148), Wrigley (134), and Kellogg (72) together accounted for 84% (2,591 of 3,088 patents) of all patented innovations in the public food and drink industry. As such, we can conclude ownership of patented innovations is highly concentrated. Also, on average, the total number of patented product innovations per firm-year is 0.55, while the corresponding averages for patented process and marketing innovations are 1.02 and 1.05, respectively. Consequently, the data suggest food and drink manufacturers do not necessarily depend on patenting product innovations for value creation, protection, or appropriation, which is indicative of price-based as opposed to quality-based competition (Vaona and Pianta, 2008). In comparison to patent ownership, trademark ownership is more common for the protection of intellectual property (6.04 trademarks per firm-year), which corresponds to the general observation by Hall et al. (2014) regarding the relative use and importance of patents, trademarks, trade secrets, and copyrights. As such, while often neglected or dismissed in the empirical literature (Schautschick and Greenhalgh, 2016), the raw data suggest it is necessary to consider trademark ownership as an alternative or supplemental method of formal intellectual property protection in the food and drink industry. In terms of propensity, trademark ownership is observed in 58% of the firm-year observations, yet 45 of the 157 firms (29%) did not register a single trademark during the 2000-2014 period. There are once again large outliers, although trademark ownership is less concentrated as compared to patent ownership. Hershey (595), Kraft Foods (575), Pepsico (536), General Mills (533), ConAgra (528), Coca-Cola (524), Kellogg (486), and Anheuser-Busch (408) owned 51% of all registered trademarks. Temporal variation in patent and trademark ownership propensity and intensity is reported in Table 3. As illustrated, there is relatively great volatility in the intensity of patent ownership. Meanwhile, the mean number of issued trademarks peaked in 2006 and then decreased by approximately 50% in 2014. In both 7

We only record issued patents. Although the underlying information may contain economic value, a rejected patent application implies USPTO did not consider the proposed invention a true invention. If so, the applicant is unable to protect the associated income stream by patenting. Subsequently, the applicant may use secrecy instead. 8 The U.S. has a two-tiered system of trademark registration: state and federal. As discussed by O’Donnell et al. (2008), federal trademark registration has both legal and economic advantages as compared to state trademark registration. As competition in the food and drink industry is not limited to local or regional (state) environments, the public manufacturers in our sample likely do not rely much on state-registered trademarks. Hence, in our study we only record federal trademarks as registered by USPTO.

International Food and Agribusiness Management Review

752

Grashuis and Dary

Volume 20, Issue 5, 2017

Table 3. Summary statistics of patent and trademark ownership data by year. Year 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

n 107 99 94 92 90 95 89 87 87 85 85 85 90 88 82

Innovation propensity

Innovation intensity

1 or more patents

1 or more trademarks

Total patents

Total trademarks

0.15 0.16 0.23 0.20 0.20 0.17 0.19 0.21 0.18 0.22 0.27 0.22 0.24 0.22 0.20

0.51 0.53 0.66 0.59 0.63 0.64 0.63 0.67 0.59 0.55 0.65 0.53 0.53 0.51 0.51

1.58 2.12 2.26 2.30 1.93 3.12 2.43 2.56 3.01 2.26 2.87 2.45 2.08 2.18 1.29

4.55 5.62 6.00 6.84 6.66 7.37 8.24 7.34 6.39 6.56 6.85 5.31 5.13 4.41 3.70

cases, the application-issue and the application-registration lag are possible explanations for the downturn at the end.9 Table 3 also reports the evolution in patent and trademark ownership propensity. As illustrated in Table 4, sectoral heterogeneity must also be addressed. Per the summary statistics, bakery manufacturers do not patent innovations, and patented innovations in new products or compositions are relatively few for meat and fruits and vegetables manufacturers in particular. Grain manufacturers engage in each type of innovation, while fats and oils manufacturers only emphasize product and process innovation. Presenting the results for the mean group comparison tests, Table 5 gives a first impression of what separates owners and non-owners of patents and trademarks in terms of firm characteristics. Because of the panel 9

In terms of the macro-environment, a more complex explanation is offered by Damanpour (2010), who suggested the overall decrease in competition has limited motivation to innovation, as well as Archibugi et al. (2013), who observed a negative impact of the most recent economic crisis on innovation input and output.

Table 4. Summary statistics of patent and trademark ownership data by sector. SIC1 200 201 202 203 204 205 206 207 208 209 1

n 81 144 95 181 105 67 122 50 356 154

Innovation propensity

Innovation intensity

1 or more patents

1 or more trademarks

Total patents

Total trademarks

0.41 0.22 0.09 0.18 0.49 0.00 0.20 0.30 0.14 0.19

0.75 0.58 0.63 0.62 0.75 0.46 0.51 0.64 0.51 0.55

11.58 0.58 0.52 0.61 4.80 0.00 1.34 4.50 2.58 0.73

15.46 3.91 4.55 3.01 12.46 3.03 7.89 3.34 6.72 2.47

SIC = standard industrial classification. International Food and Agribusiness Management Review

753

Grashuis and Dary

Volume 20, Issue 5, 2017

Table 5. Mean group comparisons of firm characteristics for innovators and non-innovators. Firm characteristic Firm age Employees Total assets (million) Total sales (million) Net income (million) Debt ratio

Patent ownership

Trademark ownership

Yes (n=275)

No (n=1,080) t-test

Yes (n=788)

No (n=567)

t-test

93.05 39,053.13 15,288.90 15,833.43 1,211.36 0.62

50.27 6,955.41 2,048.98 1,976.06 85.52 0.69

64.82 19,105.72 6,776.94 6,863.72 489.78 0.54

50.79 5,855.20 1,899.68 1,904.28 69.72 0.88

<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.9996

<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.9141

analysis, some firms are both owners and non-owners during the 2000-2014 period, which complicates the analysis. Nonetheless, it is obvious there exist significant differences in the firm characteristics of public food and drink manufacturers in relation to patent and trademark ownership. Whether indicated by patent or trademark ownership, manufacturers which patent or trademark innovations are relatively older and larger in terms of employees, assets, sales, and profit. The difference in leverage (debt ratio) is not characterized by statistical significance. 4.3 Variables Our outcome variables are patent and trademark ownership propensity and intensity, which are binary and continuous indicators, respectively (Table 6). We include the following firm characteristics: (1) past innovation; (2) firm age, which proxies knowledge accumulation; (3) total employees, which proxies both firm size and human capital; (4) past income, which proxies R&D capacity; and (5) leverage, which proxies short- and long-term perspectives.10 Excepting leverage, we hypothesize a positive relationship of each firm characteristic to the propensity and intensity of patent and trademark ownership. We address heterogeneity in the external environment by including binary variables for the years, regions, and sectors. 4.4 Panel logistic model: innovation propensity When comparing owners and non-owners of patents and trademarks, the outcome variable is obviously binary in nature. The panel binary choice model is given by đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?(đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 1|đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; , đ?&#x2018;?đ?&#x2018;?đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; ) = đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;(đ?&#x203A;˝đ?&#x203A;˝â&#x20AC;˛đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x2018;?đ?&#x2018;?đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; )

(1)

where y is the binary indicator of patent or trademark ownership for firm i in year t, x is the vector of predictors, c is the firm-specific intercept, Î˛ is the vector of parameters to be estimated, and f denotes the functional form of the model. In practice, the choice is often between the logit model and probit model, which respectively impose a logistic and a normal distribution on the data. Theoretically, it is difficult to justify the choice of one distribution or another (Greene, 2011). Here, preference is given to the logit model to facilitate comparison to other studies on innovation in the food and drink industry with similar approaches. The underlying relationship for the outcome variable is defined as đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2014; = đ?&#x203A;˝đ?&#x203A;˝ â&#x20AC;˛ đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x2018;?đ?&#x2018;?đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x20AC;đ?&#x153;&#x20AC;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;

(2)

where each symbol is as before, y* is the latent variable, and Îľ is the stochastic term. While y* is unobserved, observed variation in patent and trademark ownership is related to the latent variable in the following manner: 10

With total employees, we use a personnel-based indicator of firm size, which may not function as the ideal proxy in the differentiation of innovators and non-innovators (Damanpour, 2010). Although we exclude a financial-based indicator of firm size (total assets) to avoid multicollinearity, we believe total employees is an adequate proxy of firm size for our sample.

International Food and Agribusiness Management Review

754

Grashuis and Dary

Volume 20, Issue 5, 2017

Table 6. Overview of model variables. Variable Innovation propensity Patent ownership Trademark ownership Product innovation Process innovation Marketing innovation Innovation intensity Total patents Total trademarks Total product innovations Total process innovations Total marketing innovations Firm characteristics Ln age Debt ratio Ln size Lagged income (billion $) Macro-level characteristics Year Sector Region

Description

Source

1 if total issued patents in year t is one or more; 0 otherwise 1 if total issued trademarks in year t is one or more; 0 otherwise 1 if total patented product innovations in year t is one or more; 0 otherwise 1 if total patented process innovations in year t is one or more; 0 otherwise 1 if total patented marketing innovations in year t is one or more, 0 otherwise

USPTO

Number of issued patents in year t Number of issued trademarks in year t Number of patented product innovations in year t Number of patented process innovations in year t Number of patented marketing innovations in year t

USPTO USPTO USPTO USPTO USPTO

Natural logarithm of year t â&#x20AC;&#x201C; year of incorporation Total liabilities/total assets Natural logarithm of total employees recorded in year t Total income recorded in year t-1

Compustat Compustat Compustat Compustat

Fiscal year relating to 10-k filing, t=1, 2,..., 14 SIC201; SIC202; SIC203; SIC204; SIC205; SIC206; SIC207; SIC208; SIC209 New England; Middle Atlantic; East North Central; North Central; South Atlantic; East South Central; West South Central; Mountain; Pacific

Compustat Compustat

1SIC200;

USPTO USPTO USPTO USPTO

U.S. Census Bureau

SIC = standard industrial classification.

đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = {

1 0

đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x2C6;&#x2014; > 0 đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; â&#x2C6;&#x2014; â&#x2030;¤ 0

(3)

In choosing random over fixed effects for our panel logistic model, we considered three advantages of random effects: (1) the ability to compare between companies; (2) the ability to generalize findings and conclusions; and (3) the ability to include time-invariant predictors (Bell and Jones, 2015; Greene, 2011). Thus, empirically, the panel random effects logistic model is defined as đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?(đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 1|đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; , đ?&#x2018;?đ?&#x2018;?đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; ) = đ?&#x153;&#x2018;đ?&#x153;&#x2018;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x2018;?đ?&#x2018;?đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x2039;đ?&#x153;&#x2039; + đ?&#x153;&#x2014;đ?&#x153;&#x2014;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;?đ?&#x153;?đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x2021;đ?&#x153;&#x2021;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x20AC;đ?&#x153;&#x20AC;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;

(4)

where each symbol is as before, Ď&#x20AC; indicates the year, Ď&#x2018; indicates the region, Îť indicates the sector, Ď&#x2020; is the vector of unknown parameters to be estimated via maximum likelihood, Îź is the between-entity stochastic term, and Îľ is the within-entity stochastic term. As motivated in our introduction, Ď&#x2020; is of primary interest to our study. We estimate Equation 4 for patent ownership and trademark ownership in general as well as each type of patented innovation (product, process, and marketing) by means of the xtlogit command in STATA (StataCorp, College Station, TX, USA).

International Food and Agribusiness Management Review

755

Grashuis and Dary

Volume 20, Issue 5, 2017

4.5 Panel negative binomial model: innovation intensity The question is not only if public food and drink manufacturers use patents and trademarks to protect intellectual property, but also how much or how often. The Poisson regression model is considered to be the most appropriate for the analysis of discrete data with many zeros and small values (Greene, 2011). However, overdispersion is apparent in our data, which motivates a negative binomial model to relax the assumption of equal conditional mean and variance functions (Greene, 2011).11 As described by Hilbe (2011), the negative binomial model is specified as đ?&#x2018;&#x2019;đ?&#x2018;&#x2019; â&#x2C6;&#x2019;đ?&#x153;&#x2020;đ?&#x153;&#x2020;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;˘đ?&#x2018;˘đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; (đ?&#x153;&#x2020;đ?&#x153;&#x2020;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x2018;˘đ?&#x2018;˘đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; ) đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?đ?&#x2018;?(đ?&#x2018;&#x152;đ?&#x2018;&#x152;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; |đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; ) = đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; !

đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;

, đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = 0, 1, 2, â&#x20AC;Ś

(5)

Here, the conditional mean Îť is linked to an exponential function of a vector of predictors and its parameter estimates, (6)

đ??¸đ??¸[đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; |đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; ] = đ?&#x2018;Łđ?&#x2018;Łđ?&#x2018;Łđ?&#x2018;Łđ?&#x2018;Łđ?&#x2018;Ł[đ?&#x2018;Śđ?&#x2018;Śđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; |đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; ] = đ?&#x153;&#x2020;đ?&#x153;&#x2020;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;(đ?&#x203A;˝đ?&#x203A;˝ â&#x20AC;˛ đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; )

Because of our panel approach, Equation 6 is further specified as đ?&#x153;&#x2020;đ?&#x153;&#x2020;Ě&#x192;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;(đ?&#x203A;˝đ?&#x203A;˝â&#x20AC;˛đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x20AC;đ?&#x153;&#x20AC;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; )

(7)

which is equivalent as đ?&#x153;&#x2020;đ?&#x153;&#x2020;Ě&#x192;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = đ?&#x153;&#x2020;đ?&#x153;&#x2020;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x153;&#x2021;đ?&#x153;&#x2021;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;

(8)

where Îź represents the firm-specific intercept. Empirically, Equation 7 and 8 translate into đ?&#x153;&#x2020;đ?&#x153;&#x2020;Ě&#x192;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = đ?&#x153;&#x2018;đ?&#x153;&#x2018;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x2018;Ľđ?&#x2018;Ľđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x2018;?đ?&#x2018;?đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x2039;đ?&#x153;&#x2039; + đ?&#x153;&#x2014;đ?&#x153;&#x2014;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;?đ?&#x153;?đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x2021;đ?&#x153;&#x2021;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x20AC;đ?&#x153;&#x20AC;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;

(9)

where each symbol is as before. Again serving as our main variables of interest, the included firm characteristics are the same as in Equation 4, which implies we expect patent and trademark ownership intensity to relate to the same firm characteristics as patent and trademark ownership propensity. Equation 9 is estimated separately for the total number of patented and trademarked innovations as well as for the total number of patented product, process, and marketing innovations by means of the xtpoisson command in STATA.

5. Results and discussion 5.1 Patent and trademark ownership propensity First, we report the panel logistic model results for innovation propensity in terms of patent and trademark ownership (Table 7). Then, we report results for the three different types of patented innovation (Table 8). While presented separately, we discuss the results simultaneously. In Table 7, we report the raw coefficients, which indicates the estimated relationship of a one-unit increase in the given variable to the log odds of innovation propensity, as well as the odds ratios, which are the exponentiated values of the coefficients. In Table 8 we only report the odds ratios to conserve space. Past innovation is found to be of statistical significance in each model, except for patented innovation in food and drink products. Specifically, past innovation is estimated to multiply the odds of patent ownership propensity by a factor of 7.07. Thus, for the most part, patented innovation in the previous year increases 11

Indeed, the estimated alpha parameter for the general Poisson model is characterized by statistical significance, which implies negative binomial is the appropriate model.

International Food and Agribusiness Management Review

756

Grashuis and Dary

Volume 20, Issue 5, 2017

Table 7. Panel logistic model results for patent and trademark ownership (innovation propensity).1 Variable Intercept Past innovation (yt-1) ln firm age ln total employees Past income (t-1) Debt ratio Region binary variables Sector binary variables Year binary variables n n (groups) Wald X2 Prob>X2 McKelvey & Zavoina’s R2 1 *, **,

Model 1 – patent ownership

Model 2 – trademark ownership

Coefficient

Odds ratio

Coefficient

Odds ratio

-12.508*** (2.775) 1.956*** (0.444) 0.862** (0.381) 0.526*** (0.177) 0.678* (0.363) -0.098 (0.523) Yes Yes Yes 1,198 136 121.16 0.0000 0.83

0.000 (0.000) 7.070 (3.138) 2.369 (0.902) 1.692 (0.299) 1.971 (0.716) 0.907 (0.474)

-3.324* (1.723) 0.905*** (0.297) 0.103 (0.318) 0.649*** (0.128) 0.331 (0.209) -1.275** (0.632) Yes Yes Yes 1,198 136 92.35 0.0000 0.54

0.036 (0.062) 2.473 (0.734) 1.109 (0.353) 1.914 (0.245) 1.393 (0.292) 0.279 (0.177)

and *** indicate significant differences at 0.05, 0.01 and 0.001%, respectively.

Table 8. Panel logistic model results for product, process, and marketing innovation propensity.1 Variable

Model 3 – product innovation Odds ratio

Model 4 – process innovation Odds ratio

Model 3 – marketing innovation Odds ratio

Intercept Past innovation (yt-1) ln firm age ln total employees Past income (t-1) Debt ratio Region binary variables Sector binary variables Year binary variables n n (groups) Wald X2 Prob>X2 McKelvey & Zavoina’s R2

0.000*** (0.000) 1.577 (0.866) 5.549*** (2.982) 1.379 (0.308) 2.297*** (0.605) 0.685 (1.010) Yes Yes Yes 1,198 136 55.74 0.0144 0.91

0.000*** (0.000) 3.802*** (1.855) 2.667** (1.103) 2.062*** (0.540) 2.457* (1.177) 1.111 (0.140) Yes Yes Yes 1,198 136 85.34 0.0000 0.85

0.000*** (0.000) 5.921** (4.254) 2.786** (1.186) 1.711*** (0.317) 2.444** (0.900) 0.730 (0.542) Yes Yes Yes 1,198 136 113.55 0.0000 0.95

1 *, **,

and *** indicate significant differences at 0.05, 0.01 and 0.001%, respectively.

the probability of patented innovation in the following year, which may imply innovation and the protection and appropriation of its value is path dependent (Antonelli et al., 2013). In terms of innovation persistence, Triguero et al. (2013) reached the same conclusion for 671 food manufacturers in Spain. Like Triguero et al. (2013), we also observe a stronger lagged effect for process innovation as compared to product innovation, although the magnitude associated with marketing innovation is even higher. As such, all else equal, prior innovations in food and drink products or compositions are the least likely to spur similar innovations in the future. Another explanation for innovation persistence in general is an intra-firm knowledge spillover effect from year to year, which corresponds to the knowledge stock interpretation. Such an effect is related to the recent interest in patent citation data, which informs the quality of the patent and therefore the quality International Food and Agribusiness Management Review

757

Grashuis and Dary

Volume 20, Issue 5, 2017

of the innovation (Bernstein, 2015; Nagaoka et al., 2010; Odasso et al., 2015). In general, a high number of forward citations is interpreted as high-quality innovation, which in theory is more likely to facilitate innovation persistence. Firm age has a significant and positive relationship to the propensity to patent each type of innovation, which corresponds to general observations by Cefis and Marsili (2006). As such, the probability of patent ownership is increasing in age, which may imply innovation and protection and appropriation of its value is a long-term process associated with organizational learning (Jimenez-Jimenez and Sanz-Valle, 2011).12 With a 5.549 increase in the odds ratio, the largest magnitude of firm age is observed in relation to the propensity of patented innovations in products or compositions. However, firm age is not characterized by statistical significance in terms of the propensity to own trademarks, suggesting both young and old firms use trademarks for intellectual property protection. With total employees as its proxy, statistical significance is also observed in terms of firm size in relation to the propensity to patent process and marketing innovations. Our result is comparable to Vancauteren (2016), who measured a positive impact of employee size on the patent portfolios of Dutch food manufacturers for the 2000-2008 period. Alternatively, we can interpret total employees as human capital stock, which is necessary for product and process innovation (Berchicci, 2013). As such, a 1% increase in human capital increases the probability of process and marketing innovation by factors of 2.062 and 1.711, respectively. Also, at an odds ratio of 1.914, food and drink manufacturers are approximately twice as likely to register trademarks as size increases by 1%. Past income, which is the one-year lagged observation of net income, is observed to have a positive relationship to each type of patented innovation. The propensity to patent innovation of any type is thus correlated with equity availability, which is interpreted as an important barrier to innovation and the protection and appropriation of its value. As argued by Dâ&#x20AC;&#x2122;Este et al. (2012), cost is the primary concern when firms consider innovation. However, the coefficient corresponds to a $1 billion increase in net income, suggesting the positive relationship to patent ownership propensity is most noticeable toward the far end of the spectrum in terms of size.13 Meanwhile, the hypothesized negative relationship of leverage to innovation is not characterized by statistical significance, except for the propensity to register trademarks. Capital structure is thus determined to only be of partial importance to the protection of intellectual property in the food and drink industry. Although we included vectors of binary variables for the years, the regions, the sectors, the parameter estimates are not reported in the interest of space.14 Overall, statistical significance is observed for many but not all binary variables, which implies patented innovation in the food and drink industry is heterogeneous across time, space, and industry. Specifically, patented innovation propensity is observed to be relatively low for the year 2014, which is likely attributable to the application-issue lag. Otherwise, there is not much consistency in the statistical significance of the control variables. 5.2 Patent and trademark ownership intensity We now proceed to the panel negative binomial model results in relation to the intensity of patent and trademark ownership (Table 9 and 10). Similar to the panel logistic model results, we report the raw coefficients as well as the incidence rate ratios, which are the exponentiated values of the coefficients and indicate the estimated impact of one-unit increases in the given variable on the expected count in terms of percentages (Hilbe, 2011).15 Table 10 only contains the incidence rate ratios in order to conserve space.16 12

Squared age, which is often included in empirical specifications to test if the causal impact of age is nonlinear, proved to be nonsignificant in our model and is therefore excluded in the table. 13 Net income of the mean firm in our sample is $314 million, and $1 billion or more in revenue is only observed in 97 of the 1,355 observations (7%). 14 Full results are available upon request. 15 The interpretation of the incidence rate ratio is intuitive. The threshold is 1, which indicates no relationship between the predictor and the outcome variable. An estimate of below 1 indicates a negative relationship, and an estimate of above 1 indicates a positive relationship. 16 Full results are available upon request.

International Food and Agribusiness Management Review

758

Grashuis and Dary

Volume 20, Issue 5, 2017

Table 9. Panel negative binomial model results for patent and trademark ownership intensity.1 Model 6 – patent ownership intensity

Model 7 – trademark ownership intensity

Variable2

Coefficient

I.R.R.3

Coefficient

I.R.R.

Intercept Past innovation (yt-1) ln firm age ln total employees Past income (t-1) Debt ratio Region binary variables Sector binary variables Year binary variables N N (groups) Wald X2 Prob>X2 AIC BIC

-5.226*** (1.399) 0.010*** (0.003) 0.194*(0.110) 0.405*** (0.090) 0.019 (0.053) -0.747* (0.451) Yes Yes Yes 1,198 136 157.64 0.0000 1,928.045 2,121.404

0.005*** (0.008) 1.010*** (0.003) 1.214* (0.133) 1.499*** (0.135) 1.019 (0.054) 0.474* (0.214)

-0.879* (0.515) 0.011*** (0.002) 0.042 (0.075) 0.282*** (0.042) -0.032 (0.041) -0.643*** (0.213) Yes Yes Yes 1,198 136 244.16 0.0000 4,726.001 4,919.361

0.415* (0.214) 1.011*** (0.002) 1.043 (0.078) 1.326*** (0.056) 0.969 (0.040) 0.526*** (0.112)

and *** indicate significant differences at 0.05 and 0.001%, respectively. = Akaike information criterion; BIC = Bayesian information criterion. 3 I.R.R. = incidence rate ratio. 2 AIC

Table 10. Panel negative binomial model results for product, process, and marketing innovation intensity.1 Variable2

Model 8 – product innovation I.R.R.3

Model 9 – process innovation I.R.R.

Model 10 – marketing innovation I.R.R.

Intercept Past innovation (yt-1) ln firm age ln total employees Past income (t-1) Debt ratio Region binary variables Sector binary variables Year binary variables N N (groups) Wald X2 Prob>X2 AIC BIC

0.004*** (0.008) 1.027** (0.013) 1.243* (0.160) 1.488*** (0.203) 1.151* (0.091) 0.486 (0.305) Yes Yes Yes 1,198 136 130.64 0.0000 933.8048 1,127.164

0.000*** (0.000) 1.021*** (0.008) 1.456*** (0.196) 1.958*** (0.233) 1.057 (0.065) 0.605 (0.337) Yes Yes Yes 1,198 136 143.87 0.0000 1,343.17 1,536.53

0.000*** (0.000) 1.011** (0.005) 1.213 (0.179) 1.983*** (0.251) 0.908 (0.058)

1 *, **,

and *** indicate significant differences at 0.05, 0.01 and 0.001%, respectively. 2 AIC = Akaike information criterion; BIC = Bayesian information criterion. 3 I.R.R. = incidence rate ratio.

International Food and Agribusiness Management Review

759

Yes Yes Yes 1,198 136 141.96 0.0000 1,265.819 1,454.09

Grashuis and Dary

Volume 20, Issue 5, 2017

In relation to past innovation, the magnitude of its estimated relationship to patent and trademark ownership intensity is low as compared to propensity. With total patented innovations in products as the outcome variable, each past patented innovation is only estimated to increase the probability of more product innovations in the following year by 2.7%, which may be indicative of a poor carryover effect. For process and marketing innovations, the positive relationship to past innovation is similarly low (incidence rate ratios of 1.021 and 1.011, respectively). While significant in each case, past innovation is therefore concluded to be of primary importance to the propensity but not the intensity of patent and trademark ownership. All else equal, a 1% increase in firm age is associated with more patented innovations in products and processes (factors of 1.243 and 1.456, respectively). As age is concluded to be non-significant to trademark ownership in both propensity and intensity, the use of trademarks to protect intellectual property is similar among young and old firms, all else equal. The relationship of firm size to the intensity of patent and trademark ownership is observed to be characterized by statistical significance in each model. Moreover, firm size is the most influential firm characteristic, as determined by the magnitude of the parameter estimates. For example, patent and trademark ownership is expected to increase by 49.9 and 32.6%, respectively, for a 1% increase in firm size. Firm size is thus important to the propensity and the intensity to protect innovations by means of patents and trademarks, which lends partial support to the Schumpeterian hypothesis of firm size, R&D expenditure, market structure, and innovation. Specifically, Schumpeter (1942) hypothesized the majority of innovation activity is generated by large firms with considerable power to drive creative destruction in the market. We do not have sufficient data to determine if the large manufacturers in our sample indeed have market power. Similar to the propensity of patent and trademark ownership, firm capital structure is apparently of limited importance to intensity. Past income is only concluded to significantly relate to the total number of patented innovations in products, but the incidence rate ratio of 1.151 is almost negligible as income is measured in billions. As for leverage, its estimated relationship to each type of patented innovation is non-significant, yet its relationship to patent and trademark ownership intensity is strong, significant, and negative. Per the incidence rate ratio, a 1% increase in the debt ratio is associated with 52.6% fewer patented innovations. The relationship to total registered trademarks is comparable. All else equal, the mean food and drink manufacturer is expected to own 47.4% fewer trademarks for each 1% increase in the debt ratio.

6. Summary and conclusions Recent developments in the agri-food industry have forced or motivated food and drink manufacturers to innovate in order to remain competitive. However, it is unclear to what extent recent research on innovation in the food and drink industry, often driven by survey data on European manufacturers, is generalizable to the United States. As such, we addressed the gap in the empirical literature by using patent and trademark ownership data to study product, process, and marketing innovation in the U.S. public food and drink industry. Considering data for the 2000-2014 period, patented innovations in the manufacturing and marketing of food and drink products are the most common. We thus concluded food and drink manufacturers do not often patent innovations in food and drink products or compositions, if they engage in much product innovation at all. We also concluded food and drink manufacturers rely more on trademark ownership, which is pursued to secure the exclusive right to use words, names, or symbols associated with product, process, and marketing innovations. The data also illustrated a dichotomy as only 20% of the 157 firms in our sample patented one or more innovations during the 2000-2014 period. Furthermore, large multinationals such as Coca-Cola, Kellogg, and Pepsi own the majority of patents and trademarks, whereas smaller organizations in general do not often own patents and trademarks to protect intellectual property. We thus observed a strong concentration of intellectual property as protected by patents and trademarks, which may or may not impact the future viability of small manufacturers in the food and drink industry.

International Food and Agribusiness Management Review

760

Grashuis and Dary

Volume 20, Issue 5, 2017

Empirically, we specified a panel logistic model to study the propensity of patent and trademark ownership and a panel negative binomial model to study the intensity of patent and trademark ownership. Firm characteristics served as our main interest, and we used random effects to address heterogeneity across time, space, and industry. Per the panel logistic model and panel negative binomial model results, firm characteristics for the most part have a significant relationship to the propensity and intensity of patent and trademark ownership. Specifically, we observed a positive association to firm size as well as firm age, which is not too surprising as patent and trademark ownership is concentrated among established and large multinationals in the food and drink industry. Furthermore, variability in patent and trademark ownership propensity and intensity might be explained by lagged income and lagged innovation, which is suggestive of path dependence in R&D investment and innovation. Together, our findings raise a general perception of cost and time barriers to innovation and the protection and appropriation of its value for small manufacturers in the food and drink industry (Dâ&#x20AC;&#x2122;Este et al., 2012), which is an important consideration as 76 of the 157 firms in our sample reported an average size of 1000 or fewer employees. In all likelihood, innovation is considered to be a long-term process with significant cost and uncertain payoff, a combination only affordable to large established firms. The result is worrisome as patenting is considered to be beneficial, if not crucial, to the survival probability of small firms (e.g. Helmers and Rogers, 2011; Rosenbusch et al., 2011). There are several weaknesses and limitations to consider. First, patent ownership may not capture the full or the true extent of product innovation in the food and drink industry as intellectual property is also often protected by means of trade secrets (Cohen et al., 2000; Oâ&#x20AC;&#x2122;Donnell et al., 2008), which concern any form of confidential information with actual or potential economic value.17 Trade secret protection is traditionally pursued by means of state law (Png, 2017), but federal governance is likely to increase as U.S. Congress passed the Defend Trade Secrets Act in 2016. Obviously, trade secrets make objective observation of product and process innovation complicated as registration is nonexistent. Second, R&D expenditure often proxies innovation, but such data proved to be unreliable in our case as many firms acknowledge R&D activity yet do not explicitly report R&D expenditure, probably in the interest of secrecy. Therefore, the impact of R&D expenditure is perhaps captured in the parameter estimate of firm size, which is likely correlated with investment in long-term growth. Third, we do not have adequate data to differentiate between firm size and human capital. Ideally, total sales or total assets is used to proxy firm size and total employees is used to proxy human capital. However, total employees is correlated with both total sales and total assets, which forced us to use total employees to proxy firm size as well as human capital to avoid multicollinearity. As such, we cannot conclude with absolute certainty if the parameter estimates for total employees report the relationship of firm size or human capital to the propensity and intensity of patent and trademark ownership, although the model results with total assets in place of total employees are similar. Our results and conclusions have several implications for managers as well as policymakers. First, considering the large impact of past innovation in terms of both propensity and intensity, managers of non-innovative food and drink manufacturers may first consider engaging in some type of research partnership or collaboration to jumpstart the process. Then, once some knowledge or intellectual property is generated, a spillover effect may facilitate independent R&D and innovation. Second, as the relationship of firm size to patent and trademark ownership is positive, managers of relatively small manufacturers may consider investing in human capital to secure specific knowledge. Since leverage is only associated negatively to trademark ownership, managers may consider debt acquisition if there is intention to make patented innovations in products or processes. Of course, capital structure decisions cannot be made in a vacuum, and any investment should be contingent on its future return. Third, policy may address the apparent concentration of patent ownership by increasing opportunities for intellectual property protection for relatively small food and drink manufacturers, thus improving ex ante incentives for specific investments in intangible assets and resources. The recent implementation of the Defend Trade Secrets Act, which is a piece of federal legislation to fight trade secret misappropriation, may constitute a step in the right direction. Fourth, policy may also lend financial or 17 In terms of economic strategy, a trade secret can last forever, but the protected information is susceptible to reverse engineering (Hall et al., 2014). However, academic discussion or exploration of the relative relevance of patents, trademarks, and trade secrets in the food and drink industry is, to our knowledge, nonexistent.

International Food and Agribusiness Management Review

761

Grashuis and Dary

Volume 20, Issue 5, 2017

technical support to small food and drink manufacturers to spur innovation and thus competitiveness in the industry. One possibility is to provide tax benefits for R&D investment or subsidies for patent or trademark applications. While we produced evidence of the relationship of firm characteristics to product, process, and marketing innovation in the food and drink industry, there exist many more open research questions to be answered. For example, what percentage of firm value is composed of patents, trademarks, and trade secrets? What is the causal relationship of patented innovation to firm performance? Is patent and trademark ownership propensity and intensity similar among non-public companies? When is patent ownership complementary with trademark ownership? We recommend future research to address such questions to further our collective understanding of innovation, particularly in the food and drink industry.

References Adjemian, M.K., B.W. Brorsen, W. Hahn, T.L. Saitone and R.J. Sexton. 2016. Thinning markets in U.S. agriculture. Economic Information Bulletin Number 148. U.S. Department of Agriculture, Washington, WA, USA. Alfranca, O., R. Rama and N. von Tunzelmann. 2003. Technological fields and concentration of innovation among food and beverage multinationals. International Food and Agribusiness Management Review 5: 1-14. Allred, B. B. and W.G. Park. 2007. The influence of patent protection on firm innovation investment in manufacturing industries. Journal of International Management 13: 91-109. Antonelli, C., F. Crespi and G. Scellato. 2013. Internal and external factors in innovation persistence. Economics of Innovation and New Technology 22: 256-280. Archibugi, D., A. Filippetti and M. Frenz. 2013. The impact of the economic crisis on innovation: evidence from Europe. Technological Forecasting and Social Change 80: 1247-1260. Arrow, K.J. 1962. The economic implications of learning by doing. Review of Economic Studies 29: 155-173. Aung, M.M. and Y.S. Chang. 2014. Traceability in a food supply chain: safety and quality perspectives. Food Control 39: 172-184. Avermaete, T., J. Viaene, E.J. Morgan, E. Pitts, N. Crawford and D. Mahon. 2004. Determinants of product and process innovation in small food manufacturing firms. Trends in Food Science and Technology 15: 474-483. Baregheh, A., J. Rowley, S. Sambrook and D. Davies. 2012. Innovation in food sector SMEs. Journal of Small Business and Enterprise Development 19: 300-321. Bell, A. and K. Jones. 2015. Explaining fixed effects: random effects modeling of time-series cross-sectional and panel data. Political Science Research and Methods 3: 133-153. Berchicci, L. 2013. Towards an open R&D system: internal R&D investment, external knowledge acquisition and innovative performance. Research Policy 42: 117-127. Bernstein, S. 2015. Does going public affect innovation? The Journal of Finance 70: 1365-1403. Bhaskaran, S. 2006. Incremental innovation and business performance: small and medium-size food enterprises in a concentrated industry environment. Journal of Small Business Management 44: 64-80. Bostaph, S. 2013. Driving the market process: ‘alertness’ versus innovation and ‘creative destruction’. Quarterly Journal of Austrian Economics 16: 421. Capitanio, F., A. Coppola and S. Pascucci. 2010. Product and process innovation in the Italian food industry. Agribusiness 26: 503-518. Cefis, E. and O. Marsili. 2006. Survivor: the role of innovation in firms’ survival. Research Policy 35: 626-641. Ciliberti, S., L. Carraresi and S. Bröring. 2016. External knowledge sources as drivers for cross-industry innovation in the Italian food sector: does company size matter? International Food and Agribusiness Management Review 19: 77-98. Cohen, W.M., R.R. Nelson amd J.P. Walsh. 2000. Protecting their intellectual assets: appropriability conditions and why US manufacturing firms patent (or not). Working paper 7552. National Bureau of Economic Research, Cambridge, MA, USA. International Food and Agribusiness Management Review

762

Grashuis and Dary

Volume 20, Issue 5, 2017

D’Este, P., S. Iammarino, M. Savona and N. von Tunzelmann. 2012. What hampers innovation? Revealed barriers versus deterring barriers. Research Policy 41: 482-488. Damanpour, F. 2010. An integration of research findings of effects of firm size and market competition on product and process innovations. British Journal of Management 21: 996-1010. Flikkema, M., C. Castaldi, A.P. De Man and M. Seip. 2015. Explaining the trademark-innovation linkage: the role of patents and trademark filing strategies. In: Academy of Management Proceedings 2015(1). Academy of Management, Briarcliff Manor, NY, USA. Galizzi, G. and L. Venturini. 1996. Product innovation in the food industry: nature, characteristics and determinants. In: Economics of innovation: the case of food industry. Physica-Verlag, Heidelberg, Germany, pp. 133-153. Gehlhar, M.J., A. Regmi, S.E. Stefanou and B.L. Zoumas. 2009. Brand leadership and product innovation as firm strategies in global food markets. Journal of Product and Brand Management 18: 115-126. Gellynck, X. and B. Kühne. 2010. Horizontal and vertical networks for innovation in the traditional food sector. International Journal on Food System Dynamics 1: 123-132. Gellynck, X., B. Vermeire and J. Viaene. 2007. Innovation in food firms: contribution of regional networks within the international business context. Entrepreneurship and Regional Development 19: 209-226. Ghazalian, P.L. and A. Fakih. In press. R&D and innovation in food processing firms in transition countries. Journal of Agricultural Economics. DOI: http://doi.org/10.1111/1477-9552.12186. Giovannetti, G., G. Ricchiuti amd M. Velucchi. 2011. Size, innovation and internationalization: a survival analysis of Italian firms. Applied Economics 43: 1511-1520. Gopinath, M. and U. Vasavada. 1999. Patents, R&D, and market structure in the US food processing industry. Journal of Agricultural and Resource Economics 24: 127-139. Government of Australia. 2014. The Australian food industry: a patent analytics report. Department of Industry, Canberra, Australia. Greene, W.H. 2011. Econometric Analysis, 7th Edition. Prentice Hall, New Jersey, NJ, USA. Grunert, K.G. 2005. Food quality and safety: consumer perception and demand. European Review of Agricultural Economics 32: 369-391. Hall, B., C. Helmers, M. Rogers, and V. Sena. 2014. The choice between formal and informal intellectual property: a review. Journal of Economic Literature 52: 375-423. Helmers, C. and M. Rogers. 2011. Does patenting help high-tech start-ups? Research Policy 40: 1016-1027. Hilbe, J.M. 2011. Negative binomial regression. Cambridge University Press, Cambridge, UK. Jiménez-Jiménez, D. and R. Sanz-Valle. 2011. Innovation, organizational learning, and performance. Journal of Business Research 64: 408-417. Johnson, A.J., C.C. Dibrell and E. Hansen. 2009. Market orientation, innovativeness, and performance of food companies. Journal of Agribusiness 27: 85-106. Karantininis, K., J. Sauer and W.H. Furtan. 2010. Innovation and integration in the agri-food industry. Food Policy 35: 112-120. Kirzner, I.M. 1997. Entrepreneurial discovery and the competitive market process: an Austrian approach. Journal of Economic Literature 35: 60-85. Kirzner, I.M. 2009. The alert and creative entrepreneur: a clarification. Small Business Economics, 32: 145-152. Martinez, M.G. and J. Briz. 2000. Innovation in the Spanish food & drink industry. International Food and Agribusiness Management Review 3: 155-176. Martínez, C. and R. Rama. 2012. Home or next door? Patenting by European food and beverage multinationals. Technology Analysis and Strategic Management 24: 647-661. Minarelli, F., M. Raggi and D. Viaggi. 2015. Innovation in European food SMEs: determinants and links between types. Bio-based and Applied Economics 4: 33-53. Moser, P. 2013. Patents and innovation: evidence from economic history. The Journal of Economic Perspectives 27: 23-44. Nagaoka, S., K. Motohashi and A. Goto. 2010. Patent statistics as an innovation indicator. In: Handbook of the Economics of Innovation, Volume 2. Elsevier, Amsterdam, the Netherlands, pp. 1083-1127. O’Donnell, R.W., J.J. O’Malley, R.J. Huis and G.B. Halt. 2008. Intellectual property in the food technology industry: protecting your innovation. Springer Science and Business Media, New York, NY, USA. International Food and Agribusiness Management Review

763

Grashuis and Dary

Volume 20, Issue 5, 2017

Odasso, C., G. Scellato and E. Ughetto. 2015. Selling patents at auction: an empirical analysis of patent value. Industrial and Corporate Change 24: 417-438. Organization for Economic Cooperation and Development. 2005. Oslo manual: guidelines for collecting and interpreting innovation data. OECD Publishing, Paris, France. Png, I.L.P. 2017. Law and innovation: evidence from state trade secrets laws. Review of Economics and Statistics 99: 167-179. Posner, R.A. and W.M. Landes. 2003. The economic structure of intellectual property law. Harvard University Press, Cambridge, MA, USA. Rosenbusch, N., J. Brinckmann and A. Bausch. 2011. Is innovation always beneficial? A meta-analysis of the relationship between innovation and performance in SMEs. Journal of Business Venturing 26: 441-457. Schautschick, P., and C. Greenhalgh. 2016. Empirical studies of trade marks – the existing economic literature. Economics of innovation and new technology 25: 358-390. Schumpeter, J. 1942. Capitalism, socialism, and democracy. Routledge, London, UK. Traill, W.B. and M. Meulenberg. 2002. Innovation in the food industry. Agribusiness 18: 1-21. Triguero, Á., D. Córcoles and M.C. Cuerva. 2013. Differences in innovation between food and manufacturing firms: an analysis of persistence. Agribusiness 29: 273-292. Van Galen, M., K. Logatcheva, T. Bakker, E. Oosterkamp and G. Jukema 2013. Innovation in the food industry: an international benchmark study. LEI Report 2013-036, LEI Wageningen UR, Wageningen, the Netherlands. Vancauteren, M. 2016. The effects of human capital, R&D and firm’s innovation on patents: a panel study on Dutch food firms. The Journal of Technology Transfer 1-22. Vaona, A. and M. Pianta. 2008. Firm size and innovation in European manufacturing. Small Business Economics 30: 283-299. Weiss, C.R. and A. Wittkopp. 2005. Retailer concentration and product innovation in food manufacturing. European Review of Agricultural Economics 32: 219-244.

International Food and Agribusiness Management Review

764

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 5, 2017; DOI: 10.22434/IFAMR2016.0158 Received: 26 September 2016 / Accepted: 13 July 2017

Analyzing job satisfaction and preferences of employees: the case of horticultural companies in Germany RESEARCH ARTICLE Stephan G.H. Meyerding Postdoctoral Researcher, Department of Agricultural Economics and Rural Development, Marketing for Food and Agricultural Products, Platz der Göttinger Sieben 5, 37073 Göttingen, Germany

Abstract German horticulture, as well as horticulture and agriculture in other industrialized countries, faces increasing skilled labor shortage. Additionally family run businesses in horticulture and agriculture are lacking a new generation of entrepreneurs, leading to increased structural change. Insights about job attributes attractiveness as well as their impact on job satisfaction lead to a more transparent environment in which employers and employees can make better-informed decisions and redesign the professional environment, resulting in increased job satisfaction, performance and career sustainability. For this purpose, a survey was undertaken from August 2013 to February 2015 through a questionnaire examining the preferences and perception of employees (N=229) regarding job characteristics. The theoretical background of the study is Warr’s vitamin model, which assumes non-linear relationships between job characteristics and job satisfaction. The strongest connections with job satisfaction among employees are with future prospects and conflict between work-andfamily. The study is one of the first of its kind to provide a detailed overview of job satisfaction of different groups of employees in German horticulture. Keywords: social sustainability, human resources management, employee well-being, horticulture, vitamin model JEL code: J28, J24, J43, J81, J71, J64 Corresponding author: stephan.meyerding@uni-goettingen.de

765

Meyerding

Volume 20, Issue 5, 2017

1. Introduction Personnel costs account for approximately 40% (own calculation based on the analysis of the main operational comparison in October 2012 by the Centre for Business Management in Horticulture and Applied Research e. V.) of all costs in German horticulture. Employee-related topics will be the cause of fundamental transformation processes in German companies in all industries in the next two decades (Claβen and von Kyaw, 2007: 17), and are seen as key to the future success of horticulture in Germany (Meyerding, 2015; Schreiner et al., 2013: 73-76). Another aspect is the growing attention of consumers and society on the social dimension of sustainability (Lacy et al., 2010), the potential of which has not yet been addressed by a number of sustainability rating systems in agriculture (Meyerding, 2015). Therefore knowledge of the preferences of employees regarding certain job features and the characteristics values of these job features is significant for the development of German horticulture. There are diverse approaches to the measurement and understanding of psychological satisfaction at work (Eid and Larsen, 2008; Von Rosenstiel et al., 2000). If job satisfaction is to be measured by the subjective assessment of the values of different aspects of the work, it is essential to determine what aspects should be included in the valuation model. Specifically, it is necessary to decide whether environmental characteristics (aspects of the work) alone or additional personal features are to be included in the analysis (Warr, 2012, 2013). One way of evaluating job satisfaction is to use Herzberg’s (Herzberg, 1966; Herzberg et al., 1959) two-factor model. For example, Bitsch and Hogberg (2005) used parts of Herzberg’s model for a qualitative study in U.S. horticulture based on 31 interviews. More recently, Reiche and Sparke (2012) performed a quantitative study with 446 vocational and master craftsman scholars, adopting an innovative analytical approach, but also mainly based on Herzberg’s model (Meyerding, 2015). Although the theoretical use of Herzberg’s et al. (1959) model is widespread in business administration, personnel management literature and relevant studies, it is not supported by empirical studies (Von Rosenstiel et al., 2000). According to the author’s knowledge, there is little quantitative research in German horticulture that has analyzed the nature of job satisfaction. Many of the empirical studies examining the relation between job attributes and employee well-being and health have been inspired by Karasek’s (1979) job-demand-control model. The model postulated the importance of two particular job attributes in their effect on the well-being and health of an individual: decision latitude and job demands. The later has been described as ‘the psychological stressors involved in accomplishing the work load’ (Karasek, 1979). It refers to concepts such as complex work, time-pressure, and high working pace. Decision latitude is described as the potential control of employees over tasks together with their individual skill usage. Warr (2007) developed a conceptual framework that can be seen as a reaction and addition to Karasek’s job-demand-control model. In Warr’s vitamin model, he uses the way vitamins affect human health as a metaphor for the effect of environmental influences on well-being and mental health (Warr, 2007). Central attributes of the vitamin model are its comprehensive description of the concept of mental health. The curvilinear relations between job attributes and different kinds of well-being and mental health, the differential effects of specific job attributes and the effect of personal characteristics. A central theme of Warr is the assumption that different job attributes are associated with different dimensions of well-being in different ways. For example, job demands are assumed to be more strongly related to the comfort-anxiety dimension while job autonomy should be more associated with enthusiasm-depression. In the present study it is not distinguished between different dimensions of well-being, but focused on the concept of job satisfaction. Differential relations have been found in earlier studies using the vitamin model (De Jong and Schaufeli, 1998; De Jong et al., 1999; Jeurissen and Nyklicek, 2001; Meyerding, 2015; Warr, 1990). Few studies investigating curvilinear relations between job attributes and mental health have shown some evidence for non-linear relationships (De Jong and Schaufeli, 1998; Jeurissen and Nyklicek, 2001; Meyerding, 2015, 2016).

International Food and Agribusiness Management Review

766

Meyerding

Volume 20, Issue 5, 2017

The objective of this study is to examine the relationships between job aspects and job satisfaction in German horticulture, to shown the preferences of employees according to the job aspects in question and investigate the likely effects of personal characteristics. At the same time, it demonstrates that a job satisfaction indicator, based on the survey method used here, contains the most important aspects relevant to job satisfaction in German horticulture. In addition, the preferences with respect to the job characteristics investigated are considered. This is intended to show which areas in particular need to be considered by horticultural companies to be attractive employers today and in the future. This empirical quantitative study examines the relationships between 28 job characteristics and job and life satisfaction in German horticulture. Furthermore, the preference structure with respect to the job characteristics investigated is clarified. For this purpose, more than 229 complete records were examined. The theoretical background and the methodology of the study are based on the current version of Warr’s (2007) vitamin model, which provides 12 job features (vitamins) and their hypothetical utility functions. The utility functions describe the nature of the relationship between job features and different forms of satisfaction. The operationalization of the vitamin model and the curves of the utility functions in German horticulture have previously been tested by Meyerding (2015). In principle, satisfaction measures aim to make human emotion observable. To investigate why some people are happier than others, this article considers perspectives that are either ‘eco-centric’ or ‘person-centered.’ In the first case, the influence of the environment on the satisfaction of an individual is analyzed. In the second case the influence of a person’s own characteristics on individually experienced satisfaction is analyzed.

2. Material and methods 2.1 Job satisfaction measurement based on the vitamin model The main difference between Warr’s vitamin model and other models (Brayfield and Rothe, 1951; Herzberg et al., 1959; Weiss et al., 1967) is not the choice of features, but the idea that the expression of the characteristics does not have a linear relationship with satisfaction. Consequently, there are features that have diminishing marginal utility or negatively affect satisfaction when showing very high values (Figure 1). Satisfaction and the course of the utility function vary from person to person. A stronger link between low values for environmental aspects and subjective satisfaction has been shown in several studies. Cummins (2000) argues that subjective satisfaction is maintained at a stable individual level as long as the values of environmental characteristics are in the moderate range. Within this range, it is to be expected that people adjust their expectations and feelings, and so maintain their individual equilibrium (Headey and Wearing, 1992: 39). For each of these characteristics, a deficit is perceived as an active threat and does not represent a desirable goal.

Satisfaction

Subjective optimum

(CE)

(AD)

Environmental attribute

Figure 1. The vitamin analogy: bell-shaped curve of the utility function (AD) and diminishing marginal utility (CE) (adapted from Warr, 2013). International Food and Agribusiness Management Review

767

Meyerding

Volume 20, Issue 5, 2017

Table 1 lists the 12 features of the vitamin model. The operationalizations of these latent variables are identified by lowercase letters. The type of utility function is given by the abbreviations (CE) for diminishing marginal utility and (AD) for a bell-shaped curve (Warr, 2013). The vitamins are operationalized in this study through their 28 aspects. Meyerding (2015) has successfully tested the operationalization through confirmatory factor analysis, and the validity of the entire model in the context of German horticulture using a structural equation model. ■■ Job characteristic 1: the possibility of personal influence Two aspects of this job characteristic must be considered: intrinsic and extrinsic (Karasek, 1979). The intrinsic aspect (1a) involves the adaptation of other job characteristics, such as the opportunity for self-determination in the level of skill use, objectives, and variation from time to time. The extrinsic aspect (1b) relates to one’s influence on the organization as a whole. High perceived responsibility can lead to fears of bad decisions, cognitive overload, and fear of unpredictable events. Table 1. The 12 vitamins of Warr’s model and their 28 aspects for evaluation by employees.1 Vitamin model

Job features in the present study

Utility function

1 Control (AD)

1a Task discretion 1b Influence over the wider organization 2a Skill use 2b New learning 3a Number of job demands 3b Difficulty of job demands 3c Task coherence 3d Conflict between job demands 3e Conflict between work and home 3f Emotional dissonance (inverse) 4a Range of different tasks 5a Future predictability (excludes job tenure) 5b Clear role requirements 5c Availability of feedback 6a Amount of social contact 6b Quality of social contact 7a Pay level 8a Pleasant environment 8b Safe work practices 8c Adequate equipment 9a Value to society 9b Significance to self 10a Supervision is considerate 10b Supervisor is supportive 11a Job security 11b Good future prospects 12a Fair treatment of employees 12b The organization’s morality in society

AD AD AD CE AD AD CE CE (inverse) CE (inverse) AD AD AD AD AD AD AD CE CE CE CE CE CE CE CE CE CE CE CE

2 Skill (AD) 3 Goals (AD)

4 Variety (AD) 5 Clarity (AD)

6 People (AD) 7 Money (CE) 8 Physical security (CE)

9 Significance (CE) 10 Supervision (CE) 11 Career (CE) 12 Fairness (CE) 1

There are two possible utility functions for the vitamins and relevant job characteristics: decreasing marginal utility (CE) and a bell-shaped curve (AD); the vitamins and many job characteristics are from Warr (2007: 239-240); cf. Meyerding (2015).

International Food and Agribusiness Management Review

768

Meyerding

Volume 20, Issue 5, 2017

■■ Job characteristic 2: the possibility of using one’s skills There is a range of evidence that employees who are limited in the use of their skills are less satisfied than others. The personal appreciation of own skills (2a) is illustrated by the study of Lewin et al. (1944). The ability to learn (2b; Patterson et al., 2004), i.e. to acquire new skills, is a key feature in the model of positive progression in work presented by Spreitzer et al. (2005). ■■ Job characteristic 3: external goals At low levels of this job characteristic, few demands are placed on the individual and there is little external pressure to carry out an activity. Very high levels of externally set goals require attaining many and/or difficult goals; this can lead to feelings of being badgered, inability to maintain the quantity or quality of work output required, and possibly fear of failure (Warr, 1987). This third job characteristic has six aspects. The first aspect (3a) concerns the number of tasks. The second aspect (3b), which is associated with the first, is the possibility of requirements that are too demanding. This results in dissatisfaction according to the bell-shaped utility function. Thus, an individual optimum exists: a certain amount of goal setting, which requires a degree of effort, is desirable. However, increasing demands lead to overload (MacDonald et al., 2001). The third aspect (3c) is task identity. A specific aspect of goals in a work environment is the degree to which the individual tasks are positively related. Hackman and Oldham (1975) defined work identity as the degree to which it requires doing a task ‘completely,’ that is performing something from the beginning to the end with a visible result. The fourth aspect (3d) is role conflict. High demands are observed to result from a conflict of roles. Another form of role conflict occurs between working life and private life, and is described as work-home conflict and work-family conflict (3e; Greenhaus and Beutell, 1985). The sixth aspect (3f) is emotional dissonance. A special form of high job demands arises in situations in which the employee is expected to show an emotion without actually feeling it (Glomb et al., 2004). ■■ Job characteristic 4: variety This feature of the vitamin model includes variations (4a) in the conditions of the workplace and in the activities which are carried out by the employees. People prefer change in their experiences to find a balance between comfort and relief from repetitive routines and behaviors (Kornhauser, 1965). ■■ Job characteristic 5: clarity of the environmental situation A lack of clarity in the environmental situation is undesirable in itself. Caplan (1975) developed a ‘jobfutures-ambiguity’ scale for the first of the three aspects of this characteristic; this includes the security of career development (5a) and the expected future value of one’s own abilities. The second aspect is the clarity of the role (5b), which includes the amount of information provided on what behavior and performance levels are required (Glazer and Beehr, 2005). The third aspect is feedback (5c) on one’s own performance, which is essential to gain or maintain the personal ability to influence the environmental situation, as well as for the development and use of one’s own abilities. ■■ Job characteristic 6: contact with other people Two aspects of this feature should be considered: first, the quantity (6a) of contact and second the quality (6b) of the contact. ■■ Job characteristic 7: money The amount of salary is of personal importance, not only to secure one’s existence or a particular lifestyle, but for its social importance, ‘equal’ treatment, and as a sign of personal success (Srivastava et al., 2001). The relationship between income and satisfaction is higher in low-income groups than at a moderate level International Food and Agribusiness Management Review

769

Meyerding

Volume 20, Issue 5, 2017

(Kornhauser, 1962). Studies in this field are often based on a version of equity theory (Adams, 1963). This implies that people compare their ratio of input and income to the input-output ratio of other people. Inputs are usually defined in terms of skills, effort, qualifications, working conditions, and working hours, etc. ■■ Job characteristic 8: physical security At work, central issues of physical security (8) are the absence of danger (8b) and the presence of good working conditions, ergonomically appropriate equipment (8c), and comfortable and safe levels of temperature and noise exposure (8a), for example. Inadequate equipment can be both intrinsically undesirable and a cause of dissatisfaction, resulting in errors and interruptions in the working process (Salvendy, 2012: 708). ■■ Job characteristic 9: valued social position Professions and jobs differ in the value that is attributed to them by society but also within an organization. It transpires that job satisfaction is related to this subjective value (9a; Bradburn and Caplovitz, 1965). The ‘task significance’ scale of the job diagnostic survey records the importance that employees attach to their work tasks (9b; Hackman and Oldham, 1975). This job characteristic is open to subjective interpretations to a greater extent than others. ■■ Job characteristic 10: supportive leadership Between 1950 and 1970 (Ronan, 1970), studies of aspects of leadership were often carried out using interviews and questionnaires containing items addressing the perceptions of staff concerning the extent to which the behavior of executives was considerate (10a). The questions concerned also the support (10b) and the respect shown to the employees, compliance with the welfare of employees, and the tendency to praise and value their work. Behavior that can be defined as considerate includes a willingness to listen and to accept suggestions from employees. ■■ Job characteristic 11: career prospects A career (11) is often understood as upward advancement in the job hierarchy. Careers can, however, also develop positively for the individual in other ways, such as through career changes, or assuming an alternative role or teaching activities. Two aspects of career prospects should be considered. First, jobs differ in their job and income security (11a). The second aspect of career prospects is the possibility of taking on another role (11b). ■■ Job characteristic 12: equality The last job characteristic with diminishing marginal utility in the vitamin model includes two aspects of equality (12): fairness in the relationship between an employee and employer (12a), and fairness of the company toward society as a whole (12b). The relations to be examined may be influenced by third factors. Third factors, which only have an impact at a single point in time, can include current mood, the social environment, or even the weather. Long-term variables include demographic characteristics, such as age, education, gender, skills, and the character of employees. The characteristics also have a tendency for constant evaluation, so a person perceives both the environmental situation and subjective satisfaction more positively or more negatively at different times. In addition, the different selection of jobs as a third variable could be responsible for an observed correlation. For example, less satisfied people might tend to choose jobs with certain characteristics or not to change jobs (‘drift hypothesis’), so that a recorded association between these characteristics or characteristic values and dissatisfaction will be due, at least in part, to differences in the sample, rather than to the job characteristics

International Food and Agribusiness Management Review

770

Meyerding

Volume 20, Issue 5, 2017

themselves. Other third variables that may have an impact on an observed correlation are additional job features not included in the study. 2.2 Personal characteristics and their influence on subjective satisfaction Environmental features are only partly responsible for satisfaction. Understanding the influence of personal characteristics is important to ensure proper interpretation of the results of the measurement. ■■ Different individual comparison processes and reference values When employees are asked to evaluate a particular job characteristic, they turn to comparisons and look for reference values with which they compare their individual situations. Depending on which reference values are used, the subjective evaluation and therefore the level of satisfaction change, regardless of the objective value of the characteristic. ■■ Demographic characteristics and their influence on subjective satisfaction Overall, women tend to exhibit higher job satisfaction than men (Grandey et al., 2005). Whether a causal relationship between gender and satisfaction can be determined here is questionable. However, with regard to job characteristics, differences between the genders can be observed (Grandey et al., 2005). It has frequently been observed that older people are more satisfied than younger (Mroczek and Spiro, 2005). The reasons for the higher satisfaction of older people could be changes in their evaluation processes, but also various aspects in the values of job characteristics. ■■ Different forms of employment and their influence on subjective satisfaction A third group distribution could relate to different forms of employment with different characteristics, for example, part-time and full-time jobs, or fixed and temporary employment (Trzcinski and Holst, 2003). It is also useful to distinguish between core workers and loan or seasonal workers as seasonal workers occupy a special position in German horticulture and contribute significantly to the performance process. ■■ Influence of individual personality on subjective satisfaction Other personal factors that have an impact on subjective satisfaction are personality and innate cognitive differences. Although these personality traits are not considered in this study, but it is crucial to note that these partly innate character traits have an influence on perceived subjective satisfaction. ■■ Influence of individual abilities and level of education Also, one’s own abilities and level of education could affect subjective satisfaction (Ganzach, 1998). However, a higher level of education increases the likelihood of having a job that has beneficial levels of different job characteristics (Ganzach, 2003). 2.3 Research design As mentioned earlier an online-questionnaire was designed and distributed mainly through social media between late 2013 and early 2014. In total 229 complete data sets from employees of horticultural companies were collected. SPSS 24 (IBM, Armonk, NY, USA) was used to perform the statistical analysis. The questionnaire contained first the preference measurement for all 28 job attributes using a Likert scale from 1 for unimportant to 6 for essential. The participants were asked to think about their dream job and how important the job attributes investigated are in this dream job. The concept of a dream job was used International Food and Agribusiness Management Review

771

Meyerding

Volume 20, Issue 5, 2017

because the participants should take a step away from their current job, to think about freely what is really important to them in working life. The second part of the questionnaire was about the values of the 28 job attributes in the current job. A Kunin (1955) scale was used for this part, to help the participants to grasp the nature of the two different kind of job attributes according to the vitamin model: bell-shaped curve of the utility function or additional decrement (AD) attributes, where too high values are harmful, and diminishing marginal utility or constant effect (CE) attributes, where more of this attribute, after a subjective optimum, does not lead to an increase in subjective well-being. The Kunin scale uses emotion icons (faces) to assess the values of characteristics. This is especially necessary as they are often assumed to be linear. All variables were coded in a way that higher values of the variable represent better/more positive states of the attribute in question. As mentioned earlier, in the case of aspects for which a bell-shaped curve of the utility function is assumed (AD) (1a, 1b, 2a, 3a, 3b, 4a, 5a, 5b, 5c, 6a and 6b), the transformed variables have been used, in this case very high characteristic values were coded as very low, as both should have a negative impact on job satisfaction. Accordingly, for AD aspects, the coding for the transformed variables is as follows: 1,2,3,4,3,2,1. The two aspects concerning conflict (3d and 3e) are inversely coded due to their relationship, so that high conflict levels are indicated by low codings. In a third part of the questionnaire, the participants were asked to provide some social-demographic information such as age, education, and gender as well as data about the company they are currently working in, like branch, number of employees, etc. To rank the preference for the job attributes the mean values for the preference measurement of the first part of the questionnaire were calculated. This shows which attributes are especially important from the opinion of employees of horticultural companies. These preferences does not tell us the impact of the job attributes on job satisfaction. To show the effect of each attribute on job satisfaction the Spearman correlation coefficients for the relationship of the values of the job attributes in the current job and job satisfaction were calculated. As mentioned earlier, the variables are code in a way that higher values indicate positive states of the job attribute. This was particularly important for the bell-shaped (AD) attributes. Furthermore, personal characteristics and characteristics of the position and company the employees are working in can have an influence on job satisfaction but also on job attribute preferences and subjective perceived values (states) of the job attributes in the current job. For this reason, the Spearman correlations of the personal and structural characteristics with preferences and current job attribute evaluation are calculated as well. Another question is if the personal and structural characteristics substantially help to explain job satisfaction? To answer this question two regression models, one without and one including these additional variables are compared. To identify job attributes where action is needed, the attributes are also presented in an importance grid combining preferences for and impact of the different job attributes. If an attribute is both, important to the employees and has also a strong relation with job satisfaction, it should be under particular consideration by the employers. To see which job attributes in horticulture are negatively or positively rated by employees, the mean values of the evaluation of the job attributes in the current job are shown. For identifying future trends and possible differences between the generations the result of the present study are compared to those of a study focusing on vocational and master craftsmen scholars as well as students of horticulture science (Meyerding, 2016) are compared at the end of the result section.

International Food and Agribusiness Management Review

772

Meyerding

Volume 20, Issue 5, 2017

4. Results 4.1 Sample description Employees of horticultural companies took part in the study from August 2013 to February 2014; 229 full records are included in the analysis. In Figure 2 (left), the origins of the participants are shown on a map of Germany, using the respondents’ zip codes. The distribution on the map shows that the study participants come from all over Germany, with the eastern part less well represented than the western part. Participants in the study represent the diversity of the sectors in German horticulture (Figure 2, center). Thus, vegetable farms and fruticulture are underrepresented in the study. They account for 59% (vegetable farms) and 22% (fruticulture) of the population of workers in German horticultural production (Statistisches Bundesamt, 2006). The age distribution, shown in a histogram in Figure 2 (right), shows that the study participants are (as expected) not normally distributed: there are two accumulations around 27 years and 51 years. The first accumulation, as well as the low average age of 35 years, can be explained by the use of social media as a distribution channel, social media being predominantly the province of younger workers (Busemann, 2013). Overall, the level of education in the sample is slightly higher than would be expected in the population. The relatively high level of education within the sample can be explained by the fact that young participants have a higher level of formal education than older participants (Piopiunik and Wöβmann, 2011), and the average age of respondents is relatively low at 35 years. Furthermore, it can be suggested that people with a higher level of education are more likely to be willing to participate in an online survey generated by a university institution (Häder, 2015: 180). In the sample, the proportion of companies with more than 10 employees is slightly larger (Figure 3) than in the population as approximately 80% of the horticultural production companies have fewer than 10 employees, but the horticulture census of 2005 does not distinguish between seasonal workers and permanent employees (Statistisches Bundesamt, 2006: 72-73). Of the study participants in the observed group of employees, 42% are women. The vast majority (82%) are not self-employed, have a permanent contract (74%), and work full time (84%). Only 6% would call themselves loan or seasonal workers. Many of the study participants have a form of management responsibilities (61%).

Retail horticulture 9%

Floriculture 15%

Tree nurseries 12%

Frequency of occurrence

12 Vegetable farms 5%

Services 37%

Pomiculture 5% Multidivision companies 11%

Trade firms 6%

Figure 2. Sample description.

International Food and Agribusiness Management Review

773

Distribution of age Ø 35 year

8 6 4 2 0

40 Age

Meyerding

Volume 20, Issue 5, 2017

up to 5 6 to 10 11 to 20 21 to 50 51 to 100 > 100

Frequency (in %)

0 10 20 30 40 50 60 70 80 90 100 21 29 26 11 5 8

Number of seasonal workers

Number of employees (excluding seasonal workers)

Frequency (in %) up to 5 6 to 10 11 to 20 21 to 50 51 to 100 > 100

10 20 30 40 50 60 70 80 90 100 76 8 7 4 1 3

Figure 3. Distribution of the number of employees and seasonal workers in the horticultural companies. 4.2 Preferences of employees regarding job characteristics The study participants were asked to evaluate the 28 aspects of the 12 job characteristics on a six-point Likert scale (from 1=unimportant to 6=essential). For this purpose, they should imagine their dream job and evaluate the aspects that would be particularly important in their fictitious dream job. This wording was chosen to gain distance from their current position. The aim of the question was to determine the general preferences of employees, regardless of their current work. Figure 4 shows the mean values of these evaluations, with a small standard deviation of 1.0 to 1.2. In first place is the employer’s fair treatment of its employees (12a), followed by opportunities to expand one’s abilities and learn (2b), the considerate behavior of the supervisor toward staff (10a), and the use of one’s own skills and experience (2a). Also, emotional dissonance (3f), operationalized with the statement ‘in my work I can be completely myself’ and task identity (3c) ‘to perform tasks from the beginning to the end with a visible result,’ are very important for the participants. The work in horticulture is often of a physical nature, so appropriate equipment (8c, tools and machines) is strongly preferred by the employees. Also, the sustainability of the employer (12b), operationalized by the statement ‘the company’s dealings with society as a whole (suppliers, customers, and the environment),’ is of great importance to the employees. The amount of salary (7a) is in the middle and is of moderately minor importance. In relation to other aspects,

12a Fair treatment of employees 2b New learning 10a Considerate supervisor 2a Skill use 3f Emotional dissonance 3c Task coherence 8c Adequate equipment 10b Supportive supervision 3e Work/home conflict 11a Job security 9b Significance to self 12b Moral organization 11b Good future prospects 5c Availability of feedback 1a Task discretion 8b Safe work practices 8a Pleasant environment 5b Clear role requirements 6b Quality of social contact 9a Value to society 4a Range of different tasks 7a Pay level 3d Conflict between job demands 6a Amount of social contact 3b Difficulty of job demands 3a Number of job demands 5a Future predictability 1b Influence over organization 3.00

3.55

3.86 3.81 3.81 3.72 3.71

3.50

4.21 4.21 4.20 4.19 4.19 4.15 4.13 4.13 4.06

4.00

4.32

5.13 5.08 4.99 4.98 4.90 4.80 4.79 4.76 4.66 4.63 4.61 4.56

4.50

Figure 4. Preferences regarding aspects of job characteristics. International Food and Agribusiness Management Review

774

5.00

5.50

Meyerding

Volume 20, Issue 5, 2017

the predictability of the future with respect to the job (5a, without career prospects) and one’s impact on the organization as a whole (1b, also through trade unions or works councils) are of minor significance for the employees in German horticulture. 4.3 Influence of job characteristics on job and life satisfaction The main objective of the study is to investigate which job characteristics or aspects show the strongest influence on employee satisfaction in German horticulture. For this purpose, the relationship between the subjective evaluation of the characteristics and subjective job satisfaction was observed. Table 2 lists the correlation coefficients. As normal distribution cannot be assumed, Spearman’s rho (rs) correlation coefficient was used (Field, 2009: 179-180). Also the directions of the relationships are unknown; therefore the correlation coefficients are two-tailed (Field, 2009: 176). From Table 2, it can be seen that the job aspects also correlate with each other. Thus, an employee with a high evaluation of job autonomy (1a), for example, may be able to exert a positive influence on its variety (4a, rs=0.22; P<0.01), as well as influencing working time to a greater extent, which then leads to a lower conflict between work and family life (3a, rs=0.28; P<0.01; reverse coded, so higher values reflect less conflict). The strongest connection with job satisfaction is observed for future prospects (11b), i.e. feeling that one is moving in a positive direction. Following closely in second place comes the conflict between work and family life (3e), followed by the employer’s fair treatment of employees (12a) and appropriate equipment (8c) with the same correlation coefficient. The aspects supportive leadership (10b), considerate leadership (10a), and emotional dissonance (3f) can be found in the third place. The leadership style and the possibility of being oneself at work without having to make too great an adjustment show a strong influence on employee satisfaction in German horticulture. Ranked high in fourth place are (12b) the employer’s dealings with society as a whole (customers, suppliers, and the environment) and the personal value of the work (9b), that is, the extent to which the employees identify themselves with their work. The sustainability of the horticultural company thus has a strong relationship with employee satisfaction. Employees in horticulture want to work for the ‘good,’ because this increases one’s social position (9a; value of the job for society; rs=0.41; P<0.01) and the personal value of the work (9b; rs=0.50; P<0.01). Only after these aspects are classic job features, such as adequate working environment (8a), job security (11a), and safe working processes (8b). In midfield are aspect (2b) skill learning and the aspects (3d) task conflict, (9a) the value of the job for society, (7a) the salary and (3c) the job integrity, i.e. to carry out a task from the beginning to the end with a visible result. In the bottom midfield, there are aspects such as (1b) the impact on the organization as a whole and (1a) working autonomy, (2a) being able to use one’s own abilities, (5c) the availability of feedback, and (5b) clear definition of what is expected in the respective role. The difficulty of tasks (3b) shows no significant relationship with job satisfaction in the sample. Table 2 also shows the relationships between the individual job aspects and life satisfaction as a whole. This shows a different picture from the case with job satisfaction. Here also, (11b) good prospects has the strongest correlation, but this is not as high as for job satisfaction. In second place, just as for job satisfaction is (3e) conflict between work and family life. The following points, however, differ. The third strongest connection is (3f) emotional dissonance, followed by (3d) task conflict, and (7a) salary.

International Food and Agribusiness Management Review

775

Meyerding

Volume 20, Issue 5, 2017

Table 2. Relationships between aspects of job characteristics and job and life satisfaction.1

J Job satisfaction L Life satisfaction 2b New learning 3c Task coherence 3f Emotional dissonance 7a Pay level 8a Pleasant environment 8b Safe work practices 8c Adequate equipment 9a Value to society 9b Significance to self 10a Considerate supervisor 10b Supportive supervision 11a Job security 11b Good future prospects 12a Fair treatment of employees 12b Moral organization 3d Conflict between job demands 3e Work/home conflict 1a Task discretion 1b Influence over organization 2a Skill use 3a Number of job demands 3b Difficulty of job demands 4a Range of different tasks 5a Future predictability 5b Clear role requirements 5c Availability of feedback 6a Amount of social contact 6b Quality of social contact 1 **

10a 10b 11a 11b 12a 12b 3d

1 0.53** 0.45** 0.44** 0.50** 0.44** 0.48** 0.46** 0.51** 0.44** 0.49** 0.50** 0.50** 0.46** 0.62** 0.51** 0.49** 0.44** 0.51** 0.39** 0.39** 0.35** 0.26** 0.12 0.28** 0.27** 0.32** 0.34** 0.24** 0.25**

1 0.20** 0.23** 0.29** 0.28** 0.22** 0.25** 0.20** 0.20** 0.27** 0.19** 0.22** 0.20** 0.38** 0.24** 0.24** 0.28** 0.36** 0.23** 0.21** 0.09 0.13* -0.04 0.02 0.13* 0.07 0.16* 0.13* 0.05

1 0.56** 0.47** 0.28** 0.38** 0.39** 0.35** 0.37** 0.47** 0.46** 0.45** 0.27** 0.46** 0.42** 0.30** 0.26** 0.23** 0.30** 0.28** 0.31** -0.01 0.21** 0.13 0.08 0.17* 0.39** 0.11 0.15*

1 0.59** 0.28** 0.32** 0.37** 0.37** 0.39** 0.57** 0.42** 0.41** 0.31** 0.37** 0.42** 0.36** 0.37** 0.32** 0.24** 0.18** 0.23** 0.09 0.14* 0.12 0.03 0.17** 0.34** 0.12 0.19**

1 0.34** 0.32** 0.32** 0.41** 0.42** 0.48** 0.48** 0.43** 0.30** 0.41** 0.55** 0.46** 0.39** 0.3** 0.27** 0.23** 0.23** 0.19** 0.14* 0.10 0.10 0.33** 0.23** 0.12 0.21**

1 0.34** 0.38** 0.36** 0.35** 0.27** 0.37** 0.30** 0.33** 0.31** 0.37** 0.30** 0.29** 0.41** 0.26** 0.25** 0.17* 0.24** 0.18** 0.16* 0.17** 0.23** 0.24** 0.15* 0.16*

1 0.63** 0.57** 0.37** 0.38** 0.34** 0.32** 0.41** 0.40** 0.35** 0.30** 0.26** 0.31** 0.27** 0.24** 0.16* 0.09 0.01 0.10 0.12 0.19** 0.24** 0.18** 0.22**

1 0.65** 0.41** 0.47** 0.38** 0.32** 0.36** 0.41** 0.39** 0.43** 0.36** 0.42** 0.31** 0.24** 0.24** 0.10 0.13 0.15* 0.10 0.20** 0.21** 0.16* 0.18**

1 0.42** 0.45** 0.37** 0.40** 0.40** 0.32** 0.42** 0.53** 0.37** 0.35** 0.28** 0.26** 0.24** 0.20** 0.12 0.23** 0.23** 0.34** 0.19** 0.15* 0.26**

1 0.66** 0.39** 0.42** 0.41** 0.48** 0.36** 0.41** 0.33** 0.39** 0.35** 0.33** 0.26** 0.20** 0.20** 0.22** 0.21** 0.24** 0.22** 0.23** 0.24**

1 0.50** 0.53** 0.46** 0.54** 0.45** 0.50** 0.36** 0.37** 0.33** 0.29** 0.24** 0.10 0.16* 0.19** 0.13* 0.23** 0.25** 0.22** 0.22**

1 0.82** 0.41** 0.41** 0.74** 0.58** 0.34** 0.39** 0.38** 0.27** 0.34** 0.11 0.12 0.23** 0.23** 0.22** 0.33** 0.18** 0.24**

1 0.28** 0.30** 0.19** 0.23** 0.04 0.14* 0.15* 0.16* 0.28** 0.22** 0.18**

1 0.52** 0.42** 0.29** 0.22** 0.16* 0.23** 0.37** 0.37** 0.15* 0.30**

1 0.36** 0.34** 0.1 0.22** 0.27** 0.29** 0.38** 0.17** 0.27**

1 0.23** 0.39** 0.27** 0.21** 0.32** 0.31** 0.19** 0.32**

1 0.24** 0.36** 0.26** 0.26** 0.09 0.23** 0.24**

1 0.43** 0.26** 0.20** 0.22** 0.18** 0.19**

1 0.34** 0.20** 0.14* 0.17* 0.19**

1 0.29** 0.16* 0.23** 0.26**

1 0.31** 1 0.11 0.25** 1 0.19** 0.37** 0.54**

1 0.39** 0.40** 0.71** 0.58** 0.37** 0.32** 0.41** 0.31** 0.37** 0.14* 0.17* 0.23** 0.24** 0.26** 0.36** 0.15* 0.21**

1 0.55** 0.42** 0.42** 0.22** 0.27** 0.23** 0.19** 0.23** 0.07 0.09 0.19** 0.34** 0.19** 0.15* 0.10 0.21**

1 0.45** 0.39** 0.31** 0.35** 0.29** 0.29** 0.23** 0.12 0.10 0.19** 0.21** 0.14* 0.20** 0.10 0.17*

1 0.72** 0.43** 0.40** 0.33** 0.28** 0.28** 0.15* 0.06 0.15* 0.17* 0.25** 0.30** 0.10 0.18**

1 0.42** 0.39** 0.32** 0.34** 0.27** 0.20** 0.11 0.25** 0.20** 0.27** 0.27** 0.10 0.17*

The correlation significant at the 0.01 level (two-tailed);* The correlation significant at the 0.05 level (two-tailed).

International Food and Agribusiness Management Review

776

1 0.55** 0.32** 0.24** 0.30** 0.27** 0.16* 0.16* 0.16* 0.22** 0.25** 0.17** 0.17*

Meyerding

Volume 20, Issue 5, 2017

4.4 Influence of personal and structural characteristics on job and life satisfaction, as well as on job preferences The values of the aspects of job characteristics examined explain approximately 65% of job satisfaction in German horticulture (linear regression, R2=0.65). The rest could possibly be explained by personal characteristics. Furthermore, it is necessary to check whether different groups of employees have different preferences regarding job characteristics and whether the differences are pronounced. The following personal characteristics were analyzed: age, existing management responsibilities, education level, core workforce or loan or seasonal workers, self-employment, gender, number of employees (excluding seasonal workers), and the number of seasonal workers in the participantsâ&#x20AC;&#x2122; companies (Table 3). Table 3. Relationships between personal and structural characteristics and job aspects.1 Personal characteristics

Job aspect

Effect size (rs)

Age

8a Pleasant environment 10a Considerate supervisor 8b Safe work practices 11b Good future prospects 12a Fair treatment of employees 12b Moral organization 3b Difficulty of job demands 2b New learning 3c Task coherence 9b Significance to self 10a Considerate supervisor 10b Supportive supervision 1a Task discretion 1b Influence over organization 5c Availability of feedback 11a Job security

-0.19** -0.14* -0.19** -0.24** -0.18* -0.15* 0.24** 0.20** 0.20** 0.22** 0.16* 0.15* 0.16* 0.19** 0.19** -0.21*

2b New learning 3c Task coherence 3f Emotional dissonance 8c Adequate equipment 9a Value to society 9b Significance to self 11b Good future prospects 1b Influence over organization 2a Skill use 2b New learning 9a Value to society 11a Job security 1b Influence over the wider organization 2b New learning 3c Task coherence 3f Emotional dissonance 9a Value to society 9b Significance to self 12b Orga. morality in society 5a Future predictability 2a Skill use

0.29** 0.24** 0.23** 0.18* 0.19** 0.17* 0.17* 0.14* 0.15* 0.15* 0.14* 0.18* 0.18* -0.16* -0.24** -0.18** -0.15* -0.18** -0.21** 0.16* -0.14*

Managerial responsibility (no=0, yes=1)

Temp. employment (yes=1) Self-employed (no=0, yes=1)

Gender (w=0, m=1)

Number of employees

Number of seasonal workers 1 **

Significant at the 0.01 level (two-tailed); * significant at the 0.05 level (two-tailed). International Food and Agribusiness Management Review

777

Meyerding

Volume 20, Issue 5, 2017

Older employees in German horticulture are more frequently dissatisfied with their working lives (rs=-0.15; P<0.05) and their lives in general (rs=-0.21; P<0.01) than younger employees. Men more frequently carry leadership responsibility than women in the sample (rs=-0.18; P<0.05). Female employees are more likely to have a temporary contract than their male counterparts (rs=0.20; P<0.05). Temporary employees report low values more frequently for job security (11a). Women are also more often employed part-time (rs=0.21; P<0.05). Self-employed horticulture entrepreneurs are more often males (rs=0.20; P<0.01), have more opportunities to expand their own abilities and learn (2b), are able to conclude a task from beginning to end with a visible result (3c), and can be more completely themselves at work (3f). Self-employed persons are happier with their working lives than employees (rs=0.15; P<0.05). However, no significant relationship between selfemployment and life satisfaction can be observed. Workers in horticultural companies with more employees (excluding seasonal workers) report fewer opportunities to expand their own abilities (2b), and with the higher degree of specialization in larger organizations, perceive fewer opportunities to complete a task from beginning to end with a visible result (3c). In larger organizations, employees may be less completely themselves (emotional dissonance, 3f). It is also possible to observe connections between personal and structural characteristics and preferences with regard to job characteristics (Table 4). Executives favor aspects such as job autonomy (1a), the impact on the organization as a whole (1b), and the difficulty of the task (3b). Employees with a higher level of education put astonishingly less emphasis on the compatibility between work and family life (3e), adequate equipment (8c), and own job security (11a). Table 4. Relationships between personal and structural characteristics and preferences for job aspects.1 Personal characteristics

Job aspect

Effect size (rs)

Age Managerial responsibility (no=0, yes=1)

8c Adequate equipment 1a Task discretion 1b Influence over organization 3a Number of job demands 3b Difficulty of job demands 3e Work/home conflict 8c Adequate equipment 11a Job security 8c Adequate equipment

-0.19** 0.16* 0.26** 0.17* 0.15* -0.20** -0.15* -0.17* 0.17*

1a Task discretion 1b Influence over organization 5b Clear role requirements 2b New learning 3e Work/home conflict 3f Emotional dissonance 3c Task coherence 8a Pleasant environment 8b Safe work practices 12a Fair treatment of employees

0.15* 0.16* -0.19** -0.16* -0.27** -0.17* -0.14* -0.14* -0.17* -0.20**

Education level

Fulltime employees (yes=1) Self-employed (no=0, yes=1) Gender (w=0, m=1)

Number of employees Number of seasonal workers

1 **

Significant at the 0.01 level (two-tailed); * significant at the 0.05 level (two-tailed).

International Food and Agribusiness Management Review

778

Meyerding

Volume 20, Issue 5, 2017

For women, compared to their male counterparts, the development of their own skills (2b), the compatibility of work and family life (3e), and being able to be completely themselves at work (3f) are more important than other elements. Table 5 shows the results of the linear regressions for the complete model; they are listed once without the inclusion of personal characteristics and structural variables, and once with these additional variables. The model consisting only of job aspects has an R2 of 0.65, and thus shows only a slightly lower degree of explanation than the model that includes additional personal features and structural variables for each job and the company (R2=0.69). In the regression analysis the job attribute 11b good future prospects shows the highest impact on job satisfaction, followed by 3e work-home conflict. Other attributes such as 8a pleasant environment, 10b supportive supervision, and 4a range of different tasks also show a significant effect on subjective job satisfaction. 4.5 Comparison of influence of job aspects on job satisfaction and preferences In Figure 5, the preference measures and the impact of each job aspect on job satisfaction are juxtaposed (importance grid). Special attention should paid to the job aspects in the top right field (do first). This is where both a high influence of the aspect on job satisfaction and a high preference of employees regarding this aspect come together.

Impact on job satisfaction (Spearman's rho)

0.7 Do soon

Do first

Do sometime

Plan dedicated time

0.6

0.5

0.4

0.3

0.2

0.1 3.50

3.70

3.90

4.10

4.30

4.50

4.70

Preferences of employees (six-point Likert scale)

4.90

5.10

5.30

2b New learning 3c Task coherence 3f Emotional dissonance 7a Pay level 8a Pleasant environment 8b Safe work practices 8c Adequate equipment 9a Value to society 9b Significance to self 10a Considerate supervisor 10b Supportive supervision 11a Job security 11b Good future prospects 12a Fair treatment of employees 12b Moral organization 3d Conflict between job demands 3e Work/home conflict 1a Task discretion 1b Influence over organization 2a Skill use 3a Number of job demands 3b Difficulty of job demands 4a Range of different tasks 5a Future predictability 5b Clear role requirements 5c Availability of feedback 6a Amount of social contact 6b Quality of social contact

Figure 5. Importance grid. Cf. Reiche and Sparke, 2012.

International Food and Agribusiness Management Review

779

Meyerding

Volume 20, Issue 5, 2017

Table 5. Results of linear regressions. Model job aspects (R2=0.65)

Model job aspects + personal characteristics and structural variables (R2=0.69)

B (constant) -1.515 2b New learning 0.062 3c Task coherence 0.013 3f Emotional dissonance 0.103 7a Pay level 0.081 8a Pleasant environment 0.126 8b Safe work practices -0.070 8c Adequate equipment 0.057 9a Value to society -0.017 9b Significance to self -0.103 10a Considerate supervisor 0.014 10b Supportive supervision 0.125 11a Job security 0.032 11b Good future prospects 0.316 12a Fair treatment of employees -0.054 12b Moral organization 0.072 3d Conflict between job demands 0.028 3e Work/home conflict 0.196 1a Task discretion -0.096 1b Influence over organization 0.026 2a Skill use 0.125 3a Number of job demands 0.032 3b Difficulty of job demands -0.097 4a Range of different tasks 0.252 5a Future predictability -0.050 5b Clear role requirements 0.164 5c Availability of feedback 0.009 6a Amount of social contact 0.090 6b Quality of social contact 0.061 Number of employees Number of seasonal workers Gender (w=0, m=1) Fulltime employees (no=0, yes=1) Fix contract (no=0, yes=1) Sesonal employee (no=0, yes=1) Education level Managerial responsibility (no=0, yes=1) Age 1

S 0.526 0.070 0.072 0.066 0.055 0.067 0.076 0.068 0.066 0.083 0.084 0.075 0.056 0.059 0.076 0.072 0.068 0.060 0.115 0.091 0.104 0.111 0.122 0.131 0.095 0.112 0.092 0.122 0.120

Beta

Sig.

0.058 0.012 0.105 0.081 0.116 -0.064 0.059 -0.016 -0.092 0.016 0.151 0.033 0.331 -0.064 0.075 0.024 0.198 -0.051 0.017 0.069 0.017 -0.049 0.113 -0.029 0.083 0.005 0.041 0.030

0.004 0.381 0.857 0.120 0.143 0.062 0.353 0.400 0.801 0.216 0.865 0.096 0.568 0.000 0.478 0.320 0.683 0.001 0.403 0.776 0.232 0.774 0.429 0.056 0.600 0.143 0.926 0.464 0.613

-1.231 0.106 0.088 0.006 0.061 0.163 -0.204 0.083 -0.048 -0.040 0.025 0.139 0.131 0.363 -0.102 0.021 0.124 0.180 0.054 0.166 -0.030 -0.079 -0.091 0.166 -0.185 -0.000 0.143 0.140 0.027 -0.009 0.066 -0.096 -0.224 0.582 -0.859 -0.061 0.218 0.018

1.536 0.105 0.125 0.134 0.093 0.116 0.152 0.109 0.112 0.156 0.136 0.123 0.118 0.121 0.116 0.128 0.120 0.100 0.213 0.184 0.189 0.186 0.213 0.214 0.183 0.208 0.165 0.241 0.236 0.083 0.102 0.244 0.398 0.373 0.559 0.093 0.256 0.013

Beta

Sig.

0.100 0.080 0.006 0.058 0.147 -0.173 0.083 -0.042 -0.032 0.027 0.152 0.122 0.355 -0.119 0.020 0.105 0.177 0.026 0.108 -0.015 -0.042 -0.044 0.071 -0.096 0.000 0.090 0.063 0.012 -0.009 0.052 -0.031 -0.046 0.165 -0.141 -0.060 0.070 0.129

0.425 0.317 0.484 0.962 0.515 0.162 0.183 0.447 0.671 0.798 0.852 0.265 0.271 0.004 0.383 0.868 0.306 0.074 0.801 0.370 0.875 0.674 0.671 0.442 0.317 1.000 0.390 0.564 0.908 0.910 0.519 0.695 0.576 0.122 0.128 0.509 0.396 0.172

Dependent variable: job satisfaction.

In Figure 6, the mean values of the characteristics investigated are presented. The CE aspects were rated by the study participants on a seven-point Kunin scale, from 1=extremely low to 7=highly acceptable. On average, the aspects of task integrity (5.21), emotional dissonance (5.21), importance of the work to oneself, and the company (5.17) and the employerâ&#x20AC;&#x2122;s fair treatment of employees (5.10) were International Food and Agribusiness Management Review

780

Meyerding

Volume 20, Issue 5, 2017 1

Figure 6. Average values of job features. rated most positively. In the lowest positions are pleasant working environment (buildings, rooms, noise, temperature, etc.) at 4.66, supportive leadership (4.53), conflict between different tasks (4.48), conflict between work and family life (4.21), and finally, salary level (3.96). For the AD aspects, transformed variables were used; therefore, the averages reported are not comparable to those of the CE aspects. The aspects number of tasks (4.36), variety (4.00), difficulty of tasks (3.90), and role clarity (3.89) were assessed most positively by the study participants. In the lowest positions are job autonomy (3.72), the predictability of the future with respect to the job (without career prospects; I know what will happen in the near future) at 3.62, the impact on the company as a whole (also through trade unions and works councils) at 3.34, and the availability of feedback at 3.30. 4.6 Comparison of different groups In this study, employees of horticultural companies were analyzed. In another study by Meyerding (2016) vocational and master craftsmen scholars, as well as horticulture science students were investigated. The groups are, distinguished based on their different living circumstances, but also by characteristics such as age, level of education, and professional experience. For each group, the results of the preference measurement and the influence of the different job aspects on job and life satisfaction were examined. To gain a complete picture of the situation in German horticulture, the results of the three groups need to be compared. For this purpose, Table 6 compares the results of the various satisfaction and preference measurements. All three groups show the strongest preference for the job feature that concerns an employerâ&#x20AC;&#x2122;s fair treatment of employees. In second place, employees rank the opportunity to learn new skills, followed by considerate leadership, the use of their own abilities, and low emotional dissonance.

International Food and Agribusiness Management Review

781

Meyerding

Volume 20, Issue 5, 2017

Table 6. Comparison of the results for the three different groups. Group

Employees

Average age 35 years Sample size 229 Preferences top 12a F air treatment of employees five (5.13) 2b New learning (5.08)

Students1

24 years 205 12a Fair treatment of employees (5.21) 3f Emotional dissonance (5.11)

25 years 204 12a Fair treatment of employees (5.32) 3e Work/home conflict (5.13) 3f Emotional dissonance (5.05) 10a Considerate supervisor (5.02) 10b Supportive supervision (4.70) N/A.

10a Considerate supervisor (4.99)

2b New learning (5.07)

2a Skill use (4.98)

8c Adequate equipment (4.98)

3f Emotional dissonance (4.90)

3e Work/home conflict (4.91)

Effect on job 11b Good future prospects (0.62**) satisfaction top five 8c Adequate equipment, 12a Fair treatment of employees, and 3e Work/home conflict (0.51**) 10b Supportive supervision (0.50**) 9b Significance to self, and 12b Moral organization (0.49**) 3f E motional dissonance, and10a Supervision behaves considerately (0.50**) 1 Adapted

Vocational and master craftsman scholarsa

3f Emotional dissonance (0.52**) 10a Considerate supervisor (0.50**) 10b Supportive supervision (0.49**) 12a Fair treatment of employees (0.48**) 12b Moral organization (0.45**)

N/A.

N/A. N/A. N/A.

from Meyerding (2016).

For the vocational and master craftsman scholars, learning new skills and emotional dissonance are in second and third place, respectively. In fourth place is the provision of suitable equipment and in fifth is conflict between job and family. In the student sample (Meyerding, 2016), conflict between job and family is the second most important job feature, followed by emotional dissonance, i.e. being completely oneself at work. In fourth and fifth place are considerate and supportive leadership, respectively.

5. Discussion When it comes to employees preferences, the amount of salary (7a) is in the middle and is of moderately minor importance. This is remarkable as one would have expected that in a low-wage industry, such as horticulture, available income should have a higher priority (Diener and Biswas-Diener, 2009). The strongest connection with job satisfaction is observed for future prospects (11b), i.e. feeling that one is moving in a positive direction. This aspect is affected by the subjective perceptions of the employee more strongly than others. The employer can influence this aspect positively through positive communication, the demonstration of prospects, and active staff development. Following closely in second place comes the conflict between work and family life (3e), followed by the employerâ&#x20AC;&#x2122;s fair treatment of employees (12a) and appropriate equipment (8c) with the same correlation International Food and Agribusiness Management Review

782

Meyerding

Volume 20, Issue 5, 2017

coefficient. Having adequate equipment available in horticulture, with its primary physical work, is more important than in other industries, in which the influence of this aspect on job satisfaction is lower (Salvendy, 2012). The results show that, the leadership style and the possibility of being oneself at work without having to make too great an adjustment show a strong influence on employee satisfaction in German horticulture. The sustainability of the horticultural company has a strong relationship with employee satisfaction. Employees in horticulture want to work for the ‘good,’ because this increases one’s social position (9a; value of the job for society) and the personal value of the work (9b). Safe working processes (8b) have a stronger impact on job satisfaction in German horticulture, as in other industries (Barling et al., 2003). This is partly due to the predominantly physical work, but also the handling of hazardous materials, such as chemical pesticides. That salary can be found in the middle of the influence on job satisfaction in horticulture seems surprising because in a low-wage sector such as horticulture, it could be assumed that the influence of the amount of salary, especially for low-income people, has a relatively high impact on job satisfaction (Lazarus, 2006: 165). The difficulty of tasks (3b) shows no significant relationship with job satisfaction in the sample. Few employees in German horticulture seem to be underutilized or overburdened with respect to this aspect. The results show a familiar pattern, according to which the correlation coefficients between the job aspects and context-bound satisfaction (job satisfaction) are higher than for the context-free satisfaction (life satisfaction; Faragher et al., 2005). Older employees in German horticulture are more frequently dissatisfied with their working lives and their lives in general than younger employees. Here, horticulture is different to other industries; on average, satisfaction increases with age (until 65; Mroczek and Spiro, 2005). The different values in horticulture could be due to physical stress, which leads to difficulties with age. Thus, a negative correlation between age and job aspect (8a) pleasant working environment (buildings, rooms, noise, temperature, etc.) can be observed. The same applies to considerate leadership (10a), working safety (8b), and own future prospects (11b, Table 3). Employees with a higher level of education put astonishingly less emphasis on the compatibility between work and family life (3e), adequate equipment (8c), and own job security (11a), probably because their chances in the labor market can be assessed more positively (Häublein 2014; Piopiunik and Wöβmann, 2011). As Figure 6 shows, the average values of the subjective job attributes evaluations are within a narrow range and there is little differentiation between the individual aspects, so that meaningfulness is very limited. This phenomenon is frequently observed in the measurement of satisfaction (Warr, 2012). For the vocational and master craftsman scholars form the study by Meyerding (2016), learning new skills and emotional dissonance are in second and third place for preferences, respectively. In fourth place is the provision of suitable equipment and in fifth is conflict between job and family. At this point, it is clear that job aspects such as emotional dissonance and conflict between work and family life, reflecting an altered job setting and a preference structure that requires a cultural change within the company, become more important. The younger generation, with a higher formal education level (Piopiunik and Wöβmann, 2011), exhibits a change in work attitude and favors work aspects accompanied by an increased work-life balance.

International Food and Agribusiness Management Review

783

Meyerding

Volume 20, Issue 5, 2017

The trend described is also reflected in the students’ results. Here, conflict between job and family is the second most important job feature, followed by emotional dissonance, i.e. being completely oneself at work. In fourth and fifth place are considerate and supportive leadership, respectively. From the results, it can be deduced that for well-trained young professionals, the work-life balance and transformational employee-oriented leadership are particularly important, and that horticultural companies need to change these aspects – in particular their management culture – to ensure they are attractive employers for the new (Y) generation. Overall, for all the groups, it is particularly important that the employer deals fairly with employees and maintains supportive, considerate behavior, that they do not have to adjust too greatly at work (emotional dissonance), that they have opportunities to develop and use their skills, that adequate equipment is available, and that the conflict between work and family is as low as possible. It is interesting that the provision of appropriate equipment is given such high priority. This result has not commonly been observed in other industries (Salvendy, 2012) and indicates the high physical stress in horticulture. The twelve job characteristics could affect not only the satisfaction of employees, but also their ability to provide high performance. For example, supportive leadership (job characteristic 10) increases the satisfaction of employees and at the same time creates the conditions for high performance. In this case, higher values for supportive leadership lead to increased job satisfaction and performance (Lyubomirsky et al., 2005).

6. Conclusions For employees in German horticulture, the strongest influence on job satisfaction is exerted by good future prospects. In second place, are three aspects: suitable equipment, the employer’s fairness toward employees, and conflict between work and family life. In third place is supportive leadership and in fourth place the personal value of work and the fairness of the employer towards society. Emotional dissonance and considerate leadership are in fifth place. In this group, the aspect that stands out especially is the availability of good future prospects, underlining the importance of the employee’s feeling that he/she is moving in a positive direction (toward fulfillment of personal goals). Also, as with other groups, the provision of appropriate equipment plays a strong role, seeming to be a special feature of the industry. The results show that the sustainability of the horticultural company positively affects employee satisfaction (and/or vice versa). Furthermore, they provide an indication that ‘soft’ job aspects, which include a special form of corporate and leadership culture, show the greatest effect on employee satisfaction, particularly among the younger, highly educated employees. The results also show that the influence of ‘hard’ job aspects, such as salary, difficulty of tasks, variety, and job security, are not as highly valued. In establishing the relationships between job characteristics, job satisfaction, and work performance, the employees’ satisfaction survey could be a tool both for improving employee satisfaction and optimizing the use of human capital, thus serving company profitability. For horticultural companies, in which staff costs account for 40% of the overall costs, this potential for optimizing human resource management and personnel management is particularly interesting, especially as its importance has thus far received insufficient emphasis. There is a possibility that the utility functions of the job characteristics apply equally to the employees’ performance as well as to their levels of satisfaction, so that optimal values for satisfaction are close to the optimum long-term utilization of the performance potential of each employee. For example, underload and overload both lead to dissatisfaction and in the long-term, to suboptimal performance. This possibility of using the vitamin model has not yet been studied empirically.

International Food and Agribusiness Management Review

784

Meyerding

Volume 20, Issue 5, 2017

References Adams, J.S. 1963. Towards an understanding of inequity. The Journal of Abnormal and Social Psychology 67(5): 422-436. Barling, J., E.K. Kelloway and R.D. Iverson. 2003. Accidental outcomes: attitudinal consequences of workplace injuries. Journal of Occupational Health Psychology 8(1): 74-85. Bitsch, V. and M. Hogberg. 2005. Exploring horticultural employees’ attitudes toward their jobs: a qualitative analysis based on Herzberg’s theory of job satisfaction. Journal of Agricultural and Applied Economics 37(3): 659-671. Bradburn, N.M. and D. Caplovitz. 1965. Reports on happiness: a pilot study of behavior related to mental health. Aldine Pub. Co, Chicago, IL, USA. Brayfield, A.H. and H.F. Rothe. 1951. An index of job satisfaction. Journal of Applied Psychology 35(5): 307-311. Busemann, K. 2013. Wer nutzt was im Social Web?: Ergebnisse der ARD/ZDF-Onlinestudie 2013. [Who uses what in the social web?: Results of the ARD/ZDF-Onlinestudie 2013]. Media Perspektiven, (7-8): 391-399. Caplan, R.D. 1975. Job demands and worker health: main effects and occupational differences. U.S. Department of Health, Education, and Welfare, Public Health Service, Center for Disease Control, National Institute for Occupational Safety and Health, Washington, WA, USA. Claβen, M. and F. von Kyaw. 2007. Change Management-Studie 2008: Business Transformation – Veränderungen erfolgreich gestalten. [Change management study 2008: business transformation – making changes successful]. Capgemini, Berlin, Germany. Cummins, R.A. 2000. Objective and subjective quality of life: an interactive model. Social Indicators Research 52(1): 55-72. De Jong, J., M. Mulder and J. Nijhus. 1999. The incorporation of different demand concepts in the job demand control model: effects on health care professionals. Social Science and Medicine 63: 193-210. De Jong, J. and W.B. Schafeli. 1998. Job characteristics and well-being: a test of Warr’s vitamin model in health care workers using structural equation modelling. Journal of Organizational Behavior 19: 387-407. Diener, E. and R. Biswas-Diener. 2009. Will money increase subjective well-being?: a literature review and guide to needed research. In: Social indicators research series, edited by A.C. Michalos and E. Diener. Springer, Dordrecht, the Netherlands, pp. 119-154. Eid, M. and R.J. Larsen. 2008. The science of subjective well-being. Guilford Press, New York, NY, USA. Faragher, E.B., M. Cass and C.L. Cooper. 2005. The relationship between job satisfaction and health: a meta-analysis. Occupational and Environmental Medicine 62(2): 105-112. Field, A.P. 2009. Discovering statistics using SPSS: (and sex and drugs and rock ‘n’ roll) (3rd ed.). Introducing statistical methods. Sage Publications, Thousand Oaks, CA, USA. Ganzach, Y. 1998. Intelligence and job satisfaction. The Academy of Management Journal 41(5): 526-539. Ganzach, Y. 2003. Intelligence, education, and facets of job satisfaction. Work and Occupations 30(1): 97-122. Glazer, S. and T.A. Beehr. 2005. Consistency of implications of three role stressors across four countries. Journal of Organizational Behavior 26(5): 467-487. Glomb, T.M., J.D. Kammeyer-Mueller and M. Rotundo. 2004. Emotional labor demands and compensating wage differentials. Journal of Applied Psychology 89(4): 700-714. Grandey, A., B. Cordeiro and A. Crouter. 2005. A longitudinal and multi-source test of the work-family conflict and job satisfaction relationship. Journal of Occupational and Organizational Psychology 78(3): 305-323. Greenhaus, J.H. and N.J. Beutell. 1985. Sources of conflict between work and family roles. Academy of Management Review 10(1): 76-88. Hackman, R.J. and G.R. Oldham. 1975. Development of the job diagnostic survey. Journal of Applied Psychology 60(2): 159-170. Häder, M. 2015. Empirische Sozialforschung: Eine Einführung (3. Aufl. 2015). [Empirical social research: an introduction]. Springer Fachmedien Wiesbaden, Wiesbaden, Germany. International Food and Agribusiness Management Review

785

Meyerding

Volume 20, Issue 5, 2017

Häublein, A. 2014. Arbeitsmarkt 2013: Arbeitsmarktanalyse für Deutschland, West- und Ostdeutschland. [Labor market 2013: labor market analysis for Germany, West and East Germany.]. Bundesagentur für Arbeit, Nürnberg, Germany. Headey, B. and A.J. Wearing. 1992. Understanding happiness: a theory of subjective well-being. Longman Cheshire, Melbourne, Australia. Herzberg, F. 1966. Work and the nature of man. Thomas Y. Crowell, New York, NY, USA. Herzberg, F., B. Mausner and B.B. Synderman. 1959. The motivation to work (2nd ed.). Wiley, New York, NY, USA. Jeurissen, T. and I. Nyklicek. 2001. Testing the vitamin model of job stress in Dutch health care workers. Work and Stress 15(3): 254-264. Karasek jr, R.A. 1979. Job demands, job decision latitude, and mental strain: implications for job redesign. Administrative Science Quarterly 24(2): 285-308. Kornhauser, A. 1962. Toward an assessment of the mental health of factory workers: a Detroit study. Human Organization 21(1): 43-46. Kornhauser, A.W. 1965. Mental health of the industrial worker: a Detroit study. Wiley, New York, NY, USA. Kunin, T. 1955. The construction of a new type of attitude measure. Personnel Psychology 8: 65-77. Lacy, P., T. Cooper, R. Hayward and L. Neuberger. 2010. A New era of sustainability: CEO reflections on progress to date, challenges ahead and the impact of the journey toward a sustainable economy. Accenture, Dublin, Ireland. Lazarus, R.S. 2006. Stress and emotion: a new synthesis. Springer, New York, NY, USA. Lewin, K., T. Dembo, L. Festinger and P.S. Sears. 1944. Level of aspiration. In: Personality and behavior disorders, edited by J.M. Hunt. Roland Press, New York, NY, USA, pp. 333-378. Lyubomirsky, S., L. King and E. Diener. 2005. The benefits of frequent positive affect: does happiness lead to success? Psychological Bulletin 131(6): 803-855. MacDonald, L.A., R.A. Karasek, L. Punnett and T. Scharf. 2001. Covariation between workplace physical and psychosocial stressors: evidence and implications for occupational health research and prevention. Ergonomics 44(7): 696-718. Meyerding, S.G.H. 2015. Job characteristics and job satisfaction: a test of Warr’s vitamin model in German horticulture. The Psychologist-Manager Journal 18(2): 86-107. Meyerding, S.G.H. 2016. Job satisfaction and preferences regarding job characteristics of vocational and master craftsman scholars and horticulture students in Germany. Review of Agricultural and Applied Economics 19(1): 30-49. Mroczek, D.K. and A. Spiro III. 2005. Change in life satisfaction during adulthood: findings from the veterans affairs normative aging study. Journal of Personality and Social Psychology 88(1): 189-202. Patterson, M., P. Warr and M.A. West. 2004. Organizational climate and company productivity: the role of employee affect and employee level. CEP discussion paper: Vol. 626. Centre for Economic Performance, London, UK. Piopiunik, M. and L. Wöβmann. 2011. Humankapital in Deutschland: Verschwendung geistiger Ressourcen. [Human capital in Germany: waste of intellectual resources]. In: Der ifo Wirtschaftskompass. Zahlen – Fakten – Hintergründe [The Ifo Economic Compass. Figures – Facts – Backgrounds], edited by G. Milbradt, G. Nerb, W. Ochel, and H.-W. Sinn. Hanser, Carl., München, Germany, pp. 100-101. Reiche, S. and K. Sparke. 2012. Managing job commitment of young professionals in Germany’s green industry. DGG-Proceedings 2(3): 1-5. Ronan, W.W. 1970. Relative importance of job characteristics. Journal of Applied Psychology 54(2): 192-200. Salvendy, G. 2012. Handbook of human factors and ergonomics (4th ed.). John Wiley and Sons, Hoboken, NJ, USA. Schreiner, M., M. Altmann, M. Stenger, M. Korn, W. Dirksmeyer, S. Ludwig-Ohm , U. Gossen, J. Graf. 2013. Zukunftsstrategie Gartenbau: Abschlussbericht zum Zukunftskongress Gartenbau am 11./12. September 2013 in Berlin [Future strategy horticulture: final report on the future conference on horticulture on 11/12. September 2013 in Berlin]. Bundesministerium für Ernährung und Landwirtschaft, Bonn, Germany.

International Food and Agribusiness Management Review

786

Meyerding

Volume 20, Issue 5, 2017

Spreitzer, G., K. Sutcliffe, J. Dutton, S. Sonenshein and A.M. Grant. 2005. A socially embedded model of thriving at work. Organization Science 16(5): 537-549. Srivastava, A., E.A. Locke and K.M. Bartol. 2001. Money and subjective well-being: it’s not the money, it’s the motives. Journal of Personality and Social Psychology 80(6): 959-971. Statistisches Bundesamt. 2006. Land und Forstwirtschaft, Fischerei: Gartenbauerhebung 2005 (Fachserie 3). Statistisches Bundesamt, Wiesbaden, Germany. Trzcinski, E. and E. Holst. 2003. Wochenbericht des DIW Berlin 35/03: Hohe Lebenszufriedenheit teilzeitbeschäftigter Mütter (Wochenbericht des DIW Berlin No. 35/03). [Weekly report of DIW Berlin 35/03: high quality of life of part-time mothers]. DWI, Berlin, Germany. Von Rosenstiel, L., H.M. Kehr and G.W. Maier. 2000. Motivation and volition in pursuing personal work goals. In: Motivational psychology of human development. developing motivation and motivating development, edited by J. Heckhausen. Amsterdam, the Netherlands, pp. 287-305. Warr, P. 1987. Work, unemployment and mental health. Oxford Science Publications. Oxford University Press, Oxford, UK. Warr, P. 1990. The measurement of well-being and other aspects of mental health. Journal of Occupational Psychology 63: 193-210. Warr, P. 2007. Work, happiness, and unhappiness. Lawrence Erlbaum Associates, Mahwah, NJ, USA. Warr, P. 2012. How to think about and measure psychological well-being. University of Sheffield, Sheffield, UK. Warr, P. 2013. Jobs and job-holders: two sources of happiness and unhappiness. In: Oxford library of psychology. the Oxford handbook of happiness, edited by S.A. David. Oxford University Press, Oxford, UK, pp. 733-750. Weiss, D.J., R.V. Dawis, G.W. England and L.H. Lofquist. 1967. Manual for the Minnesota satisfaction questionnaire. University of Minnesota. Industrial Relations Center. Bulletin 45: Vol. 22. Work Adjustment Project Industrial Relations Center University of Minnesota, Minneapolis, MN, USA.

International Food and Agribusiness Management Review

787