Energize Phoenix: Year Three Appendices

ENERGY EFFICIENCY ON AN URBAN SCALE: YR1

APPENDICES

YR2

YR3

APPENDIX C

14

RESULTS OF THE RESIDENTIAL CONTRACTOR SURVEY (Will Heasley, Michelle Schwartz, Mara DeFilippis, Mick Dalrymple)

APPENDIX D

20

RESULTS OF THE YEAR THREE COMMERCIAL CONTRACTOR SURVEY (Will Heasley, Mick Dalrymple, Michelle Schwartz, Mara DeFilippis)

APPENDIX E

PRINCIPAL AUTHORS

27

BEHAVIORAL ELEMENTS OF ENERGY USE IN ENERGIZE PHOENIX (Michelle Shiota, Anna Berlin, Samantha Neufeld)

Mick Dalrymple (Co-Principal Investigator) with the assistance of Rob Melnick (Principal Investigator) Michelle Schwartz (Editor)

APPENDIX F

54

COMMERCIAL PARTICIPATION FACTORS (Michael Kuby, Elizabeth Mack, Scott Kelley)

APPENDIX G

58

IMPLEMENTING TWO HOME ENERGY INFORMATION (HEI) DASHBOARD FIELD EXPERIMENTS (Harvey Bryan, Aleksasha Webster, Karla Grijalva, Shaily Rungta)

Global Institute of Sustainability, Arizona State University

APPENDIX H

66

ENERGY SAVINGS EVALUATION OF COMMERCIAL UPGRADE MEASURES THROUGH INDIVIDUAL PROJECT ANALYSIS AND UTILITY BILL MODELING (T. Agami Reddy, Karthik Thalappully, Marcus Myers, Oscar Solache Nishizaki)

APPENDIX I

89

DESCRIPTIVE, INFERENTIAL AND ECONOMETRIC ANALYSIS OF ENERGIZE PHOENIX PARTICIPATION AND SAVINGS (Tim James, Alex Castelazo, Anthony Evans)

TABLE OF CONTENTS APPENDIX A

2

6

SPATIAL AND SPATIO-TEMPORAL CLUSTERING ANALYSIS OF PROJECT LOCATIONS (Elizabeth Mack, Scott Kelley, Michael Kuby)

106

ENERGIZE PHOENIX FINANCE PROGRAM EVALUATION (Andrew Conlin)

MARKETING AND COMMUNICATIONS FINAL PROGRAM DETAIL (Michelle McGinty, Denise Resnik)

APPENDIX B

APPENDIX J

APPENDIX K

117

ENERGIZE PHOENIX, 2010-2013: AN ECONOMIC IMPACT ANALYSIS (Anthony Evans, Alex Castelazo, Tim James)

ENERGIZE PHOENIX IS A PARTNERSHIP OF

Back to Main Table of Contents

APPENDIX A MARKETING AND COMMUNICATIONS FINAL PROGRAM DETAIL

5. AZRE – City of Phoenix Makes Changes to Commercial Real Estate Program – 5.4.12

The following observations and statistics delineate the marketing lessons learned and marketing activities of Energize Phoenix over the three-year grant period of June, 2010 – April, 2013:

7. Downtown Phoenix Journal – From the Wire, Energize Phoenix Hosts Energy Fair – 3.22.12

MOST SUCCESSFUL INITIATIVES

9. Get Your Phx – Double Your Rebate for Energy Improvements – 3.20.12

6. InBusiness Magazine – Time to Get Energized – April 2012

8. Arizona Republic – Free Energy Audits Offered on Rail Route – 3.23.12

• Brand Development and Implementation

10. USA Today – A Fair Deal: Save Twice on Energy Fixes – 3.18.12

• Save Money Messaging • Door Hanger Campaign

11. Arizona Republic – A Fair Deal: Save Twice on Energy Fixes – 3.18.12

• EPHX Events (It’s Easy With Energize Phoenix and Go Green Like Your Grocer)

12. Valley Forward – Balancing Limits Potential – 2.2.12

• Community Newsletter Advertising and Outreach

13. BOMA Greater Phoenix – Energize Phoenix Commercial Financing – 1.23.12

• Contractor Program Promotion

14. AIA Arizona News – 1.19.12

• Strategic Alliances

15. U.S. Green Building Arizona Chapter – The city of Phoenix and National Bank of Arizona Announce Commercial Financing for EPHX – 1.12.12

• Media Relations LESSONS LEARNED

16. Local First AZ – Energize Phoenix Means Big Rebates – 1.11.12

• Secure Energy Saving Super Heroes early in the program to leverage the case study and human element for program promotion

17. Phoenix Business Journal – Energize Phoenix attempting to boost residential participation – 1.6.12

• Create an energy concierge for one-on-one outreach to alleviate questions and perception of red tape

18. Midtown Muse – Spring 2012 – 1.2.12

• Engage key neighborhood ambassadors to supplement word-of-mouth marketing

19. Grow AZ – Energize Phoenix Expands Program Boundaries and Announces New Commercial Financing – 12.12.11

TOTAL IMPRESSIONS SUMMARY

20. DT Blog – Energize Phoenix Announces New Commercial Financing, Expanded Program Boundaries – 12.12.11

Media Relations – 5,147,931 impressions

21. Get Your Phoenix – Energize Phoenix – 12.9.11

Advertising – 55,000,536 impressions

22. Valley Forward Newsletter – EPHX Commercial Financing – December 2011

Web – 89,378 page views Collateral – 73,119 pieces

23. Energize Phoenix – Carnation Association Newsletter – December 2011

MARKETING EFFORTS DETAIL: OCTOBER 2010 – APRIL 2013

24. The Atlantic Cities – In Arizona, Reducing Water and Energy Use Through Peer Pressure – 10.31.11

MEDIA RELATIONS (TOTAL IMPRESSIONS 5,147,931) 1. Epcon – Do You Own a Home or Business in Phoenix – 10.12

25. Phoenix Green Chamber – Energize Phoenix – 10.11.11

2. Arizona Republic – Energize Phoenix to Double Some Rebates – 6.30.12

26. AZ Builder’s Exchange – PPP Aims to Leverage $9M into $20M in Energy Efficiency Upgrades Along Green Rail Corridor – 10.5.10

3. Arizona Builder’s Exchange – Energize Phoenix Increases Cap – 5.11.12

27. Light Rail Connect – APS Home Energy Check-up Rebate Offer – September 2011

4. Phoenix Business Journal – Energize Phoenix Alters Loan Program – 5.5.12 Energy Efficiency on an Urban Scale

28. The Arizona Republic – Phoenix light rail areas – 9.28.11 2

Global Institute of Sustainability

Back to Main Table of Contents

29. Green Living AZ Magazine – Home Energy Check-up – September 2011

• More Dough Carnation Association (monthly impressions 1,000)

30. Prensa Hispana – Energize Phoenix – September 2011

• Cool Savings

31. The Phoenix Republic – Energize Phoenix Rebate – 8.20.11

• Easy Money

32. AZ Builder’s Exchange – Update on the Energize Phoenix Program – 8.1.11

• Smart Money

33. Green Living AZ Magazine – Energy Saving Tips – July 2011

• More Dough

34. Arizona Republic – Phoenix to sell solar energy – 7.23.11

Pierson Place Historic District (quarterly impressions 1,000)

35. Arizona Republic – Bright days ahead for solar – 7.23.11

• General business card sized ad

36. Relumination.com – Energize Phoenix Project Revamps Inefficient Arizona Neighborhoods – 2.3.11

• Smart Money ¼ page ad Willo District (monthly impressions 1,000)

37. Save Money Live Healthy – Phoenix Energy Audits and the Energize Phoenix Program – 1.18.11

• Cool Savings • Easy Money

38. Green ID – Phoenix Energy audits and the Energize Phoenix program – 1.28.11

• Smart Money

39. Arizona Republic – Energy program along light rail to bring jobs – 10.29.10

• More Dough • It’s Easy with Energize Phoenix

40. Downtown Devil – Energize Phoenix program to promote green jobs, sustainability along light rail route – 10.27.10

SPANISH ADS – SEPTEMBER 2011 - OCTOBER 2012 & JANUARY 2013 – APRIL 2013 (TOTAL IMPRESSIONS 12,975,000)

41. TBJ – Energize Phoenix revs up $25M in grants – 10.26.10 42. Energize Phoenix Radio Interviews (Univision - Spanish/ Phoenix Today Radio - English)

La Voz (weekly impressions 82,500) • Cool Savings

ADVERTISING – 55,000,536 IMPRESSIONS

• Easy Money

LIGHT RAIL ADS – JULY 2011 – OCTOBER 2012 & JANUARY 2013 – APRIL 2013 (TOTAL IMPRESSIONS 41,977,536)

• Smart Money

• 5 Kiosks (monthly impressions 2,209,344)

• More Dough

• Cool Savings

• It’s Easy with Energize Phoenix Prensa Hispana (weekly impressions 350,000)

• More Dough

• Cool Savings

• Easy Money

• Easy Money

• Smart Money

• Smart Money

• Energy Fair

• More Dough

NEIGHBORHOOD ASSOCIATION NEWSLETTERS – JULY 2011 – OCTOBER 2012 & JANUARY 2013 – APRIL 2013 (TOTAL IMPRESSIONS 48,000)

• It’s Easy with Energize Phoenix WEB (TOTAL PAGE VIEWS 89,378) • Launch website – Oct. 2010

Midtown Muse (quarterly impressions 1,000)

• Commercial Financing page launch – November 2011

• Cool Savings

• Commercial Financing snipe- January 2012

• Easy Money

• Submit DSIRE.com inquiry; pending confirmation – January 2012

• Smart Money Energy Efficiency on an Urban Scale

3

Global Institute of Sustainability

Back to Main Table of Contents

• It’s Easy with Energize Phoenix event page – March 2012

– Arizona Opportunities Industrialization Center (AZOIC)

• Track Google Analytics weekly

– Midtown Museum District event

SOCIAL MEDIA

– Willo Neighborhood Association

• Facebook (Currently 131 “Likes”) and Twitter (1,704 followers) accounts

– Phoenix Historic Neighborhood Coalition – Pierson Place Association

• Purchase Flipcam to capture footage for social media channels

– Sky Harbor Neighborhood Association – Rogue Green

• Film and edit Energy Saving Super Hero Story – Residential example (Jablonski Family) December 2011 and January 2012 – posted to YouTube

– Sierra Club Energy Committee – Eastlake Neighborhood Association

• Deck Park Vista; awaiting to post – March 2012

– Garfield Organization Neighborhood Association

• Super L Ranch; awaiting to post – March 2012

– Carnation Association of Neighbors

• It’s Easy with Energize Phoenix; awaiting to post – March 2012

– Go Green Phoenix – Scottsdale Green Building Lecture (outside Corridor)

COMMUNITY EVENTS • Purchased iPad to enter addresses into GIS tool at events

• In-Corridor Exhibition Events

• Strategic allies – professional and community organization partners that help us spread the word

– Keep Phoenix Beautiful – City of Phoenix Earth Day (x3) – Phoenix Energy Awareness Expo

– Blooming Rock

– Phoenix 10K

– Greater Phoenix Green Chamber

– Go Green Phoenix

– Carnation Association

– Wilson School Health Fair

– Midtown Museum District

– Tamale Festival

– Willo Historic District

– Phoenix Park N’Swap

– Roque Green

– ACT Kids Health Fair

– Go Green Phoenix

– Green Vendor Showcase

– Get Your PHX

– Maricopa County Health Fair

– Local First

• EPHX Hosted Events

– Downtown Voices Coalition

– It’s Easy with Energize Phoenix – Energy Fair

• Sponsorships

• 500 attendees; 130 sign ups

– Phoenix 10K

• Go Green Like Your Grocer

– Keep Phoenix Beautiful – City of Phoenix Earth Day (x3)

COLLATERAL (TOTAL PIECES 73,119)

– Go Green Phoenix

• Door Hangers – General Info – Distributed October 10, 2011 – 3,000

– Wilson School Health Fair – Ranch Market Health Fair

• Door Hangers – It’s Easy with Energize Phoenix – First distribution 3/8 & 3/9 and second 3/19 & 3/20 – 14,500

– Arizona Manufacturing Council

• Residential Brochures (8,000)

• In-Corridor Community Presentations

– Spanish – 500

– Central City Village Planning Committee Energy Efficiency on an Urban Scale

– English – 7,500 4

Global Institute of Sustainability

Back to Main Table of Contents

• Residential Brochure Refresh (13,000) – Spanish – 3,000 – English – 10,000 • Residential Brochure (10,000) – Spanish – 3,000 – English – 7,000 • Commercial Brochures – Business and Small Business (10,500) – First run – 9,000 – Second run – 1,500 • HVAC Brochure – 1,500 • Business Finance Brochure – 500 • Energize Phoenix Postcards – 3,000 • Tote Bags – 2,000 • Magnets – 5,000 • Banners – 3 • DOE Energy Saver Booklets – 1,000 • T-Shirts/Polos – 150 • Custom Table Cloths – 2 • Window Decals – 1,000 • Contractor Badges and Lanyards – 300 • Yard Signs -50 • Foam Core Boards – All EPHX ads – 7 • Ad posters – 80 • 11 x 17 Laminated Maps – 25 • It’s Easy with Energize Phoenix Confirmation and Reservation Cards – 500 • It’s Easy with Energize Phoenix Flyers – 500 • Energize Phoenix Welcome Signs – 2

Energy Efficiency on an Urban Scale

5

Global Institute of Sustainability

Back to Main Table of Contents

submit applications around the same time, thus producing clustering in both space and time.

APPENDIX B SPATIAL AND SPATIO-TEMPORAL CLUSTERING ANALYSIS OF PROJECT LOCATIONS

Results of the analysis highlight different spatial signatures in project clustering as well as different spatio-temporal signatures in project clusters. This clustering propensity does differ depending on the breakdown of projects. This is particularly true when the contractors with the largest number of completed energy upgrade projects are evaluated and compared to all other contractors. When clustering is examined for the six most prolific contractors combined, no clustering is evident. On an individual level however, three of these more prolific contractors exhibit spatial clustering.

INTRODUCTION There are many ways Energize Phoenix participants become aware of the program, such as contractor solicitation, advertisements, and announcements at neighborhood meetings. These methods of advertising the program, as well as the sales strategy of contractors, may leave a geographic signature among businesses and residents that choose to participate. If such a signature exists, one may expect to see clusters of residential or commercial projects. One might also expect to see clusters of projects at distinct distances from other clusters as information travels farther from the initial upgrade adoption sites in the neighborhood or business district.

This clustering may be indicative of their sales strategy, which tended to place high importance on door-to-door sales. While many customers did not answer the survey question about how they heard about the Energize Phoenix project, those that did and that were located in the most significant spatial clusters said far more often that they heard about the project from contractors than from various forms of localized “word of mouth.” From a spatio-temporal perspective as well, the signature of individual contractors is quite distinct from the global patterns across all contractors. This pattern may be related to their method of making customer contacts through a dedicated sales representative.

The degree to which program adoptions occur as a result of spread by word of mouth may have implications for the level of publicity needed by contractors, the degree of targeting in marketing strategies or the mix of marketing strategies employed. The extent of such transmission may also have implications for the likelihood that energy behavior norms can spread via neighborhood social networks. The degree to which program adoptions cluster may also have implications for the efficacy of particular sales strategies. For example, it is anticipated that face-to-face methods of contact such as door-to-door sales may be more effective than more passive strategies (door hangers or web-advertising) at persuading customers to adopt energy upgrade projects.

DATA For each of the following analyses, commercial projects were analyzed in two ways, at the point level and by zones. The spatiotemporal clustering analysis used point-level data collected for each commercial project upgrade and the spatial cluster analysis used zonal data. The zonal analysis broke the Energize Phoenix corridor into zones of approximately 1/16 square-mile each, based on Public Land Survey Systems quarter-quarter sections.

To reduce potential bias in results, zones on the Corridor boundary that were significantly smaller than 1/16 square-mile were excluded from the analysis. A total of 550 projects were analyzed. Projects were broken into nine groups:

In order to evaluate the presence of clustering, and the potential efficacy of publicizing the program and particular sales strategies, two types of analyses were conducted. The first type of analysis, spatial analysis, evaluated clustering in space only. This helps determine if adopters of energy upgrades are located in close proximity to one another. Spatiotemporal analysis was also conducted to evaluate if the energy upgrades that are close to one another in space are also close to one another in time. Closeness in space may indicate spread of information about the project by word of mouth.

A particular spatio-temporal signature may indicate use of a particular sales strategy. For example, a salesperson may canvas an area using a door-to-door strategy for one week and then move down the road a mile two weeks later and use the same door-to-door strategy. Applicants that hear about the program at a particular time in the same neighborhood may

Energy Efficiency on an Urban Scale

• All contractors irrespective of the number of projects completed • A group combining the six contractors with the most completed commercial projects • All other contractors besides those six with the most completed commercial projects • Six different groups, one for each of the commercial contractors with the most completed commercial projects.

6

Global Institute of Sustainability

Back to Main Table of Contents

BOX 1

SPATIAL CLUSTER ANALYSIS The hypothesis of interest for the spatial clustering analysis was as follows: Are there portions of the Energize Phoenix corridor with a larger than anticipated number of businesses participating in energy upgrades given the spatial distribution of businesses within the Corridor? To address this question two popular tests of spatial clustering were used. The first test was a global Moran’s I test (see Box 1 for a description). This test tells us whether there is clustering in the Corridor. The second test used is the local Moran. This test is a complement to the global Moran because it can tell us the location of specific clusters within the Corridor.

What is a Moran’s test?

For a study area divided up into zones, each with a value of some variable, a Moran’s I test tells whether there is statistically significant clustering of zones with similar values. The output of the test is a statistic ranging between -1 and 1. A Moran’s I of 1 indicates perfect clustering. In the case where all zones have a value of 1 (e.g., black) or 0 (e.g., white), the black and white zones are separated into two completely separate regions. At a value of 0, the pattern of black and white zones is random, while at -1, the pattern exhibits “anti-clustering” – the black and white zones are arranged in a perfect “checkerboard” pattern such that they touch their own color as little as possible. Examples are shown below. Statistical significance values can be attached to Moran’s I tests.

Both of these tests use zonal data. As discussed above, the commercial project data were aggregated to quarterquarter sections located in the Energize Phoenix corridor.

These aggregated data were then converted to rates using the information on business presence in 2009 derived from the National Establishment Time Series Database (NETS), which is a point-based dataset that shows the location of all businesses within the Energize Phoenix corridor. This conversion gives the number of commercial upgrades per business in each quarter-quarter section. This step is necessary to control for the fact that some portions of the Corridor have more businesses than others, and are thus predisposed to have more upgrade adoptions. By controlling for initial business presence with rates, we can better determine if particular portions of the Energize Phoenix corridor are more likely to have upgrades than might be expected otherwise.

TABLE 1: GLOBAL MORAN’S I TEST FOR SPATIAL AUTOCORRELATION

Table 1 below shows the results of the global Moran’s I test for the nine subdivisions of businesses mentioned previously. The Z-value indicates where the clustering is positive or negative.

As described in Box 1, a checkerboard pattern is indicative of negative spatial clustering. In other words, observations tend to repel one another. This is not uncommon for businesses that must have a certain market size to be profitable (retailers, chain restaurants, etc.). You would not expect to find two JC Penny stores next to one another. A positive z-value indicates positive spatial clustering. In other words, observations are located close to one another in space where closeness is defined by a spatial weights matrix. The size and sign of the Z-value only matter however if the Z-value is statistically significant. In this particular study, a p-value of 0.05 or less indicates significant spatial clustering.

*significant at p=0.05 level, **significant at p = 0.01 level.

A spatial weights matrix determines the neighbors of a spatial unit. There are several versions of spatial weights matrices.

Energy Efficiency on an Urban Scale

7

Global Institute of Sustainability

Back to Main Table of Contents

The one used to produce the results in this study is a firstorder queen weights matrix. This defines neighbors in the same way that a queen moves on a chessboard. Thus, quarterquarter sections that touch the quarter-quarter section at any point are considered neighbors.

project activity. Figure 1 is a graphic that shows the results of the local Moran analysis for the nine divisions of projects of interest in this study. Across all contractors and projects, many low clustering values are found near the edge of the Corridor boundary, along with only one high value in the northern part of the Corridor. Interestingly, quarter-quarter sections with lower participation rates surround the highhigh cluster area. Individual contractors that completed 30 or more projects show differing patterns, but their clusters do not appear to dictate the overall spatial clustering pattern for all contractors combined. Some of the more active individual contractors have scattered high-high clustering areas across the Corridor, while others appear to concentrate their efforts in specific areas.

Table 1 below shows that across all projects, there is positive spatial clustering. In other words, the spatial distribution of participation rates across all quarter-quarter sections is not random and businesses that adopted upgrades are located near other businesses that adopted upgrades (the Z-value is positive). If we break the businesses down by contractor however, the results of the global Moran indicate clustering in some cases but not others.

At a more disaggregate level of analysis; there is variation in spatial clustering. There is spatial clustering across all contractors except for the six most prolific contractors. There is no spatial clustering if we group the six most prolific contractors together. This lack of clustering reflects varied degrees of spatial autocorrelation across these six individual contractors. Projects completed by Contractor B, Contractor D, and Contractor E all show a greater level of positive and significant clustering compared to Contractor A, Contractor C, and Contractor F, which exhibit no clustering.

FIGURE 1: LOCAL MORAN RESULTS

After determining at a global level the presence or absence of clustering, results for the local Moran were computed to evaluate the locations of clusters within the Energize Phoenix corridor. In this regard, the local Moran is particularly useful because it categorizes quarter-quarter sections into one of five categories:

Figure 2 identifies the quarter-quarter sections with high clustering surrounded by other quarter-quarter sections with high clustering for each of the six contractors and plots them on one map. These are actually cluster centers and not the entire cluster for ease of visual interpretation. This shows where each of the individual contractors concentrated many of their projects, and it appears that many areas were claimed as service areas by multiple contractors. Other quarterquarter sections are identified as high-high cluster areas for multiple contractors. This likely indicates either competition or cooperation in these areas between the larger contractors.

1. No significant clustering 2. High-high: Section that has a high level of completed projects and its neighbors also have a high number of completed projects 3. Low-high: Section that has a low level of completed projects and its neighbors have a significantly high number of completed projects 4. High-low: Section that has a high level of completed projects and its neighbors have a low number of completed projects 5: Low-low: Section that has a low level of completed projects and its neighbors have a low number of completed projects The high-high sections of the Corridor are of particular interest for this study because they identify the hot spots of completed

Energy Efficiency on an Urban Scale

8

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 2: LOCATION OF HIGH-HIGH CLUSTERS

BY CONTRACTOR

the product of differences in sales strategies, differences in the use of referral incentives, and differences in the ways in which contractors assess potential participants.

The Knox statistic was used to identify statistically significant spatio-temporal interaction. Box 2 below provides a more detailed description of both spatio-temporal interaction and the computation of the Knox statistic. This particular global statistic was selected over other options because more localized patterns in the data may be identified by changing the critical distances and times, which are inputs needed to compute the test statistic. For this particular analysis, the critical times range from 1 to 50 days at intervals of 5 days.

The critical distances range between roughly half a mile and five and a half miles.

BOX 2

Contractor C is the only contractor that does not share a high-high clustering area with another contractor, and its clustering area is also uniquely in the heart of the commercial central business district of Phoenix. In contrast, Contractor B and Contractor E concentrated their projects in the industrial and warehouse district east of downtown. Contractor A has scattered clusters across the Corridor, while Contractor D and Contractor F’s are in both commercial and industrial areas away from downtown. Since none of these contractors hold offices in this area, it appears that they are specifically choosing to focus on selling energy upgrades to certain business types in specific areas of the Energize Phoenix corridor.

What is the Knox Statistic?

This is a point-based statistic for spatio-temporal interaction. It computes the number of events that are close to each other in both space and time. The table below highlights how spatio-temporal interaction is different from purely spatial or purely temporal clustering.

The user specifies a critical distance and time within which they would like to investigate spatio-temporal clustering. Next, a count of the number of events that are within both this critical distance and time is computed. If the sum is large enough, we can say that the points are very unlikely to be randomly distributed and are therefore considered close in both space and time.

SPATIO-TEMPORAL CLUSTER ANALYSIS After analyzing whether completed projects are close to one another in space, projects were analyzed for spatio-temporal interaction. This is important because it can tell us if projects are close to one another in both space and time. For example, how many completed projects are within 1 mile and 5 days of each other? Is this number of completed projects more than we would expect from a random distribution of projects, and more specifically, is it so much more that it is unlikely to happen by chance?

Figures 3 through 11 below display the results for multiple Knox tests performed with different critical distances and times for each of the nine subdivisions of the commercial projects mentioned previously. The dotted line in each of these graphs represents a p-value threshold of 0.05. Any movement of the lines above this threshold indicates no significant spatio-temporal clustering.

The identification of spatio-temporal clustering at specific critical distances and times is helpful because it can generate hypotheses about the processes that produce these patterns, such as the ways in which Energize Phoenix participants learned about and decided to participate in the project.

Differences in the critical distances and times at which space-time clustering exists between contractors may also be

Energy Efficiency on an Urban Scale

9

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 5: KNOX RESULTS FOR ALL OTHER CONTRACTORS

Overall, these figures highlight significant space-time clustering. An interesting finding at the contractor level of the analysis was different spatio-temporal signatures across contractors. Contractor A (Figure 6) had significant spacetime clustering across all critical distances and times up until five miles. Contractor A is quite active in several locations in the Corridor and even appears to share territories with four contractors including Contractor E, Contractor B, Contractor F, and Contractor D. FIGURE 3: KNOX RESULTS FOR ALL CONTRACTORS

FIGURE 6: KNOX RESULTS FOR CONTRACTOR A

FIGURE 4: KNOX RESULTS FOR 6 MOST PROLIFIC CONTRACTORS

FIGURE 7: KNOX RESULTS FOR CONTRACTOR B

Energy Efficiency on an Urban Scale

10

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 8: KNOX RESULTS FOR CONTRACTOR C

FIGURE 11: KNOX RESULTS FOR CONTRACTOR F

Contractor C and Contractor F exhibit clustering across all temporal intervals and short distances. Contractor D on the other hand only exhibits significant spatio-temporal clustering at short distances and short temporal instances. Contractor E and Contractor B exhibit clustering at medium distances.

Contractor E exhibits clustering across all time periods except for 26 and 31 days, however, while Contractor B only exhibits clustering at longer temporal intervals.

FIGURE 9: KNOX RESULTS FOR CONTRACTOR D

The other contractors operate in more limited spaces within the Corridor and the spatio-temporal results reflect this. The distances at which space-time clustering takes place are shorter, which perhaps reflect this limited clustering in the Corridor. The temporal distances at which the space-time clustering is relevant are longer, which perhaps suggests repeated visits to all business in their respective territories over a long period of time. Contractor D was the only exception to this trend, as it appeared to operate within a very small territory within a limited period of time.

FIGURE 10: KNOX RESULTS FOR CONTRACTOR E

CONTRACTOR SURVEY ANALYSIS In an effort to explain the spatial and spatio-temporal patterns uncovered above, the contractor survey results were analyzed to add context. Specifically, differences in contractor sales strategy, use of referral incentives, manner of making the initial sales contact, women in the sales force, and/or the manner in which contractors assessed prospective upgrade customers may help explain clustering of businesses in space and in space and time. Similar to the analysis above, survey results were analyzed in eight groups: the six most prolific contractors individually, all contractors together, and all other contractors. Unfortunately, Contractor D did not answer a

Energy Efficiency on an Urban Scale

11

Global Institute of Sustainability

Back to Main Table of Contents

large portion of the survey questions. Therefore, the following analysis discusses the survey results for the same set of contractors highlighted above, with the exception of Contractor D.

door sales would seem to dictate a fairly tight concentration in space and time.

The survey results about sales contacts (Table 3) are also important to explaining the results of the cluster analysis above. The six most prolific contractors extracted for additional analysis use a dedicated sales representative as their preferred method of contact instead of telemarketing, auditor contact, installer contact, a combination of contacts, or some other method of contact. These contractors differ from the other contractors that completed the survey, of which only half used a dedicated sales representative to make sales contacts.

One interesting finding from the survey analysis was that the sales strategy profile (Table 2) for the most prolific contractors is very different than the other contractors participating in the Energize Phoenix project. Specifically, door-to-door sales were particularly important for the most prolific group (most of which rated door-to-door 7 out of 7 in importance in their sales strategy) and not for the other contractors (an average rating of 3). This finding suggests that door-to-door sales are a more successful way to obtain upgrade participants than other sales strategies; it may also help explain the spatial clustering uncovered above.

TABLE 2: COMMERCIAL CONTRACTOR RESULTS FOR

TABLE 3: COMMERCIAL CONTRACTOR RESULTS FOR CUSTOMER CONTACT STRATEGY LIMITATIONS There is one limitation to this cluster analysis that is worthy of mention, as it may impact the findings. This limitation is with regard to the geocoding of the upgrade projects. The projects were geocoded to a building rooftop in the case of an exact address match. For other matches, projects were geocoded to the street level. Since Phoenix streets are laid out according to township and range boundaries, this means that geocoding to the street level effectively geocodes projects to the boundaries of quarter-quarter sections in some cases. Although this was not the case for all projects, any geocoding error at the street level may place a project in one quarter-quarter section as opposed to another. Given the small number of projects geocoded in this manner, this is not likely to be a large issue, but still worthy of mention.

SALES STRATEGY A competing hypothesis for the process that produces spatial clustering is word of mouth among customers. If word of mouth were the main factor, and if “seeing is believing,” then we might expect that frequent face-to-face contact, possibly involving a visit to see the installed upgrades of a neighboring business, could also account for spatial clustering. While we cannot dismiss this explanation definitively, certain factors suggest that door-to-door sales are the larger part of the explanation. First, we identified the commercial customers in the high-high clusters and checked their Energize Phoenix surveys. While many of the customers did not answer the question about “how did you hear about the Energize Phoenix program?” the overwhelming majority who answered this question in the high-high clusters said they heard about it in one way or another from the contractors. Second, if customer word of mouth were important, then we might not see as much space-time clustering over short time spans, because word of mouth and witnessing the upgrade results in person could follow significantly later in time, whereas efficiency in door-to-

Energy Efficiency on an Urban Scale

CONCLUSIONS In summary, this study finds evidence of both spatial and spatio-temporal clustering. The clustering propensity does differ depending on the breakdown of projects. This is particularly true when the contractors with the largest number of completed energy upgrade projects are evaluated and compared to all other contractors. When clustering is examined for the six most prolific contractors combined, no clustering is evident. On an individual level however, three of these more prolific contractors exhibit spatial clustering.

12

Global Institute of Sustainability

Back to Main Table of Contents

This clustering may be indicative of their sales strategy, which tended to place high importance on door-to-door sales.

From a spatio-temporal perspective as well, the signature of individual contractors is quite distinct from the global patterns across all contractors. This pattern may be related to their method of making customer contacts through a dedicated sales representative.

Energy Efficiency on an Urban Scale

13

Global Institute of Sustainability

Back to Main Table of Contents

Summary notes: A majority of the participating residential contractors were smaller companies having between 3-5 employees engaged in energy efficiency upgrade work. Nearly half were mid-sized companies with 6-15 employees, plus a couple of larger outliers with 20 and 45 employees respectively.

APPENDIX C RESULTS OF THE RESIDENTIAL CONTRACTOR SURVEY Residential contractors identified customer referrals and incentives for those referrals as important marketing tools.

They also identified community events as a valuable marketing channel.

QUESTION 4: NUMBER OF EMPLOYEES COMPANY-WIDE ENGAGED IN ENERGY EFFICIENCY UPGRADE WORK? (N=17)

These are the best ways to increase participation in energy efficiency upgrades, according to an Energize Phoenix survey of its participating residential contractors conducted in late February/early March 2013. The survey was intended to gain a better understanding of contractors’ business characteristics and marketing techniques for addressing the residential energy efficiency market.

The majority of residential contractors surveyed only did a small percentage of their overall work within the Energize Phoenix Corridor. This may be due to contractors having a less robust sales effort in the Corridor compared with other areas where there is potential for greater revenue per project.

Summary notes: When the question was framed as to how many employees are engaged in energy efficiency upgrade work ‘COMPANY-WIDE’, the results shift slightly higher, indicating that some contractors also conduct business out-of-state. The outliers also shifted, with three companies having between 57-100 employees engaged in energy efficiency upgrade work company-wide.

Another finding of the survey was that for residential property owners who are considering installing energy efficiency upgrades in their homes, the biggest impediment is money.

Upfront costs and out-of-pocket expenses remain as significant barriers in making energy efficiency sales, despite increased incentives offered through the Energize Phoenix program.

QUESTION 5: PLEASE SELECT TECHNICAL CERTIFICATIONS HELD BY EMPLOYEES IN ARIZONA (SELECT ALL THAT APPLY). (N=16)

SURVEY RESULTS Out of 32 residential contractors, 18 responded to the survey.

However, not all 18 answered all of the questions. The number of responses is indicated as (n) for each question.

QUESTION 3: NUMBER OF EMPLOYEES IN ARIZONA ENGAGED IN ENERGY EFFICIENCY OR ENERGY RETROFIT RELATED WORK? (N=17)

X axis = Percentage of the contractors who responded to this question

Summary Notes: BPI Building Analyst Professional is the most popular certification amongst participating contractors with a 100% response, which is not surprising as it is a pre-requisite to contractor participation in the AZ Home Performance with Energy Star program, upon which Energize Phoenix programs are based. BPI Envelope Professional follows with 37.5%, and then 31.3% of contractors have at least one employee with Energy Efficiency on an Urban Scale

14

Global Institute of Sustainability

Back to Main Table of Contents

QUESTION 7: WHAT PERCENTAGE, OF YOUR ENERGIZE PHOENIX WORK IS PERFORMED BY SUB-CONTRACTORS? (N=18)

a RESNET HERS Rater certification. NABCEP PV Installer, LEED Green Associate and LEED Accredited Professional each garnered 6.3%. The remaining certifications in the survey, mainly oriented toward the commercial building market, were all at 0.0%. This result coincides with very little overlap of the contractor base between the residential and commercial programs.

QUESTION 6: WHAT PERCENT OF YOUR COMPANY’S OVERALL WORK IN ARIZONA BETWEEN MARCH 2011 AND TODAY HAS BEEN ENERGIZE PHOENIX RELATED? (N=18)

Y axis = Percentage of work performed by sub-contractors X axis = Percentage of residential contractors who responded to question

Summary notes: One third of participating residential contractors surveyed do all of their own work, while nearly a third (27.8%) sub-contract out between 1 – 10 percent of their Energize Phoenix work. The remaining contractors sub-contract out anywhere between 21 and 90 percent of their Energize Phoenix work. This indicates a range of business models employed and, possibly, different marketing techniques. Some contractors appear to operate as fullservice contractors while others may operate either as general contractors, auditors or predominantly sales organizations.

Operating as a full service contractor could be an indication of the historical roots or strengths of the company, or could reflect a desire to have tight quality control by maintaining all work in-house. Sub-contracting a portion of the work may indicate better economies of scale in fulfilling lower-demand tasks, tasks with high equipment costs, or tasks that require specific expertise (such as HVAC replacement).

Y axis = Percentage of company’s overall work in the Energize Phoenix corridor X axis = Percentage of the contractors who responded to this question

Summary Notes: The survey demonstrates that energy efficiency projects within the Energize Phoenix Corridor were a relatively small percentage of the total business for most participating residential contractors. There were 12 contractors (66.7% of those surveyed) whose Energize Phoenix work comprised between 1-10% of their total work.

This is consistent with comments made by some contractors in an Energize Phoenix contractor marketing meeting in Year Two, that they viewed Energize Phoenix as an opportunity to secure projects to fill in the slow season while they focused their marketing during the high hot season on wealthier parts of the Valley with larger-sized projects. One contractor each performed between 21-30%, 31-40%, and 41-50% of their work in the Energize Phoenix Corridor, and three other approved contractors who participated in the survey did not do any Energize Phoenix projects.

Energy Efficiency on an Urban Scale

QUESTION 8: WHAT PERCENTAGE OF YOUR SALES & MARKETING STAFF SPEAK SPANISH? (N=18)

15

Global Institute of Sustainability

Back to Main Table of Contents

Summary notes: Language skills were of research interest because of the presence of a large Spanish-speaking population in the Energize Phoenix Corridor, including Spanish-speaking home and business owners. Language and associated cultural nuances could play a role in sales success.

Almost two thirds of companies had at least one Spanishspeaker on their sales and marketing staff. Seven out of 18 had between 1 -10 percent of their staff who spoke Spanish, 4 other contractors had anywhere between 21- 100 percent who spoke Spanish. Seven out of eighteen residential contractors did not have anyone on their sales & marketing staff who spoke Spanish.

Summary Notes: It may be valuable to know how upgrade contractors communicate with homeowners and how the presence of women in the sales force influences this communication. A majority of the contractors surveyed did not have women on their sales force. However, a relatively significant proportion of contractors had between 1–10 percent women on their sales force. The rest had a higher percentage of women on their sales force, which may be due to some of the participating companies being owned by women.

QUESTION 11: IS YOUR COMPANY WORKING ON ANY ENERGIZE PHOENIX PROJECTS THAT YOU FEEL WOULD MAKE A GOOD CASE STUDY FOR ENERGY AND/OR FINANCIAL SAVINGS? (N=17)

QUESTION 9: WHAT PERCENTAGE OF YOUR PROJECT MANAGEMENT STAFF, SPEAK SPANISH? (N=18)

Summary Notes: Nearly a third (29.4%) of the contractors surveyed had projects they considered as good candidates for further research as a case study. One contractor indicated that they worked on multiple homes that saw ‘duct/shell’ leakage cut by as much as 50%. The contractor added that customer feedback from the same participants was excellent with respect to increased comfort and utility bill reductions.

On the other hand, there were some contractors who were unsuccessful in breaking into the Energize Phoenix market to perform home energy check-ups. “Even with the rebate money available, people seemed very hesitant to undertake any upgrade projects. The incentives did not seem to significantly improve the odds of us winning these jobs”.

Summary notes: Seven out of eighteen residential contractors did not have any Spanish-speaking project managers. Six had between 1-10 percent of project managers who spoke Spanish and the remaining four contractors had anywhere between 21100 percent of project managers who spoke Spanish. While not necessarily impacting sales, this could impact communication during the implementation process, which could impact upgrade effectiveness.

QUESTION 10: WHAT PERCENTAGE OF YOUR SALES FORCE IS FEMALE? (N=18)

Y axis = Percentage of female sales force X axis = Percentage of the contractors who responded to this question

Energy Efficiency on an Urban Scale

16

Global Institute of Sustainability

Back to Main Table of Contents

QUESTION 12: SCORE YOUR COMPANY’S EXPERTISE IN THE FOLLOWING AREAS. DO NOT INCLUDE THE EXPERTISE OF SUB-CONTRACTORS/PARTNERS. (N=6)

QUESTION 14: IF YOU OR YOUR STAFF WANTED TO ATTEND A PROFESSIONAL DEVELOPMENT SESSION(S), WHICH WOULD BE THE BEST TIME(S) TO SCHEDULE THE SESSION(S). (N=6)

X axis = Average ranking score (1 = no value, 7 = high value) Y axis = number of contractor responses

Only six contractors responded to this question. Their responses reflect the priority of energy conservation measures included in the AZ Home Performance with Energy Star program.

Summary notes: The majority of contractors who responded to this question would prefer to attend any professional development session(s) in the afternoon Monday – Friday.

This coincides with the time of day when it is hottest in Valley attics. Area contractors tend to start as early as possible during the busy season to complete attic work before the hottest part of the day.

QUESTION 13: IF PROFESSIONAL DEVELOPMENT SESSION(S) WERE MADE AVAILABLE FOR ENERGIZE PHOENIX CONTRACTORS, WHICH TOPICS WOULD BE OF THE MOST BENEFIT TO YOUR COMPANY? (N=6)

QUESTION 15: WHAT WOULD BE THE BEST MONTHS FOR STAFF TO ATTEND PROFESSIONAL DEVELOPMENT SESSIONS? (N=6)

X axis = Average ranking score (1 = no value, 7 = high value)

Summary Notes: The six contractors who responded to this question expressed an interest in professional development.

The most desired sessions were: Advanced Building Diagnostics, Sales & Marketing of Energy Efficiency, New Energy Efficient Technologies, Available Tax Credits and Incentives and Promoting Energy Efficient Behavior, respectively.

Energy Efficiency on an Urban Scale

Y Axis = Percentage of contractors

Summary notes: The majority of contractors who responded to this question would prefer to have professional development sessions during the months of November, December, and January. During these months, demand for contractor services is typically lower.

17

Global Institute of Sustainability

Back to Main Table of Contents

QUESTION 16: IF PROFESSIONAL DEVELOPMENT SESSIONS WERE MADE AVAILABLE FOR ENERGIZE PHOENIX CONTRACTORS, WHAT WOULD BE THE IDEAL SESSION LENGTH, GIVEN YOUR STAFF’S SCHEDULE AND COMMITMENTS? (N=5)

QUESTION 18: WHAT WERE THE PRIMARY METHOD(S) OF MARKETING THAT YOUR COMPANY USED IN THE ENERGIZE PHOENIX CORRIDOR THAT RESULTED IN RETROFITS? (N=6)

X axis = Ranking score (1 = no value, 7 = high value)

Summary Notes: The most successful residential marketing methods were: customer referrals, incentives for customer referrals, community events and door hangers. Door-to-door sales were not ranked very high in importance in this question, but a separate analysis (see Appendix B: Spatial and SpatioTemporal Clustering Analysis of Project Locations) indicates that door-to-door sales were particularly important for the commercial contractors who performed the greatest number of commercial projects. It may prove beneficial for residential contractors to increase their door-to-door sales efforts as well. Interestingly, while both residential and commercial contractors rated customer referrals as important, only the residential contractors seem to have employed customer incentives for referrals. This may be due to a perceived effectiveness of incentives as a motivation when offered to an individual versus a company.

Y axis = Number of contractor responses X axis = Length of professional development session(s)

Summary notes: While only a few contractors responded to this question, the majority would prefer professional development sessions of 1 – 2 hours. The remaining contractors would prefer more in-depth sessions ranging from 4 hours to 2 days.

QUESTION 17: WHAT ARE THE MOST SIGNIFICANT CHALLENGES IN CONVINCING HOMEOWNERS TO DO RETROFITS? (N=6)

QUESTION 19: WHAT HAVE BEEN THE MOST SUCCESSFUL ASPECTS OF THE ENERGIZE PHOENIX PROGRAM THAT YOU HAVE EXPERIENCED? (N=4)

X axis = Ranking score (1 = no value, 7 = high value)

• Getting money in the hands of homeowners who want to spend it on energy efficiency upgrades.

Summary Notes: With ratings of high to moderately-high challenge, the most significant barriers in convincing homeowner to undertake upgrades were: Home value upside down, out of pocket expense, economic uncertainty, employment statues and awareness of program, respectively.

Financial considerations consistently appeared as the most significant challenge when considering whether or not to complete an energy efficiency upgrade.

• Additional rebates for homeowners.

• The energy savings the homeowners were able to partake of in this program.

• Construction.

QUESTION 20: IN YOUR EXPERIENCE, WHAT HAVE BEEN THE MOST CHALLENGING ASPECTS OF THE ENERGIZE PHOENIX PROGRAM? (N=4) • Out of pocket for the homeowner until the rebates were issued.

• Rebates taking much too long and contacting those in charge with EPHX.

Energy Efficiency on an Urban Scale

18

Global Institute of Sustainability

Back to Main Table of Contents

• Wait time in finding out if homeowners qualify for the higher tier programs.

• Paperwork obligation.

QUESTION 21: PLEASE PROVIDE ANY SUGGESTIONS THAT YOU HAVE FROM YOUR ENERGIZE PHOENIX EXPERIENCE THAT COULD THAT COULD BE USED TO IMPROVE OTHER ENERGY EFFICIENCY PROGRAMS. (N=4) • Awareness. Large project rebates for apartments.

• Have the rebates come directly to the contractor.

• Faster rebate turn around times, and quicker communication with staff. Contractors are busy, and when we get a chance to address an issue, it’s important that it’s resolved or addressed at the time - not one week or two weeks later.

Also we had trouble sending the rebate documents viaemail, as the files were too large due to the great amount of paperwork. There needs to be some sort of easier online sharing process (NO FAX as it’s outdated and many contractors, including us, do not use fax).

• Keep it going!

Energy Efficiency on an Urban Scale

19

Global Institute of Sustainability

Back to Main Table of Contents

are cited by numerous contractors as the most successful aspect of the project.

APPENDIX D RESULTS OF THE YEAR THREE COMMERCIAL CONTRACTOR SURVEY

The most important consideration at the end of the day is the financial benefit of doing an energy efficiency upgrade. If there is a relatively quick pay off and customers feel confidant in their return on investment, then business and property owners are much more likely to hire a contractor to complete an upgrade on their property.

424 buildings upgraded, is substantial… but what would it take to convince even more business owners to upgrade their property?

Even when rebates might cover nearly all of their expenses, the owners of commercial properties may still hesitate to invest in energy efficiency upgrades, according to contractors. Many business owners do not budget for these upgrades, and they are often skeptical that they will see an acceptable return on investment. Contractors view this as one of the primary obstacles for their sales & marketing staff to overcome.

SURVEY RESULTS Out of 59 commercial contractors, 58 responded to the survey.

However, not all 58 answered all of the questions. The number of responses is indicated as (n) for each question.

QUESTION 3: SIZE OF COMPANY? (N=58)

In addition, the time and effort to sign up for the Energize Phoenix commercial program could have deterred some property owners from participating. However, one contractor also noted the benefit of quick responses to rebate and application questions within the program, making it easier to overcome the paperwork challenge.

These are some of the key findings of an Energize Phoenix survey of its participating commercial contractors. The survey, conducted in late February/early March 2013, was a followup to an initial survey conducted during the first year of the project.

The intent of the survey was to learn more about commercial contractors’ marketing methods in an attempt to help understand and explain patterns of business participation in the Energize Phoenix commercial programs.

Summary notes: A clear majority of participating commercial contractors had 10 or more employees (65.5%). Nearly 20 percent of those surveyed had between 5 – 10 employees and the rest had four or less.

Interestingly, even though most contractors ranked ‘customer referral’ and ‘word of mouth’ as having the most value in their sales strategy, most did not offer any type of incentives for customer referrals. One potential reason may be conflict of interest policies within customers’ businesses that would prohibit employees from accepting incentives from vendors.

Of those contractors who did offer referral incentives, offerings ranged from cash payments and discounts on future work to free equipment and gift cards.

QUESTION 4: FOR THE ENERGIZE PHOENIX PROJECT, CHOOSE THE STATEMENT THAT BEST DESCRIBES YOUR SALES FORCE. (N=57)

While challenges and barriers will likely continue to exist within the energy efficiency upgrade market, many contractors identified a number of successful aspects of the Energize Phoenix Program in helping overcome these challenges.

Providing customers with energy savings options and the ability to save money at a very low upfront cost stand out. The rebate and incentive amounts available through the program Energy Efficiency on an Urban Scale

X axis = Percentage of Contractors

20

Global Institute of Sustainability

Back to Main Table of Contents

Summary Notes: Participating contractors were asked to describe the primary areas of expertise within their sales force. A large number of companies (22.8%) had a dedicated sales force with some training in energy efficiency, while a limited number of companies (7%) relied mainly on a sales force consisting of energy experts with some sales training.

The majority of contractors (61.4%) utilized a combination of both sales experts and energy experts. The remaining (8.8%) did not specify.

Summary notes: The pattern for Spanish-speaking project management staff is quite similar to that of Spanish speaking sales staff. Just over half the companies surveyed had zero Spanish-speaking employees on their project management staff, while just over a third had between 1–10% Spanish speaking project managers. While not necessarily impacting sales, this could impact communication during the implementation process, which could impact upgrade effectiveness.

QUESTION 7: WHAT PERCENTAGE OF YOUR SALES FORCE IS FEMALE? (N=58)

QUESTION 5: WHAT PERCENTAGE OF YOUR SALES STAFF SPEAKS SPANISH? (N=58)

Y axis = Percentage of women in sales force X axis = Number of contractors

Summary Notes: This question was asked because second year preliminary results showed that women-owned businesses were participating at a rate much lower than would be expected relative to their presence in the Corridor. However, that finding did not persist through the third year. The majority of contractors have very few women on the sales force.

Summary notes: Language skills were of research interest because of the presence of a large Spanish-speaking population in the EP Corridor, including Spanish-speaking business owners. Language and associated cultural nuances could play a role in sales success. A majority (51.7%) of commercial contractors surveyed did not have any Spanish-speaking sales representatives on their sales staff. Over one third had between 1-10 percent Spanish-speaking sales representatives and the rest had a significantly larger percentage.

QUESTION 8: HOW DID YOU MAKE THE INITIAL CUSTOMER CONTACTS? (N=47)

QUESTION 6: WHAT PERCENTAGE OF YOUR PROJECT MANAGEMENT STAFF SPEAKS SPANISH? (N=58)

Y axis = Percent of contractors

Summary Notes: The survey demonstrates an overwhelming majority of initial contacts, (78.7%) were made by dedicated sales representatives. However, installers and energy auditors also played a role in making initial contacts with (29.8%) and (21.3%), respectively. Telemarketing played a rather minor Energy Efficiency on an Urban Scale

21

Global Institute of Sustainability

Back to Main Table of Contents

Summary Notes: Interestingly, while contractors listed both ‘word of mouth’ and ‘customer referrals’ as having high importance in their sales strategies (See Question 9), only 27.3% actually offer incentives for customer referrals. The vast majority of contractors surveyed (72.7%) offer no incentives for customer referrals. This may be because those contractors do not feel referral incentives are necessary (i.e.: their work speaks for itself and will influence customers to speak for it), or simply that they had not yet tapped into that marketing device. Alternatively, businesses may have purchasing and employee conflict of interest policies that prohibit employees from accepting gifts from vendors. At the same time, since vendor incentives are not a common revenue stream, company acceptance of an incentive may create the perception of causing more accounting work than it is worth.

role according to survey results. Additional analysis by the Geography team found that the six most prolific contractors used dedicated sales representatives to a much higher degree than all other contractors. (See Appendix B: Spatial and Spatio-Temporal Clustering Analysis of Project Locations.)

QUESTION 9: PLEASE SCORE THE FOLLOWING ON A SCALE OF 1 – 7 ACCORDING TO ITS IMPORTANCE IN YOUR SALES STRATEGY. (N=56)

The following is a list of survey responses to this question.

• Free equipment.

• Percentage.

• Solar project only.

X axis = Average ranking score (1 = no value, 7 = high value)

• $50.00 referral.

Summary Notes: Both “Customer referral” and “Word of mouth” stand out as having been ranked the highest importance in overall contractor sales strategies. However, separate analysis (See Appendix B: Spatial and SpatioTemporal Clustering Analysis of Project Locations) indicates that door-to-door sales were particularly important for the six commercial contractors with the greatest number of projects. In addition, there were some contractors who identified ‘business networking groups’ and ‘email marketing’ as valuable sales tools. One contractor commented, “We are business-to-business and leveraged this program in addition to our typical services”.

• Cash, gift cards, etc.

• $250.00 • Discounts on next service call.

• Cash or product incentives for sales to referred customers.

• Discounts on work.

• $$$ Percentage.

• It completely depends on the referral and the situation.

QUESTION 12: TO WHAT EXTENT ARE EACH OF THE FOLLOWING A BARRIER TO ENCOURAGING PARTICIPANTS TO GET A RETROFIT? (N=52)

QUESTION 11: DO YOU OFFER ANY INCENTIVES TO CUSTOMERS FOR REFERRING OTHER CUSTOMERS TO YOU? (IF SO, WHAT TYPE?) (N=55)

Y axis = Percent of commercial contractors

Energy Efficiency on an Urban Scale

X axis = Average ranking score (1 = no value, 7 = high value)

22

Global Institute of Sustainability

Back to Main Table of Contents

Summary Notes: Out of pocket expense is perceived as the greatest barrier in the decision making process as to whether or not to get an upgrade. Financial health of company, tenant lacking authority to make upgrades and tenant not sharing in the full financial benefit of making upgrades on the property are also ranked fairly high as barriers in encouraging business owners to get energy efficiency retrofits.

Summary notes: Logistical barriers (such as cost, effort, and building characteristics) and motivational barriers (lack of interest and too many competing obligations) combined forces to keep more organizations from signing up, according to contractors.

QUESTION 15: WOULD HAVING ACCESS TO UTILITY BILLS BE HELPFUL TO PROVIDE ACCURATE SAVINGS ESTIMATIONS? (N=55)

QUESTION 13: IN YOUR EXPERIENCE, TO WHAT DEGREE DID EACH OF THE FOLLOWING, ACT AS A BARRIER TO A BUSINESS’ OR ORGANIZATION DECISION TO GET ENERGY EFFICIENCY UPGRADES? (N=54)

Y axis = Percentage of contractors

Summary notes: This question was included to help understand inaccuracies in contractor energy savings estimation and identify opportunities to improve accuracy.

Many of the contractors noted that on most or all of the projects, the customer provides utility bills. This is not surprising, as this is ultimately required before receiving an Energize Phoenix incentive check. However, other contractors suggest that having access to utility bills upfront would enable them to provide more accurate analysis in their sales effort.

One contractor responded by saying “This would be beneficial, as long as deceptive phone solicitors did not have access to the same info. Our biggest barrier is a factor of the customer already being harassed by a telephone solicitor.”

X axis = Average ranking score (1 = no barrier, 7 = high barrier)

Summary notes: Consistent with other data collected, even with increased incentives, the upfront cost is the greatest barrier in the decision to get energy efficiency upgrades. Another barrier is the prospective company being just too busy to take the time to consider the benefit of having an energy efficiency upgrade on their property. Other significant barriers are ‘sign-up effort’ (time, scheduling and paper work) and the ‘belief that upgrades would not save enough money.’ One contractor noted, “Businesses do not have funds allocated for energy savings upgrades.” So, annual budgeting cycles and priorities may also serve as a barrier to opportunistic decision-making.

QUESTION 16: WHAT HAVE BEEN THE MOST CHALLENGING ASPECTS OF THE ENERGIZE PHOENIX PROGRAM ACCORDING TO YOUR EXPERIENCE? (N=45)

QUESTION 14: WHEN PEOPLE DID NOT SIGN UP, WAS IT DUE TO… (N=53)

• Paperwork.

• Getting customers to fill out surveys.

• A significant amount of paperwork to be completed by the contractor and the customer.

• Lots of paperwork and process.

• Difficult to get utility bill information to provide the information necessary to determine potential savings.

Getting the paperwork to the right person (often out of town office) for signature

Y axis = Percentage of contractors

Energy Efficiency on an Urban Scale

23

Global Institute of Sustainability

Back to Main Table of Contents

• Getting the information from the owner as to what he wants to do. They are naive in the process and the upfront costs. The majority of people think its free money.

• Finding good size customers in this boundary who are willing to do a project.

• Initial Leads.

• Many Customers are somewhat apprehensive in regards to the financial incentives.

• Response.

• Winning jobs.

• Convincing people of the ‘energy savings’ and to make the upfront investment. Many don’t see the value and the ‘ROI’.

• Nothing that comes to mind it has all been good.

• Return on investment.

• Not too much of a challenge overall.

• Lack of initial funds for measures.

• We haven’t really had any.

• Making sure customers have sufficient funds to follow through with the project.

QUESTION 17: WHAT HAVE BEEN THE MOST SUCCESSFUL ASPECTS OF THE ENERGIZE PHOENIX PROGRAM THAT YOU HAVE EXPERIENCED? (N=44)

• Time it takes to get pre apps approved.

• The lead time-for approval was the greatest challenge.

• Being able to provide customers with energy savings options.

• The application turn around time and uncertainty of the rebate.

• Lowering peoples utility usage. Saving money.

• ROI and energy savings.

• Getting paid the EPHX portion of the retrofit.

• The price match making the customers projects more affordable.

• Waiting for APS to get the funding. The program dragged out and due to the delay, two customers decided not to move forward with the program.

• Customer out of pocket being reduced and or totally covered.

• Dealing with APS.

• Very low cost to customer.

• Getting the right information from APS – Custom, not custom – What will the rebate amount be?

• Monetary incentive.

• The amount of money given every year.

• Product-replacement counts.

• The rebates that are offered to incentivize customers.

• Corporate not pulling the trigger.

• 100% rebate.

• Many of the large buildings, greater than 50,000 square feet, were managed by just a few facilities management companies who did not really cooperate. Also, budget constraints with large companies.

• Rebates.

• Rebates amount.

• Very good final rebate amounts.

• The attitude that because they lease, they are not wanting to do any improvement. Amount of paperwork.

• The rebate amounts for customers.

• Assuring the customer that we will take care of their facility, due to them having friends with bad experiences with other contractors that didn’t do or lied about what they were going to do.

• The customer did receive their additional funds as promised.

• Dual Rebates.

• Fantastic rebate matching. This allowed us to do projects with extremely favorable paybacks. Communication was also excellent with both APS and Energize Phoenix.

• Not speaking Spanish, distrust from businesses, finding a willing party that has not been approached by many before us.

• Great visibility and cooperation with APS, through the rebate application process.

• We do not have enough sales people to push sales beyond clients that we were already working with on other services.

• The funds matching and that the agency behind it has powerful and known legitimacy.

• Our scope of work within the program is limited.

• Marketing the program.

• Quick response to rebate questions and apps.

• Not enough customers are aware of the program.

• The financial paybacks were significant, though that didn’t often make the sale for us.

• Finding Customers.

Energy Efficiency on an Urban Scale

24

Global Institute of Sustainability

Back to Main Table of Contents

• The program was a good one, however, with all the additional paperwork needed it was difficult to implement for smaller customers where margins are very thin.

The amount of paperwork for small customer that it is the same as a large customer. For smaller customers, a simplified acquisition threshold amount should be determined, and paperwork requirements should be less.

This would result in less of an administrative burden therefore, doing the quicker and smaller jobs would be more attractive and result in more locations being retrofitted by the contractors.

• We have had quite a struggle.

• Have not experienced any.

• Not very successful.

• We have not done any.

• None yet.

• None.

• Anyone can install “low hanging fruit measures”. It takes a program like EPHX to allow the installation of “deep retrofit” energy saving measures.

• Less paperwork shorter payment timelines.

• Awareness of the program.

• Speed up the time for the app process.

• Getting customers to make a decision faster.

• Figure out how to get a preliminary answer in a week to the contractor.

• The ability to provide LARGE energy savings for large buildings down town and assist in a 300+ LED Pole light retrofit.

• Less turnaround time between applications. Pre application should be accepted and analyzed for the customer so that we can offer a more accurate proposal of a rebate to the customer without having to wait for until submitting to see if there is enough funding. Reserve funding would also be helpful.

• Being able to show savings over the years.

• Once we have completed a project, hearing back from the customer, that they have indeed saved utility usage.

• More money towards the energy upgrades themselves rather than research.

• Those customers that are motivated by energy efficiency.

• VFD Sale.

• If the pre approval process actually secured and guaranteed money. Better T8 to T8 rebates. Better qualification of contractors.

• Community Events.

• Following APS Lead.

• Clearly state that the funds paid to the customer, are for them to pay the contractor not simply pocket the money.

• Energy Efficiency of equipment installed. Smooth process once in place.

• 3rd party payment or Joint check.

• The flow of application submittal process.

• Un-complicate the owner-tenant utility bill release process.

Incentivize cooperation between tenant and landlord.

• Ease of use.

• Ease of working with Energize Phoenix employees.

• APS opening rebates for more customer classifications such as apartment complexes.

• Energize Phoenix’s attention to the project.

• Perhaps the city putting out more projects for bid.

• Making it an easy way to have both APS and Energize phoenix employees help to inform the customer with the projects and opportunities.

• Mandate an ASHRAE Level I audit be done and ALL potential ECMs be reported to the customer (not just the low hanging fruit). Combining lighting savings with HVAC savings provides a substantial energy saving impact with the best possible simple payback, ROI, IRR.

• Quick response and easy to work with.

QUESTION 18: PLEASE PROVIDE ANY SUGGESTIONS YOU HAVE TO MAKE THE ENERGIZE PHOENIX PROGRAM MORE EFFICIENT OR EFFECTIVE. (N=34)

• Just one thing more communication on who we go to for what when program is started.

• Standardize most of the rebates to a fixed amount i.e.: replacing a 4 tube troffer with an LED troffer $200 easier and faster for everyone involved.

• Not sure if this is already provided, but have representatives from the City of Phoenix within the Energize Phoenix, meet with potential Customers to better explain the program benefits.

• Make it easier to complete paperwork and engage in the program.

Energy Efficiency on an Urban Scale

25

Global Institute of Sustainability

Back to Main Table of Contents

• Energize Phoenix, City, to directly mail to commercial buildings the benefits of Energy retrofit and possible rebates complete with a list of certified Contractors.

• Educate the community.

• Providing more information /knowledge to the customer their first time calling in to Energize Phoenix. My experience with the phone calls I receive are that the client thinks there is “free money” out there... and in a sense it is... however, their knowledge of the process and the savings is under estimated.

• Get the utility companies involved in the process of information on a level that makes sense to the customer.

Most of the time, the information is just passed along and there is no true explanation or incentive for the customer to investigate any efficiency upgrades.

• Perhaps flyers being distributed inside the utility bills?

• It would be easier if APS were a little more, user-friendly.

• Expand the Energize Phoenix Area.

• Expanding the geographical borders. We have many buildings as regular customers in Phoenix but not within the specified borders.

• Expand area.

• Allowing it to all of the city of Phoenix. Extending the program for more years.

• Keep doing it – don’t end it!

• Keep it going!

• None – Make it last longer?

• None.

• Great job to the team.

Energy Efficiency on an Urban Scale

26

Global Institute of Sustainability

Back to Main Table of Contents

for the financial “bottom line” play an important role in motivating energy-efficiency upgrades. However, greater focus on saving money as a motivation for conserving energy was associated with higher baseline electricity consumption, rather than lower consumption.

APPENDIX E BEHAVIORAL ELEMENTS OF ENERGY USE AND PARTICIPATION IN ENERGIZE PHOENIX ENERGIZE PHOENIX SUMMATIVE REPORT: BEHAVIOR TEAM EXECUTIVE SUMMARY

• Pro-Environment Attitudes/Beliefs Have Little Impact on Behavior: In the Residential study, intention to conserve energy in the future was associated with higher likelihood of obtaining an energy-efficiency checkup.

However, other pro-environment attitudes were associated with lower checkup likelihood. Attitudes and beliefs did not predict pursuit of residential upgrades, business participation in EP, or baseline energy use in any study.

The Behavior Change team’s mandate is to analyze the roles of psychosocial variables such as demographics, household and business characteristics, beliefs, attitudes, and motivations in predicting participation in the Energize Phoenix (EP) program, as well as patterns of energy use. The research presented in this report addresses three core aims: 1) Assess possible predictors of participation in EP upgrade programs among residential homeowners and businesses, including survey respondent demographics, household/ business characteristics, attitudes and beliefs about conservation, motivations to conserve energy, and ways respondents had learned about EP.

• Motives Matter More: Unlike attitudes and beliefs, self-reported motivations for conserving energy were consistently associated with decisions to participate in EP, and with baseline electricity consumption. Some of these effects were striking and counterintuitive. Among residential homeowners, high social motives (e.g., to be seen as an eco-friendly person) were associated with lower likelihood of pursuing en energy-efficiency checkup.

Among commercial organizations, higher business-related motivation (e.g., save money, making the business more competitive; market the business as environmentally responsible) strongly predicted greater likelihood of participation. However, relative intensity of the motivation to save money, as a reason for conserving energy, was consistently associated with higher levels of electricity consumption. It may be that households and businesses that face higher electrical bills each month find the possibility of saving money especially compelling. But this consistent finding does suggest that motivation to save money had not previously been effective in driving energy-efficient behavior. In contrast, relative intensity of motivation to preserve the environment and act responsibly in ecological matters was associated with lower levels of consumption, at least in residential settings.

2) Assess possible predictors of baseline energy usage by residential homeowners, businesses, and residents of two multi-unit housing complexes, with an emphasis on attitudes and motivations.

3) Examine the impact of “TED” – in-home, real-time feedback devices – on electricity usage among residents of two multi-unit housing complexes: Arizona State University’s Taylor Place undergraduate residence hall and the Sidney P. Osborn subsidized apartment complex.

The following key points are suggested by the overall pattern of findings presented in this report: • Kids Matter: The presence of children in the home offers some motivation for homeowners to consider energyefficiency upgrades; single family households were more likely to receive a checkup than households consisting of roommates – the reference group. However, children also present significant constraints. Households with a greater number of children in the home were less likely to obtain a checkup, and also tended to consume more electricity than households with no or few children.

• Get the Word Out: Eligible homeowners who had heard about the EP program through a greater number of channels were more likely to seek out an energy-efficiency checkup (although a similar effect was not observed for actually getting an upgrade). Among businesses, those who heard about the EP program from a contractor were more likely to participate; other channels did not appear to have the same impact.

• Money is Complicated: As a resource, money may support pursuit of energy-efficiency upgrades. Households in which the respondent worked full-time, and those with higher household incomes, were more likely to receive an upgrade through the EP program. For businesses, implications Energy Efficiency on an Urban Scale

27

Global Institute of Sustainability

Back to Main Table of Contents

EP commercial program did not have an independently defined checkup stage comparable to that in the residential program). We had access to energy use data from the local utility for a subset of these businesses.

One set of analyses was aimed at identifying predictors of participation in the EP program – i.e. receiving an upgrade. As with the Residential project, another set of analyses sought to identify key predictors of baseline energy use.

• Limited Effect of Real-Time Electricity Use Feedback Dashboards: In two studies, one with an undergraduate residence hall and one in a low-income, publicly subsidized apartment complex, we found little evidence that the presence of a real-time electricity use feedback device led to reduced consumption. No effect was found in the residence hall; a small effect was observed in the apartment complex, but during the winter months when demand for electricity is typically low. However, findings suggest that use of the real-time, immediate feedback display option on these devices may lead to greater savings than alternative settings (e.g., month-to-date usage).

3. Dashboard Project. The central aim of the dashboard project was to examine the effects of “The Energy Detective” or “TED” – an in-home, real electricity use feedback device (also known as a Home Energy Information (HEI) device) – on actual electricity consumption among residents of two multi-unit housing complexes: (a) the Taylor Place residence hall on the Downtown campus of Arizona State University, and (b) the Sidney P. Osborn (SPO ) assisted public housing development, also in downtown Phoenix. In both complexes, a randomly selected subset of participating rooms/apartments received a TED display device, as well as an energy Measuring Transmitting Unit (MTU) that measured energy use for the duration of the project. Data for this project were gathered by way of surveys and energy use information, collected either directly from TED data logging devices (Taylor Place) or from the complex management (SPO). In primary analyses, effects of the Dashboard devices were assessed by comparing the experimental versus control groups’ electricity usage. In additional analyses, the extent to which various psychosocial variables predicted baseline energy use and/or Dashboard-related savings was examined as well.

GENERAL INTRODUCTION: AIMS AND APPROACH The Energize Phoenix Behavior Change team’s mandate is to analyze the roles of psychosocial variables such as demographics, household and business characteristics, beliefs, attitudes, and motivations in predicting participation in the EP program, as well as patterns of energy use and savings. Our approach was founded on existing theory and empirical work from the domains of social psychology, health psychology, and sustainability/conservation psychology, all of which seek both an understanding of the causes of human behavior and the pursuit of effective tools for socially meaningful behavioral intervention.

Findings in this report reflect data from three projects within the Energize Phoenix program: 1. Residential Project. Behavioral data for the Residential Project were gathered through surveys completed by residential homeowners in the target geographical area who (a) received upgrades through the EP program, (b) received a home energy checkup but had not gone on to complete an upgrade as of March 31, 2013, or (c) did not receive a checkup, but were eligible for the program and willing to take the survey. We had access to energy use data from the local utility for a subset of these households.

One set of analyses was aimed at identifying predictors of participation in the EP program, both at the checkup and the upgrade level. Another set of analyses sought to identify key predictors of baseline energy usage.

Across projects, survey questions focused on five specific categories of predictors, each of which may play an important role in explaining decision-making about energy consumption and investment in energy efficiency: • Demographics. Demographic variables such as gender, ethnicity, age, education level, employment status, and political orientation are commonly used as the “first line” of predictors in behavioral research. These predictors sometimes account for substantial amounts of variability in behavioral outcomes (i.e., why some people behave differently from others). Demographic factors may not actually cause behaviors, in the sense that being one ethnicity or another directly causes one to use more or less electricity. However, demographics are associated

2. Commercial Project. Behavioral data for the Commercial Project were gathered through surveys completed by decision-makers of businesses in the target geographical area that either (a) received an upgrade through the EP program, or (b) had not received an upgrade as of March 31, 2013 but were still willing to complete a survey (the Energy Efficiency on an Urban Scale

28

Global Institute of Sustainability

Back to Main Table of Contents

with elaborate profiles of social roles, resources (social, economic, and informational), and constraints that are likely to have causal effects. As a result, demographic effects can provide important clues to the underlying causes of behavior. In addition, demographic effects can help policy-makers make informed decisions about how to approach various populations in behavioral interventions.

articles and radio/TV programs, websites, social media, contractors, community organizations, and word-of-mouth.

We conducted analyses to see whether any particular channel(s) appeared especially effective in reaching participants and predicting participation in the EP program.

Findings presented in this report represent collaborative efforts with the EP ASU Geography team, Economics team and Dashboard team, based upon data collected in collaboration with City of Phoenix and Arizona Public Service (APS) EP program staff. In addition to these empirical findings, we offer a brief summary of key points in previous research on the psychology of energy efficiency and sustainable behavior (see D.1 below), and summaries of implications of the Behavior team analyses.

• Household and Business Characteristics. Household characteristics such as income and household composition, and business characteristics such as corporate status and hours of operation, often present incentives and/or constraints around energy use and investment in energy efficiency. We also assessed features of the physical plant, such as lighting and equipment, associated with patterns of energy use likely to predict interest in upgrades.

RESIDENTIAL PROJECT

• Attitudes/Beliefs. Social Psychological models of behavior have traditionally emphasized attitudes (i.e., positive versus negative affective feelings about a behavior) and beliefs about the need for, control over, and consequences of behavior as important predictors of behavior itself.

Unfortunately, the actual evidence regarding impact of attitudes/beliefs on behavioral outcomes in Social and Health Psychology is mixed at best. However, we felt it important to examine the predictive power of explicitly held attitudes and beliefs about the environment, global warming, and energy conservation as potential predictors of EP participation and energy use outcomes.

DATA COLLECTION: SAMPLE AND PROCEDURES The Residential Project survey was accompanied by a voluntary waiver which, with a respondent’s permission, would permit APS to release a respondent’s energy usage data to ASU. Survey respondents were approached using a number of methods. First, undergraduate research assistants were trained and ASU-approved per City of Phoenix requirements for door-to-door survey administration, and then assigned a set of residential addresses within the program target area. In this way, surveyors attempted to obtain a completed English- or Spanish-language survey from every home in the target area before program launch or early in the program.

If an adult resident of the household was home, the surveyor asked him or her to complete the survey. If no one answered the door, or if the resident was unable to complete the survey at that time, the surveyor left a stamped, addressed envelope with the respondent so they could return the survey via mail. Additionally, surveyors set up booths at a variety of local events (e.g., Willo Neighborhood Historic Home Tour, Phoenix 5K) and invited attendees to complete the survey at those events. Finally, surveys were included as an optional component of the participant program application package administered by contractors. Per federal Better Buildings grant program reporting requirements, the utility data waiver was mandatory for program participants.

• Motives for Conserving Energy. A newer approach to predicting behavior asks what motivations people have for engaging in that behavior – what they believe they stand to gain. This is of particular interest in the psychology of sustainable behavior because people may have a wide variety of reasons for conservation, including saving money, preserving the environment, protecting one’s offspring and future kin, and improving one’s social status.

Importantly, these varying motives may have intersecting or even conflicting implications for a given behavior, with one motive promoting the behavior but another discouraging it. Also, the relative importance of certain kinds of motivations may tell us something about which motivations actually predict behavioral follow-through, and which do not.

Survey Respondents. Through these procedures, residential respondents completed 929 surveys. From this initial set of surveys, 61 were removed as duplicates (same address and/ or email), and 172 were removed because the respondent’s stated address was not within the program target area.

• How Respondents Heard About the EP Program. The Energize Phoenix program was marketed to participants through a wide variety of channels, including news Energy Efficiency on an Urban Scale

29

Global Institute of Sustainability

Back to Main Table of Contents

In addition, surveys from the 168 respondents who selfidentified as renters were removed from analyses, as only four renters participated in the EP program and we anticipated that the process for these four households was likely very different from the process for homeowners. Thus, the analyses below predicting EP program participation reflect data from the remaining 566 surveys, completed by homeowners giving an address in the EP target area.

following changes were made based on this examination:

Of the total sample, 96.5% of respondents completed the survey in English, and the remainder in Spanish.

Approximately half of respondents were female. With respect to ethnicity, 58% self-identified as Non-Hispanic White/ Caucasian, and 13% as Hispanic/Latino; the remainder indicated another ethnicity or declined to provide this information. Approximately a third of respondents selfidentified as Democrats, 13% as Republicans, and 14% as Independents; notably, nearly a third of respondents declined to provide information about political party. Approximately 60% of respondents had a college degree (including 2-year degrees), and about 40% of respondents’ partners had a college degree. Approximately 60% of respondents stated that they worked full-time.

• Education (participant and partner): The eight response options for level of education were collapsed into two - No College Degree (including “some college” but without any degree) vs. College Degree (including two-year degrees and professional schools). For partner education, “No Partner” was a third category.

• Ethnicity: Responses were collapsed into three categories – Non-Hispanic White/Caucasian, Hispanic/Latino, and Other/Declined.

• Political Affiliation: Responses were condensed into three categories - Democrat/Green, Republican, and Other/ Declined.

• Income: 12 of the 598 respondents indicated an income greater than $300,000, in some cases in the several millions. In order to avoid the problems commonly associated with outliers in statistical analyses, the income for these respondents was entered as $300,000, effectively capping income at this level.

• Children in Household: Answers to the three questions about number of children of different ages were summed to create a “Total Number of Children” variable.

With regard to household composition, 35% of respondents reported living as a couple with no children, and 22% as single adults with no children. Twenty-nine percent (29%) reported living in a single-family household with children (with either one adult or a couple). The remaining respondents reported living with roommates or with extended family. The average respondent had lived just under 30 years in Arizona, and about 14 years in their current home. Mean household income was $94,525 per year (median = $75,000).

It is important to note that this income level is considerably higher than the known level for the area, likely reflecting the removal of renters from the survey as well as a general tendency for higher-SES individuals to complete surveys and participate in studies. Thus, caution should be used in considering whether the present findings might generalize to low-income and lower-middle-income populations.

Preliminary Analyses and Subscale Development: Attitudes Items. The 21 initial, seven-point scale Likert-type items assessing attitudes toward the environment and conservation were subjected to a Principal Components Analysis (PCA), with the aim of identifying conceptually meaningful sets of items that could be treated as subscales in this sample.

Unfortunately the PCA failed to identify statistical patterns among all original items that would point the way to meaningful composites. Based on analysis of the conceptual meaning of the individual items, we created six attitude subscales used to predict participation in EP as well as patterns of energy use and savings. Cronbach’s alpha was calculated for each subscale as evidence of the coherence/ internal consistency among items within that subscale; values near or greater than .70 are typically considered acceptable for three-item scales and all met this criterion, so all were retained as potential predictors.

Data Cleaning and Reduction: Demographics and Household Characteristics. Demographic and household characteristic data were examined to assess validity of responses and address problems with outlier values likely to have an undue influence on analyses. Where possible and conceptually valid, categorical responses (e.g., political party) were collapsed into a smaller number of categories, in order to increase statistical power in logistic regression (“logit”) analyses. The Energy Efficiency on an Urban Scale

• “Global Warming”: Three items assessing agreement that the world is experiencing an environmental crisis caused by human behavior (Mean = 5.80, SD = 1.58, Cronbach’s alpha = .92).

30

Global Institute of Sustainability

Back to Main Table of Contents

• “Valuing Conservation”: Three items assessing personal effort to conserve energy and natural resources (Mean = 5.51, SD = 1.26, Cronbach’s alpha = .75).

respected others say energy conservation is important (Cronbach’s alpha = .80).

Self-report measures of motivation often show ceiling effects, such that most people give a very high rating of every motive.

In anticipation of this problem, we conducted an additional logit analysis in which ipsatized single item scores (i.e., standardized to the individual’s own mean and standard deviation across the nine motivation items) for all nine original items were entered as predictors.

• “Personal and Societal Efficacy”: Three items assessing belief that humans (including the self) can change energy habits and solve environmental problems (Mean= 5.75, SD = 1.24, Cronbach’s alpha = .75).

• “Human Supremacy”: Three items assessing belief that modern science will solve environmental problems, nature exists for human use, and economic development is more important than preserving the environment (Mean = 3.24, SD = 1.49, Cronbach’s alpha = .62).

Data Reduction: How Respondents Heard About the Program. Responses to the question about whether and how respondents had heard about the Energize Phoenix program were combined into five non-mutually-exclusive binary (yes/ no) variables: Program Marketing (e.g., newspaper/newsletter ads, door hangers, or EP websites); Contractor; Community (e.g., school, church, friend or neighbor); Other; and None.

• “Negative Emotions”: Three items assessing the experience of guilt, anger, and sadness when thinking about the effects of energy use on the environment. (Mean = 4.22, SD = 1.79, Cronbach’s alpha = .86).

RESULTS

• “Intent to Conserve”: Two items assessing intended effort to conserve energy in the next few months (Mean = 5.36, SD = 1.54, Cronbach’s alpha = .71).

The Behavior team had a series of research questions it sought to shed light on through Energize Phoenix research.

Each question is designated with a letter code indicating to which program activities it pertained; specifically, R=Residential, C=Commercial and D=Dashboards.

The remaining four items did not load adequately on any subscale, and were excluded from further analyses.

R.2: WHAT PERSON-LEVEL VARIABLES PREDICT HOW LIKELY SOMEONE IS TO PARTICIPATE IN THE ENERGIZE PHOENIX PROGRAM, AND AT WHAT LEVEL (HOME ENERGY CHECKUP VS. UPGRADE)?

Preliminary Analyses and Subscale Development: Motivation Items. The nine initial items assessing the respondent’s motivations for conserving energy were subjected to a Principal Components Analysis (PCA), again with the aim of identifying conceptually meaningful sets of items that could be treated as subscales. As with the attitude items, the PCA was conducted using data from all available questionnaires. In a preliminary PCA, examination of the scree plot of eigenvalues clearly indicated a two-factor solution. We therefore conducted a follow-up PCA forcing two factors and using direct oblimin rotation to maximize simple structure (i.e., tight fit of individual items to one factor or the other). Based on the PCA, combined with analysis of the conceptual meaning of the items, we created two motivation subscales.

Of the 566 respondents who completed surveys, 205 had received or would later receive a home energy checkup through the Energize Phoenix program, as of March 31, 2013. Of those who received a checkup, 111 went on to complete at least one home upgrade by the same cut-off date.

Two sets of simultaneous logistic regression (“logit”) analyses were conducted to identify significant predictors of each level of participation, relative to the preceding level. In these analyses several possible predictors of a dichotomous outcome are entered into the statistical model simultaneously; for each predictor, the analysis provides a test of the statistical significance of the effect, as well as a “weight” estimating the magnitude of the effect in terms of change in log-odds of the positive (i.e., getting the checkup vs. not) outcome. While logit produces fairly accurate estimates of significance (the p-value, or probability that the observed effect reflects sampling error rather than a “real” effect), the weights should be interpreted with caution as they depend greatly on the set of predictors and sample in the current model.

• “Responsibility Motivation”: Six items assessing motivations to save money, preserve the environment, do the “right thing,” protect future generations, preserve national security, and protect his/her family’s future (Cronbach’s alpha = .83).

• “Social Motivation”: Three items assessing motivation to keep up with what others are doing, to self-present as an environmentally responsible person, and because Energy Efficiency on an Urban Scale

31

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 1: ANALYSIS STEPS PREDICTING RESIDENTIAL AND COMMERCIAL PARTICIPATION

a final logit model including all significant predictors from the more constrained models, suggesting that the effects of Ethnicity could be explained in terms of differences in household characteristics, attitudes, and/or channels of program exposure among those of different ethnicities. Also, having a partner with a college degree was associated with significantly greater probability of checkup level participation, compared with not having a partner (Wald = 5.51, p = .019, B = 0.80); the same effect was not observed for respondents whose partners did not have a college degree.

FIGURE 2: HOME ENERGY CHECKUP-LEVEL PARTICIPATION

BY ETHNICITY

Source: ASU Global Institute of Sustainability

In the interest of maximizing statistical power, separate logit analyses were conducted for each of the major categories of predictors. This raises the possibility that predictors from different categories, each identified as significant, will account for overlapping effects in participation outcomes. Thus, as an additional step, we entered only those predictors that had emerged as significant in the more constrained models (including models examining how respondents had heard about the program, see below) into a final logit analysis, to see whether one or more predictors would drop out of the model.

Effect size statistics and p-values in the text below are taken from the initial, more constrained models, but results of the final model are always summarized as well.

Source: ASU Global Institute of Sustainability **p < 01, *p < .05. Denotes a significant difference between percentage of respondents in this Ethnicity category that received a home energy checkup, and the percentage among non-Hispanic Whites.

Upgrade-Level Participation. Demographic variables were less useful in predicting the step from getting a checkup to completing an upgrade. The demographics-focused model did not fit the data significantly better than an empty model, although it accounted for about 10% of variability in upgrade outcomes (Cox & Snell R2 = .092). Note that statistical power decreases sharply from predicting checkup-level participation to predicting upgrade-level participation, so comparable effect sizes in predicting the two levels of participation may still differ greatly in terms of statistical p-vales. The only effect that even approached significance was for respondent employment status. Specifically, respondents who did not work full time were marginally less likely to obtain an upgrade than those who did work full time (Wald = 3.44, p = .064, B = -0.71) – an effect that held up in the final analysis with significant predictors from the separate categories.

Demographics Checkup-Level Participation. Demographics logit analyses included seven categorical predictors (Sex, Ethnicity, Political Affiliation, Respondent Education, Partner Education, Respondent Employment Status, and Partner Employment Status) as well as one continuous predictor – Age. In terms of predicting home energy checkup-level participation, the model including these predictors fit the data significantly better than an “empty” model with only an intercept. However, it accounted for less than 10% of variability in checkup-level participation vs. non-participation (Cox & Snell R 2 = .090).

Of the predictors, only two significantly predicted checkuplevel participation (relative to no checkup). With respect to respondent Ethnicity, Non-Hispanic Whites/Caucasians were significantly more likely to participate at the checkup level than both Hispanics/Latinos (Wald = 11.16, p = .001, B = -1.45) and those reporting another ethnicity or declining to report (Wald = 4.06, p = .044, B = -0.58). Importantly, however, the effect of Ethnicity was no longer significant in Energy Efficiency on an Urban Scale

Household Characteristics Checkup-Level Participation. Household characteristics analyses included three categorical predictors (landscape type, pool ownership, and household composition) as well 32

Global Institute of Sustainability

Back to Main Table of Contents

as six continuous predictors (household income (in $1000s), number of adults in the home, number of children in the home, years in home, years residing the state of Arizona, and years expecting to live in Arizona). In terms of predicting checkuplevel participation, the model including these predictors fit the data significantly better than an “empty” model with only an intercept. However, it only accounted for approximately 15% of variability in checkup outcomes (Cox & Snell R 2 = .152).

logit analysis, the Pool effect remained significant, although the landscape effect did not. Taken together, these effects suggest that those who were already “doing the right thing” from a conservation standpoint (i.e., no pool, water-efficient landscaping) may have seen less need to seek out additional information on how they could conserve energy further.

FIGURE 3: HOME ENERGY CHECKUP-LEVEL PARTICIPATION

BY HOUSEHOLD COMPOSITION

One important finding from this analysis is that household income was not at all associated with probability of getting a checkup (Wald = .060, p = .807, B <.001). However, various aspects of household composition did emerge as predictors of getting a checkup. First, longer duration of residence in the current home was associated with lower probability of checkuplevel participation (Wald = 3.92, p = .048, B = -0.03), although this effect did not remain significant in a final model with all significant predictors from previous models.

Household composition broadly speaking appeared to be an important predictor of checkup-level participation. Rates were lower among households with more adults (Wald = 7.16, p = .007, B = -0.61) as well as those with more children (Wald = 5.12, p = .024, B = -0.41). Household Composition as a categorical variable also emerged as an overall significant predictor (Wald = 13.58, p = .009). Specifically, with “Roommates” as the reference group, households consisting of adult couples (Wald = 4.53, p = .033, B = 2.28) or single families (Wald = 5.03, p = .025, B = 2.47) were significantly more likely to have received a checkup. Although the effect of number of children was no longer significant in the final logit across significant predictors, the other effects remained at least marginally significant. Taken as a whole, these findings suggest that those in smaller and more stable family unit arrangements (i.e., couples or single parents with few children) may be most likely to seek an energy-efficiency checkup.

Source: ASU Global Institute of Sustainability *p < .05. Denotes a significant difference between these two Household Composition categories in terms of percentage of respondents that received a home energy checkup.

FIGURE 4: HOME ENERGY CHECKUP-LEVEL PARTICIPATION

BY LANDSCAPE TYPE

Source: ASU Global Institute of Sustainability

In addition, two outdoor properties of the home were significantly associated with getting a checkup. First, those in homes without pools were less likely to receive a checkup than those with a pool (Wald = 15.45, p < .001, B = -0.99). Second, landscape type had a strong and significant overall effect in predicting checkup-level participation (Wald = 8.40, p = .039).

Relative to no landscaping (“Neither”), no specific type of landscaping was significantly different. However, examination of the means for each landscape type suggests that those whose homes had primarily grass-based landscaping were more likely to receive a checkup than those with a landscape that demanded less water. In the final, more comprehensive Energy Efficiency on an Urban Scale

Upgrade-Level Participation. Some household characteristics did predict the transition from checkup-level participation to upgrade, with this model fitting the data significantly better than an empty model without predictors, and accounting for about 15% of variability in upgrade outcomes (Cox & Snell R 2 = .154). Two predictors emerged as significant; both remained significant or marginally significant in the final logit analysis as well. First, whereas income failed to predict checkup-level participation, greater income was significantly associated with greater likelihood of upgrade once a checkup had been completed (Wald = 6.34, p = .012, B = .008). Presumably the importance of income increases when making a decision 33

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 5: MEAN ATTITUDES BY RESIDENTIAL CHECKUPLEVEL PARTICIPATION STATUS

about actually carrying out the upgrade, which could involve significant costs to homeowners despite the EP subsidy. Also, not having a pool was significantly associated with greater likelihood of upgrade (Wald = 4.75, p = .029, B = 0.83). It is important to note that while traditional pool motors are high energy consumers, they were not part of the whole home performance program, and pool motor upgrades were not eligible for EP rebates (although the local utility provides incentives). As a result, homeowners with pools (and the associated energy costs) may have had greater incentive to have a home energy checkup, but the contractors performing the checkups may not have emphasized pool motor upgrades in their proposed solutions, and even if they did, homeowners would have less incentive to receive such an upgrade through the EP program.

Source: ASU Global Institute of Sustainability **p < 01, *p < .05, + p < .10. Denotes a significant or marginally significant effect of this attitude in predicting checkup-level participation in the logistic regression model with all attitude scales entered as predictors.

Attitudes

Upgrade-Level Participation. Notably, attitudes did not predict making the jump from checkup-level participation to actually getting an upgrade. The logit model including the attitude variables in predicting the difference between checkup-only and upgrade participants did not fit the data significantly better than an intercept-only model, nor did any single attitude subscale emerge as significant, or even marginally significant.

Checkup-Level Participation. Attitudes emerged as significant predictors of receiving a checkup through the Energize Phoenix program; the model including these predictors fit the data significantly better than an “empty” model with only an intercept, although it predicted less than 10% of variability in checkup-level participation (Cox & Snell R 2 = .090). Higher intention to conserve was significantly and strongly associated with greater probability of checkup-level participation (Wald = 31.78, p <.001, B = 0.49).

Motives for Conserving Energy Checkup-Level Participation. Examined as composites, motives emerged as significant predictors of receiving a checkup through the Energize Phoenix program; the model including these predictors fit the data significantly better than an “empty” model with only an intercept, although it predicted less than 2% of variability in checkup-level participation (Cox & Snell R 2 = .019). Specifically, higher scores on social (i.e., to keep up with what others are doing; to be seen as an environmentally responsible person) were associated with lower likelihood of checkup (Wald = 9.51, p = .002, B = -0.18). Notably, higher scores on the social motives composite were significantly associated with more pro-environmental ratings on the attitudes that emerged as significant predictors of checkup in the analysis above, and the effect of social motives was no longer significant in the final logit analysis that included these attitude measures. This likely reflects an impact of self-presentation on the attitudes items, suggesting that these items were influenced by desire to be seen as highly pro-environment, in a way that may not translate to pro-environmental behavior. No effect was observed for the responsibility motives composite. The logit model with ipsatized motives as predictors did not fit the data significantly better than an empty model, and no individual item emerged as a significant predictor.

Remarkably, other pro-environment attitudes actually predicted lower probability of checkup-level participation. Specifically, higher belief in human supremacy over the environment was associated with higher likelihood of checkup (Wald = 9.37, p = .002, B = 0.22), and higher scores on valuing conservation were associated with lower likelihood of checkup (Wald = 3.98, p = .046, B = -0.20). Moreover, both greater belief in global warming (Wald = 3.02, p = .082, B = -0.14) and greater negative emotion when thinking about the effects of human energy use (Wald = 3.60, p = .058, B = -0.13) were associated with lower likelihood of getting a checkup at the marginal level of significance. It is important to note that only two of these effects remained significant in the final logit model (intention to conserve and human supremacy). However, one reasonable possibility, consistent with the landscaping and pool effects above, is that individuals who identify strongly with proenvironmental attitudes may believe (rightly or wrongly) that they have already taken major steps to conserve resources, and do not see a need for a checkup to identify more potential improvements. The one positive effect is for the more futureoriented intention to increase one’s efforts to conserve energy.

Energy Efficiency on an Urban Scale

34

Global Institute of Sustainability

Back to Main Table of Contents

Upgrade-Level Participation. In general, motivations to conserve (both raw score composites and ipsatized individual motive items) were not strong predictors the transition from checkup-level participation to upgrade-level participation.

Neither model fit the data better than an empty model.

However, the ipsatized motivation to “keep up” with what others are doing emerged as significant (Wald = 5.53, p = .019, B = 0.82), and the motive of promoting national security (Wald = 3.27, p = .071, B = 0.47) as marginally significant when only motives were entered as predictors, and these remained marginally significant in the final model with all significant predictors from the more constrained models.

and December); neutral months (March, April, May, and October); and hot months (June, July, August, and September).

When a household’s monthly usage was below 100 kWh, that month was treated as missing; no more than one missing month was allowed per seasonal composite, or the entire composite was treated as missing.

Five multiple regression models were used to examine respondent- and household-level predictors of each of the three seasonal baseline electricity use composites. As with the analyses predicting Energize Phoenix program participation, one model included demographic characteristics of the respondent as predictor variables; one included household characteristics; one included the respondent’s attitudes and beliefs about conservation; and two included the respondent’s motivations for conserving energy. Analyses in each model used only those households for which all necessary survey data were available.

R.5: WHAT PERSON-LEVEL VARIABLES ARE STATISTICALLY ASSOCIATED WITH ENERGY USE AND ENERGY SAVINGS? In order to examine the predictors of baseline energy usage, we examined the statistical association of person- and household-level variables derived from the residential survey with energy usage during the year of 2010 – prior to the implementation of any residential upgrades subsidized

by Energize Phoenix. Of the 566 survey respondents (205 participants and 361 non-participants) whose data were used in analyses predicting program participation, it proved possible to obtain complete 2010 electricity billing data from the local utility for 167 households (108 participants and 59 non-participants). One household was removed due to consistent, implausibly low usage levels (< 20 kWh/month), leaving a total of 166 households.

Demographics These models included seven categorical predictors (respondent sex, ethnicity, and political affiliation, respondent education, partner education, respondent employment status, and partner employment status) as well as one continuous predictor – respondent age. Analyses predicting the cold month usage composite included 113 households; those predicting the neutral and hot month composites included 110 households.

None of the three overall models predicting seasonal energy usage was significant: for cold months F(11, 101) = 1.56, R 2 = .146, n.s.; for neutral months F(11, 98) = 1.48, R 2 = .142, n.s.; and for hot months F(11, 98) = 1.28, R 2 = .125, n.s.. However, individual predictors occasionally emerged as significant. Households in which the survey respondent identified as neither White nor Hispanic used less electricity than those in which the respondent was White (the reference group) in both the cold month (B = -343.93, p = .026) and Neutral Month (B = -314.12, p = .037) analyses.

Because the utility’s monthly billing cycles vary from household to household, and daily usage data were not available, it was necessary to transform the raw billing data to create energy usage estimates for actual calendar months that could be compared across households. For each household, this was done by (a) calculating mean daily usage in kWh for each billing cycle, and then (2) forming an estimate of each month’s use by multiplying the daily mean from each of the two relevant bills by the number of days in that month covered by each bill, and then summing these two figures. While this approach is not error-free, it substantially reduced the noise caused by variable billing cycles without introducing any systematic bias.

Also, households in which the respondent identified as republican used more electricity during cold months than households in which the respondent was libertarian, independent, or another political affiliation (the reference group; B = 396.02, p = .031). This effect was not observed during the neutral months. No demographic variable significantly predicted energy use during the hot months.

Temperatures in Phoenix fluctuate dramatically over the course of a calendar year, driving comparable fluctuations in demand for electricity. Because we anticipated that the predictors of baseline energy use might differ across seasons, we formed three different energy use composites for each household: cold months (January, February, November, Energy Efficiency on an Urban Scale

Household Characteristics These models included three categorical predictors (landscape 35

Global Institute of Sustainability

Back to Main Table of Contents

type, pool ownership, and household composition) as well as six continuous predictors (household income (in $1000s), number of adults in the home, number of children in the home, years in home, years residing the state of Arizona, and years expecting to live in Arizona). Analyses predicting the cold month usage composite included 101 households; those predicting the neutral and hot month composites included 99 households.

overall models using attitudes to predict seasonal energy usage was significant: for cold months F(6, 130) = 0.98, R 2 = .043, n.s.; for neutral months F(6, 127) = 1.45, R 2 = .064, n.s.; and for hot months F(6, 127) = 1.19, R 2 = .053, n.s..

Within these models, no individual attitude composite was found to have a significant effect.

Motives for Conserving Energy Two separate models examined the extent to which energyconservation motives predicted each of the three seasonal energy use composites. For each season, one model included only the two motivation composites created from individual motivation items: the Responsibility Motives composite and the social motives composite. Analyses predicting cold month usage included 134 households; those predicting neutral and hot month usage included 131 households. None of these three overall models was significant: for cold months F(2, 131) = 0.17, R 2 = .003, n.s.; for neutral months F(2, 128) = 0.80, R 2 = .001, n.s.; and for hot months F(2, 128) = 1.20, R 2 = .018, n.s.. In these models, neither of the motivation composites proved to be a significant predictor.

All three overall models predicting seasonal energy usage were significant: for cold months F(14, 86) = 2.32, R 2 = .274, p = .009; for neutral months F(14, 84) = 3.05, R 2 = .337, p = .001; and for hot months F(14, 84) = 3.40, R 2 = .361, p < 001. For all three seasonal composites, households with greater numbers of children (but not numbers of adults) consumed more electricity: for cold months B = 170.11, p = .029; for neutral months B = 168.72, p = .019; and for hot months B = 227.41, p = .072.

Greater number of yearsâ&#x20AC;&#x2122; residence in the current home also predicted greater baseline electricity use in all three seasonal composites: for cold months B = 12.00, p = .081; for neutral months B = 13.51, p = .037; and for hot months B = 22.04, p = .054. However, greater number of yearsâ&#x20AC;&#x2122; residence in the state of Arizona was associated with lower household electricity usage during the neutral (B = -6.60, p = .036) and hot months (B = -11.24, p = .043), likely reflecting acclimation and/or adaptation of lifestyle to the late-spring and summer heat of this region.

However, all three models using the individual ipsatized (i.e., relative) motivation item ratings as predictors were significant: for cold months F(9, 112) = 2.33, R 2 = .158, p = .019; for neutral months F(9, 109) = 2.26, R 2 = .157, p = .023; and for hot months F(9, 109) = 2.34, R 2 = .162, p = 019. A number of consistent patterns emerged in terms of specific motivations significantly associated with energy use. The relative importance of preserving the environment was associated with lower energy use in all three models: for cold months B = -364.26, p = .005; for neutral months B = -281.19, p = .023; and for hot months B = -441.71, p = .044.

Finally, households with primarily desert-based landscaping used less electricity than households with no landscaping (the reference group), but only during the hot months (B = -1588.78, p = .024). Household income was not a significant predictor of electricity usage in any of the three seasonal composites.

Remarkably, higher relative importance of saving money as a motive for conserving energy was associated with higher baseline energy usage in all three models: for cold months B = 234.10, p = .008; for neutral months B = 270.23, p = .002; and for hot months B = 441.55, p = .004.

Attitudes Overall, explicit conservation-related attitudes and beliefs failed to predict actual electricity usage. Analyses predicting the cold month usage composite included 137 households; those predicting the neutral and hot month composites included 134 households. These analyses included composites assessing: belief in global warming; personal value of conservation; belief in personal and societal efficacy to change energy use; belief in human supremacy over environmental issues and problems; negative emotions when thinking about the effects of energy use on the environment; and intention to conserve energy in the next few months. None of the three Energy Efficiency on an Urban Scale

Finally, higher relative importance of making sure future generations do not inherit our environmental problems was also associated with higher baseline use in all three models: for cold months B = 251.45, p = .007; for neutral months B = 235.19, p = .009; and for hot months B = 379.16, p = .017. One possibility is that this motivation is more salient for adults with children, and as noted above, households with more children were found to consume more electricity in these analyses.

36

Global Institute of Sustainability

Back to Main Table of Contents

R.6: HOW DID PARTICIPANTS FIND OUT ABOUT THE PROGRAM? DOES THIS PREDICT LEVEL OF PARTICIPATION (CHECKUP ONLY, UPGRADE, NEITHER) AND/OR ENERGY SAVINGS?

effects remained significant in the final logit model including significant predictors from the five more constrained models.

However, the decision to upgrade was not significantly predicted by how participants had heard about the EP program, either in terms of total number of channels (Spearman’s r = -.016, p = .827) or in terms of individual channels in the logit analysis. The implication is that multichannel marketing is important for getting energy efficiency program prospects “in the door” and that other factors influence their decision to follow through on an upgrade.

The majority of survey respondents (66.8%) had heard about the Energize Phoenix program from at least one source by the time they completed the survey. Respondents were equally likely to have heard about the program from members of their communities as from formal marketing efforts, consistent with prior evidence that word-of-mouth and communitybased social marketing are crucial components in behavioral intervention campaigns. A smaller proportion of respondents had heard about the program from a contractor.

R.7: IS THERE A REBOUND EFFECT, SUCH THAT GETTING THE UPGRADES INCENTIVIZES PARTICIPANTS TO USE MORE ENERGY IN OTHER WAYS?

FIGURE 6: HOW RESIDENTIAL SURVEY RESPONDENTS HAD HEARD ABOUT EP

Addressing this question will require follow-up data, not yet available, with participants who received upgrades. Thus we are unable to address this question at this time.

R.8: IS THERE A SPILLOVER EFFECT, SUCH THAT GETTING UPGRADES IS CORRELATED WITH A CHANGE IN PARTICIPANTS’ SAVING/CONSERVING IN OTHER WAYS? Addressing this question will require a follow-up survey, not yet complete, with participants who received upgrades. Thus we are unable to provide data on this question at this time.

SUMMARY

Source: ASU Global Institute of Sustainability

Program Participation: Although the correlational nature of this study necessitates great caution in inferences about causal direction, the findings presented above suggest that psychosocial factors can be useful predictors of people’s decisions to seek a home energy checkup and/or pursue an upgrade, information which may be useful in shaping programs and their marketing. The specific predictors of these two levels of participation differed in very important ways.

This is consistent with research from other areas of behavior change; for example, the well-known “Stages of Change” model in health psychology states that individuals at different levels of the change process (e.g., thinking about quitting smoking versus planning a quit attempt) are motivated and supported in different ways. The present results suggest a similar approach may be needed for promoting residential energy-efficiency upgrades and other investments.

***p < 001. Denotes a significant effect of having heard about EP through this channel (versus not having heard about EP through this channel) on probability of receiving a home energy checkup.

On average, respondents had heard about the EP program through 0.68 channels (SD = 0.61; includes participants who had not heard about the program at all). The number of ways respondents had heard about the program was a strong and significant predictor of receiving a checkup (Spearman’s r = .43, p < .001). A simultaneous logit analysis with the four channels as well as total number of channels fit the data significantly better than an empty model, and predicted more than 20% of variability in checkup vs. non-participation outcomes (Cox & Snell R 2 = .194). This effect was not specific to any particular channel of information. In a simultaneous logit analysis, all four channels were strongly, independently, and significantly associated with greater likelihood of

At the checkup level, demographic and household factors proved to be important predictors of participation. Adult couples without children and single families with children were most likely to receive checkups. Households with fewer children, and those in which the respondent was a

checkup-level participation: marketing (Wald = 20.55, p < .001, B = 1.14), contractor (Wald = 52.53, p < .001, B = 2.99), community (Wald = 54.23, p < .001, B = 1.98) and other (Wald = 12.58, p < .001, B = 1.02). Notably, all four Energy Efficiency on an Urban Scale

37

Global Institute of Sustainability

Back to Main Table of Contents

non-Hispanic White, were also more likely to receive a checkup.

Taken together, these factors may reflect a combination of household stability with available time, such that established single-family households with no or few children had a combination of motivation and opportunity that facilitated seeking out a checkup.

these effects, however, household characteristics also had surprisingly little predictive utility. Attitudes also failed to predict historical patterns of energy use.

Of the four types of predictors examined in these analyses, motivations for conserving energy appeared to have the most powerful effects. Higher motivation to preserve the environment, relative to other motives, was strongly associated with lower baseline energy usage. In contrast, higher relative motivation to save money was associated with higher baseline energy usage. Of course, it may be that those who are currently spending more on their monthly electricity bills are more motivated by the promise of financial savings. However, this finding also suggests that primarily financial motives are not, in and of themselves, very effective in promoting energy conservation behavior. This effect is consistent with a rich body of social psychology literature suggesting that external motivations for a behavior (such as the promise of material reward) may actually begin to replace and inhibit internal motivations (such as enjoyment), eventually reducing the motivation to engage in that behavior once the reward is no longer on the table. Finally, higher motivation to make the world better for future generations was also associated with higher baseline energy use – a point discussed under “caveats,” below.

Attitudes also had strong, yet surprising, implications for probability of checkup-level participation. Although higher intention to conserve was associated with higher likelihood of getting a checkup, several other pro-environment attitudes were associated with lower likelihood of getting a checkup.

Higher scores on social motivation to conserve energy (e.g., to be like other people; to be seen as eco-friendly), as well as certain household characteristics suggesting pro-environment decisions (e.g., water-conserving landscaping; not having a pool) were associated with lower checkup rates. Taken together, these effects may suggest a kind of person who strongly self-identifies as eco-friendly, but believes (rightly or wrongly) that he/she has already taken sufficient steps in this regard, and is not motivated to seek out additional ways to improve energy efficiency in the home.

Program marketing also had a clear impact at the checkup stage. The more ways respondents had heard about the program, the more likely they were to pursue a checkup. It is worth noting that most surveys were conducted quite early in the program period, leaving ample time for households to seek a checkup after completing the survey itself if they chose to do so. Importantly, household income was not a significant predictor of checkup level participation.

Caveats: Two caveats are especially important for interpreting these findings. First, as noted above, these data are not based on an experimental design, and thus inferences about the direction of causality are not warranted. One reasonable possibility, which we were unable to explore, is that some observed associations were actually accounted for by structural variables not included in this data set. For example, it may be that households in which the respondent placed greater value on protecting future generations from current damage to the environment were more likely to be households with children, and that households with more children simply live in larger homes. Thus, the effects of the “future generations” motivation and of number of children in the household on electricity usage may actually reflect size of the home.

Baseline Electricity Use: A different set of factors predicted actually getting an energy-efficiency upgrade, once the checkup was complete. Resources appeared to play a much more important role at this stage – both higher household income and respondent full-time employment were associated with greater likelihood of upgrade-level participation. In contrast, demographics, attitudes and motivations, and Energize Phoenix program exposure were less important predictor.

The predictors of baseline (i.e., prior to any upgrades) energy usage also differed in important ways from the predictors of program participation. Demographic variables were of little use in predicting electricity consumption, especially during the hot months. Households with more children, and those that had lived in the current home for a longer duration, tended to use more electricity. However, those who lived in Arizona longer used less energy in neutral and hot months. Beyond Energy Efficiency on an Urban Scale

Second, although most Energize Phoenix program participants completed the survey, only a small proportion of eligible nonparticipants did so. As a result, some of the observed effects in predicting participation may reflect biases in our control group (i.e., non-participants) rather than “real” effects. It is possible, for example, that individuals with strong proenvironment attitudes were more likely to fill out a survey even 38

Global Institute of Sustainability

Back to Main Table of Contents

if they had no intention of participating, whereas less ecofriendly non-participants did not bother. It is still noteworthy that these individuals did not then go on to request a checkup after completing the survey, as the residential surveys were distributed early in the project period and most who completed one would have had ample time to do so. However, the nonparticipants in the current sample likely differ from the larger population of non-participants in important ways, and this may have influenced our findings.

non-participant comparison group drawn from the National Establishment Time Series (NETS) industry database (see Appendix F: Commercial Participation Factors).

Moreover, because many survey respondents skipped questions in completing the surveys, and logit analysis uses listwise deletion (i.e., if even one data point needed for an analysis is missing for a given business, that business is removed from the analysis entirely), the analyses below are based on much smaller sample sizes. This raises further questions about the representativeness of the sample, and indicates that considerable caution is needed in interpreting these results.

COMMERCIAL PROJECT DATA COLLECTION: SAMPLE AND PROCEDURES As with the Residential Surveys, Commercial Survey respondents were approached by a number of methods. First, research assistants were trained and ASU-approved per City requirements for door-to-door survey administration, and then assigned a set of commercial addresses within the program target area. In this way, surveyors attempted to obtain a completed survey and waiver from every business in the target area. If no employee was willing to complete the survey at that time, the surveyor left a stamped, addressed envelope so they could return the survey via mail. Surveys were also included as an optional component of the participant program application package administered by contractors. Per federal Better Buildings grant program reporting requirements, the utility data waiver was mandatory for program participants.

Surveyors went out again at the end of the second year and beginning of the third year specifically to organizations that had completed upgrades but had not returned a survey in order to make a final attempt to secure one.

Preliminary Analyses and Subscale Development: Attitudes Items. The ten initial items assessing attitudes toward the environment and conservation were subjected to a Principal Components Analysis (PCA), with the aim of identifying conceptually meaningful sets of items that could be treated as subscales in this sample. Unfortunately the PCA failed to identify statistical patterns among items that would point the way to meaningful composites. Based on analysis of the conceptual meaning of the items, we created three attitude subscales and two individual items used to predict participation in EP, as well as patterns of energy use and savings.

• “Climate Change Beliefs”: Two items assessing agreement that the world is experiencing an environmental crisis caused by human behavior (inter-item r = .64).

• “Valuing Conservation”: Two items assessing agreement that it is important to conserve energy and natural resources (inter-item r = .61).

After removing redundant surveys from the same business (the more complete survey was retained, or if both were complete, the first one completed was retained) and surveys from businesses that both did not own the building and were not allowed to improve it, 318 usable surveys were available. Of these, 226 were completed by respondents from businesses that received an upgrade by March 31, 2013, and 92 were completed by respondents from businesses that did not participate as of this cut-off date. It is important to note that this indicates an extremely low compliance rate among non-participant businesses, raising the possibility that the non-participant comparison group in these survey analyses is biased relative to all possible non-participant businesses in important ways. We discuss implications of this possible bias in the Commercial Project analysis summary, and crossreference our findings with those of the Geography Team, which conducted more limited but similar analyses using a Energy Efficiency on an Urban Scale

• “Personal and Societal Efficacy”: Two items assessing belief that humans are capable of changing the effects they have on the environment (inter-item r = .52).

• “Company Efficacy”: One item assessing belief that the company is able to change its energy use.

• “Prioritization of Economic Growth,” One item assessing belief that economic growth is more important than environmental protection.

The remaining two items were not included in subsequent analyses.

Preliminary Analyses and Subscale Development: Motivation Items. The nine initial items assessing the business’s 39

Global Institute of Sustainability

Back to Main Table of Contents

motivations for conserving energy were subjected to a Principal Components Analysis (PCA), again with the aim of identifying conceptually meaningful sets of items that could be treated as subscales. As with the attitude items, the PCA was conducted using data from all available Commercial questionnaires. In a preliminary PCA, examination of the scree plot of eigenvalues clearly indicated a two-factor solution.

We therefore conducted a follow-up PCA forcing two factors and using direct oblimin rotation to maximize simple structure (i.e., tight fit of individual items to one factor or the other).

Notably, this PCA analysis indicated a quite different factor structure for motivation than was observed among residential participants. Based on the PCA analysis, and considering the conceptual meaning of the items, we created two motivation subscales, entered into one binary logistic regression analysis predicting participation:

the observed effect reflects sampling error rather than a “real” effect), the weights should be interpreted with caution, as they depend greatly on the specific predictors and sample in a particular analysis.

In the first logit analysis, variables reflecting features of the business that involve decision-making structure, nature of the business per se, or demographics of the owners were entered simultaneously. Predictor variables included: legal status of business (corporation vs. other); property status (own vs. lease), only place of business, small business, woman-owned business, minority-owned business, veteran-owned business, whether or not the business has a formal policy promoting alternative transportation, whether or not the business has a formal sustainability policy, whether or not the business currently takes specific measures to improve electricity efficiency, whether or not the business had received any offers for other energy incentives/rebates in the preceding two years, and whether or not the business had taken advantage of any other offers in the past two years (all Y/N). This analysis included 139 businesses (98 participants and 41 nonparticipants) for which all necessary data were available.

• “Environmental Motivation”: Three items assessing motivations to preserve the environment, to protect future generations, and to do the “right thing” (Cronbach’s alpha = .89).

• “Business Motivation”: Three items assessing motivation to keep up with other businesses, to be more competitive, and to follow industry leaders (Cronbach’s alpha = .79).

This model accounted for approximately 20% of variability in participation outcomes (Cox & Snell R 2 = .194), classifying 78% of businesses correctly, and a number of individual predictors emerged as significant. Specifically:

As noted in the section on the Residential Program, above, self-report measures of motivation often show ceiling effects, such that most people give a very high rating of every motive.

In anticipation of this problem, we conducted an additional logit analysis in which ipsatized (i.e., standardized to the individual’s own mean and standard deviation across the nine motivation items) individual item scores for all nine original items were entered as predictors.

• Legal Status was one such predictor, with corporations showing greater likelihood of participation than other types of entities (e.g., sole proprietorships, partnerships, LLCs; Wald = 9.34, p = .002, B = 1.51).

• Businesses that owned their building, rather than leasing it, were more likely to participate in the program (Wald = 8.33, p = .004, B = 1.35).

RESULTS C.3: WHAT BUSINESS-LEVEL VARIABLES AND BUILDING CHARACTERISTICS PREDICT HOW LIKELY AN ORGANIZATION IS TO PARTICIPATE IN THE ENERGIZE PHOENIX PROGRAM?

• Businesses that reported already taking specific electricity efficiency steps were somewhat more likely to participate than those that did not, although this effect only approached significance (Wald = 2.63, p = .105, B = 0.81).

Implications of business-level variables for EP participation were examined in three logistic regression (“logit”) analyses. In these analyses, several possible predictors of a dichotomous outcome are entered into the statistical model; for each predictor, the analysis provides a test of the statistical significance of the effect, as well as a “weight” estimating the magnitude of the effect in terms of change in log-odds of participation. While logit produces fairly accurate estimates of significance (i.e., p-value or the probability that Energy Efficiency on an Urban Scale

• Woman-Owned businesses appeared less likely to participate than other businesses (Wald = 5.09, p = .024, B = -1.48).

In a second logit analysis these same predictors were entered along with features of the physical plant expected to influence energy demand. Specifically, the second logit added: lighting 40

Global Institute of Sustainability

Back to Main Table of Contents

type (Y/N for each – incandescent, fluorescent, compact fluorescent/CFL, high pressure sodium, LED, Halogen) and major equipment (Y/N for each – pumps/motors, ovens/ kilns, refrigerators, computers/TVs, compressed air, process equipment, cooling towers), each as a binary Y/N variable. The analysis was performed using backward stepwise regression.

In this version of logistic regression, all predictors are included in the model on the first “step.” In subsequent steps, one non-significant predictor at a time is removed from the model, simplifying it and increasing statistical power (by increasing the case-to-predictor ratio as much as possible). This process continues until all predictors have a p-value below a given threshold (in this analysis, p = .10) indicating that all remaining predictors are significant.

of processing equipment, all of the predictors that were significant in the preceding analysis remained significant.

Legal status of corporation was associated with significantly higher probability of participating (Wald = 8.17, p = .004, B = 1.09), as was building ownership relative to leasing (Wald = 6.98, p = .008, B = 0.92). Again, women-owned businesses appeared less likely to participate (Wald = 4.16, p = .041, B = -1.06). Finally, use of incandescent lighting was still associated with higher likelihood of participation (Wald = 6.91, p = .009, B = 1.04).

C.7: ARE ORGANIZATIONS THAT ALREADY ENCOURAGE GREEN BEHAVIOR MORE LIKELY TO UPGRADE? ARE THEY MORE SUCCESSFUL OVERALL IN REDUCING THEIR ENERGY CONSUMPTION? As noted above (see C.6), analyses based on our survey data did not find that businesses with formal sustainability or alternative transportation policies were more likely to participate in the Energize Phoenix program than businesses without such policies. Businesses that reported having previously taken concrete steps to improve their energy efficiency were somewhat more likely to participate, although this effect did not reach conventional levels of significance.

This type of analysis has both strengths and limitations. On one hand it can accommodate a larger number of predictors while still retaining reasonable statistical power; on the other hand, it risks “overfitting” the model, capitalizing on chance characteristics in a particular sample that may not generalize to other populations. For these reasons we chose to run both the more constrained (above) and more inclusive versions of the model, and compare findings. Predictors that were significant in the more constrained model, but lost in the larger model, may indicate apparent business-level characteristics that were actually explained by features of energy demands in the physical plant.

C.8: DO THE ENVIRONMENTAL ATTITUDES AND BELIEFS OF THE BUSINESS’S DECISION-MAKERS RELATE TO THE DECISION TO UPGRADE? Attitudes Analysis of attitudes was based on a subsample of 169 businesses (117 participants, 52 non-participants) for which the necessary data were available. In this subsample, the logit model including attitude variables did not improve significantly upon an empty model with no predictors, and none of the attitude variables significantly predicted participation.

The second logit analysis included 139 businesses (97 participants and 40 non-participants) for which all necessary data were available. The model converged in the 27th step.

At this point, four predictors were significant – legal status (corporation vs. other), property status (own vs. lease), women-owned status, and use of incandescent lighting – and a fifth predictor, use of processing equipment, was marginally significant.

Motives for Conserving Energy Analysis of motivations for conserving energy as predictors of participation was based upon a subsample of 196 businesses (140 participants, 56 non-participants) for which the necessary data were available. First, a simultaneous logit model including only the two motivations composites as predictors was significantly better than a model with no predictors, but accounted for only a small amount of variability in participation outcomes (Cox & Snell R 2 = .037).

Specifically, the business motivation composite emerged as a significant predictor of participation (Wald = 7.14, p = .008, B = 0.28). The effect of environmental motivation was not significant.

In the third logit analysis, these five predictors were entered into a single, simultaneous logit model predicting participation. This analysis used a larger sample, as only data on these five predictors were needed for inclusion, making it possible to obtain a better estimate of the effect sizes associated with these predictors. The analysis included 198 businesses (141 participants and 57 non-participants) for which the necessary data were available.

The model accounted for approximately 15% of variability in participation outcomes (Cox & Snell R 2 = .146), classifying 77% of businesses correctly. With the exception of use Energy Efficiency on an Urban Scale

41

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 7: RELATIVE MOTIVATIONS TO CONSERVE BY COMMERCIAL PARTICIPATION STATUS

and motivations as measured through the commercial survey with energy usage during the year of 2010 – prior to the implementation of any upgrades subsidized by Energize Phoenix. Analyses included 152 businesses for which survey and 2010 energy use data (provided by the local utility) were both available.

Because the utility’s monthly billing cycles vary from business to business, and daily usage data were not available, it was necessary to transform the raw billing data to create energy usage estimates for actual calendar months that could be compared across businesses. This was done in the same way as in the residential energy use analyses above. For each business (a) mean daily usage in kWh was calculated for each billing cycle, and (2) an estimate of each month’s use was calculated by multiplying the daily mean from each of the two relevant bills by the number of days in that month covered by each bill, and then summing these two figures. Also as in the residential analyses, we formed three different energy use composites for each business: cold months (January, February, November, and December); neutral months (March, April, May, and October); and hot months (June, July, August, and September).

Source: ASU Global Institute of Sustainability *p < 05, +p < .10. Denotes a significant or marginally significant effect of this relative motivation in predicting commercial participation status, in the logistic regression model with all relative motivations entered as predictors.

Second, all ipsatized individual item scores are entered as predictors in a separate logit model (186 businesses; 134 participants, 52 non-participants). This model is a significant improvement on a model with only an intercept and accounts for a greater amount of variability in participation outcomes – approximately 10% (Cox & Snell R 2 = .098). In this analysis, relative intensities of four motivations were associated with greater likelihood of participating: promoting the business as environmentally responsible (Wald = 5.85, p = .016, B = 0.83); saving money on electricity bills (Wald = 4.64, p = .031, B = 0.60); making the business more competitive (Wald = 4.51, p = .034, B = 0.65); and keeping up with what other businesses are doing (Wald = 2.85, p = .091, B = 0.47).

Attitudes Analyses uncovered no evidence that respondent attitudes toward conservation were associated with 2010 energy consumption by his/her business. None of the three overall models predicting energy use from attitudes was significant: for cold months F(5, 116) = 0.48, R 2 = .020, n.s.; for neutral months F(5, 115) = 0.46, R 2 = .020, n.s.; and for hot months F(5, 115) = 0.44, R 2 = .019, n.s.. None of the individual attitude composites had a significant effect in any model.

Taken together, these findings suggest that businesses may be motivated to take advantage of energy efficiency subsidy programs such as Energize Phoenix primarily in order to save money and promote their business, rather than by proenvironmental motives per se. Importantly, these motivations remained at least marginally significant in a follow-up logit model including the business characteristics identified in section C.3 above, indicating that the motivational effects do not overlap with effects of more objective business characteristics.

Motives for Conserving Energy In contrast, the respondent’s motives for conserving energy emerged as powerful predictors of 2010 electricity use.

The three models in which the business and environmental motivation composites were entered as predictors were all significant: for cold months F(2, 149) = 6.93, R 2 = .085, p = .001; for neutral months F(2, 148) = 7.15, R 2 = .088, p = .001; and for hot months F(2, 148) = 7.63, R 2 = .093, p = 001. For all three seasonal composites, higher scores on the business motivation composite were associated with higher 2010 electricity usage: for cold months B = 28290.79, p = .003; for neutral months B = 32478.79, p = .002; and for hot months B = 39498.18, p = .001. In contrast, environmental motivation composite scores did

C.9: DO MORE PRO-ENVIRONMENTAL ATTITUDES AND BELIEFS PREDICT LOWER HISTORICAL ENERGY CONSUMPTION, AND/OR HIGHER POST-CHECKUP AND POST-UPGRADE ENERGY SAVINGS? In order to examine the predictors of baseline energy usage, we examined the statistical association of respondent attitudes Energy Efficiency on an Urban Scale

42

Global Institute of Sustainability

Back to Main Table of Contents

not significantly predict electricity usage in any of the three seasonal models.

27% of participation outcomes (Cox & Snell R 2 = .273).

Having heard about the program from a contractor emerged as a marginally significant predictor of participation (Wald = 3.63, p = .057, B = 1.34). Not surprisingly, not having heard about the program at all (Wald = 5.14, p = .023, B = -1.62) was negatively associated with probability of participation.

C.10: DOES AN ORGANIZATION’S DESIRE TO LOOK “GREEN” INFLUENCE UPGRADE DECISIONS? As noted above (see C.8), greater intensity of motivation to market the business as environmentally responsible, relative to the other motives measured in the survey, was significantly and uniquely associated with participation in Energize Phoenix.

SUMMARY Program Participation: Analyses of the predictors of business participation in Energize Phoenix offered some new insights into the key factors that drive such investments. At least in these analyses, features of the physical plant (with the exception of incandescent lighting) had little impact on probability of participation. Demographic variables (with one exception under “caveats” below) and attitudes also generally failed to predict participation. Key predictors included aspects of the business that involved legal structure sophistication (legal status as a corporation) and decision-making autonomy (ownership of the building), as well as motivations involving promotion and enhancement of the business itself (e.g., saving money, making the business more competitive, ability to market the business as “green”). Pro-environment attitudes, motivations, and policies had little to do with the decision to pursue an energy-efficiency upgrade. Rather, this decision appears to rely on pragmatic, bottom-line issues – a finding with clear implications for marketing future such programs to local businesses.

C.13: HOW DID ORGANIZATIONS FIND OUT ABOUT THE PROGRAM? IS THERE A CORRELATION WITH THEIR LEVEL OF PARTICIPATION (CHECKUP OR UPGRADES)? WITH HOW MUCH ENERGY THEY SAVE? Data were available from 267 businesses regarding how they had heard about the Energize Phoenix program. Of these, 56 (21%) reported not having heard about the program at all prior to taking the survey; 49 (18%) had heard about EP through formal marketing channels (newspaper ads, web sites, etc.); 129 (49%) had heard about the program from a contractor; 25 (9%) from family, friends, or a community member; and 28 (10%) from some other source.

FIGURE 8: PARTICIPATION RATES BY HOW BUSINESSES HAD HEARD ABOUT EP PROGRAM

Having heard about the program from a contractor was also associated with higher likelihood of participation, whereas other forms of marketing and exposure did not appear to have this effect. This diverges from the findings with the residential study, in which all channels of program promotion and an increase in the number of channels of exposure were associated with higher probability of checkup level participation. Business decision-makers may be more swayed by endorsements of the program by actual contractors, along with the detailed information that contractors are able to provide. The results also likely reflect the contrasting program marketing approaches to the two sectors.

Source: ASU Global Institute of Sustainability +p < .10. Denotes a marginally significant effect of having heard about EP through this channel on probability of participating in EP.

In a logit analysis with a single predictor of participation status (178 businesses included; 122 participants, 56 non-participants), total number of ways participants had heard about the program was not significantly related to the decision to participate (Wald = 0.10, p = .749). Implications

Baseline Energy Use: As observed for program participation, explicit pro-environmental attitudes and beliefs attitudes failed to predict baseline (2010) energy use by businesses included in these analyses. Motivations were much more powerful in this regard. Strikingly, however, higher businessfocused motivation was associated with higher baseline electricity usage, not lower usage. Again, this suggests that

of specific channels for participation were examined using a separate, follow-up logit analysis in which the five response options above (four channels plus “had not heard”) were entered simultaneously as predictors (101 business included; 55 participants, 46 non-participants). This model improved significantly upon the empty model and accounted for over Energy Efficiency on an Urban Scale

43

Global Institute of Sustainability

Back to Main Table of Contents

energy-efficiency attempts by businesses are fundamentally about the bottom line. Whether financially-driven motivations are effective in promoting successful energy conservation is unclear.

its consequences more directly and precisely. The feedback principle has been applied to enhance performance and improve behavior in a variety of domains, including education and health promotion. Increasingly, researchers are interested in whether real-time feedback on energy consumption helps people to alter their behavior in the energy-efficiency domain as well.

Caveats: Similar cautionary notes are needed here as in the summary of the residential study. The correlational study design makes it difficult to infer causal direction, even when a significant predictor of program participation or baseline energy use is observed. Also, the present sample of nonparticipants is probably quite biased compared to the total population of non-participating but eligible businesses. For example, although these analyses suggest that women-owned businesses were less likely to participate, it is highly probable that women business owners who chose not to participate in the program were more likely than men to agree to complete the survey anyway. It is useful to compare the findings from these analyses with those of the Geography Team (see Appendix F: Commercial Participation Factors), who conducted similar analyses using the NETS system to provide information about non-participating businesses. To the extent that the two analyses offer converging findings, those findings should be quite reliable. Where the findings differ, the NETS-based findings may be more trustworthy.

Most studies so far have examined the impact of feedback devices on electricity usage by single-family, middle- and upper-middle class homes. In two randomized, controlled studies executed in coordination with the Dashboard team as well as City of Phoenix and ASU partners, the behavior team examined the impact of TED device feedback on electricity consumption in quite different settings: a college residence hall and a low-income subsidized housing apartment complex, both in the light rail corridor that served as the target area for Energize Phoenix.

DATA COLLECTION: SAMPLE AND PROCEDURES Taylor Place University Residence Hall The Taylor Place residence hall is located on the Downtown Phoenix campus of Arizona State University. In total, 121 dorm rooms on four floors of the residence hall were included in the present study. Floors were loosely segregated by major; most residents on two of the floors were journalism/mass communication majors, and most residents of the other two floors were public programs or criminal justice majors, although there was some mixing across floors. Approximately two-thirds of residents in these rooms were women, and one-third were men. Residents ranged in age from 17 to 21 years, with a mean just over 18 years. Although many residents declined to provide information regarding their ethnicity, approximately half of those who did self-identified as non-Hispanic White, approximately a third as mixedethnicity, just over 10% as Hispanic/Latino, and the rest as another ethnicity. Similarly, many residents declined to provide information about their political affiliation. Among those who did, approximately a third identified as Democrats, and another third as Republicans, with the remainder identifying as Independent, Green, or having another affiliation.

DASHBOARD PROJECT The Energy Detective (TED) device, is an in-home electricity monitoring system that allows residents to track their realtime usage in kilowatt-hours (kWh) and/or dollars spent, among other tracking functions. The TED unit consists of three distinct but linkable components: a set of clamps and a transmitter that installed within the home’s electrical panel to wirelessly transmit electricity use information; a “Gateway” wall plug that receives and records this information; and a “Display” that makes the information visually available to residents. Thus, the full system allows people to view the amount of electricity they are currently using, and to track changes in electricity consumption that are linked to use of specific appliances.

An extensive body of research has documented the helpful role that real-time feedback can play in behavior modification (see Appendix XX-3 for a brief review of this literature). Especially when the consequences of a behavior are otherwise fairly distal and aggregated over long periods of time, as is the case with energy use (i.e., electricity bills aggregate a month’s usage, and reflect consumption from weeks earlier), real-time feedback gives people the chance to map their behavior to Energy Efficiency on an Urban Scale

Prior to the start of the semester, TED Gateways were installed in all 121 rooms and set to record energy use from lighting and plug-in devices. TED real-time displays were also installed in half of the rooms on each floor using a “checkerboard” approach, such that alternating rooms on each side of a hallway received a device, and either one’s room or the one 44

Global Institute of Sustainability

Back to Main Table of Contents

across the hall had a display. In addition, each floor was assigned quasi-experimentally to one of two conditions: designation as a “community floor” with discussions of energy conservation to be organized by the floor’s residential assistant; or as a non-community floor (see sections D.5 and D.6 under “Results,” below for more detail on this aspect of the intervention). One floor representing each academic major was assigned to each condition, in order to avoid confounding these two factors. Of the 121 rooms, 28 had a TED display and were located on a community floor, 30 had a TED display but were not on a community floor, 31 did not have a TED display but were on a community floor, and 32 did not have a TED display and were not on a community floor.

replacement survey in early October 2012, with an explanation of the error and a request to complete it again.

Surveys were identified by room and resident’s date of birth, allowing us to identify particular individuals and match their surveys to other data. Both residents of a given room had the opportunity to complete a survey, and in many cases both did so. However, data analyses in this study were conducted using room as level of analysis, because the key dependent variable (energy use) and independent variable (TED Display Presence) were both room-level variables. As a result, it was necessary to choose a single survey to represent each room.

If only one resident had completed the survey, that one was used. If both residents had completed the survey, but only one resident completed a follow-up survey in December (not used in the present analyses), that resident’s survey was used to represent the room. If both residents had completed the main survey, and neither or both had completed the follow-up survey, then the survey representing the room was selected using a random number generator. Finally, if both an initial (August 2012) and replacement (October 2012) survey were available from the same individual, the initial survey was retained to represent the room. In this manner, 82 surveys were chosen for inclusion (no survey was available from 39 rooms). Of these, 56 were from the original administration in August, and 26 from the follow-up in October.

FIGURE 9: TAYLOR PLACE DASHBOARD STUDY EXPERIMENTAL DESIGN

A variety of energy-savings resources were provided to residents in all participating rooms, and thus served as constants for the study. First, each room was provided with a “smart” power strip that included “master” and “slave” outlets, such that power to any devices plugged into the “slave” outlets was cut automatically when the “master” device was turned off. Second, residents were informed that their residential assistants had “Kill-A-Watt” devices available for check-out. These are small electricity load monitors that plug into a wall outlet; any electrical device can beplugged into the Kill-A-Watt, which then shows how much electricity the device consumes. Third, all residents received five emails from the behavior team over the course of the Fall Semester, each containing two energy-saving tips (e.g., turning up the temperature on mini-refrigerators; exchanging traditional holiday light strings for energyefficient LED lights).

Source: ASU Global Institute of Sustainability

Residents moved into their rooms in the middle of August, 2012. The Taylor Place Resident Survey (see Appendix XX-4) awaited each resident in his or her room at that time, with a request to complete it and return it to the floor’s residential assistant. The surveys included questions on resident demographics, number of hours per day the room is typically occupied, presence and use of eight specific electrical devices in the room (laptop and desktop computers, mini-refrigerator, microwave, electric kettle, television, fan, and hair dryer), two questions on frequency of talking with others about energy usage, two items on thinking about energy usage when turning on/off electrical devices, nine items on motives for conserving energy, and eight items assessing conservationrelated attitudes. Unfortunately, many of the initial, completed surveys were discarded by a residential assistant from a non-participating floor of the residence hall, who was unaware of the study, and these were not retrievable. In order to maximize available survey data, we provided residents with a Energy Efficiency on an Urban Scale

At the end of the semester, after residents had vacated the residence hall for the holiday break, daily electricity use data were downloaded from the TED Gateways in each room.

Mean daily usage from the months of September, October, 45

Global Institute of Sustainability

Back to Main Table of Contents

and November 2012 was averaged for each room to create a Fall semester consumption index. These months were selected because residents were in the room for the entire month (avoiding move-in and move-out months of August and December), and because they represent a range of weather conditions in the Phoenix area, from hot and humid (September) to temperate and dry (November). This threemonth index was used as the dependent/outcome measure in all analyses of energy use.

(June-July 2012), leaving 23 households in this condition.

Similarly, 11 households in the EdTEDBud condition were unreachable or withdrew from the study at this time, leaving 17 households in this condition.

TED feedback “dashboards” were installed in the units of remaining households in the EdTED and EdTEDBud conditions, and Gateways installed in all participating households, between June 14 and July 12 of 2012. Electrical contractors made the necessary adjustments to the unit’s electrical panel, and installed the units.

Sidney P. Osborn Public Housing Development The Sidney P. Osborn complex is a publicly subsidized, lowrise housing development located in downtown Phoenix. The complex has 146 units with 2-5 bedrooms each across 26 buildings, available for rent by families with annual incomes below a threshold determined by household size (e.g., $56,600 for a 4-person household). Resident households receive “allowances” of between $22 and $111 per month for electricity, depending upon month of year and number of bedrooms, but are required to pay any electricity charges above this amount. Because the complex is owned and managed by the city of Phoenix, it was possible to obtain electricity usage records for all units.

All participating households were offered personalized, in-home energy conservation education sessions between June 27 and August 8. During these visits the research team administered an initial survey, performed an inventory of major electrical appliances and devices, measuring the energy used by each item and sharing that information with the participant, and provided information about steps a resident could take to reduce their energy consumption (e.g. closing curtains during peak sunlight hours). Not all households took advantage of this opportunity: 13 (50%) of the households in the EdOnly condition, 19 (83%) of those still in the EdTED condition, and 13 (76%) of those still in the EdTEDBud condition received such visits. Only data from these 45 households were ultimately included in data analyses.

Participation in the study was limited to those in 2-4 bedroom units, those speaking English or Spanish as the primary language, and to those who had lived in their unit for at least 12 months and who intended to remain in the unit for at least six more months. This latter criterion was needed to ensure that pre-post intervention comparisons of energy use could be made for the same household.

All remaining households were visited again between September 19 and October 17 to administer a second survey, to verify that the TED displays were working (when applicable), and to answer any questions the household might have about energy use. The devices were removed and an exit survey administered between January 15 and February 15, 2013.

Households that completed the study and returned the TED device at this time were given a gift basket containing $75 worth of energy-saving devices and supplies.

In May 2012, participants were recruited through evening presentations at the complex, as well as door-to-door visits by members of the research team (all visits were completed in English or Spanish, as needed). By these means, 82 households were recruited for participation. Household was treated as the unit of analysis, and these 82 households were randomly assigned to three experimental conditions: (1) Education Session Only Control (“EDOnly,” n = 26); (2) Education Session + TED Device (“EdTED,” n = 28); and (3) Education + TED + household-specific Budgeting Information (“EdTEDBud,” n = 28) conditions. In this final condition, households received tailored electricity use budgeting recommendations in addition to the TED display, with the aim of helping households manage their usage more effectively.

Five households in the EdTED condition were unreachable or withdrew from the study at the time of TED device installation Energy Efficiency on an Urban Scale

Adjusting usage data for weather and building characteristics.

The initial step was to recode as missing any months where the total kWh used was implausibly low for an occupied apartment. We used the rationale that minimum usage for an occupied apartment would include at least one refrigerator running at all times, plus one CFL light for 8 hours a day. Using the energy usage calculator at (http://visualization.geblogs. com/visualization/appliances/) we calculated that this would amount to approximately 140 kWh/month. All values below that 140 recoded as missing (-99). Additionally one apartment had a usage in February 2012 that was approximately 10 times the expected value for an apartment of that size during that month and was also recoded as missing (-98).

46

Global Institute of Sustainability

Back to Main Table of Contents

Next, to account for building characteristics we computed two separate regression models, one for hot months (70 degrees and above) and one for cold months (below 70 degrees; there were no months with a mean of 70 degrees). Each data point (per month and per apartment) was treated as an independent observation. For each model, we estimated a regression line based on mean monthly temperature, Floor level (first or second floor), Orientation (N, S, E or W), Unit location within building (outside, or middle), Number of occupants, and Square footage of a given apartment and saved the residuals.

Because the models accounted for the intercept (constant) as well as the effects of each predictor, some degree of baseline usage was partialed out of the residuals. Because baseline usage is of conceptual importance, we re-adjusted for this by adding back in the mean of the predicted usage values for the cold (M = 364.20) and hot (M = 1132.69) models to the residuals for each month. This process resulted in a mean cold-month usage of 364.20 (sd = 195.07), and a mean hotmonth usage of 1132.69 (sd = 346.95; because the sum of the residuals is, by definition, zero the mean of the adjusted scores was necessarily equal to the mean of the predicted scores that had been added back in).

• Pro-Conservation Attitude: Principal components analysis was used to examine the factor structure of the eight conservation attitudes items. Examination of the scree plot strongly indicated a single-factor solution. However, the two reverse-scored items loaded negatively on this factor, even after reversing the scores so that higher values indicated more pro-conservation attitudes.

Anticipating that this indicated a response set on the part of participants completing the survey, we excluded these items, and created an index by averaging ratings of the remaining six items (Cronbach’s alpha = .83).

• Responsibility Motives: Based upon the principal components analysis conducted with the Residential Program survey participants (a much larger sample), the nine motives for conserving energy items were aggregated into two indices of Responsibility Motives and Social Motives. The Responsibility Motives index assesses responsibility-focused motives for conserving energy, such as saving money, preserving the environment, doing the “right thing,” and protecting future generations (Cronbach’s alpha = .83). See the “Data Collection” subsection of the Residential Project section of this report for

RESULTS

details on the items included in this index.

D.1: WHAT ARE THE DOCUMENTED KNOWNS AND UNKNOWNS REGARDING HUMAN BEHAVIOR LINKS TO RESIDENTIAL ENERGY EFFICIENCY?

• Social Motives: Also based on preliminary analysis from the Residential Survey, three items were averaged to create an index of social motives for conservation

Appendix XX-3, “Residential Energy Efficiency and the Psychology of Sustainable Behavior,” offers a brief summary of key approaches to the promotion of sustainable behavior that are well-documented in empirical research. These approaches commonly build on the rich literature on behavior change more broadly speaking, in the social psychology and health psychology fields. The review emphasizes evidence regarding in-home, real-time feedback devices, as these are the focus of the present study. However, several other empiricallysupported approaches are addressed as well.

(Cronbach’s alpha = .85). These items assessed desire to keep up with what others are doing, desire to be seen as environmentally responsible, and belief that people the respondent respects say conservation is important, as motives for saving energy. See the Residential Project section of this report for further details.

A single simultaneous regression analysis was used to examine the implications of non-manipulated person- and room-level variables on energy use in Taylor Place rooms in

D.2: WHAT PERSON-LEVEL VARIABLES PREDICT ENERGY USE LEVELS AMONG RESIDENTS IN MULTI-UNIT RESIDENTIAL COMPLEXES?

the study. This analysis was limited to 68 rooms for which

Taylor Place

deletion). Predictors in this analysis included participant sex,

survey data were available, and all necessary questions had been answered (SPSS regression procedures use listwise total number of electrical devices, number of hours per day

Mean daily electricity usage in the Taylor Place rooms was 1.91 kWh (SD = .76) across the three target months (SeptemberNovember 2012). In preparation for predicting energy use, three indices were created to reflect pro-environment attitudes and motivations for conserving energy.

Energy Efficiency on an Urban Scale

the room was typically occupied, the index of pro-conservation attitudes, and the two motivation indices. The overall model was marginally significant, accounting for approximately 17% of variability in energy use (R = .417, p = .057).

47

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 10: SCATTERPLOT PREDICTING ENERGY USAGE BY RESPONSIBILITY MOTIVATION

These regression models included the following predictors: respondent Gender, respondent Age, Duration of Residence in the unit, typical Thermostat Setting (on hot days), number of hours per day the unit was typically empty, perceived frequency of electrical bill overages, difficulty paying the electricity bill, how well respondents understand how much energy their electrical appliances/devices use, how well participants understand what they can do to save energy, and rank ordering of three motives for conserving energy (saving money, preserving nature, for children’s future). This last set of predictors was taken from a single task in which participants were shown pictures (e.g., dollar bills, children) illustrating each of these three reasons plus a fourth – because leaders of the community say it is important – and asked to rank their importance as reasons for conserving energy. Rank importance of “leaders” was not included in the analyses, as it is perfectly predicted by the other three “rank” items causing a multicollinearity problem for regression, and because it showed the least variability (nearly all respondents rated it as the least important motivation). Data for these variables were drawn from the first (demographics, duration of residence in unit, typical thermostat settings, hours per day unit is empty) and second (remaining predictors) surveys completed by participants during in-home visits. Thus, it is important to remember that the “predictors” were actually measured after the periods of energy use that they are being used to predict when interpreting the results – causal inferences are not easily justified.

Source: ASU Global Institute of Sustainability

The only significant predictor of energy use was number of electronic devices in the room (β = .249, p = .046). However, higher scores on the Responsibility Motives index were associated with lower electricity consumption at the marginal level of significance (β = -.364, p = .084), suggesting that respondents’ pro-conservation motives did indeed carry over into energy-conserving behavior. No other predictors in the

The regression model predicting Cold-month electricity usage was not significant, F(12, 22) = 1.73, p = .128), and no individual predictors emerged as significant, although the effect of difficulty paying the electrical bill was marginally significant (β = 0.38, p = .051). Not surprisingly, respondents from households that used more electricity tended to report greater difficulty paying the bill.

model were significant.

Sidney P. Osborn Most participant surveys were completed by the female head of household. Analyses were performed using only the 45 households that were reachable for the second in-home visit, during which the second survey was administered.

The regression model predicting Neutral-month electricity usage was significant, F(12, 22) = 2.33, p = .041. Typical hot-day thermostat setting emerged as a significant predictor of energy usage (β = -0.38, p = .030), with those setting the thermostat to a higher temperature using less electricity, even though these were moderate-temperature months by Arizona standards. In addition, households in which the respondent was older used significantly less electricity (β = -0.43, p = .022), and those that had lived in the unit for a longer period of time used significantly more electricity (β = 0.36, p = .035). No other predictors were significantly associated with Neutral-month energy use.

Simultaneous regressions were used to examine the predictors of baseline (i.e., pre-intervention) electricity use. Because we anticipated that, in Phoenix, the predictors of energy use may differ in important ways between the hot summer months and more temperate months, we ran three separate analyses: one predicting mean monthly electricity usage during January, February, November, and December of 2011 (“Cold” months), one predicting mean monthly electricity usage during March, April, May, and October of 2011 (“Neutral” months), and one predicting mean monthly usage during June-September of 2011 (“Hot” months).

Energy Efficiency on an Urban Scale

48

Global Institute of Sustainability

Back to Main Table of Contents

The regression model predicting Hot-month electricity usage was also significant, F(12, 22) = 2.44, p = .033. Not surprisingly, typical hot-day thermostat setting emerged as a significant predictor of energy usage (β = -0.42, p = .016), with those setting the thermostat to a higher temperature using less electricity. Number of hours per day in which the unit was typically empty also predicted usage (β = -0.36, p = .046), suggesting that participants may have remembered to turn the air conditioning off or set it to a higher temperature when they were outside the home. Finally, the rank orders of saving money (β = -1.09, p = .042) and of preserving nature (β = -0.82, p = .079) as motivations for conserving energy were significant predictors of energy use. Remarkably, in both cases, respondents who rated these motivations as more important (i.e., rank closer to 1) tended to come from households that used more electricity, resulting in a negative regression weight.

was not significant, nor did TED Display Presence interact significantly with any other predictors in the model. However, the interaction of Display Presence with Sex approached significance, F(1, 102) = 2.64, p = .108. Specifically, men who had a TED Display in their rooms appeared to use somewhat less energy then men without TED Displays, whereas this was not the case for women.

FIGURE 11: ENERGY USAGE BY TED DISPLAY PRESENCE AND RESPONDENT SEX

D.3: DOES HAVING A REAL-TIME ENERGY USE FEEDBACK “DASHBOARD” HELP USERS SAVE ENERGY? DOES THIS EFFECT DEPEND UPON USER DEMOGRAPHICS?

Source: ASU Global Institute of Sustainability

Taylor Place In order to assess whether the effect of the TED displays on actual electricity usage might be detected after accounting for other resident characteristics, we repeated the above analysis entering Hours per Day the room was occupied, the ProConservation Attitudes index, the Responsibility Motivation index, and the Social Motivation index as covariates.

Importantly, this analysis was restricted in statistical power as it was limited to the 68 rooms for which both survey and energy use data were available. In this analysis, the main effect of having a TED Display on energy usage was still not significant, nor was this effect significantly moderated by any other variable. However, Responsibility Motivation again emerged as a significant predictor in this analysis, with higher Responsibility Motivation associated with lower energy usage, F(1, 56) = 4.38, p = .041, partial eta-squared = .072.

No other predictors, or interactions among factors, were significant.

The impact of the TED feedback dashboards was assessed in terms of two outcome variables – one proximal and one ultimate. The proximal outcome was an index formed by averaging the five survey items assessing talking with others about energy usage, and thinking about energy use when actually using electrical devices (Cronbach’s alpha = .82) – an index of the practical salience of energy use during the course of the study. The second outcome was the three-month index of actual electricity use. In each case, analyses were three-way Analyses of Variance (ANOVAs) with room as the unit of analysis and TED Display Presence, Community Floor Status, and Sex entered as predictors.

The analysis predicting Energy Salience was limited to 79 rooms for which the necessary survey data were available. TED Display Presence did not significantly predict Energy Salience, nor was the effect of TED Display Presence significantly moderated by any other variable. In this analysis, resident Sex was the only variable to have a significant effect, F(1, 71) = 5.26, p = .025, partial eta squared - .069, with women (M = 2.62) reporting higher salience than men (M = 1.86).

Sidney P. Osborn The impact of the TED devices on electricity usage was examined using mixed-model Analyses of Variance (ANOVAs) in which Time was treated as a within-subjects variable and Condition (EdOnly vs. EdTED vs. EdTEDBud) as a between-subjects variable, with respondent Sex, Age, and duration of residence in unit entered as covariates. Two sets of analyses were conducted – one with average monthly

The analysis predicting actual electricity usage included 110 rooms for which the necessary data were available (including resident sex). The predictors entered into the ANOVA model were TED Display Presence, Community Floor Status, and Sex. The main effect of having a TED display on electricity usage Energy Efficiency on an Urban Scale

49

Global Institute of Sustainability

Back to Main Table of Contents

electricity usage during August and September (“Cooling Months”) as the dependent variable, and one with average usage during November and December (“Heating Months”) as the dependent variable, in case the effects of having a dashboard differed across these seasons. In each case, the Time variable contrasted the relevant period in 2011, before the intervention, with the same period in 2012, when the intervention was underway.

dramatically. This discrepancy is all the more surprising as the units do not have electric heating, although residents may own portable space heaters. We are hesitant to over-interpret this effect in the absence of further information. However, one reasonable possibility is that feedback on energy use has less impact during the intense heat of Phoenix summers, when air conditioning accounts for the bulk of energy consumption but is necessary for physical comfort. In contrast, energy use feedback may have greater impact on use of other kinds of electrical devices, which are seen as luxuries rather than necessities and which account for a greater proportion of variability in energy use during seasons with more moderate temperatures.

FIGURE 12: MONTHLY ELECTRICITY USE DURING HEATING MONTHS, ESTIMATED MARGINAL MEANS BY CONDITION

D.4: WHAT TYPE OF FEEDBACK MESSAGES INCREASES ENERGY CONSERVATION (ENVIRONMENTAL VS. FINANCIAL)? IS THIS CORRELATED WITH USER DEMOGRAPHICS? Sidney P. Osborn TED feedback devices can be set to provide feedback in a number of formats, and study participants were encouraged to try different formats and select the one they found most helpful. At the end of the study, study staff checked which setting each device had been on and recorded it for analysis.

Source: ASU Global Institute of Sustainability Note: The significant Time x Condition interaction indicates that the effect of time on electricity use (i.e., pre-intervention vs. during the intervention period) differed significantly across the three experimental conditions.

FIGURE 13: TED FEEDBACK DISPLAY SETTING OPTIONS

For the “Cooling Months” of August and September, the TED devices did not appear to have an effect on electricity usage; the Time x Condition interaction was non-significant and extremely small, F(2, 38) = 0.19, p = .826. However, the TED feedback devices did appear to have an effect on electricity usage during the “Heating Months” of November and December. During this period the Time x Condition interaction was significant, F(2, 37) = 3.42, p = .043, indicating differences among the experimental conditions in pre-post intervention change in electricity use. Although the pairwise pre-to-during-intervention contrasts were not significant for any single condition, likely due to the very small cell sizes, examination of the estimated marginal means indicates that electricity usage in the EdOnly condition increased during this period from 2011 to 2012, whereas usage in the EdTED and EdTEDBud conditions decreased from 2011 to 2012.

Source: ASU Global Institute of Sustainability

The most commonly used setting was “Real-Time,” which shows dollars per hour being used as well as current watts in use. Alternative, but less frequently used, settings aggregate various indices of energy usage (dollars, kWh, or CO2 output) over a longer period of time, or give feedback on voltage rather than kilowatt-hours. Because more immediate, precise feedback should have a greater impact on behavior change than more aggregate feedback, we wished to compare the effects of TED devices set to Real-Time display versus other

It is surprising that an effect of the TED devices was observed during the “Heating” months rather than the “Cooling” months, when air conditioning is necessary for physical comfort in Phoenix and electricity use typically rises Energy Efficiency on an Urban Scale

possible settings.

In order to address this question, we used two mixedmodel ANOVAs similar to those used in D.3 above, this time 50

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 14: ELECTRICITY USE DURING HEATING MONTHS, ESTIMATED MARGINAL MEANS BY DISPLAY SETTING

combining the two TED Conditions but conducting separate analyses for households whose TEDs were set to Real-Time feedback, versus households whose TEDs were set to another feedback setting. In each case, the TED condition was contrasted with the EdOnly control condition. Because effects of the TED devices were only observed during the winter Heating months, analyses were limited to those months.

The effect of TED devices set to give Real-Time feedback was pronounced; the Time x Condition interaction for this test was highly significant, F(1, 27) = 11.50, p = .002. Estimated marginal means for electricity use among the EdOnly control condition rose from 309.85 kWh in 2011 to 341.66 in 2012; means among TED users who had set their devices to give Real-Time feedback dropped from 443.22 in 2011 to 367.54 in 2012. In contrast, TED devices set to other feedback settings had no significant effect, F(1, 19) = 0.21, p = .652 for the Time x Condition interaction. In this case, estimated marginal means for electricity usage increased from pre-test to the intervention period in both the Education Only and the TED conditions. These analyses suggest that the TED devices were only effective at promoting electricity conservation among our participants when set to display Real-Time energy usage.

Source: ASU Global Institute of Sustainability The significant Time x Condition interaction indicates that the effect of Time (i.e., pre-intervention vs. during the intervention period) on Heating Month electricity usage differed significantly between the Education Only households and the subset of TED Display households who had the display set to the “real-time” setting. This interaction was not significant when comparing the Education Only households to households whose TED devices were set to other feedback settings.

Unfortunately, security concerns prevented supervision of the community-focused aspect of the intervention by the Behavior team, and there is considerable evidence that this aspect of the intervention was not implemented as designed. At a midsemester check-in, one Community Floor resident assistant indicated that he was not holding floor meetings at all, but simply emailing important information to residents or speaking to them individually and informally; the other resident assistant indicated that she had mentioned the energy conservation issues at the first floor meeting, but forgotten to do so in subsequent meetings. It quickly became clear that the resident assistants were not motivated to carry out this aspect of the intervention.

D.5: TO WHAT EXTENT DOES SPECIFIC TRAINING/INFO ON HOUSEHOLD ENERGY USE AND CONSERVATION AFFECT ENERGY SAVINGS? Taylor Place The Taylor Place Dashboard study included two floors assigned to receive a community-focused intervention, aimed at increasing discussion about energy conservation among residents. The resident assistants on this floor agreed to facilitate these discussions as part of their regular floor meetings with residents, and received a document outlining specific content for the discussions: the energy conservation tips emailed by the study team; experience with and observations about the TED devices; and ways in which residents attempted to reduce their energy usage. The intent of this manipulation was twofold: (1) to establish a social norm around energy conservation, and (2) to build a sense of community in the endeavor to conserve energy. Both social norms and community-based social movements have been the target of successful conservation-focused interventions in previous research (see Appendix XX for more detail).

This aspect of the intervention sought to implement these strategies, to see whether they had an independent effect on energy use and/or enhanced the impact of the TED displays.

Energy Efficiency on an Urban Scale

To help compensate for these problems, members of the behavior team attended floor meetings on the two Community Floors in November 2012, which had been convened specifically for that purpose. During the meetings team staff covered the content initially planned for this aspect of the intervention. However, it was likely too late in the semester to achieve the intended goals of the floor meetings, i.e. establishing norms and developing a sense of community around the issue. As noted above (see D.3), analyses did not uncover evidence that the “Community Floor” intervention had any effect on energy savings. The main effect of Community Floor Status was not significant, in any analysis, in predicting either Energy Salience or Electricity Usage.

51

Global Institute of Sustainability

Back to Main Table of Contents

Sidney P. Osborn

pay any electricity bill at all. The intent of this condition was to provide a specific goal for electricity usage, potentially enhancing the effect of the real-time feedback provided by the TED display. However, analyses described above (see section D3 above, Figure XX) did not find that the impact of the TED device differed between the EdTED and EdTEDBud conditions. Thus, adding budgeting information (EdTedBud) did not appear to enhance the energy-saving effects of the TED devices (EdTed).

A personalized, in-home, energy conservation training was provided to all households participating in the Sidney P. Osborn study, and thus served as a constant. In the EdOnly condition, energy use increased somewhat from pre- to postintervention. However, this increase is comparable to the mean change for the corresponding period of time among households in the same complex that did not participate in the study at all (74 households), and thus did not receive the brief education intervention. Thus, this study did not find evidence that the in-home energy conservation instructional visits had any impact on energy usage.

SUMMARY Person Level-Predictors of Electricity Use: Two key findings emerged from the analyses predicting baseline electricity usage. First, human behavior matters. In the Taylor Place study, number of electrical devices used in the room was a strong predictor of electricity consumption. While this may seem obvious, it is important to remember that the decision to own an electricity-consuming device is, in fact, a choice, and potentially an important target of intervention.

In the Sidney P. Osborn study, thermostat setting was an important predictor of energy use, at least during the seasons of the year when air conditioning is necessary in the Phoenix area. Again, this behavioral effect offers a clear target for intervention, and is already the focus of many attempts.

FIGURE 15: MEAN ELECTRICITY USE, EDONLY HOUSEHOLDS VS. NON-PARTICIPANTS

Source: ASU Global Institute of Sustainability

Second, as with the Residential and Commercial studies, motivations for energy conservation proved much more important in predicting electricity consumption than proenvironment attitudes and beliefs. In the Sidney P. Osborn sample, higher rank order of saving money as a motive for conserving was associated with higher baseline electricity use, consistent with the findings of the Residential study discussed earlier. A somewhat different pattern emerged with the college students in the Taylor Place residence hall, however. For these students, higher “responsibility motives” were associated with lower electricity use. This motivation index included the motivation to save money, but also more explicitly pro-environment motivations; these two categories of motivation covaried more closely in the college student sample than in homeowner adults. One speculative possibility, important to explore in future research, is that young adults do not yet see the responsibility to be careful with money and the responsibility to take care of the environment as distinct and unrelated. This global “do the right thing” motivation may be a more potent driver of moderate energy consumption than the explicitly financial motivation found to be problematic in our other samples.

D.6: DO SPECIFIC ADDITIONAL FEEDBACK/INFORMATION MESSAGES AND THE PRESENCE/ABSENCE OF A REAL-TIME ENERGY USE FEEDBACK DASHBOARD INTERACT? Taylor Place The design of the intervention in the Taylor Place study systematically crossed the TED Display variable with the “Community Floor” variable, facilitating analyses asking whether the community floor intervention enhanced the impact of having a TED display (i.e., a Display Condition x Community Floor interaction). As noted above (see D.3), analyses did not uncover any evidence that the “Community Floor” intervention interacted with TED Dashboard Status in this way, in predicting either Energy Salience or three-month Energy Usage.

Sidney P. Osborn Households in the “EdTEDBud” condition of this study received tailored electricity budgeting information in addition to their TED display. This budget indicated how many kilowatt-hours participants could consume per month and stay within the electricity allotment for their unit, thereby avoiding having to Energy Efficiency on an Urban Scale

52

Global Institute of Sustainability

Back to Main Table of Contents

Effects of Real-Time Feedback Devices: Unfortunately, these analyses offer limited support for the effectiveness of the TED devices in reducing electricity consumption among residents of multi-unit housing complexes. The two kinds of complexes included in the Dashboard study present different challenges and motivational structures than typical, singlefamily homes. In Taylor Place, students do not have to pay an electricity bill that is tied to their use, so the feedback system is drastically different. In the publicly subsidized Sidney P. Osborn project, residents are also somewhat buffered from the financial consequences of energy use. Whatever the causes, there was no main effect of TED device presence in Taylor Place. There was a hint that men were more likely to benefit from the devices than women; whether this was due to a ceiling effect for salience of energy usage among the women, or the greater appeal of these technology “toys” to the male students, cannot be determined with the present data. There was an effect of TED device presence in the SPO study, but only during the coolest months of the year. One important finding was that the real-time, immediate feedback settings were more effective at promoting electricity savings than other settings; a pattern quite consistent with earlier research.

Caveat: The primary limitation of these two studies is their very small sample size. Because the SPO study, in particular, had so few households in each condition, null findings may reflect a lack of statistical power rather than the true absence of an effect. Also, it is important to recognize the limitations of the “Community Floor” intervention in the Taylor Place study.

This aspect of the intervention essentially failed to take place, so it cannot be inferred that educational- and communitybased interventions would not have enhanced the impact of TED device presence, or even had independent effects on dorm residents’ energy consumption.

Energy Efficiency on an Urban Scale

53

Global Institute of Sustainability

Back to Main Table of Contents

• Had a much larger percentage of minority-owned businesses (7.7% of participants vs. 2.1% of nonparticipants, see Figure D).

APPENDIX F COMMERCIAL PARTICIPATION FACTORS For designing successful programs for retrofitting commercial buildings to save energy, it is important to understand what kinds of businesses and other organizations choose to participate in these programs. In this section, we compare the commercial establishments that participated in Energize Phoenix (EP) with others in the Corridor that did not retrofit.

This analysis uses the National Establishment Time Series (NETS) Database for 2009, the most recent year data are available to us. An establishment is “an economic unit at a single geographic location, where business is conducted or where services or industrial operations are performed.”

• Had a larger percentage of woman-owned businesses (8.3% of participants vs. 5.0% of non-participants, See Figure D).

In addition, certain economic sectors, or types of businesses, had a higher participation rate than others. The United States, Canada, and Mexico use the North American Industrial Classification System (NAICS) to categorize output and employment according to the nature of the final product or service supplied. Figure E compares the percentage of EP participants with a given NAICS code with the percentage of non-participants. Several blue-collar industries such as manufacturing, wholesale, transport, and warehousing signed up for retrofits at a disproportionately high rate, as did some customer-oriented sectors such as retail, real estate, education services, lodging and food services, and miscellaneous other services (including many non-profit or religious organizations). In contrast, business services and information companies (NAICS 51-56) retrofitted at a disproportionately lower percentage than they make up of the composition of the Corridor as a whole, with the exception of real estate companies (NAICS 53). This low participation rate of many kinds of business services is somewhat surprising given the technical expertise available in these kinds of information, scientific, and management companies. It is possible that some of these businesses may have retrofitted previously, did not own or were not allowed to retrofit their facilities, or may never have heard about the program.

Using GIS, we identified 21,745 establishments within the Energize Phoenix corridor. We then classified each establishment as an EP Participant (=1) or not (=0). Unique establishments receiving multiple retrofits were considered as a single participant. Companies, organizations, or government agencies, however, may consist of multiple establishments, and if a company/agency received retrofits at different establishments, each establishment was considered a separate EP participant.

Data on EP commercial participants were supplemented with information from the NETS database. We were able to successfully match 325 unique EP commercial participants to their entry in the NETS Database, and added 50 additional EP participants from our own commercial survey that were not in the NETS Database. The following analysis, therefore, studies 21,795 establishments, of which 375, or 1.72%, got at least one retrofit. While the sample is quite large and representative of the Corridor, the analysis is limited to the business characteristics reported in the NETS Database.

FIGURE A

In contrast to the entire population of commercial establishments in the Corridor that did not get an Energize Phoenix retrofit, the sample of EP participants that chose to retrofit: • Had more employees (Figure A).

• Had higher percentages of non-profit organizations and corporations, and lower percentages of sole proprietorships and partnerships (Figure B).

• Had higher percentages of headquarters and branch locations, and lower percentage of stand-alone establishments that were the only place of business (Figure C).

Energy Efficiency on an Urban Scale

54

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE B

FIGURE E

FIGURE C

LOGIT ANALYSIS: A MULTIVARIATE ANALYSIS OF THE RETROFIT DECISION The bar charts in Figures A-E are useful for descriptive purposes to paint a profile of the commercial establishments that did and did not participate in Energize Phoenix. Their limitation, however, is that they show one dimension at a time, and donâ&#x20AC;&#x2122;t take the other factors into account. For instance, woman-owned establishments participated at a higher rate than non-women-owned businesses. There are several potential explanations for this higher participation rate besides female ownership, which only a multivariate analysis can assess. This result may be an artifact of the industries in which women have higher ownership rates. For instance, in the Energize Phoenix corridor, women own disproportionately more businesses in manufacturing, retail, and food and lodging, all of which had a higher propensity to get retrofits, and fewer in finance/insurance and real estate. The participation rate may be a factor of legal organization. Women-owned businesses in the Corridor are more often organized legally as sole-proprietorships and less often as corporations or nonprofits, and more often minority-owned, as compared with all commercial establishments in the Corridor. These same kinds of alternative explanations also apply to the other variables when viewed in isolation.

FIGURE D

To disentangle these factors and determine which variables are most strongly associated with deciding to upgrade their facilities, we conducted a multivariate statistical analysis Energy Efficiency on an Urban Scale

55

Global Institute of Sustainability

Back to Main Table of Contents

TABLE 1. LOGIT REGRESSION COEFFICIENTS RESULTS

using logistic regression, or “logit,” analysis which is designed to help explain discrete, yes-or-no (0 or 1) choices. Here the dependent variable is whether or not an establishment in the Corridor got at least one retrofit (1) or not (0) under the EP program. The variables in Figure A-E are the independent variables in the logit analysis for explaining and predicting participation. The advantage of a logit analysis is its ability to estimate the partial effects of each variable on the retrofit decision while statistically “holding the other variables constant.” In effect, the logit analysis allows us to assess each factor’s effect on the retrofit decision assuming all other factors equal across the entire sample of businesses in the Corridor.

After eliminating establishments with missing data or that were in categories where 0% or 100% of members retrofitted, there remained 4,185 establishments in the Corridor, of which 225 or 5.38% pursued energy retrofit projects. In terms of goodness of fit, the resulting logit model explained only a small portion of the retrofit decision process: the Cox & Snell and Nagelkerke pseudo-R2 values were .041 and .120 respectively (with 1.0 being perfect). This is not unusual for discrete choice models, which rarely account for even half of the variability in choice, especially when dealing with a lowprobability occurrence, and the model is able to successfully predict 94.6% of retrofit decisions.

1

The -2 Log Likelihood value was 1577.13.

Non-significant variables in model: Woman-owned, NAICS_3, NAICS_42, NAICS_48_49, NAICS_61, NAICS_71, NAICS_81, HQ, Branch, Partnership.

• Wholesaling businesses (NAICS 42) were 2.4 times as likely to retrofit

Most importantly, the model found 14 variables that were significantly associated with the relatively rare decision to participate in EP retrofit program and 8 variables that were not related, all else being equal, using the widely used scientific standard of having less than a 5% chance of happening by chance (Table 1). The number of employees was significant, with each employee increasing the probability of retrofitting by .001%. Minority ownership was significant and doubles the probability or retrofitting. However womanownership did not significantly increase the probability of retrofitting with more than 95% confidence. Organization as a corporation was significant and increased the likelihood of retrofitting by 2.1 times, while non-profit status did so by 3.5 times, relative to a sole proprietorship.

• Retail (NAICS 44-45) and manufacturing (NAICS 31-33) were about 1.9 times as likely to retrofit • Real estate (NAICS 53) and lodging/food services (NAICS 72) were about 1.6 times as likely to retrofit Other economic sectors were less likely to participate, all else being equal, relative to construction: • Information businesses (NAICS 51) were 70% less likely to participate (or .304 times as likely) • Business services in NAICS codes 52 (finance/insurance) and 54 (professional, scientific, and technical services) were 80% less likely to participate.

• Administrative/support/waste management/remediation services (NAICS 56) and health care (NAICS 62) were about 55-60% less likely to participate.

Equally interesting were the variables that were not statistically associated with a higher likelihood to retrofit.

Energy Efficiency on an Urban Scale

56

Global Institute of Sustainability

Back to Main Table of Contents

None of the geographic/hierarchical types of establishment were significantly related to retrofitting; headquarters and branch locations were not significantly different than establishments that were its only place of business. As such, the decision to retrofit a facility appears to be independent of whether the business is stand-alone or not.

energy retrofit projects. There may be a large number of important factors that were left out of this analysis because the variables were not available in the NETS database, or it may simply be a business decision that is inherently difficult to predict. Nevertheless, because of the large sample size, a number of variables were found to be significantly related to the participation decision. This information can be useful to agencies designing programs and contractors seeking customers in a couple of ways. The variables positively associated with retrofitting show where the programs have been successfully targeted. Energize Phoenix was more successful in reaching and signing up larger establishments, corporations, non-profit organizations, minority-owned businesses, and businesses in manufacturing, retail, real estate, lodging and food services. At the same time, the variables negatively associated with participating highlight types of businesses that are either not being successfully reached by the programâ&#x20AC;&#x2122;s marketing or not being successfully convinced to proceed. Alternatively, they may be less likely to own their own facilities or be allowed to retrofit a leased facility, or may benefit less from retrofitting than other kinds of businesses. Future programs in the Phoenix area need to do a better job of attracting information industries, business services other than real estate, and health care companies, or perhaps the office buildings in which they are located.

Legal organization as a partnership was also not significantly different than a sole proprietorship, implying that the degree of legal separation from individual owner (alone or in partner with other individuals) may be related somehow to the propensity to participate in these kinds of programs.

Corporations and non-profits are legally more separated from individual owners and are significantly more likely to retrofit, relative to sole proprietors and partnerships, for which taxes and other liabilities flow to the individual owners. It is possible that corporate or non-profit ownership provides a greater degree of protection against the risks associated with these kinds of investments, although those risks are small given the amount of incentives offered. Non-profit companies may be more attuned to retrofitting for societal reasons. Corporations may also have more experienced and sophisticated management than partnerships and sole proprietorships.

Finally, several of the other business types were not significantly different than the construction industry, which participated at a slightly lower percentage than the Corridor as a whole. NAICS codes 48-49 (transportation and warehousing), 61 (education services), 71 (arts, entertainment, and recreation), and 81 (other services) were not significantly different than construction or the Corridor in general.

While women-owned businesses were more likely to participate than establishments not owned by women, the difference was below the 95% confidence level scientists generally like to have. Based on this analysis, holding the type of industry, number of employees, minority ownership, and our other variables constant, our confidence in a higher likelihood of retrofitting by women-owned businesses is about 80%. Further analysis is needed to reach a more definitive conclusion. We refer readers to the sections on the contractor surveys and on the commercial surveys for two different perspectives on marketing efforts to and by women and on the attitudes of business decision-makers who completed longer surveys as part of the Energize Phoenix project.

In summary, the results of the logit analysis do not explain very much of the variability in business participation in Energy Efficiency on an Urban Scale

57

Global Institute of Sustainability

Back to Main Table of Contents

INTRODUCTION

APPENDIX G

The following summarizes the execution and some of the results of two Home Energy Information (HEI) or “Dashboard” experiments conducted over the final year of the three-year Energize Phoenix program. Further HEI study background information can be found at http://energize.asu.edu in the Year One Report Appendix I: Research to Inform Design of Residential Energy Use Behavior Change Study and Year Two Report Appendix F: Logistical Anatomy of Executing a Home Energy Information (HEI) Dashboard Field Experiment. A full compendium of project documents, processes and results for the Year Three studies is available by emailing energize@asu. edu. Additional behavioral results of the third year experiments are available in Appendix XX: Behavioral Elements of Energy Use and Participation in Energize Phoenix.

IMPLEMENTING TWO HOME ENERGY INFORMATION (HEI) “DASHBOARD” FIELD EXPERIMENTS TABLE OF CONTENTS INTRODUCTION

58

RECAP OF YEAR TWO

58

YEAR THREE REGROUP: TWO NEW PROJECTS, STUDY RE-DESIGN AND EXECUTION

59

SIDNEY P. OSBORN STUDY

59

SITE DESCRIPTION

59

STUDY DESIGN

59

STUDY TIMELINE

60

STUDY PROCESS

60

RECAP OF YEAR TWO

DATA ANALYSIS

61

PARTICIPANT SURVEY ANALYSIS

61

TAYLOR PLACE RESIDENCE HALL STUDY

62

SITE DESCRIPTION

62

PILOT STUDIES

63

STUDY TIMELINE

63

TED INSTALLATION

64

COLLECTION OF DATA

64

DATA ANALYSIS

64

During the second year of Energize Phoenix, the original single-family and small multi-family rental Home Energy Information (HEI) experiment was halted and new studies were developed. The Dashboard team learned a tremendous amount about working as a team player within a partnership of large and complex entities including a city municipality, large utility and university. The lessons arose out of the many challenges and setbacks encountered with installation of real-time, in-home energy usage feedback devices in mostly older single- and multi-family homes within the Energize Phoenix corridor (see Year Two Report Appendix F: Logistical Anatomy of Executing a Home Energy Information (HEI) Dashboard Field Experiment).

CONCLUSION 65 REFERENCES 65

The Dashboard team was able to quickly regroup midstream and identify multiple other study possibilities. Working with the partners, the situation inspired the team to think outside the box and imagine potential locations where there would be fewer variables and where the issue of placing equipment in the utility side of the electrical panel did not exist. This spurred the exploration of City of Phoenix housing projects, in some of which the individual apartment unit meters are actually sub-meters owned by the City. Another potential site was identified as Taylor Place residence hall, owned by Arizona State University and managed by Capstone On-Campus, with electric sub-panels where the TED electric amp clamps could be installed away from any electric meters.

Energy Efficiency on an Urban Scale

58

Global Institute of Sustainability

Back to Main Table of Contents

Appendix G Table of Contents

YEAR THREE REGROUP: TWO NEW PROJECTS, STUDY REDESIGN AND EXECUTION

FIGURE 2. APARTMENT BLOCKS CONSISTING OF THREE UPPER AND THREE LOWER UNITS, WHICH CAN HAVE BETWEEN TWO AND FIVE BEDROOMS

Sidney P. Osborn Housing Complex As discussed in the Energize Phoenix Year 2 report, the Sidney P. Osborn complex was deemed the city housing complex best suited for the final year’s study.

Taylor Place Residence Hall The Dashboard Team also used the Taylor Place residence hall, located within Arizona State University’s Downtown Phoenix campus as the other site for the final year’s study.

The area of each apartment type, in square feet, is specified in Table 1.

SIDNEY P. OSBORN STUDY Site Description

TABLE 1. APARTMENT TYPE AREA IN SQUARE FEET

The Sidney P. Osborn housing complex was built in 1966 and includes 26 buildings with a total of 145 apartments. Each apartment building is made of un-insulated concrete masonry units (CMU) and single-pane, clear glass windows. All the units are mechanically air-conditioned, not evaporatively cooled, which provides greater potential for electrical energy savings for this study. The units are gas-heated (see Figures 1 & 2).

Included in their rent, Sidney P. Osborn residents receive a free nominal monthly dollar allotment for electricity usage at $0.10/kWh, which varies by month and apartment size (see Tables 2). They are billed for any monthly electricity use that exceeds the specified allotment amount calculated by the City of Phoenix’s Housing Department.

FIGURE 1. SIDNEY P. OSBORN SITE PLAN

TABLE 2. MONTHLY DOLLAR ALLOTMENT FOR 2011 AND 2012

The City of Phoenix’s Housing Department manually reads meters each month for billing purposes. Residents are charged 10 cents for every kWh used above the month’s allotment for their unit size. The residents are not provided their actual energy usage or their allotment amount on their monthly electricity bills. Therefore, most residents do not have a clear understanding of their electric energy usage and billing. Over half the residents exceed their allotments in winter months and over three quarters exceed them during summer months.

STUDY DESIGN The Sidney P. Osborn study aimed to estimate the impact of real-time energy feedback through an in-home energy display, called The Energy Detective (TED), in conjunction with an information and motivational intervention provided to the residents. In addition, it investigated the effectiveness of Energy Efficiency on an Urban Scale

59

Global Institute of Sustainability

Back to Main Table of Contents

Appendix G Table of Contents

FIGURE 3. SIDNEY P. OSBORN STUDY TIMELINE

different conditions of feedback interventions given to the residents. In particular: • Analysis and comparison of the following group conditions to determine if there were any significant differences in energy use: – Education only (Group 1) – Education plus an in-home energy display (Group 2) – Education, the in-home display and an added motivational intervention consisting of budgeting information (Group 3) – Control group which did not receive any information or interventions.

• Analysis of the effect of orientation and position (ground level vs. upper level) of the apartment on energy reduction/ increase due to feedback interventions.

STUDY PROCESS Prior to the recruitment phase, the Dashboard team worked closely with members of the Behavioral team to develop the project experimental design, during which all participant-related materials were prepared in both English and Spanish, and approvals obtained from the Institutional Review Board (IRB). Materials included permission forms, interview questionnaires, informational flyers, educational scripts, TED operating instructions, appliance inventory forms, as well as energy budget information. Actual recruitment consisted of two evening pizza parties where the project was presented to potential participants. This was followed with several days of door-to-door follow-up and further recruitment. A total of 82 potential participants were recruited and were then randomly assigned to groups.

TED displays were then installed by a licensed electrician in apartments assigned to either Group 2 & 3. Then, all participating households were given targeted education.

Depending on the group assignment, this may have involved additional surveying, energy saving tips, TED operating instructions and energy budget information. During the data collection phase, the Dashboard team made periodic visits to participating households, which typically involved checking on project participation and TED operating status.

Finally, data analysis involved comparing actual monthly energy usage received from the city with the data collected by the TEDs. The data was regressed using average monthly temperatures for the corresponding monthly energy consumption. This regression was performed separately for each apartment using a Three-Parameter Model Regression tool.

Because of the limited information provided in monthly billing, the study also aimed to: • Foster awareness among participating residents of their own patterns of residential electricity consumption and understanding of energy use related savings.

• Analyze the participant surveys collected during the study to determine residents’ understanding of their energy consumption and monthly billing.

• Analyze the participant survey results with their energy reduction or increase to determine a trend or identify an explanation for particular findings in the results.

STUDY TIMELINE The study’s entire process, from recruitment through device de-installation, ran from May 7, 2012 through February 15, 2013. TED device installation was still underway during June and July 2012, though most of the apartments had received their device by the beginning of July. For analysis purposes, the actual study months were considered to be from July 2012 until the end of December 2012, i.e., six months (see Figure 3).

Energy Efficiency on an Urban Scale

60

Global Institute of Sustainability

Back to Main Table of Contents

Appendix G Table of Contents

DATA ANALYSIS

FIGURE 5. PERCENTAGE REDUCTION (INCREASE) IN ENERGY USAGE (STUDY VERSUS BASELINE)

Energy consumption and its corresponding billed usage from the pre-study period of July through December, 2011 was weather-corrected to 2012 weather conditions and compared to the measured energy consumption for the same months in 2012, which constituted the study period.

Energy savings comparison. Figure 4 shows the predicted versus measured energy consumption during the study period.

The energy consumption from the months of July to December for both predicted and measured usage of all units within each group is summed up to calculate the baseline and study period energy consumption, respectively, for that group. Figure 4 illustrates that Group 1 (consisting of 9 households) had an increase in consumption of 3009 kWh in total during the study period. Group 2 (consisting of 15 households) had a savings of 2667 kWh. Group 3 (consisting of 10 households) had a savings of 524 kWh. The Control Group (consisting of 49 households) experienced an increase of 4634 kWh.

PARTICIPANT SURVEY ANALYSIS General post-study participant understanding of energy According to the participant survey taken at the end of the project, almost all of the 34 participants reported at poststudy that they better understood the relationship between their electrical devices’ energy usage and how much money they could save by practicing what was suggested to them during education sessions. About 76% of the participants felt that they had benefitted from the program, whereas 24% of the participants felt they had little benefit.

FIGURE 4. BASELINE VERSUS STUDY ENERGY USE – COMPARISON BETWEEN GROUPS

Participant interaction with the TED device The following survey topics were asked only of the TED group participants, which consisted of Groups 2 & 3: Understanding information on display screen. 36% of the participants mentioned that they understood the device very well, whereas 52% of the participants mentioned an acceptable understanding.

The corresponding percentage changes in usage are shown in Figure 5. Group 3 experienced an insignificant energy reduction of 0.9% during the study period and its savings percentage was less than Group 2’s 3.3%. Group 1, the ‘education only’ group, incurred a higher increase of 6.7% than the control group’s 1.7% increase.

Display setting preferred. 24% of the participants preferred the real-time use setting on the display. 40% of the participants preferred the ‘recent usage’ setting on the display. 28% of the participants preferred the ‘month-to-date’ setting.

Frequency of viewing displayed information. Participants were asked how frequently the observed the display during the last week of the experiment. 32% of the participants mentioned they did not look at the device at all, 32% mentioned that they looked at it 1-3 times in the week, 4% mentioned that they looked at the device 4-6 times in the week, 16% looked at it daily and only one participant looked at the device several times a day.

Comparing device interaction survey results with energy savings. Out of the 36% of participants who mentioned

Energy Efficiency on an Urban Scale

61

Global Institute of Sustainability

Back to Main Table of Contents

Appendix G Table of Contents

TABLE 3. PARTICIPANT SURVEY- ENERGY SAVINGS OR LOSS VERSUS SAVINGS STRATEGIES MENTIONED DURING SURVEY

that they understood the device very well, 75% of them had savings in their energy usage. Out of the 52% of participants who mentioned that they interacted with the TED through the last week of the study, 66% had savings.

Understanding of and using budgeting information. This part of the survey was asked of Group 3 participants only. Out of the 10 participants who received the budgeting information, 6 compared this information with the screen during the last month of the study. Out of the 10 participants, only 5 knew which screen on the display was to be compared with the budgeting information. The overall understanding of the budgeting information was difficult for the residents to understand. Though the participants were re-educated regarding this information during the follow-up session in September, more than half of the participants did not fully understand how to use it.

Energy saving strategies mentioned by participants. All the participants were asked about their general understanding of what would help them save energy in their household.

Table 3 gives a summary of the strategies mentioned by each participant in Groups 2 & 3. The majority of the participants mentioned disconnecting appliance and electrical devices when not in use, especially to avoid ‘phantom’ loads. The majority also mentioned turning off lights and the television in order to avoid wasting energy. This is consistent with the Behavioral team findings that indicate participants with displays were able to reduce their non-HVAC loads (see Appendix XX: Behavioral Elements of Energy Use and Participation in Energize Phoenix) The next most mentioned strategy was to keep the air conditioner on auto mode or to change thermostat setting to avoid excess usage when not needed.

TAYLOR PLACE RESIDENCE HALL STUDY Site Description The Dashboard team also utilized Taylor Place residence hall, located within Arizona State University’s Downtown Phoenix campus as a site for third year experimentation.

FIGURE 6. TAYLOR PLACE RESIDENCE HALL, WEST VIEW (SOURCE: ASU-TAYLORPLACE.COM/PHOTOS)

The building is a two-tower, 352,000 square foot, mixed-use facility with amenities for student life on campus, including retail areas, a fitness center, a dining hall, and common areas at each level of the towers. Between the two towers, Taylor Place holds 1,200 beds. The north tower’s typical floor layout has 22 dorm rooms per floor that each include a shared living/ kitchen area with two private sleeping rooms. The south tower’s typical floor layout has 32 dorm rooms that include shared living, kitchen, and sleeping. These dorm rooms can be single or double occupancy.

Energy Efficiency on an Urban Scale

62

Global Institute of Sustainability

Back to Main Table of Contents

Appendix G Table of Contents

Pilot Studies

• The installation of the TED devices for the pilot began on June 13, 2012.

During the winter & spring of 2012, two different pilot studies were run in parallel to explore the potential for energy savings enabled by the ThinkEco modlet (which is a receptacle level monitoring and switching device) as well as The Energy Detective (TED) feedback devices. The primarily goal was to test these two different types of real-time energy-use devices while also troubleshooting possible future issues during the main period of the study to follow in fall 2012 at this same facility.

• The MTU devices were installed for each of the 22 rooms of the floor.

• Gateways and Displays were installed in 19 participant rooms, since there were 2 vacant units and 1 resident Opt-out. Of the 19 rooms, 2 become vacant later in the pilot study.

Findings from TED pilot study

ThinkEco pilot study

• Testing was done by turning on and off lights and appliances plugged into the outlets to determine the speed of transmission. The readings on the display were not updating close enough to real-time after installation.

This pilot was divided into two phases: • The first installation took place February 2012 in one vacant room as a preliminary test prior to the summer pilot since this technology was new. Only 4 modlets were plugged into one vacant room and appliances were installed to test communication and oversee possible interaction issues.

• The data was not being transmitted accurately to the gateway. Vast amounts of data were dropped during transmission.

Outcomes

• The second installation was completed in June 2012, to include the testing of a gateway device to facilitate the scalable collection of all the data from the modlets from one floor.

• Due to the slow speed of data transmission and the amount of dropped data, the team determined that in-line filters would be required for the fall semester study installation.

Findings from ThinkEco pilot study

• TED devices would be able to provide lights and plug-load energy usage per room and per floor (HVAC thermal energy was provided by a downtown district cooling loop and individual dorm HVAC fans would not be measured due to limited equipment quantities.).

• The installation of the modlets in one room prior to installation of the gateway was successful. Disaggregated data from each of the outlets could be obtained and seen live using the interactive screen in a computer.

• The installation of the gateways in the residence hall was more challenging. While coordinating the installation and connection of the gateway to the network service in the building with the University’s IT Department, the server and the security teams indicated that, due to potential security protocols and issues like the potential for bringing down the wireless network and/or creation of problems with students trying to connect to the wireless network, the connection of the gateway to the network system would not be feasible. Therefore, it was determined that the ThinkEco device would not be suitable for use in the fall project.

• Taylor Place administration requested that the team limit all interaction with the students during the semester, except for a welcome session and emails during the semester.

STUDY TIMELINE The study’s entire process, from pilot study through device deinstallation, took place from February 2012 through December 21, 2013. For analysis, the actual study months were considered to be from mid-August 2012 until mid-December 2012, i.e., four months (see Figure 7).

TED pilot study • Per Taylor Place’s Director of Operation’s advice, the installation of energy-use devices for the June pilot was undertaken on the 3rd floor of the north tower of the complex. The fall full study took place on the South tower.

Energy Efficiency on an Urban Scale

63

Global Institute of Sustainability

Back to Main Table of Contents

Appendix G Table of Contents

FIGURE 7. TAYLOR PLACE STUDY TIMELINE

COLLECTION OF DATA From the 128 devices installed, the data received and stored in each of the gateways during the period of study was extracted and complied in one data file. However, during this process some data was found incorrect, incomplete or null from the gateways. From the 128 devices installed, 107 presented useful data and 21 data sets were eliminated (see Table 4).

TABLE 4. NUMBER OF INSTALLED DEVICE VS. USEFUL DATA

TED Installation

DATA ANALYSIS

The installation of the TEDs in the student residences took place from July 26th to August 3rd of 2012 on the 8th, 9th, 11th and 12th floors of the South tower. The number of rooms per floor in this tower is 32. It is important to note that the energy use measured in this study corresponds only to the lighting and plug loads of the student rooms participating in the study. Displays were installed in alternating rooms, based on the studyâ&#x20AC;&#x2122;s design. A one-page information flyer with the different display options and functionality of the display was provided in the participantâ&#x20AC;&#x2122;s rooms. As part of the study, all rooms received a smart power strip to use. See Figure 8 for floor plan, display/no display layout.

1) The total average energy consumption for lighting and plug loads during the 4 months of the period of study in all the floors was 0.07kWh less for the rooms that had a TED display compared to those ones that did not have one in their rooms (see Figure 9).

FIGURE 9. TOTAL AVERAGE ENERGY USE ON FLOORS AMONG ROOMS WITH DISPLAY AND NO DISPLAY

FIGURE 8. DISPLAY (SHADED), NO-DISPLAY LAYOUT IN ROOMS OF SOUTH TOWER FLOOR PLAN

2) The total average energy use for lighting and plug loads during the period of study per floor is summarized in the Figure 10. Floor 8 total average energy use for lighting and plug loads during the four months was the highest among all the floors, in both the display and no display categories.

Floor 9 had the lowest total average energy use for rooms with no display and Floor 12 for rooms with display. Floors 8 and 12 showed energy savings with displays, while Floors 9 and 11 did not.

Energy Efficiency on an Urban Scale

64

Global Institute of Sustainability

Back to Main Table of Contents

Appendix G Table of Contents

FIGURE 10. TOTAL AVERAGE ENERGY USE PER FLOORS AMONG ROOMS WITH DISPLAY AND NO DISPLAY (KWH)

• By signing 117 renters to the first dashboard experiment (while this experiment did not go forward), many landlords (whose permission was required for renter participation) became exposed to the program.

• Coordinating 82 households in the Sidney P. Osborn Housing Complex in the second dashboard experiment involved interacting with residents, spouses, children, extended families and other residents, many of whom received considerable education about energy savings.

A number of documents, such as energy saving tips and appliance inventories, were generated that the City of Phoenix’s Housing Department could use in other complexes.

CONCLUSION

• Many students in the Taylor Place residence hall, away from home for the first time, gained their first knowledge about the energy consumption of various personal household devices.

Both study projects showed only a slight energy savings benefit for low-income housing residents and dorm residents that received real-time feedback devices vs. those that did not. Thus, the hypothesis that real-time feedback devices could achieve meaningful energy savings could not be confirmed in the two settings studied. The challenges of educating two populations in short order that previously did not possess even aggregate historical energy usage information nor paid their own full energy bills proved daunting. Such information and motivations might be considered pre-requisites to affecting behavior change through real-time feedback. Indeed, the original hypothesis involved testing the behavior change potential in single family rental homes, where renters would have historical knowledge of energy usage and financial motivation to save. Several anecdotal comments made by Sidney P. Osborn residents during exit interviews did suggest some subjects were beginning to understand the potential application of real-time feedback devices but they apparently were not significant enough (or in time) to alter the outcome. Several studies of the energy impact of real-time feedback devices published by others during the time period did validate the energy savings potential of these devices.1,2 Thus, it is believed the potential exists for these devices in appropriate settings.

• Several graduate students were trained in the use of real-time feedback devices and project management, and two masters level theses were completed on this subject.

Several of these students have plans to work professionally in this subject area.

• Researchers have plans to write several academic papers on this project, especially on the challenges that field projects like this confront in attempting to operate in a community setting.

References 1. Ehrhardt-Martinez, K., K.A. Donnelly, & J.A. Laitner. 2010. Advanced Metering Initiatives and Residential Feedback Programs: A MetaReview for Household Electricity-Saving Opportunities, American Council for an Energy-Efficient Economy, Washington, D.C. 2. Foster, B., & S. Mazur-Stommen. 2012. Results from Recent Real-Time Feedback Studies, American Council for an EnergyEfficient Economy, Washington, D.C.

While the original and re-structured hypotheses could not be confirmed, a number of positive outcomes did emerge from the HEI projects, including: • Door-to-door dashboard recruitment by students during the summer of 2011 wearing bright green t-shirts branded with the Energize Phoenix logo on the front and the program marketing motto “Save Money by Saving Energy” on the back, which helped to introduce the Energize Phoenix to hundreds of residents in the Corridor.

Energy Efficiency on an Urban Scale

65

Global Institute of Sustainability

Back to Main Table of Contents

APPENDIX D: CASE STUDY: LEVEL 2 ANALYSIS OF GARAGE LIGHTING RETROFIT

85

D1. BACKGROUND

85

D2. OBJECTIVE

85

D3. METHODOLOGY

85

TABLE OF CONTENTS

D4. RESULTS OF SAVINGS MEASUREMENTS

86

ABSTRACT 67

D5. COMPARISON WITH UTILITY BILL ANALYSIS

87

1. INTRODUCTION

67

D6. CONCLUSION

2. OBJECTIVES

67

3. OVERALL APPROACH

68

4. LEVEL 1 ANALYSIS METHODOLOGY

69

4.1 DATA SCREENING AND BINNING

69

4.2 AUTOMATION OF SAVINGS CALCULATION

70

4.3 ANALYSIS RESULTS

71

4.4 POSSIBLE CAUSES FOR DISCREPANCY

74

APPENDIX H ENERGY SAVINGS EVALUATION OF COMMERCIAL UPGRADE MEASURES THROUGH INDIVIDUAL PROJECT ANALYSIS AND UTILITY BILL MODELING

5. LEVEL 2 ANALYSIS

75

6. LEVEL 3 ANALYSIS

77

6.1. BACKGROUND

77

6.2. UPGRADES SUGGESTED AND PREDICTED ENERGY SAVINGS

78

6.3. ANALYSIS AND SAVINGS VERIFICATION

79

6.4. CONCLUSION

79

7. PAYBACK ANALYSIS

80

8. CUMULATIVE SAVINGS OVER TIME

80

This report summarizes the efforts and findings of the commercial team during the three year course of the Energize Phoenix program. We wish to acknowledge useful contributions from all the members of the team, especially, Alex Castelazo and Prof. Patrick Phelan. An undergraduate student, Hara Kumar, was also involved in the last stages of this work, especially related to the creep analysis. The contribution of former graduate students, Shreya Agnihotri and Sadiq Jubran is also acknowledged.

9. SUMMARY AND SUGGESTIONS FOR FUTURE ENERGY CONSERVATION PROGRAMS 81 REFERENCES 81 APPENDIX A: BASELINE MODEL DEVELOPMENT AND UNCERTAINTY

82

APPENDIX B. MODEL IMPROVEMENT AFTER ADJUSTING FOR UTILITY BILL READ DATES 83

B1. PROBLEM STATEMENT

83

B2. COMPARISON OF MODEL GOODNESS-OF-FIT

83

B3. ENERGY SAVINGS COMPARISON

83

B4. CONCLUSION

83

APPENDIX C: ENERGY CREEP ANALYSIS

84

C1. BACKGROUND

84

C2. ISSUE INVESTIGATED

84

C3. CALCULATION METHODOLOGY

84

C4. RESULTS AND CONCLUSION

84

Energy Efficiency on an Urban Scale

87

66

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

ABSTRACT

buildings included residential, multi-family, commercial/ industrial, etc. The program is one of 41 from across the United States which are supported by the U.S Department of Energy’s Better Buildings Neighborhood Program and the American Recovery and Reinvestment Act of 2009 in order to test new models for scaling energy efficiency and to create jobs. Historically, energy efficiency programs have faced a trio of interconnected forces — technical, economic, and socio-behavioral — which continue to hinder mass-market scaling. Inter-disciplinary research within the framework of EP is aimed at understanding and helping resolve these barriers.

Research projects cover numerous facets (such as behavioral and attitudinal differences between participating and nonparticipating homeowners and business owners, contractor marketing methods, the effects of energy feedback devices coupled with other education or budgeting information, spatiotemporal trends in participation rates, econometric modeling of savings, and economic impact analysis), all of which were meant to study the above influences and at assuring that energy efficiency programs realize their full potential.

Energize Phoenix (EP) was a three year energy efficiency program conducted by a joint collaboration between the City of Phoenix, Arizona State University and a large electricity provider. The intent was to improve energy efficiency in residential, multi-family, commercial/industrial buildings located in a portion of the city of Phoenix around the light rail corridor. There are several facets to the EP program with engineering-based verification of the energy savings due to commercial buildings upgrades being one of them, and the focus of this report.

Various issues had to be addressed such as incomplete data, spurious data behavior, multiple upgrade projects in the same facility, weather normalization using well-accepted change point models applied to utility bill data, and baseline model uncertainty. Considering the nature and characteristics of utility bill data, the evaluation methodology initially adopted was labor intensive and involved a manual data screening procedure of projects on an individual basis, and then one-by-one baseline modeling and savings assessment. An automated process was developed in order to reduce the labor required to analyze and update energy savings in hundreds of buildings on a periodic basis. This report presents results of analyzing close to 560 completed upgrade projects with the primary objectives of measuring savings from Energize Phoenix commercial projects and comparing evaluated savings with those predicted by the contractors prior to the upgrades.

Reasons for these differences are discussed, and follow-up investigations into this discrepancy are also documented.

The uncertainty in the savings determined of all the projects has also been computed and reported. This report ends with conclusions and suggestions for further investigation needed to improve the accuracy and reliability of determining energy savings in allied large scale energy efficiency programs. Four appendices describe specific allied investigations undertaken in support of some of the issues identified during the course of this study. A separate technical report assembles, in further detail, the complete analysis work done by the non-residential analysis team in the framework of the EP project.

EP was a contractor-driven program in which participants receive incentives to encourage energy efficiency projects.

Prospective buildings were not pre-selected by ASU, APS or the City of Phoenix; instead, the task of initiating contact and convincing customers to join the EP program rested with the contractor. Customers utilizing Energize Phoenix were able to match the incentive provided by APS through its ongoing Solutions for Business energy efficiency program with an additional incentive from EP. EP restricted incentives such that the sum of the incentives could not exceed 100% of the incremental project cost to the customer. This report is narrowly focused in its scope on research work performed to better understand the energy savings achieved by Energize Phoenix in commercial projects and the accuracy of the contractor predictions of those savings.

2. OBJECTIVES The primary objective of the EP commercial team was to analyze the data and to quantify the energy savings achieved in the commercial buildings which underwent upgrades incentivized by the EP program. These savings were then compared to the savings predicted by the energy contractors during the project sales process. Note that the contractors used either custom audits or prescriptive guidelines that rely on equipment counts (such as lights) to predict savings. In addition, contractors utilizing the Small Business program, a streamlined, direct-install program meant for small businesses and schools, employed standard software and

1. INTRODUCTION Energize Phoenix (EP) was a three year energy efficiency program led by a joint collaboration of three major institutions –the City of Phoenix, Arizona State University (ASU) and Arizona Public Service (APS) – the state’s largest electricity provider. The main goal of the program was to improve the energy efficiency in the buildings located around the Phoenix light rail corridor and to create jobs. The participating Energy Efficiency on an Urban Scale

67

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

other proprietary tools supplied by a third party contractor. The energy savings comparison between predicted and evaluated by the EP team helps to assess the overall effectiveness of the upgrades and the accuracy of the savings projections of the program as a whole. Other allied issues were also investigated; for example, whether certain contractors tended to consistently over-predict savings as compared to others, and the reasons for doing so.

upgrade. Hence, the term “evaluated savings” is used in this report when such an M&V path is followed.. The commercial team’s entire analysis process is described in more detail in Appendix A. Figure 1 depicts these three levels of analysis in a succinct manner. This report presents the results of our Level 1 and Level 2 analyses along with Level 3 results for one of the two buildings studied, while the technical report (Reddy et al., 2013) includes the Level 3 results of the other building.

There were hundreds of commercial buildings which underwent energy upgrades through the 3-yr duration of the program.

These projects varied vastly in characteristics like business type, size, etc. Four different approaches are suggested in standard protocol documents such as IPMVP (2010) and in ASHRAE Guideline 14 (2002). However, the monthly utility bill analysis approach was deemed to be the only realistic method to determine savings in a program this big with limited personnel involved in the measurement and verification (M&V) process. Further, since EP was an ongoing program and contractors often specialized in certain types of retrofits, buildings underwent upgrades on a continuing basis, and so savings calculations had to be redone at frequent intervals.

Utility bill data for all projects (old as well as recent) was provided by the utility on a quarterly basis, and it was logical to recalculate and update the savings for all buildings every three months. This prompted an additional objective, namely to simplify and automate the savings analysis methodology as far as possible so that future energy conservation programs similar to EP could reduce M&V analysis costs.

Because of the error introduced in such a general approach (called Level 1 analysis) which does not involve inspecting the buildings individually, it was decided to conduct a limited number of in-depth analyses in buildings where large discrepancies were found between evaluated and contractorpredicted savings. This approach (referred to as Level 2) was meant to provide some degree of credibility in our speculation as to the observed differences, and allow us to correct the data as appropriate. Due to the large number of projects, our approach was to sample a sub-set of the completed upgrade projects and verify the savings predicted by the contractor through follow-up field visits, installing in-situ equipment and monitoring for a relatively short period of time. The degree of over- or under- prediction of the savings could then be determined more accurately, and the causes for any such discrepancies identified. This would provide useful feedback to APS and to the contractor, and suggest ways by which future upgrade savings estimations can be improved.

FIGURE 1. THE THREE LEVELS OF ANALYSIS APPROACHES UNDERTAKEN IN THE EP PROGRAM

BY THE COMMERCIAL TEAM

3. OVERALL APPROACH Because of time constraints, the baseline electricity consumption prior to the implementation of energy upgrades could not be determined by in-situ measurement. Hence, the whole building analysis approach, which is one of the four general M&V approaches widely followed by the professional M&V community (see for example, ASHRAE Guideline 14, 2002 or IPMVP, 2010) was adopted. The approach involves relying on a whole year of utility bill data prior to the upgrade to establish a baseline model of energy use against monthly mean outdoor temperature. Such monthly utility bill data was made available from the APS customer billing database. The model was then applied to measured outdoor temperature during the post-upgrade period, and the sum of the monthly differences between these model predictions and the actual measured utility bills during the post-upgrade period were taken to be the “evaluated” upgrade energy savings. Note that energy savings cannot be directly measured but are often inferred from measurements performed before and after the Energy Efficiency on an Urban Scale

68

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

FIGURE 2. TIME SERIES PLOT OF MONTHLY UTILITY DATA FOR OVER THREE YEARS BEFORE THE UPGRADE AND ONE YEAR AFTER UPGRADE FOR A SPECIFIC EP BUILDING

Finally, Level 3 involved an in-depth energy analysis of a few selected projects so as to evaluate energy savings and to provide recommendations for additional potential energy conservation measures. Level 3 analysis involved developing a calibrated detailed simulation model of the energy use in the building based on owner-provided architectural drawings, energy audit reports, usage data and project applications. This is consistent with another standard M&V approach described in such documents as ASHRAE Guideline 14 (2002) and IPMVP (2010). Sub-monitoring the energy use and indoor environment was also done to calibrate the model. The primary objective was to determine quantitatively the effect of individual energy efficiency upgrades on overall energy consumption and to identify other possible energy conservation measures (ECMs).

The data visualization step allowed identification of anomalous behavior and grouping of buildings into bins, as illustrated in Figure 3. Bin A consisted of buildings where there were missing or inadequate pre-upgrade data (i.e., less than twelve utility bills). Buildings with abnormal data patterns were placed in Bin B. Three of the common generic cases encountered are illustrated in Figure 3. Some buildings exhibited an increase in energy use after the upgrade, some had abnormal spikes, and others had markedly different seasonal variation patterns. Bin C consisted of buildings which did not have at least six months of post-upgrade data, in which case the calculation of energy savings was deferred until more utility bill data was forthcoming. Finally, those buildings which did not fall in any of the above three bins were placed in Bin D, for which the savings were determined.

Additional manual screening criteria for data quality had to be empirically framed as shown in Table 1. For example, if predicted savings were less than 1% of the energy use, our analysis procedure was deemed to be unsuitable. An example of anomalous behavior which warranted placing a project in Bin B was a case in which the contractor-predicted savings exceeded the total energy use of the building. This screening process made the whole analysis labor intensive, but it needed to be done only once per project.

It is important to note that APS currently utilizes a third party contractor to conduct the measurement and verification of energy savings for each of the APS energy efficiency programs, rather than relying on contractor predictions. The results of that M&V work are integrated into the annual Energy Efficiency report that is submitted each year to the Arizona Corporation Commission (ACC) to document the claimed savings from APS’s energy efficiency programs. The savings estimates made by the third party M&V contractor are based on M&V industry standard practices involving extensive measurement and verification activities employing statistical sampling techniques. Activities include field metering of the specific equipment installed, on-site inspection of the equipment, customer surveys to estimate run times, engineering models to simulate electricity use, and comparison to estimated usage by standard efficiency equipment. The EP commercial team’s analysis was conducted independently of that of the third party contractor at APS’ request.

4. LEVEL 1 ANALYSIS METHODOLOGY 4.1 Data screening and binning The first step involved ascertaining consistency of energy use over the years. This was conveniently done by simply generating time series plots (see Figure 2) of historic utility bills, and looking at them visually. Some of the projects showed considerable variation in usage pattern which made it necessary to manually screen all individual projects for data quality. This also led to the decision to use only one year of data immediately prior to the upgrade as the baseline period since, as is well known, energy use patterns in commercial buildings tend to change over time.

Energy Efficiency on an Urban Scale

FIGURE 3. ILLUSTRATIVE EXAMPLES OF ABNORMAL DATA BEHAVIOR PERTINENT TO BIN A AND BIN B IDENTIFIED DURING THE VISUAL SCREENING PROCESS

69

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

Since there were several buildings which fell into Bin B, phone calls to the facility managers or owners of a subset of these buildings were also undertaken in order to identify possible reasons and to reconcile the odd behavior. If the behavior could be explained convincingly or if appropriate rectifying action could be taken, these buildings were moved to Bin D, otherwise they were moved to Bin A. A possible factor causing some of the anomalous behavior could be attributed to the fact that the team was unable to perform account matching with the master meter of the facility. Such data was not made available due to privacy reasons.

There were several buildings which qualified for EP incentives involving multiple energy upgrades. These were treated as single projects using the simple approach illustrated in Figure 4. The data period in-between the first and the last upgrades was simply excluded from the analysis since, in most cases, these multiple upgrades were done within a few months of each other. All the upgrades were treated as one single upgrade with the post upgrade period assumed to start after the last upgrade was completed. The sum total of all the contractorsâ&#x20AC;&#x2122; savings predictions for the building was taken to be the overall predicted savings.

TABLE 1. DATA SCREENING AND BINNING CRITERIA EMPLOYED FOR SCREENING

As the number of projects increased and since savings had to be recalculated at quarterly intervals as more data was forthcoming, it was critical to automate the process as much as possible. The automation scheme which evolved is shown in Figure 5. Note that there are still two steps which require manual screening.

FIGURE 5. FLOWCHART OF THE AUTOMATED ROUTINE DEVELOPED TO DETERMINE ENERGY SAVINGS FROM NUMEROUS UPGRADED BUILDINGS IN THE FRAMEWORK OF EP PROGRAM

4.2 AUTOMATION OF SAVINGS CALCULATION Savings were evaluated by comparing the energy consumption between corresponding months of pre and post upgrade periods. A large number of projects showed weather dependency where the total energy consumption was influenced by cooling and heating loads of the building. Thus, the influence of the weather had to be taken into account for these projects in order to properly evaluate the upgrade savings. The billing cycle does not usually match with calendar months, and since read dates were available, the utility bills of each project were adjusted to match calendar months as this would simplify the analysis considerably. A detailed evaluation of analysis results with and without such an adjustment is described in Appendix B. FIGURE 4. A HYPOTHETICAL BUILDING SHOWING MULTIPLE-UPGRADE PROJECTS. ENERGY SAVINGS WERE SIMPLY DETERMINED USING PRE-UPGRADE AND POST-UPGRADE PERIODS AS SHOWN

To facilitate the screening in the automation process, a visual template (shown in Figure 6) was developed. This involved generating scatter plots of monthly energy use versus outdoor temperature as well as annual time series plots superimposed on each other for energy use both prior to and after the upgrades.

Energy Efficiency on an Urban Scale

70

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

FIGURE 6. PLOTS FROM THE VISUALIZATION TEMPLATE MEANT AS AN AID TO PERFORM MANUAL SCREENING OF DATA

which the results were analyzed, and are reported here. Over time, as more data is forthcoming, the 171 projects under Bin C will also be moved to Bin D and will become part of the pool of buildings where energy savings will be determined.

Nonetheless, 150 projects which fell into Bins A and B had to be kept aside due to data quality and inadequate data issues, and savings from these projects are not part of the results reported here.

Figure 7 is a pie chart showing the relative distribution (as percentages) of specific ECMs implemented (lighting, controls, pumps/motors, HVAC, windows, food refrigeration and solar water heating). Lighting upgrades were by far the most frequent, accounting for 70% of the total projects, primarily because of the relative low effort required in both estimating and installing the upgrades, as well as the relatively high incentives and the minimal disruption to participantsâ&#x20AC;&#x2122; operations.

FIGURE 7. PIE CHART DEPICTING THE PERCENTAGES OF SPECIFIC ENERGY CONSERVATION MEASURES (ECMS) PERFORMED BY ECM-TYPE FOR THE 557 PROJECTS. LIGHTING ACCOUNTED FOR 70% OF THE TOTAL NUMBER OF ECMS EMPLOYED IN PROJECTS

The methodology for developing the baseline model is described in Appendix A. It is consistent with the modeling procedures advocated in the engineering literature involving identifying the best change point regression model among several different model formulations with outdoor temperature as the independent variable. A FORTRAN program was developed specifically for the purpose of the EP commercial building analysis effort which incorporated the widely used Inverse Modeling Toolkit (IMT) computer code (Kissock, Haberl and Claridge 2002) as a subroutine. The program reads the utility bill data for a specific building along with outdoor temperature, and assigns it to the pertinent bin. If the building falls into Bin D, the program then identifies the best change point model among several possible functional forms, calculates savings for that building, and does this for all the buildings in the database. The total savings are then determined along with the contractor-predicted savings.

Finally, the automated routine generates pertinent summary statistics and graphics of the entire program savings.

TABLE 2. SUMMARY TABLE OF RELEVANT OVERALL STATISTICS (AS OF END OF APRIL 2013).

4.3 ANALYSIS RESULTS As of end of April 2013, 557 non-residential upgrade projects were completed. Pertinent statistics are assembled in Table 2. Of all these projects, only 236 projects fell into Bin D for

Energy Efficiency on an Urban Scale

71

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

TABLE 3. SUMMARY RESULTS OF SAVINGS FRACTIONS COMPUTED FOLLOWING EQUATIONS (1) AND (2)

The savings fraction can be calculated in two ways, which would yield different results because of large differences in the sizes of the upgrade projects. The two definitions are given below, and analysis results following both methods are presented. (a) Aggregated savings fraction determined as the ratio of the aggregated energy savings summed across all projects divided by the aggregated baseline energy use: â&#x2C6;&#x2018;Ni = 1 Annual Savings (kWh)i Aggregated Savings Fraction = N â&#x2C6;&#x2018;i = 1 Annual Baseline (kWh)i Equation 1 where N is the total projects for which the savings are determined. This is tantamount to considering all the upgrades as having happened in one large hypothetical building. This analysis may introduce some error as some of the projects with large energy consumption can bias the analysis results.

The results of the Level 1 analysis are summarized in the following section. (i) There is a major discrepancy between the total savings predicted by the contractors and those determined from weather normalized savings calculations, referred to as evaluated savings (see Figure 9 and Table 3). Both savings calculation methods (namely, the total fractional savings method and average of the individual project method) reveal this discrepancy. While the total fractional method suggests a fractional savings of 5.5% of the baseline energy use, the contractor-predicted savings fraction was 9.8%.

Similarly, the average of individual fractional savings method resulted in evaluated fractional savings of 10.4% while the contractor-predicted savings turned out to be 22.4%. Thus, both methods point to a considerable over-estimation of savings by the energy contractors.

(b) Average of individual fractional savings of individual projects calculated as: Savings (kWh) Average of Individual Annual Baseline (kWh) i Fractional Savings = N Equation 2 In addition, the percentile values of individual fractional savings at 10%, 25%, 50%, 75% and 90% were also calculated since this would yield insights into the distribution of savings across the numerous projects (which is a nongaussian distribution).

FIGURE 9. COMPARISON OF EP EVALUATED AND CONTRACTOR-PREDICTED FRACTIONAL SAVINGS FOR 201 PROJECTS ANALYZED COMPUTED TWO DIFFERENT WAYS (EQS. 1 AND 2)

The aggregated savings fraction for the 236 Bin D projects was 7.2%. The average of fractional savings of individual projects amounted to 10.0%.

A data audit of contractor predictions revealed that 201 of the 236 analyzed projects had usable data regarding final contractor savings predictions. Using the 201 projects, researchers re-calculated total and average evaluated energy savings and contractor-predicted energy savings for those projects. The results are summarized in Table 3. It should be noted that many of the 35 projects dropped were multipleupgrade projects, some were large projects and all utilized the prescriptive application of the Business program.

Energy Efficiency on an Urban Scale

72

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

(ii) The comparison of contractor-predicted savings with aggregated evaluated savings by both weather corrected and non-weather corrected analysis methods is provided

by Figure 10. Note that the weather corrected and nonweather corrected (direct month to month comparison) methods yielded results which were relatively close. With weather correction, the total fractional savings came out to 5.5 % whereas the non-weather corrected resulted in a 5.7% savings. These values are close, but we still recommend the use of weather correction as a general approach since it would be of general relevance, i.e., in places where outdoor temperature is an influencing factor of energy consumption in a building.

savings uncertainties are relatively small. This is better illustrated by Fig. 12 which is a sorted distribution of 201 projects. Only 2 projects have uncertainties greater than 200 % (the y-axis scale has been cut off at 40% to provide greater resolution for most of the other projects which had lower uncertainties). (v) The percentiles reported in Table 3 also provide some insights. Firstly, the contractor predictions are never negative for the 10% percentile (as one would expect) while the evaluated savings are about -0.9% (negative meaning that energy use increased after the ECM was installed). This could be caused by increases in energy use in the building unrelated to the upgrade. However, it is likely that energy use in other buildings could also have declined as a result of dynamic changes in the way the building is operated unrelated to the upgrade.

One would then expect these trends to compensate each other to some extent when a sample size consisting of 201 buildings is analyzed. The differences between the contractor-predicted and EP- evaluated savings are roughly of the order of 2:1 for all other percentiles. The contractors over-predict savings by 100% relative to evaluated savings.

FIGURE 10. COMPARISON OF CONTRACTOR-PREDICTED TOTAL ANNUAL SAVINGS WITH EP EVALUATED AGGREGATED SAVINGS FRACTION (BY BOTH WEATHER CORRECTED AND NON-WEATHER CORRECTED METHODS). N=201

FIGURE 11. PLOT DEPICTING DIFFERENCES IN EP EVALUATED ENERGY SAVINGS PERCENTAGE AND THE ASSOCIATED RELATIVE UNCERTAINTY FOR INDIVIDUAL PROJECTS. THE RELATIVE UNCERTAINTIES ARE GENERALLY SMALL THOUGH THERE ARE SOME EXCEPTIONS. THIS PLOT APPLIES FOR 201 PROJECTS SORTED INTO BIN D

(iii) Table 3 also assembles results of analyzing the lightingonly projects. Again, the differences between EPevaluated and contractor-predicted savings fraction are significant, with the difference being very similar to those from all projects since lighting upgrade is by far the most dominant type of energy upgrade. (iv) Figure 11 depicts the evaluated savings percentage (i.e. energy savings divided by baseline energy use) on an annual basis for all individual projects along with the associated fractional uncertainty (i.e. energy savings uncertainty divided by energy savings). The uncertainties of the change point models, characterized by their coefficient of variation of the root mean squared error (CV-RMSE) are generally large. However, the baseline model is used to predict energy use each month for the 12 months of the year, and so the uncertainty of the summed values is lower. The relevant formulae are given in various publications (Reddy and Claridge, 2000 or ASHRAE 14, 2002). Except for a small number of buildings, the

Energy Efficiency on an Urban Scale

73

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

FIGURE 12. PLOT DEPICTING THE RELATIVE UNCERTAINTIES IN INCREASING ORDER. THIS PLOT IS SHOWN FOR 199 PROJECTS OUT OF THE 201 PROJECTS AS 2 PROJECTS HAD UNCERTAINTIES GREATER THAN 200%.

THE RELATIVE UNCERTAINTIES ARE GENERALLY SMALL FOR THE MAJORITY OF THE PROJECTS

fraction for the 201 projects with usable contractor predictions falling in Bin D. It can be seen that most of the projects are below the expected ratio line of 1 with a few projects assuming large negative values. These fractional and negative projects drive the noted discrepancy. Figure 15 is the same plot redrawn with a narrower y-axis range so as to provide better resolution.

FIGURE 14. SORTED DISTRIBUTION PLOT OF THE SAVINGS RATIO FOR ALL THE 201 PROJECTS ANALYZED

Possible causes for the discrepancy stated under (i) above were investigated. Figure 13 is a plot of the distributions in annual energy savings fraction for the 201 projects predicted by the contractors and those determined by the weather normalized approach. The predicted savings fractions have a noticeably wide distribution across the various projects, exhibiting a long positive tail. On the other hand, the evaluated savings have a tighter distribution, and peak around 15% savings. However, many of the projects show negative savings which is a contributor to why the total EP evaluated energy savings fraction turns out to be so much lower than those predicted by the contractors.

FIGURE 15. SAME AS FIG. 14 BUT WITH NARROWER Y-AXIS SCALE (WITHOUT OUTLIER POINTS) FOR BETTER VIEWING OF THE DISTRIBUTION

FIGURE 13. COMPARISON OF THE FREQUENCY DISTRIBUTIONS OF ANNUAL SAVINGS PERCENTAGE AS A PERCENTAGE OF BASELINE CONSUMPTION PREDICTED BY THE CONTRACTOR AND AS PER EP EVALUATED SAVINGS DATA ANALYSIS. DATA IS FROM 201 PROJECTS WITH WEATHER CORRECTION

4.4 Possible Causes for Discrepancy We also investigated possible causes of differences between predicted and evaluated savings distributions. The following potential causes were identified. (a) The way visual binning was performed A possible reason for the discrepancy could be due to bias introduced in the analysis due to the visual manner by which buildings are sorted into the various bins (as described earlier). The savings analysis was repeated without such binning before the audit of contractor savings predictions.

Both the contractor-predicted savings and the evaluated savings fractions decreased. However, this analysis could not be repeated post-audit due to data constraints, so it is not possible to make conclusions. It is recommended that, though time consuming, screening and binning be conducted. It is

Another way of presenting this discrepancy is shown in Figure 14 as a distribution where the ratio of evaluated to contractorpredicted savings are sorted by magnitude of the savings Energy Efficiency on an Urban Scale

74

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

also recommended that analysis be run on the data both with binning and without binning to investigate if any possible bias has been introduced.

and #7, performed a high number of projects, and consistently over-predicted savings greatly. Figure 18 shows the same plot sorted by the number of projects done by the contractor (the contractor id is consistent with Figure 17).

(b) Difference due to single ECM versus multiple ECM upgrades

FIGURE 17. DIFFERENCE BETWEEN PREDICTED AND EVALUATED SAVINGS FOR 12 CONTRACTORS SORTED BY THE PREDICTED SAVINGS. THE NUMBER OF PROJECTS IS INDICATED ABOVE THE INDIVIDUAL BARS

In another effort to isolate the cause for the discrepancy between contractor-predicted and EP evaluated savings projects were divided into single ECM and multiple ECM categories. Figure 16 provides a direct comparison of the contractor-predicted and evaluated savings percentages. For single projects, the savings predictions by the contractors amounts to 9.2% of baseline energy use, while analysis suggests 3.4% evaluated, a larger discrepancy than for all projects combined. Of the 162 single-ECM projects analyzed, 161 were lighting upgrade projects, suggesting a possible connection specifically between lighting projects and estimation accuracy. . Further investigations were undertaken as part of Level 2 analysis, and are reported later on in this document. For projects adopting multiple ECMs, the discrepancy between contractor predictions and evaluated savings were much less than for all projects combined, (11.0% versus 9.8%).

FIGURE 18. DIFFERENCE BETWEEN CONTRACTORPREDICTED AND EP EVALUATED SAVINGS FOR 12 CONTRACTORS SORTED BY THE NUMBER OF PROJECTS. THE NUMBER OF PROJECTS IS INDICATED ABOVE THE INDIVIDUAL BARS

FIGURE 16. COMPARISON OF CONTRACTOR-PREDICTED SAVINGS AND EP EVALUATED SAVINGS WITH WEATHER CORRECTION FOR SINGLE PROJECTS (162 PROJECTS) AND MULTIPLE ECM PROJECTS (39 PROJECTS).THE PERCENTAGE SAVINGS COMPARED TO THE BASELINE IS ALSO SHOWN IN THE GRAPH

5. LEVEL 2 ANALYSIS In the case of lighting projects, the contractor-predicted savings were calculated as the kW reduction multiplied by the number of hours of operation. The calculation of the kW reduction appears fairly straightforward since it entails counting the number of fixtures and using engineering formulae to account for ballast and other effects. The number of hours, on the other hand, is an estimate, often supplied by the building owner.

(c) Contractor bias in savings estimation Another investigation involved determining whether certain contractors tended to consistently over-predict energy savings.

While there were 36 different contractors in total, there were 12 who undertook numerous projects or projects with large energy savings. The results of this study are summarized in Figure 17. We note that most of the contractors over-predicted savings as compared to evaluated savings with varying degrees. In particular, four contractors, contractor #1, #5, #6 Energy Efficiency on an Urban Scale

Level 2 analysis was meant to resolve any noticed discrepancy with the aid of direct measurements. In the Level 1 analysis, any lighting projects were found to have contractor kWh saving predictions which amounted to more than 70% of the baseline consumption for the building per utility data provided by APS. 75

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

One primary reason for such high percentages may have been the lack of a proper framework for matching an account with the master meter at the facility, but the differences could be due to other causes as well.

of the baseline energy use and it was found that the utility bills provided were only from one electric meter while the facility had four electric meters. As mentioned, there was no framework to match master meters with participant accounts.

FIGURE 19. LEVEL 2 ANALYSIS PERFORMED ON VARIOUS SITES FOR ASCERTAINING DISCREPANCY IN ESTIMATED HOURS OF OPERATION BY THE CONTRACTOR AND THE ACTUAL MEASURED HOURS OF OPERATION. THE NUMBERS ABOVE THE BARS ARE THE NUMERICAL VALUES OF THE WEEKLY HOURS OF OPERATION

Another potential source of contractor estimation error is inaccurate assessment of pre-upgrade equipment conditions.

Field measurements and internal inspection of fixtures on one project determined that, while the owner and/or contractor had assumed that all existing ballasts consisted of older, inefficient magnetic technology, at least some of the ballasts had been replaced with newer electronic versions during regular maintenance as ballasts had burned out. While a 100% pre-upgrade audit is not a cost-effective solution, this example illustrates the importance of pre-upgrade inspections with appropriate sampling. (APS has a statistical samplingbased inspection process in place for its programs and, for Energize Phoenix, APS agreed to have 100% of projects inspected. The percentage of fixtures/equipment inspected at each project and the depth of inspection is not known.)

Even though the customer confirms in writing on the incentive application that the operating hours estimated by the contractor are correct, inaccurate operating hours were hypothesized to be a major cause of savings prediction discrepancies. The procedure adopted to test this hypothesis was to install data loggers to determine hours of operation, and then compare them to the values submitted by the contractors. Out of numerous sites visited, twelve sites granted permission to install monitoring devices. Out of the twelve sites, eleven generated usable data. The data loggers were installed for a minimum of two weeks. The position of the data logger was carefully selected to avoid interference by human element or sunlight. The results of the analysis are shown in Figure 19. The results show that eight of the sites visited exhibited differences greater than 10% between reported and measured values of operating hours. Five of the sites had major discrepancies.

Other identified sources of contractor estimation error revealed during the data audit process include simple math errors, reading data from the wrong cells on worksheets (such as kW savings rather than kWh savings) and varying quality of estimation tools. The tools used by contractors to generate savings predictions varied widely in sophistication, usability and utility. Some involved room by room detailed equipment counts, specs and operating hours, and also included predicted HVAC savings from individual lighting heat load reductions. Some were little more than summary word processing documents. Some contractor spreadsheets and the contractor savings reporting spreadsheet provided by EP for prescriptive projects would have benefitted from having formula-driven calculations and cross-checks to minimize math errors.

The causes of discrepancies discussed above could greatly skew either contractor predictions or analysis results, and so it is important for programs to put procedures in place to ensure proper quality control. Other potential causes include the “rebound effect” (using savings to buy more energy-using equipment or being less diligent on behavioral conservation habits) and “energy creep” (a gradual increase in installed plug loads over time) in a facility after the upgrades were completed. The latter issue has been investigated and found to be unimportant for the Energize Phoenix project (reported in Appendix C).

For sites #6, 7, 9, 10 and 11, the cause for the observed discrepancies could not be determined. For two of these projects, the actual operating hours were less than half of the value used to predict energy savings. Since the predicted savings depend upon a direct multiplication by this value, inaccuracies lead to a directly proportionate overestimation of the savings.

Another source of discrepancy in the savings calculations could be the quality of the utility bill data itself and how it was designated in the database. A field visit was made to another facility where predicted savings were over 100% Energy Efficiency on an Urban Scale

76

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

6. LEVEL 3 ANALYSIS

FIGURE 21. EQUEST MODEL OF THE AZRS BUILDING.

6.1. Background The objective of Level 3 analysis was to perform a detailed energy analysis of a few selected projects involving the use of a detailed building energy simulation program. Two sites were selected: a. The Arizona State Retirement System Building (AZRS), and b. Mercado D building of Arizona State University This report only describes analysis work done on the first one, while that for the second building can be found in the Technical Report under preparation (Reddy et al., 2013).

Though the latter was a relatively small building, there were several similar buildings on the same campus which suffered from HVAC units which reached their end of life status.

Further, these buildings were repurposed several times, and so the rooftop HVAC units were added on incrementally resulting in excess cooling capacity, improper ducting runs and poor air distribution in spaces. The effect of a complete redesign of the HVAC units and of the air distribution system on the energy efficiency of the entire building was studied and is documented in Reddy et al (2013).

TABLE 4. AZRS BUILDING.

Arizona State Retirement System (AZRS) was one of the first projects, and also one of the largest, to be completed using EP incentives. Table 4 assembles some pertinent facts about the building. AZRS is part of the 3300 tower building in Downtown Phoenix (see Fig. 20.). The project involved replacement of chillers and converting to a chilled water variable speed pumping system. These upgrades were predicted to save about 1.26 million kWh of energy annually.

FIGURE 22. COMPARISON OF THE ACTUAL BASELINE MONTHLY UTILITY BILLS AND THE ENERGY CONSUMPTION OF BASELINE COOLING PLANT CALCULATED BY THE ENERGY MODEL FOR THE AZRS BUILDING. NOTE, THAT THE ENERGY MODEL ONLY CONSIDERED A FRACTION OF THE TOTAL ENERGY CONSUMPTION OF THE BUILDING TO ESTIMATE THE SAVINGS.

FIGURE 20. PHOTOGRAPH OF THE ARIZONA STATE RETIREMENT SYSTEM BUILDING (AZRS)

TABLE 4. AZRS BUILDING

The analysis reports and procedures used by the contractors during the design phase were requested, acquired and used to reanalyze and verify both the energy model and the savings, using actual monitored data. For predicting the savings, Energy Efficiency on an Urban Scale

77

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

FIGURE 24. WATER SIDE PLANT EQUIPMENT AS MODELED IN EQUEST SOFTWARE

the contractor had developed an energy model using eQuest simulation software (2010) (see Figure 21). However, the model was specifically developed for the cooling plant and for the HVAC system with little effort put into capturing the building loads accurately, which is a rather lengthy and tedious process. The time series plots shown in Figure 22 clearly indicate that the actual utility bills of the building during the baseline period are much higher than that of the cooling plant energy use generated from the eQuest model simulation. However, it is interesting to note that the patterns of variability are quite similar for both plots. The objective of this analysis was specifically to ascertain whether such an approach did yield the savings actually observed once the upgrades were completed.

The eQuest model (2010) for the water side plant equipment is shown in Figure 24. Table 5 assembles the month-bymonth disaggregated baseline consumption and the annual energy savings as predicted by the simulation model. Note that the predicted annual savings are 4,996,000-3,730,000 = 1,266,000 kWh, which is about 25.3% of the baseline value before the upgrades.

6.2. Upgrades Suggested and Predicted Energy Savings The following upgrades were identified and implemented by the contractors: • Replace the two (2) existing 900 ton Trane Centrifugal Chillers with four (4) new 350 Ton McQuay Frictionless Centrifugal Chillers (see Figure 23). Piping can be easily modified to accept the four new chillers. Existing power supply is adequate with minor modifications. Existing pad size is adequate with minor pad addition to accept an additional chiller per pad. The new chillers will primarily operate below 50% of existing kW/Ton conditions.

FIGURE 25. DISAGGREGATED END USE ENERGY BEFORE AND AFTER THE UPGRADES AS PREDICTED BY THE EQUEST SIMULATION MODEL DEVELOPED

• Re-pipe the central plant from a variable primary system to a primary secondary system. This new piping configuration will provide chilled water throughout the building in the most efficient manner possible; meaning water will be pumped only as require by cooling demand.

• Relocate HX within the piping system to optimize efficiency and utilize excess capacity of cooling Towers.

• Reprogram existing Alerton BACNET controls to maximize system staging and operate central plant in the most efficient manner, while maintaining tenant comfort.

FIGURE 26. PREDICTED ENERGY SAVINGS CONTRIBUTIONS FOR THE AZRS BUILDING

FIGURE 23. PHOTO OF THE NEW CENTRIFUGAL CHILLER

Energy Efficiency on an Urban Scale

78

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

Figure 25 shows the same data in a more visually meaningful manner. As expected from the nature of the upgrades, the space cooling, pumps and auxiliary and the vent fans energy uses before and after the upgrades are distinctly different.

The others (area lights, miscellaneous equipment and heat rejection) show no change. Figure 26 depicts the savings from the three individual end-uses. Note that space cooling energy reduction accounts for 76% of the total savings with pumps and auxiliary accounting for much of the remaining savings contribution.

difference is less than 10%, which is quite close in view of the other uncertainties present.

FIGURE 27. THE 5P CHANGE POINT BASELINE MODEL FITTED FOR THE AZRS BUILDING

TABLE 5.BASELINE AND PROPOSED ENERGY CONSUMPTION CALCULATED USING ENERGY MODEL

FIGURE 28. COMPARISON OF CONTRACTOR PREDICTED AND MEASURED ENERGY SAVINGS FOR THE AZRS BUILDING

6.3. Analysis and savings verification As stated earlier, the eQuest model developed by the contractors was not built to reflect the whole building energy use data. The reasoning was that this was not necessary since only cooling plant and HVAC related upgrades were being performed, and it would be adequate to model them properly to reflect energy use prior to the upgrade. Rather than trying to retune the baseline eQuest model to fit actual data, it was decided to simply evaluate the energy savings as a result of the upgrades and determine whether evaluated savings were consistent with predicted savings.

6.4. Conclusion Whenever possible, developing a detailed calibrated model is the best way to analyze energy savings in a building which undergoes multiple upgrades. However, the process is tedious (which translates to labor costs) and requires some amount of expertise both in the use of the software program and in the calibration process. Further, the needed monitoring data may not be available. The norm is to calibrate the model as accurately to the whole building energy use as possible and only then can one obtain an accurate prediction of the effect of different ECMs. The AZRS building model, though it was not well calibrated to the whole building, was modeled accurately enough in terms of the performance of the cooling plant and the distribution system that the energy savings predicted matched quite closely with those determined from actual utility bills before and after the upgrades.

A change point model (with five parameters) was found to best capture the monthly variations of the whole building energy use prior to the upgrades. This is consistent with the fact that the building had both electric heating and cooling sources.

Figure 27 shows how well the 5P model fits actual utility bills. Figure 28 allows comparison of the energy savings as predicted by the contractors and that determined from our analysis (termed evaluated savings). Note that there is a small difference (about 0.9%), with the contractor estimating a 9.7% savings while we found 8.8%. In relative terms, this

Energy Efficiency on an Urban Scale

79

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

7. PAYBACK ANALYSIS

FIGURE 29. PROJECTED YEARS FOR PAYBACK DETERMINED

BY VARIOUS METHODS

A simple payback analysis was also performed in order to obtain a conservative projection of the effectiveness of the Energize Phoenix program. Two types of simple payback values were calculated: one without EP rebates, and another with EP rebates. From Table 6, note that for the 236 Bin D projects the projected payback is 9.5 years without rebates and 7.4 years with rebates. Note that there is an additional rebate given by Arizona Public Service on top of the Energize Phoenix rebate, which could not be taken into account since the rebate amounts were not available for our analysis.

TABLE 6. PAYBACK ANALYSIS FOR 236 COMPLETED PROJECTS

8. CUMULATIVE SAVINGS OVER TIME Another aspect studied related to how energy savings accrued over the course of the three year EP project as a result of energy upgrades. These savings would depend on the number of projects implemented over time and how they performed subsequently. Figure 30 shows the cumulative plot based on contractor-predicted savings for the 236 projects from bin D. The cumulative energy savings for each month was determined by prorating annual energy savings predicted by the contractor, and adding the total historical savings up to that month. The plot reveals an exponential rise as one would expect. The corresponding number of projects implemented is also shown.

In essence, the payback periods are very high, especially given that the great majority of the EP upgrades were lighting upgrades which tend to have 1-2 year payback. If it had not been for the APS and EP rebates provided and/or for overpredicted savings, it is very likely that these lighting upgrades with total project cost payback periods of 9.5 years would not have been acceptable to the building owners. These calculations, however, do not include what the industry refers to as Non-Energy Benefits (NEBS), such as carbon reduction, increased property valuations and environmental benefits.

Other NEBS, such as comfort, durability, indoor air quality, and safety and their resulting impacts on health and productivity may alter the financial payback equation substantially. Since many of these benefits are very complex to estimate financially and some are broader societal benefits that are not captured directly by the building owner, they are unlikely to drive the building ownerâ&#x20AC;&#x2122;s decision-making process. The behavioral team separately analyzed organizational attitudes and motivations for participating in the program.

The plot serves to emphasize the ever-increasing disparity between projected and evaluated energy savings.

FIGURE 30. THE CUMULATIVE ADDITION OF CONTRACTORPREDICTED SAVINGS AND THE EVALUATED SAVINGS FOR THE 236 BIN D PROJECTS

These differences between payback periods using evaluated and contractor-predicted savings directly attributed to the under-performance of lighting projects. The results are also plotted in Figure 29 for better visualization.

Energy Efficiency on an Urban Scale

80

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

9. SUMMARY AND SUGGESTIONS FOR FUTURE ENERGY CONSERVATION PROGRAMS

using utility bills would generally have relative uncertainties much lower than the savings themselves at the individual building level. So, this study reinforces the prevalent outlook in the professional energy service industry that utility bill analysis is a simple and inexpensive way to evaluate energy savings and provide the necessary risk alleviation safeguard which financing entities may stipulate as a condition for providing funding.

In summary, the major conclusion of the EP commercial analysis effort is that contractor Predicted savings are much higher than those actually determined from analysis of actual utility billing data. While the former was around 9.8% of the baseline energy use for the entire program till end of April 2013, the EP evaluated savings were only 5.5%.

The discrepancy in average savings of individual projects was 22.4% versus 10.4%. Higher discrepancy was found in lighting-only projects than in projects with multiple ECMs.

In closing, the direct sources of discrepancy between the contractor-predicted savings and evaluated savings could come from:

Contractor bias accounts for some, if not much, of the observed differences in savings. One suggestion is that contractors be provided with utility bills or smart meter data of the facility at the time of estimating savings. This would better inform the contractor and minimize inappropriately high savings predictions. Another suggestion is that contractors be provided with a suite of savings estimation tools and incentives to use them (such as faster processing time for projects that utilize the provided tools). This could not only reduce estimation errors but also reduce program administration costs analyzing savings predictions from a wide range of tools of varying quality. Also, contractors should be provided feedback on how their predicted savings stack up against evaluated savings, i.e. inform them of the performance-based evaluated savings. This education would enable them to make quality control corrections to their audit process so as to produce more realistic savings predictions.

(i) Contractors’ estimates of operating hours being inaccurate even though the customer is required to sign off on the contractor’s estimate of the number of operating hours at the facility; (ii) Instances of high burnout in the pre-upgrade case for lighting upgrades while energy savings predictions presume fully operational lighting even when it is known that burnout has occurred; (iii) Discrepancies in input wattage between installed fixtures & database values. The database values reflect averages for each particular lighting technology and provide conservative estimates for consumption, and (iv) Errors from inadequate contractor prediction processes and tools, and lack of training of contractor staff in the proper use of those processes and tools,

From an evaluation standpoint, all building data should be screened in order to identify and remove spurious data spikes and patterns. The discrepancy between evaluated and contractor-predicted savings fraction increases greatly when binning is not undertaken. This leads to a tentative conclusion: in order to have sufficient confidence in savings analysis results, it is strongly advisable to first screen the data even though it is a time consuming process. The process could be somewhat automated, but this effort was deemed to be beyond the limited resources and time commitment of the commercial team of the EP program.

Pertinent to (ii) above is the inaccurate assessment of preupgrade equipment conditions.

If energy efficiency programs are to be scaled substantially, large portfolio financing is one enabler to reach scale.

Financing sources need predictable returns in order to invest without requirements for the high risk premiums warranted by uncertainty. A possible means of reconciling predicted versus actual performance, either by installing data loggers or by field visit surveys has been suggested. The proliferation of interval data from smart meters also opens up new possibilities for increasing estimation accuracy at the individual building level and through analysis of “Big Data” at the program level.

Further, even though the differences were not large in the EP program, it is advisable when evaluating savings using preand post-upgrade utility bills to perform weather normalization in a routine manner. The savings estimations would likely be more realistic and program managers would tend to view these corrections as warranted.

References ASHRAE Guideline 14 (2002), ASHRAE Guideline 12-2002.

Measurement of Energy and Demand Savings. American Society of Heating, Refrigeration and Air-Conditioning Engineers, Atlanta, GA.

Finally, change point models used for weather normalization Energy Efficiency on an Urban Scale

81

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

eQuest (2010). Building energy simulation software program, http://www.doe2.com/equest/

6. Sum the individual monthly savings to determine cumulative (or annual) savings and percentage savings.

Haberl J.S. and Culp, C. (2007). “Measurement and Verification of Energy Savings,” in Energy Management Handbook (Edited by W.C. Turner), Fairmont Press.

7. Compare the model goodness-of-fit (using the coefficient of variation of the root mean square error or CV-RMSE) with the percentage savings determined.

IPMVP (2010). International Performance Measurement and Verification Protocol, Volume 1. Efficiency Valuation Organization, U.S. Department of Commerce, Springfield, VA.

The model approach is statistical in nature, involving identifying a regression model of monthly energy use against monthly mean outdoor temperature using the monthly mean temperature model (Kissock, Reddy and Claridge 1998).

The ambient temperature is chosen as the only independent variable because of the easy availability of the data, the difficulty in acquiring other data, and to avoid statistical difficulty arising from a small data set (only 12 data points) and multi-collinearity with environmental indices such as ambient humidity and solar radiation.

Kissock, J. K.,Haberl, J. S. and Claridge, D. E (2002) “Development of a Toolkit for Calculating Linear, Change-Point Linear and Multiple-Linear Inverse Building Energy Analysis Models,” ASHRAE RP 1050, American Society of Heating, Refrigeration and Air-Conditioning Engineers, Atlanta, GA. Kissock, J.K., Reddy, T.A. and Claridge, D.E. (1998). “AmbientTemperature Regression Analysis for Estimating Retrofit Savings in Commercial Buildings,” ASME Journal of Solar Energy Engineering, vol. 120, no. 3, p. 168.

Another significant parameter to be considered is the uncertainty in the baseline model for a specific site characterized by the CV-RMSE (coefficient of variation of the root mean square error) of the model. This allows direct insights into the statistical soundness of the associated savings deduced. The CV-RMSE is a rough measure of the fractional (or percentage) uncertainty in the baseline model compared to the mean baseline energy use. A 10% CV-RMSE would imply that model uncertainty is 10% of the mean annual pre-upgrade energy use. If the savings fraction is less than about twice the CV-RMSE (corresponding to approximately 95% confidence level), then one is unjustified statistically in placing too much confidence in the associated savings estimated at that site. Adding this filter criterion to the analysis would have further reduced the total number eligible projects within the EP program. So for a single project all models were evaluated as shown in Fig. A1, and the model with the least CV-RMSE was chosen as the best fit baseline model.

Reddy, T.A., Claridge, D.E. (2000), Uncertainty of “Measured” Energy Savings from Statistical Baseline Models. HVAC&R Research, Journal Vol. 6, no. 1, pp. 3-20, American Society of Heating, Refrigeration and Air-Conditioning Engineers, Atlanta, GA. Reddy, T.A., K. Thalapully, M. Myers and O. Nishizaki (2013).

Final Technical Report by the Non-Residential Analysis Team of the Energize Phoenix Program, in preparation, Arizona State University, Tempe, AZ. APPENDIX A: BASELINE MODEL DEVELOPMENT AND UNCERTAINTY The savings methodology adopted is consistent with the one suggested in the professional literature (see for example, Haberl and Culp 2007).The process includes the following steps: 1. Acquire monthly energy use data (from utility bills) and data on influential variables (limited in this study to outdoor dry-bulb temperature) during the pre-upgrade period.

FIGURE A1. PROCESS OF DETERMINING THE BEST FIT REGRESSION MODEL FOR A SPECIFIC PROJECT INVOLVES FITTING ALL FORMS OF CHANGE POINT MODELS AND IDENTIFYING THE ONE WITH THE LEAST ROOT MEAN SQUARE ERROR (RMSE)

2. Develop a regression model of pre-upgrade energy use as a function of influential variables - this is the “baseline model”.

3. Acquire date of energy use (from utility bills) and influential variables during post upgrade period.

4. Use the values of influential variables from the post upgrade period (from step 3) in the pre upgrade model (from step 2) to predict how much energy the building would have consumed on a monthly if it had not been upgraded.

5. Subtract measured post upgrade energy use (step 3) from the predicted pre-upgrade energy use (step 4) to estimate savings on a monthly basis.

Energy Efficiency on an Urban Scale

82

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

APPENDIX B. MODEL IMPROVEMENT AFTER ADJUSTING FOR UTILITY BILL READ DATES

TABLE B1. ENERGY SAVINGS FRACTIONS DETERMINED

BY ADJUSTED AND NON-ADJUSTED UTILITY BILL DATA (151 PROJECTS)

B1. Problem Statement Utility bills are generally read every month, but the exact dates can vary by a few days from one month to the next.

In a large project such as Energize Phoenix involving hundreds of commercial buildings, the read dates for each building are spread out throughout the month. Taking this into consideration while automating the baseline modeling, and then calculating savings for each building, is rather error-prone and tedious especially in the calculation of the associated monthly mean ambient dry-bulb temperatures needed for weather normalization as described in Appendix A. A simplified approach is to assume the utility bills to correspond to the calendar month and use the corresponding monthly mean ambient dry-bulb temperatures which will then be the same for all projects. A better approach, and the one adopted in the Energize Phoenix program, is to weight the utility bill reading during adjacent months with the number of days in each consecutive month, and thereby, obtain adjusted utility bills corresponding to the calendar months.

This appendix reports on the differences between the adjusted utility bill results and those based on the raw or non-adjusted ones.

B4. Conclusion From the analysis results obtained, it is recommended that utility bills adjusted to match calendar months should be used to calculate energy savings. The change points models identified are clearly superior, while the energy saving results from both methods are different enough to warrant the extra step of utility bill adjustment. Hence, it is highly recommended that energy conservation programs, akin to Energize Phoenix, adopt this methodology.

FIGURE B1. COMPARISON OF CHANGE POINT MODEL CV_RMSE SORTED BY PROJECT ID BETWEEN THE RAW OR UNADJUSTED UTILITY BILL DATA AND THE REVISED UTILITY BILLS ADJUSTED TO CALENDAR MONTHS)

B2. Comparison of Model Goodness-of-Fit Figure B1 assembles the results of analyzing 151 projects by both methods. The CV_RMSE (coefficient of variation root mean square error) of the individual projects is determined from both data sets. It is evident the CV_RMSE, which is a direct indicator of the goodness-of-fit of the model, has reduced considerably from the unadjusted data set (termed â&#x20AC;&#x2DC;From actual utility billsâ&#x20AC;&#x2122;). A better way to illustrate this improvement is provided by Figure B2 where the projects have been sorted by increasing CV_RMSE values of the unadjusted utility bill results. Except for 5 projects, the improvement in the resulting change point models for the adjusted data set is very striking.

FIGURE B2. SAME AS FIG. B1 BUT SORTED BY CV_RMSE OF THE UNADJUSTED UTILITY DATA. THE IMPROVEMENT IN MODEL FIT WHEN USING ADJUSTED DATA IS CLEARLY NOTICEABLE

B3. Energy Savings Comparison Table B1 assembles the results of the energy savings by both methods. The results using adjusted utility bills are obviously the ones where one would place more confidence. It is clear that there is an important bias in savings between both data sets, and so, the extra step in adjusting the utility bills to correspond to calendar months is warranted.

Energy Efficiency on an Urban Scale

83

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

APPENDIX C: ENERGY CREEP ANALYSIS

Figure C1 shows the results of the analysis for the 204 projects. Here, the projects with positive increase from 2009 would suggest the presence of energy creep while those with negative creep would show a decrease in energy consumption. The graph is almost symmetric indicating that there is no observable energy creep during this period across all the projects.

C1. Background Energy creep of a building can be defined as a gradual increase in the energy consumption in the building over time due to increase in plug loads (computers, lights…) or other operational parameters such as changing thermostat set points, increased number of occupants. This factor does not (rather, should not) include changes in other operational parameters such as longer operating hours of the building, or installing additional HVAC equipment.

Figure C2 plots summary results across all 204 projects analyzed. All three average indices are negative suggesting in fact that 2010 energy use is very slightly lower than that of 2009. The median difference between 2009 and 2010 for all the projects is only 0.9%, which is well within the uncertainty range of the models, and can be effectively interpreted as zero. The same conclusion applies to the mean and the total savings fraction also. Thus, we would conclude that though there are differences in individual projects, taken as a whole, there is no statistical evidence to conclude that creep is a factor which needs to be considered in the commercial building data set analyzed within the framework of the Energize Phoenix program.

The study of energy creep is important in a program such as the Energize Phoenix (EP) program. It can mask some of the predicted energy savings due to implemented energy conservation measures, and help undermine the credibility of energy conservation programs. In other words, if a building had shown a continuous energy increase in the years prior to the implementation of the upgrades, then the effect of the upgrades would be under-predicted if the effect of energy creep were neglected.

C2. Issue Investigated

APPENDIX D: CASE STUDY: LEVEL 2 ANALYSIS OF GARAGE LIGHTING RETROFIT

We wanted to determine whether the commercial buildings which underwent upgrades under the EP program suffered from this problem or not. This could have a direct bearing on the evaluated savings determined from the analysis methodology.

D1. Background A lighting upgrade which involved replacing 378 T12 light fixtures 8’ long with 756 T8 light fixtures 4’ long was performed in the three levels of the underground Parking Garage of the Arizona State University Nursing and Health Innovation (NH1) Building. Electrical and luminance measurements, both before and after the upgrade, were conducted. The main intent was to physically measure and study the effects of energy upgrades on both illumination levels and energy consumption.

C3. Calculation Methodology Since the EP program started in 2011, the baseline year was chosen to be 2010, and all projects in Bin D which had complete data were taken as the population set. The objective was to simply determine whether energy use during 2010 was statistically different from 2009 or not. This would involve developing a baseline change point model for 2010 for each building and using it to predict energy use for 2009. The difference between the energy use between the two years was attributed to energy creep in that building. Though we realize that random effects would introduce bias and uncertainty in our analysis at the individual building level, taken over the whole population of buildings, this random effect is likely to get smoothened out.

FIGURE C1. DISTRIBUTION OF PERCENTAGE ENERGY CHANGE FROM 2009 TO 2010 FOR THE 204 PROJECTS ANALYZED FOR CREEP

C4. Results and Conclusion The total number of projects analyzed here is 204 as opposed to 236 which was the total number of projects in the final analysis (see main body of report). This discrepancy is because of data abnormality in the other 32 projects which led us to reject these projects.

Energy Efficiency on an Urban Scale

84

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

FIGURE C2. COMPARISON OF THE FRACTIONAL CHANGE IN ENERGY CONSUMPTION FROM 2009 TO 2010 FOR THE 204 PROJECTS ANALYZED. NEGATIVE SAVINGS IMPLY A REDUCTION IN ENERGY USE OR A NEGATIVE CREEP

D3. Methodology The power savings determined by the measurements were directly compared with the value predicted by the contractor.

The contractor-predicted savings were determined with feedback from the customer about the details of individual lamps, such as quantity, type, operating hours, etc. This information was then entered into a standard spread sheet to compare it with the upgraded equipment ratings to determine the energy savings. This method is typical of the methodology adopted in most of the EP projects of such type.

The commercial team noted that this simple spreadsheet method also added cooling load savings (resulting from the decrease in the heat output of the lights), with predicted cooling load savings of approximately 15,000 kWh/year. This is incorrect since the parking garage was not air-conditioned, and so the credit associated with cooling load reduction does not apply in this case. It is also worth mentioning that this discrepancy was rectified by the contractor itself and a more accurate value, which didnâ&#x20AC;&#x2122;t include the cooling load savings, was provided to us with which to perform the utility bill analysis. Such estimation errors can be avoided with fairly simple documentation tools and procedures and, if possible, by physical inspection by the concerned authorities to avert such mistakes.

D2. Objective The objective of this study was to sub-meter the energy savings and the lighting quality improvement due to retrofits.

This building had numerous advantages which made it ideal for measurements. Normally, getting access to one of the commercial projects for performing measurements was a difficult task. Being an ASU building, the team had easier access to this structure. The nature of the upgrade was also another advantage. The parking garage had lights which operated 24 hours a day throughout the year and so the measurements could be taken for a small period which could then be extrapolated for the whole year. The study was

The lumen levels were also measured at various locations in the garage both before and after the upgrade. The measurement was made using a standard light meter and the measurements were done using procedures prescribed in IES standards (Refer to full technical report for more information).

FIGURE D1. THE ARIZONA STATE UNIVERSITY NURSING AND HEALTH INNOVATION (NH1)

Finally the savings for this particular project were determined from a utility bill analysis. The results were compared with the contractor predictions and the measurement results made by the EP team to validate the results. A few concerns and inferences are reported below.

D4. Results of Savings Measurements From Table D1 it is evident that there has been a reduction in the power consumption in both measured as well as the contractor predictions due to the lighting upgrades. The garage level 1 upgrades were not fully complete at the time of the post-upgrade site measurements and the lamps upgraded were wired to different electrical circuits. Hence, these values were intentionally not used in our analysis. The overall savings were thus determined for the levels 2 and 3 of the parking garage assuming year-round operation. The savings result as a percentage decrease is shown in Table D2.

conducted in year 2 of the Energize Phoenix project and has been documented in detail in year 2 report. In year 3, an additional analysis was done using the utility data to verify and compare with the earlier results. This is the main focus of this report.

Energy Efficiency on an Urban Scale

85

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

TABLE D1. MEASURED POWER AND CONTRACTORPREDICTED POWER AT VARIOUS LEVELS

The utility data for the project was obtained in year 3 of the Energize Phoenix project. This allowed re-evaluation of the savings predictions by the contractor against analysis results.

Note that the data obtained was for the whole building and the consumption of the garage could not be isolated from it.

The contractor-predicted savings fraction was about 4% of

TABLE D2. THE ANNUAL SAVINGS ESTIMATED BY MEASUREMENT AND CONTRACTOR PREDICTIONS The lumen levels increased after the lighting upgrade. The readings for one of the floors is assembled in Figure D2 for illustrative purposes. The figure also indicates the lumen levels for four locations marked (shown in red on the grid map). The pre and post upgrade lumen levels for these locations are shown in the adjacent bar chart (one for each location). For instance, the location A/B had an increase in lumen levels in all three orientations in which lumen levels were measured.

The result is also visually obvious from the pre and post upgrade photographs of Figure D3 corresponding to garage

level 1.These results are typical of all the three garage levels where lighting values where measured. We had 8 points for

each floor where the readings were measured. The complete details of the measurements can be found are in the EP summative report for year 2.

FIGURE D2. PRE AND POST UPGRADE LIGHTING LEVELS FOR FOUR LOCATIONS AT A SPECIFIC GARAGE LEVEL. THE PLAN IS SHOWN ON THE LEFT WITH THE RED DOT INDICATING THE LOCATIONS WHERE THE MEASUREMENTS WERE TAKEN. THE READINGS FOR THESE LOCATIONS ARE SHOWN IN THE HISTOGRAMS ON THE RIGHT. FIGURE D3. PHOTOGRAPHIC COMPARISON OF LUMEN LEVELS BEFORE AND AFTER RETROFIT

baseline consumption. This suggests that the total energy consumption of the parking garage was a minor fraction of the

D5. Comparison with Utility Bill Analysis FIGURE D4. ENERGY CONSUMPTION HISTORY OF THE NH1 BUILDING

Pre-upgrade picture of garage parking level 1

Energy Efficiency on an Urban Scale

86

Post-upgrade picture of garage parking level 1

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

whole building. So, it is highly likely that the savings value we would determine may not be sensitive enough to discern such a small difference. We also studied the three years of utility

between the energy savings predicted by the contractor and that from our analysis. This may be due to one or more of the following reasons. a) In the initial pre upgrade measurements, two of the individual fixtures were inspected and it was found that

bills prior to the retrofit, and found good consistency in the usage pattern (see Figure D4). We also made sure that there have been no other major energy upgrades in the building.

Thus, the savings determined from a utility bill analysis can be assumed to be solely due to the parking garage energy upgrades. one of them had an electronic ballast and the other one had a magnetic ballast (which consumes considerably more power). Since there were no data available on the actual numbers of electronic and magnetic ballasts present in the circuit before the retrofit, the contractor had assumed (or the owner had indicated to the contractor) that all the lamps had magnetic ballasts.

Figure D5 is a scatter plot of the utility bills versus outdoor temperature. The scatter is quite consistent, and is well fit by a 5 parameter change point model as shown.

FIGURE D5. SCATTER PLOT OF THE UTILITY BILLS VERSUS OUTDOOR TEMPERATURE TO WHICH A 5P CHANGE POINT REGRESSION MODEL HAS BEEN IDENTIFIED Finally, the savings results obtained by different analysis methods are summarized in Figure D6. It is clear that energy savings determined by our analysis (1.8%) are only half of that predicted by the contractor. The effect of weather normalization is relatively minor.

b) A few lighting circuits had emergency lamps connected to them. These lamps run on batteries and only use power to charge the batteries when the battery charge is below a threshold level. It was assumed that the batteries were fully charged all the time, since these lamps turn on only when there is a power outage. So their consumption was assumed to be negligible when doing the measurements.

But, perhaps not coincidentally, the circuits with emergency lamps showed much less reduction in post retrofit consumption than expected. Since the power consumption of these lamps could not be separated out in our measurements, this might be another reason why evaluated savings are lower than those of the contractor.

FIGURE D6. ENERGY SAVINGS BY CONTRACTOR PREDICTIONS AND WEATHER CORRECTED AND NON-WEATHER CORRECTED METHODS D6. Conclusion The measured power consumption decreased, as is evident

c) Our measurement of electrical power did not involve a 3-phase power measurement but, rather, separate readings of current and voltage. Such an approach gives only an approximate estimate of the actual power consumption value and is not always accurate. This might also be a reason for the difference between the values predicted by the contractor and the measured ones. from measurements and utility bill analysis. There is also a considerable increase in the lumen levels from pre upgrade levels. However, there was a large (about 50%) discrepancy Energy Efficiency on an Urban Scale

d) Finally, recall that the lighting upgrades proposed by the contractor were only 4% of the baseline energy use of the

87

Global Institute of Sustainability

Back to Main Table of Contents

Appendix H Table of Contents

building. The savings results predicted by the utility bill analysis are likely to have uncertainties larger than the amount of savings being evaluated, and so we cannot place too much confidence on the utility bill analysis results.

In summary, the overall conclusions subject to the caveats stated above are: 1) Post-upgrade illumination levels are consistently and significantly higher than those prior to the retrofit.

2) The post upgrade illumination levels exceed IES recommended parking garage lighting levels. (A large percentage of the pre lighting levels were below the recommended levels) 3) The evaluated savings were much lower than those predicted by the contractor. Using direct measurements, post-upgrade power consumption savings were evaluated at 5% (for two parking garage levels) whereas the savings predicted by the contractor were 30 %.

4) A utility bill analysis suggests that the upgrades saved only half of what was predicted by the contractor for the facility (3.9 % to 1.8%).

Energy Efficiency on an Urban Scale

88

Global Institute of Sustainability

Back to Main Table of Contents

• Just under half of all single-family residences implement three types of ECMs, with a further 24.2% implementing two ECMs.

APPENDIX I DESCRIPTIVE, INFERENTIAL AND ECONOMETRIC ANALYSIS OF ENERGIZE PHOENIX PARTICIPATION AND SAVINGS EXECUTIVE SUMMARY

• The most popular combination is air sealing, duct sealing and insulation (38.8%).

This report analyzes data collected by the Energize Phoenix program between 2010 and 2013 to describe participants’ characteristics and quantify the impact of the program on energy usage.

• The total contractor-estimated annual energy savings for single-family residences ranges from 832 kWh to 12,281 kWh, with a mean of 3,208.5 kWh and a median of 2,662 kWh.

The characteristics of commercial and residential participants and the energy conservation measures (ECMs) implemented are as follows:

Matched pairs of mean energy usage for three time horizons – months 1-3, 4-6 and 1-6 immediately before and after the installation of the ECM(s) show:

Commercial

• A decline in mean energy usage of -9.77, -8.06, and -8.91 watt hours per square foot per day, for months 1-3, 4-6 and 1-6 respectively compared to the same pre-ECM period1 for commercial participants.

• 54% of the commercial participants are drawn from four NAICS business sectors – retail trade, real estate, other services (excluding public administration), and accommodation and food services.

• A decline in mean energy usage of 1.91 kWh per day for months 1-6 pre- and post-ECM for residential participants.

• Light bulbs and fixtures is the most popular ECM, installed by 87% of commercial participants, followed by lighting controls (17.2%). Over 8% of commercial applicants have taken advantage of HVAC, and 7.8% pumps and motors.

Shade screens account for 6.8% of all commercial ECMs, and refrigeration 3.1%. The program includes only two commercial water heater ECMs.

Econometric methods used to consider the full range of influences on energy consumption by participants shows: Commercial • The initial effect of a set of ECMs is a reduction in energy usage intensity of 3.24 watt hours per square foot per day (5%) ceteris paribus.

• The total contractor-estimated potential annual energy savings for all commercial ECMs implemented ranges from 813 kWh to 2,718,983 kWh, with a mean of 106,896 kWh and a median of 24,285 kWh.

• The long-run impact of a set of ECMs is 10.79 watt hours per square foot per day (17%). Long-run savings rates are essentially achieved in approximately 8 months ceteris paribus.

Residential • Approximately 70% of residential heads of households work full-time, compared to 3% unemployed and seeking employment.

Residential

• Over 90% of all heads of households are either White Caucasian (the biggest ethnic group), Hispanic Latino or of multiple races.

• The long-run effect is 4.72 kWh per day (12%), essentially achieved in approximately 2 months ceteris paribus.

• The ECM effect immediately after installation is 2.57 kWh per day (6.7%) ceteris paribus.

The long-run impacts of ECMs are used to calculate annualized energy and dollar savings for completed upgrades and forecasted completions by September, 2013. These are:

• Duct sealing is the most popular single-family residence ECM (90.9%). Air sealing is implemented as part of 81.3% of all single-family residence ECMs. Insulation is implemented by 77.2% of single-family residences, and sun shades by 24.7%. The least popular measures are water heaters (12.3%), HVAC replacement (11.9%), and HVAC tune-ups (1.4%).

Energy Efficiency on an Urban Scale

• Residential annual energy savings – 689,120 kWh. • Residential annual dollar savings – $73,943.

The inferential testing does not account for any other influences on energy usage such as changes in temperature and fluctuations in the general economy.

1

89

Global Institute of Sustainability

Back to Main Table of Contents

Furthermore, given the low participation of multi-family residential properties, these are grouped in the analysis with commercial participants at the City of Phoenix’s request.

• Commercial annual energy savings – 134,320,000 kWh. • Commercial annual dollar savings – $12,558,920.

• Total Energize Phoenix energy savings – 135,009,120 kWh.

Section 2 provides participant characteristics and initial descriptive statistics for the commercial and residential groups.

• Total Energize Phoenix dollar savings – $12,632,863.

1. PROGRAM REACH

Section 3 describes energy usage patterns before and after the installation of ECMs by residential and commercial participants.

The Energize Phoenix program consists of energy conservation measures (ECMs) targeted at single-family residential, multifamily residential and commercial buildings located in the Phoenix Light Rail corridor.

Section 4 describes the inferential testing and econometric analysis of the energy usage.

Up to and including March 31, 2013, a total of 424 commercial organizations participated in the program, including 11 multifamily residential properties (representing 246 residential units). A further 219 single-family residences also participated in the program.

The costs per kWh and emission savings are described in Section 5.

2. PARTICIPANT CHARACTERISTICS AND DESCRIPTIVE STATISTICS

This report contains a series of descriptive statistics for program participants and calculates the total energy savings using a simple and then more sophisticated regression analysis.

Commercial Figures 1 and 2 categorize Energize Phoenix’s commercial participants in terms of business type and building type. The 2-digit North American Industry Classification System (NAICS) is used for business type,2 and the Commercial Buildings Energy Consumption Survey (CBECS) for building type.3

Participant data is collected from the City of Phoenix, APS and participants in four ways:

FIGURE 1: NUMBER OF COMMERCIAL EP PARTICIPANTS BY 2-DIGIT NAICS SECTOR

• Commercial and residential logs – these are collected by the City of Phoenix and contain records of each energy efficiency project describing, among other things, the customer address, the rebate amount, and the date of the ECM installations.

• PDF and CSG files - PDF files are specific to the commercial program, and contain details about the ECMs installed.

CSG files arrive direct from APS, are specific to the residential program, and contain information about home physical attributes, the ECMs installed and the estimated kWh saved.

Source: Seidman Research Institute analysis

• APS energy usage data for each location.

• Commercial and residential surveys – these are collected

by ASU’s behavioral research team, contractors and city staff and primarily contain attitudinal data for each participant.

NAICS is the standard used by Federal statistical agencies to classify business establishments for the collection, analysis, and publishing of statistical data related to the U.S. business economy.

2

Despite the best efforts of the ASU team, some of the log data, PDF/CSG files, read data and/or surveys are missing for specific participants. As a result, this report focuses on participants with completed records.

Energy Efficiency on an Urban Scale

The Commercial Buildings Energy Consumption Survey (CBECS) is a national sample survey that collects information on the stock of U.S. commercial buildings, their energy-related building characteristics, and their energy consumption and expenditures.

Encompassing any building in which at least half of the floor space is used for a nonresidential, non-industrial, or non-agricultural, it includes building types that are not traditionally considered “commercial”, such as schools, correctional institutions, and places used for religious worship.

3

90

Global Institute of Sustainability

Back to Main Table of Contents

Figure 1 illustrates that 54% all commercial participants are drawn from four business sectors – retail trade, real estate, other services (excluding public administration), and accommodation and food services. Retail trade is a broad industry super-sector comprised of all forms of retail businesses including department stores, car dealerships, and hardware stores. Public administration, arts/entertainment and recreation, transportation and warehousing, and the information sectors account for only 8.5% of all commercial participants.

The commercial program ECMs can be classified into seven types – HVAC, light bulbs and fixtures, lighting controls, pumps and motors, refrigeration, shade screens, and water heating. Figure 3 summarizes the type of ECM implemented by commercial participants. Light bulbs and fixtures is the most popular ECM, installed by 87% of commercial participants, followed by lighting controls (17.2%). Over 8% of commercial applicants have taken advantage of HVAC, and 7.8% pumps and motors. Shade screens account for 6.8% of all commercial ECMs, and refrigeration 3.1%. The program includes only two commercial water heater ECMs.

Figure 1’s results should perhaps be compared with the NAICS classification of all commercial organizations in the target corridor to confirm the representative nature of the participants.

TABLE 1: CROSS-TABULATION BETWEEN NAICS AND CBECS CLASSIFICATIONS

FIGURE 2: NUMBER OF COMMERCIAL EP PARTICIPANTS BY CBECS BUILDING TYPE

Source: Seidman Research Institute analysis

FIGURE 3: TYPES OF COMMERCIAL ECM IMPLEMENTED Source: Seidman Research Institute analysis

Figure 2 groups commercial participants by CBECS building type. Over 30% of participants are categorized as offices – that is, buildings used for general office space or professional services, including medical offices that do not operate any diagnostic equipment. Approximately 40% of participants are based in service, mercantile or food premises. Warehousing accounts for 7.3%, hotels and lodging 5.2% and schools 3.3%. A small number of multi-family buildings (3.1%) are also included, reflecting a City of Phoenix request to include multi-family properties with commercial ones.

Source: Seidman Research Institute analysis

Most commercial participants implement just one type of ECM (76.2%). 17.5% of commercial participants implement two ECMs; and 6.4% three or four ECMs. Multiple ECM installations can occur simultaneously or as separate projects in the timeline of the Energize Phoenix program.

Table 1 cross-tabulates the NAICS and CBECS classifications, to identify some inconsistencies, particularly in relation to the interpretation of retail/mercantile and food business sector building types. The NAICS business classification appears to provide greater detail for CBECS’ broader office and service building types and is therefore used for additional descriptive analysis.

Energy Efficiency on an Urban Scale

91

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 5: NAICS CLASSIFICATION OF COMMERCIAL PARTICIPANTS WITH TWO OR MORE ECMS

Figure 4 illustrates the total participation rate for various types of two-ECM project. This shows that lighting (as bulbs and fixtures or controls) features in seven of the nine pairedECM combinations implemented.

Figure 5 illustrates the NAICS 2-digit business sector classification of commercial participants with two or more ECMs. Consistent with Figure 1, this suggests that the main beneficiaries are the retail trade sector, accommodation and food services, real estate, and other services (except public administration).

Figures 6 and 7 examine the NAICS code classifications for commercial participants implementing two ECMs, and those implementing three or four ECMs.

Source: Seidman Research Institute analysis

FIGURE 6: NAICS CLASSIFICATION OF COMMERCIAL PARTICIPANTS WITH TWO ECMS

Figure 6 suggests that the retail, accommodation and food, other services, and real estate sectors account for almost 60% of the paired-ECM combinations.

27 commercial participants benefit from three or four ECMs, illustrated in Figure 7. Nine of this subset (33%) is drawn from retail trade. Real estate accounts for 4 participants, accommodation and food 3 participants, and educational or health care 2 participants each. The other seven sectors listed account for 1 participant each.

FIGURE 4: NUMBER OF TWO-ECM COMBINATIONS BY TYPE

Source: Seidman Research Institute analysis

FIGURE 7: NAICS CLASSIFICATION OF COMMERCIAL PARTICIPANTS WITH THREE OR FOUR ECMS

Source: Seidman Research Institute analysis

Source: Seidman Research Institute analysis

Energy Efficiency on an Urban Scale

92

Global Institute of Sustainability

Back to Main Table of Contents

Contractors provide prospective commercial participants with an estimate of energy savings for specific ECMs as part of the sales process or through a contracted energy audit.

The contractors’ total estimated annual energy savings for commercial participants ranges from 813 kWh to 2,718,983 kWh, with a mean of 106,896 kWh and a median of 24,285 kWh. The standard deviation for the entire commercial sample’s energy usage savings is very high at 255,443.2 kWh, demonstrating the significant extent to which individual commercial estimated energy savings vary. This is related to a high correlation between estimated energy savings and building size (Pearson correlation coefficient of 0.595, with a two-tailed p value < .000).

The mean space for commercial participants is 65,217 square feet, with a median of 9,000 square feet and a range of 500 square feet to 2,190,237 square feet. Dividing the commercial participants into quartiles, the percentiles for this variable are as follows: • 25th percentile: 3,200 square feet.

• 50th percentile: 9,000 square feet.

• 75th percentile: 30,370.75 square feet.

TABLE 3: CONTRACTORS’ ESTIMATED ANNUAL ENERGY SAVINGS (KWH) FOR MULTIPLE-ECM PROJECTS

Table 2 illustrates the contractors’ estimated annual kWh savings for specific types of ECMs, projected during the sales or energy audit process. The table only relates to the 76.2% of commercial participants implementing a single ECM. TABLE 2: CONTRACTORS’ ESTIMATED ANNUAL ENERGY SAVINGS (KWH) FOR SINGLE-ECM PROJECTS

Source: Seidman Research Institute analysis

Table 2 suggests that commercial participants are projected to save an annual mean of 89,765 kWh and an annual median of 19,667 kWh from light bulb and fixtures ECMs. This is the only ECM with a sufficient sample size allowing for any form of projected annual energy saving generalization to be made. The variation in the range of estimated savings for light bulbs and fixtures is quite large, with a Pearson two-tailed correlation coefficient of 0.695 (p value <.000) between the contractors’ estimated energy savings and the size of a commercial business for this sub-group.

The sample size of each of the other five ECM types listed is too small to draw any conclusions. No commercial participant has exclusively implemented a refrigeration or shade screen ECM.

Source: Seidman Research Institute analysis

Table 4 summarizes the contractors’ mean and median estimated annual energy savings by square feet quartile. This suggests that the largest buildings in absolute terms account for the greatest estimated annual energy savings. However, an estimate of the annual mean and median kWh saved per quartile mean square footage, illustrated in Table 5, suggests that the contractors estimate greater kWh savings per square foot for smaller rather than larger commercial buildings. That

Table 3 illustrates the contractors’ estimated annual energy savings for different combinations of multiple ECM projects.

Caution should be exercised in making any generalizations about specific combinations of ECMs, given their respective low sample sizes. For all combinations of multiple ECMs, the mean estimated energy saving is 161,680 kWh, and the median 36,984 kWh. However, given the variance in the data, it is difficult to draw any meaningful conclusions.

Energy Efficiency on an Urban Scale

93

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 8: TOTAL COST PER SQUARE FOOT FOR ALL COMMERCIAL PARTICIPANTS

is, the smaller first quartile has an estimated annual mean saving of 6.56 kWh per square foot, falling to 5.02 kWh for the second quartile, 2.93 kWh for the third quartile, and 1.42 kWh for the fourth quartile.

TABLE 4: CONTRACTORSâ&#x20AC;&#x2122; ESTIMATED ANNUAL ENERGY SAVINGS (KWH) BY SQUARE FEET VARIABLE QUARTILES

Source: Seidman Research Institute analysis

The mean total cost of the ECM(s) implemented per commercial applicant is $54,621.49, with a median of $6,993.10 and a range of $110 to $2,569,968.

Source: Seidman Research Institute analysis

Table 6 presents further analyses of the total cost in terms of the number of ECMs completed and the square feet quartiles.

The table suggests a mean cost per square foot of $3.01 for the smaller first quartile, falling to $1.62 for the second quartile, $0.95 for the third quartile and $0.79 for the largest square feet fourth quartile.

The mean program rebate for all 424 commercial participants is $14,029, with a median of $2,605.71, a range of $27.50 to $362,293.01 and a standard deviation of 35,636.8.4 TABLE 5: CONTRACTORSâ&#x20AC;&#x2122; MEAN AND MEDIAN ESTIMATED ANNUAL ENERGY SAVINGS (KWH) BY SQUARE FEET

The significant differences between mean and median costs for one, two or three ECMs in Table 6 illustrate the variance in the range of total costs for each grouping. As a result, the median cost for each number grouping of ECMs is in all probability a more reliable statistic. Only two commercial participants benefit from four types of ECMs, undermining the value of their mean and median total cost calculations.

Source: Seidman Research Institute analysis

The total cost and program rebates per square foot for all 424 participants are compared in Figure 8. This suggests that the total cost per square foot is less than $1 for approximately 62% of commercial participants, and less than $2 for 83% of commercial participants. At the top end, the total cost per square foot is $5 or more for less than 7% of commercial participants. Figure 8 also suggests that the total rebate per square foot is less than $1 for over 90% of commercial applicants. No rebate of $10 or more per square foot has been awarded to a commercial participant.5

TABLE 6: TOTAL COST OF COMMERCIAL INSTALLATION

BY SQUARE FEET QUARTILES AND NUMBER OF ECMS

Source: Seidman Research Institute analysis

Single-Family Residences 46.1% of all single family residence decision-makers are female, compared to 53.9% male. Figure 9 summarizes the key demographics of participants. This suggests that

Total program rebate amount is the amount paid by the City of Phoenix. It does not include any rebate from APS. 4

The analysis excludes City of Phoenix-owned properties that are in-progress as of March 31, 2013, two of which are expected to receive rebates greater than $40 per square foot.

5

Energy Efficiency on an Urban Scale

94

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 10: NUMBER OF OCCUPANTS PER SINGLE-FAMILY RESIDENCE PARTICIPANT

approximately 70% of residential heads of households work full-time, compared to 3% unemployed and seeking employment. 17% are unemployed but are not seeking employment due to retirement, disability or because they are a stay-at-home partner. Over 90% of all heads of households are either White Caucasian (the biggest ethnic group), Hispanic Latino or of multiple races. The mean and median age is 51 years; and over 8 out of 10 also have a college education.

FIGURE 9: DEMOGRAPHICS OF SINGLE-FAMILY RESIDENCE HEADS OF HOUSEHOLDS

Source: Seidman Research Institute analysis

Figure 10 suggests that almost 1 in 4 homes are occupied by a single person. Over half of all single-family residences have two occupants. 11.2% of participant homes have three occupants, 9% four occupants and 3.3% five or more occupants.

Figure 11 suggests 3 bedrooms and 1 or 2 bathrooms are the norm among participants.

FIGURE 11: NUMBER OF BEDROOMS AND BATHROOMS PER SINGLE-FAMILY RESIDENCE PARTICIPANT Source: Seidman Research Institute analysis

According to the residential survey data, the annual household income ranges from $19,500 to $300,000, with a mean of $121,783 and a median of $100,000. Dividing this variable into quartiles, this produces: • 25th percentile: $64,000.

• 50th percentile: $100,000.

• 75th percentile $150,000.

Source: Seidman Research Institute analysis

These income ranges appear quite high given the general Energize Phoenix target area and are therefore not used as part of any additional descriptive analysis in this report.

FIGURE 12: WALL-TYPE PER SINGLE-FAMILY RESIDENCE

Participants have resided in Arizona for up to 65 years, with a mean term of residence of 27 years, and a median term of residence of 30 years.

Figures 10 to 13 describe the single-family residence participants in terms of the number of occupants and key property characteristics.

Source: Seidman Research Institute analysis

Energy Efficiency on an Urban Scale

95

Global Institute of Sustainability

Back to Main Table of Contents

Approximately 59% of participants’ homes have a wooden frame, compared to 38% with masonry.

tune-up, insulation, shade screens, and water heater (solar or other). Figure 15 summarizes the type of ECMs implemented by single-family residential participants. They are also all given 10 CFLs for contractor- or self-installation. However, lighting as an ECM is excluded from the single-family residence analysis as there is no way of confirming that participants actually replaced their old bulbs.

Figure 13 describes the homes’ roof insulation as predominantly fair (61%) or good (25%). The average ceiling height of participants’ homes is 8.5 feet, with a median of 8 feet, and a range of 7.7 to 12 feet.

FIGURE 13: CONDITION OF ROOF INSULATION PRE-ECM PER SINGLE-FAMILY RESIDENCE

FIGURE 15: ECM BY TYPE FOR SINGLE-FAMILY RESIDENCES

Source: Seidman Research Institute analysis

Source: Seidman Research Institute analysis

Duct sealing is the most popular single-family residence ECM (90.9%). Air sealing is implemented as part of 81.3% of all single-family residence ECMs. Insulation is implemented by 77.2% of single-family residences, and sun shades by 24.7%.

The least popular measures are water heaters (12.3%), HVAC replacement (11.9%), and HVAC tune-ups (1.4%).

FIGURE 14: FRONT-FACING DIRECTION OF SINGLE-FAMILY RESIDENCES

Figure 16 illustrates the number of ECMs installed per single family residence. Just under half of all single-family residences benefit from three types of ECMs, while 24.2% benefit from two ECMs. 18.3% have installed four types of ECMs, compared to 3.7% who have installed only one ECM. 5.0% of homes benefit from five different types of ECMs, compared to 0.9% of homes who have taken advantage of six ECMs. No participant has taken advantage of all seven ECMs available because the HVAC replacement and HVAC tune-up are mutually exclusive.

Source: Seidman Research Institute analysis

Figure 14 summarizes the front-facing direction of the participating properties. This shows that 48% front-face south and 42% front-face north.

Landscaping and pool data is not recorded by 59.4% of all residential participants. Focusing exclusively on those residents answering these questions, 40.4% have a pool and 59.6% do not; and 10.1% have a desert landscape, 48.3% a grass landscape, and 39.3% a mixture of desert and grass.

65.3% are one-story buildings, 12.8% two-story buildings and the balance unrecorded.

Single-family residences are offered seven principal types of ECM – air sealing, duct sealing, HVAC replacement, HVAC

Energy Efficiency on an Urban Scale

96

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 16: TOTAL NUMBER OF ECMS BY SINGLE-FAMILY RESIDENCE

The contractors’ total estimated annual energy savings for single-family residences recorded in the energy audits ranges from 832 kWh to 12,281 kWh, with a mean of 3,208 kWh and a median of 2,662 kWh. This is a more reliable statistic than the means provided for commercial participants, although house size again is a key contributory factor.6 The standard deviation for the entire single-family residential sample’s estimated energy savings is 1,685 kWh. Table 7 illustrates the mean and median contractors’ estimated annual energy savings by number of ECMs implemented. In absolute terms, the more ECMs installed, the greater the contractors’ mean and median estimated energy savings.

Source: Seidman Research Institute analysis

TABLE 7: CONTRACTORS’ ESTIMATED ANNUAL ENERGY SAVINGS (KWH) BY NUMBER OF ECM INSTALLATIONS

Figure 17 illustrates the different combinations of single and multiple ECMs implemented at single-family residences.

The most popular combination is air sealing, duct sealing and insulation (38.8%). Other popular combinations are air sealing, duct sealing, insulation, and shade screens (11.0%), air sealing and duct sealing (7.3%), and duct sealing and water heater (6.8%). There are no single instances of shade screen ECMs, HVAC replacements, HVAC tune-ups, or water heaters. Any multiple ECM combination not featured in Figure 17 has also not been implemented as part of the Energize Phoenix program.

Source: Seidman Research Institute analysis

Table 8 examines the contractors’ estimated annual energy savings per single or multiple ECM option. It is difficult to draw many conclusions from this table, as the sample sizes for most ECM combinations are statistically very low. The air, duct and insulation package is an exception, with mean estimated energy savings of 2,626 kWh and a median of 2,349 kWh. The air, duct, insulation, and shade screen combination has mean estimated energy savings of 3,937 kWh and a median of 3,573 kWh. The sample sizes for all other combinations are less than 20.

FIGURE 17: TOTAL UPTAKE OF SINGLE AND MULTIPLE-ECM SINGLE FAMILY RESIDENCE OPTIONS

The mean estimated energy savings for all single or multiple ECM combinations is 2,981 kWh, with a median of 2,349 kWh and range of 832 - 12,281 kWh. Table 9 illustrates the mean and median total project cost by number of ECMs installed. This strangely suggests that the mean and median cost of 2 ECMs is greater than 3 ECMs. The reason for this apparent anomaly is that 15 of the 53 projects with 2 ECMs include water heaters (including solar water heaters), and these projects are recorded as costing $10,104.25 on average.

Source: Seidman Research Institute analysis

There is a correlation between conditioned volume of a single-family residence and a contractor’s estimated energy savings – Pearson correlation coefficient of 0.315 with a two-tailed significance of <.000) 6

Energy Efficiency on an Urban Scale

97

Global Institute of Sustainability

Back to Main Table of Contents

TABLE 8: CONTRACTORS’ ESTIMATED ANNUAL ENERGY SAVINGS (KWH) BY TYPE OF SINGLE OR MULTIPLE ECM INSTALLATION

TABLE 9: TOTAL PROJECT COST ($) BY NUMBER OF ECMS INSTALLED

Source: Seidman Research Institute analysis

Table 10 summarizes the mean and median cost of specific ECM combinations, and the cost per estimated annual kWh savings. Once again, caution should be exercised drawing conclusions from this table given the low level of participation for many of the ECM combinations listed. The mean cost of all ECM combinations implemented is $4,379.54, with a median of $2,778.64 and range of $250 - $20,754.

Table 10 also divides the mean total project cost for each single or multiple ECM by the contractors’ mean estimated annual energy savings to arrive at an estimated cost for first year kWh savings for each combination of ECMs implemented.

This ranges from 30 cents for air sealing to $4.28 for Energize Phoenix’s residential duct sealing and HVAC replacement. The mean project costs per mean estimated annual kWh saved for all ECMs implemented is $1.13.

Conditioned volume data in cubic feet is missing for approximately a quarter of all single family residence properties. The mean conditioned volume for the data supplied is 14,956 cubic feet, with a median of 14,012 cubic feet and range of 1,000 - 33,000 cubic feet.

Table 11 describes the Energize Phoenix program rebate amount given by ECM type.7 The small sample size for almost all of the ECM combinations listed again negates any possibility of single or multiple ECM-specific conclusions.

Nevertheless, the mean Energize Phoenix program rebate for all ECMs is $2,243.59, with a median of $1,365.97 and range of $62.50 - $14,609.94.

Source: Seidman Research Institute analysis Program rebate is the amount paid by the City of Phoenix. It does not include any customer rebate received from APS. 7

Energy Efficiency on an Urban Scale

98

Global Institute of Sustainability

Back to Main Table of Contents

TABLE 10: TOTAL PROJECT COST ($) BY TYPE OF ECM INSTALLATION

TABLE 11: FINAL REBATE ($) BY TYPE OF ECM INSTALLATION

Source: Seidman Research Institute analysis

Source: Seidman Research Institute analysis

Energy Efficiency on an Urban Scale

99

Global Institute of Sustainability

Back to Main Table of Contents

3. AGGREGATE ENERGY USAGE

Figure 18 requires careful interpretation. The two time series are not coincident because the aggregate usage for commercial buildings in the entire City of Phoenix APS territory is based on billing data. The difference between when the meter is read and when the usage appears on the customer’s bill is reflected in the leftward shift of the EP Commercial Gross time series. The series for EP Commercial Gross traces aggregate energy usage for participants but the data set is incomplete for some participants for some of the periods. For example, in the earlier months of Figure 18, there are 22% fewer participants included in the aggregate calculations than at comparable months near the end of the sample period.

This biases aggregate energy usage downwards for the early periods which, in turn, mask the effects of ECMs on aggregate energy consumption in later periods. Further, ECMs are completed at different points in time for each participant, so their effect is an emerging one in Figure 18. To allow for these issues and also the seasonal variation in energy usage evident in the figure, a twelve-month moving average for aggregate energy usage is calculated in Figure 18. This demonstrates an obvious rise in energy usage beginning in 2011 and a slight decline starting during the middle of 2012. It is difficult to attribute any decline in energy usage to the ECMs because Figure 18 does not account for variations in temperature from year to year. These effects are addressed in Section 4.

The following are analyses of energy usage using data provided by Arizona Public Service Company (APS) for each commercial and residential program participant. The raw data contains the following pieces of information: • A code identifying each participant.

• The date the meter is read.

• The number of days in the period corresponding to the read date.

• The kWh consumed during each period.

To construct a (calendar) monthly time-series for each participant, the mean daily energy usage for each meter reading period is calculated. Metering read periods are the number of days between meter readings which vary from month to month and generally do not coincide with calendar months. Aggregate calendar month usage data is therefore constructed using summations of mean daily usage for days falling within each respective calendar month.8 Commercial An index of aggregate energy usage across all commercial participants in the program (n = 424) and across all commercial buildings within APS territory in the City of Phoenix is illustrated in Figure 18. The index is constructed using January 2008 energy usage as the base month for both series.

FIGURE 19: CITY OF PHOENIX VS. EP COMMERCIAL PARTICIPANTS WITH COMPLETE TIME SERIES DATA - AGGREGATE MONTHLY ENERGY CONSUMPTION

FIGURE 18: CITY OF PHOENIX COMMERCIAL VS. ALL EP COMMERCIAL PARTICIPANTS AGGREGATE MONTHLY ENERGY CONSUMPTION

Source: Seidman Research Institute analysis

Figure 19 illustrates the aggregate energy usage for commercial participants with a complete time-series of energy usage data (n = 304). That is, their energy usage for every month from January, 2008 through April 2013 is available.9

Source: Seidman Research Institute analysis

For example, read period 1 may run from January 20th through February 10th with an average daily energy usage of 100 kWh. Read period 2 may run from February 11th through March 5th with an average daily energy usage of 150 kWh. So, to calculate the aggregate usage for the month of February, we add the first 10 days in the month at 100 kWh/day (1,000 kWh) to the last 18 days in the month at 150 kWh/day (2,700 kWh) to arrive at an aggregate usage of 3,700 kWh for the month of February.

8

Similar to Figure 18, a 12-month moving average for commercial participants is calculated in Figure 19. This demonstrates a clear downward trend in energy usage

Out of 424 commercial participants, complete time series data is only available for 304 participants.

9

Energy Efficiency on an Urban Scale

100

Global Institute of Sustainability

Back to Main Table of Contents

FIGURE 21: CITY OF PHOENIX RESIDENTIAL VS. EP RESIDENTIAL PARTICIPANTS WITH COMPLETE TIME SERIES - AGGREGATE MONTHLY ENERGY CONSUMPTION

throughout the entire time period. While the issue of missing observations is eliminated from Figure 19, other effects such as weather and the economy remain unaccounted for. To isolate the effects of energy efficiency measures on energy usage, a more sophisticated econometric model is required.

This is discussed in detail in Section 4.

Single-Family Residential Aggregate energy usage across all single-family residential participants (n = 219) is illustrated in Figure 19.

FIGURE 20: CITY OF PHOENIX VS. ALL EP RESIDENTIAL PARTICIPANTS AGGREGATE MONTHLY ENERGY CONSUMPTION

Source: Seidman Research Institute analysis

Figure 21 illustrates the energy usage for residential participants with energy usage data available for every month from January, 2008 through April, 2013 (n = 141).10 A 12-month moving average exhibits no discernible trend in energy usage for residential program participants. As with the commercial figures, isolation of the effects of ECMs on energy usage requires an econometric analysis, which is discussed in Section 4.

4. ENERGY USAGE CHANGES –INFERENTIAL TESTS AND ECONOMETRICS

Source: Seidman Research Institute analysis

Commercial – Inferential Testing Figure 20 also requires careful interpretation as the singlefamily residential data is incomplete. That is, there are 33% fewer participants included in the aggregate calculations at the beginning of the time horizon than in comparable months at the end. This biases aggregate energy usage downwards for the early periods which, in turn, masks the effects of ECMs on aggregate energy consumption among participants in the later periods. ECMs are also completed at different points in time for each participant, so the effect of ECMs is an emerging one.

The 12-month moving average in Figure 20 reveals a positive general trend in energy usage largely due to the small number of observations at the beginning of the time series.

To seek broad evidence of a statistically-significant reduction in energy usage, inferential statistical tests are conducted examining differences in mean energy usage before and after the installation of ECMs for all participants. In particular, the mean energy usage in mean watt hours per square foot per day consumed by participant is calculated, matched in pairs across the sample and then compared with the same time horizon in the prior year for three distinct periods: • Months 1-3 immediately after the installation of the ECM(s).

• Months 4-6 immediately after the installation of the ECM(s).

• Months 1-6 immediately after the installation of the ECM(s).

The mean results are reported in Table 12.

The omission of all projects with erroneous or incomplete read data reduces the sample to 369 participants. Additionally, omitting projects that lack at least 6 months pre- or postECM data reduces the sample to 144 commercial participants.

The same 144 participants are used in all three inferential statistical tests.

Complete time series data (January 2008 through April 2013) is only available for 141 of the residential participants.

10

Energy Efficiency on an Urban Scale

101

Global Institute of Sustainability

Back to Main Table of Contents

TABLE 12: COMMERCIAL PRE-/POST-ECM ENERGY USAGE ANALYSIS

Commercial – Econometric Analysis To investigate more fully the effects on energy usage reductions among the commercial participants, panel data econometric methods are used to estimate the shortand long-run effects of ECMs while controlling for the heterogeneity among participants and seasonal variations in weather conditions.

Source: Seidman Research Institute analysis

Table 12 shows a decline in mean energy usage for months 1-3, 4-6, and 1-6 pre- and post-retrofit. For instance, in the first 3 months post-retrofit, a reduction of 9.77 mean watt hours per square foot per day is estimated compared to the equivalent pre-retrofit period. This decreases in months 4-6 to a reduction of 8.06 watt hours, but nevertheless remains evidence of the effectiveness of the installation of ECMs in the commercial sector. This conclusion is based on a series of inferential tests that do not make any allowance for other influences on energy usage (such as, for instance, the weather). This is the subject of the econometric work included later in this report.

There is usable cross sectional data on 342 participants for 64 (approximately) monthly read cycles between January 208 and March 2013 – that is, a maximum of 21,888 observations.

However, this is an unbalanced panel because there is missing data for some of the cross-sectional variables and time series energy usage. After extensive examination of alternatives, watt hours per square foot per day (WHSQFTDAY) is used as the dependent variable. The model that performs best in terms of its statistical properties is one that subsumes the cross-sectional elements within a series of intercept dummies through a fixed effects panel least squares estimation of the form outlined in Table 14.

A series of hypothesis tests using the matched pairs across the sub-samples for different ECMs are also conducted on the changes in mean watts house per square foot per day consumed. The relevant p-values are reported in Table 13. For cases where lighting alone and where controls and pumps/motors are installed, the effect of the ECM is weaker though time as evidenced by the higher mean difference for months 1-3 compared to months 4-6. The mean differences for controls and lighting or controls-only are not statistically significant. Some caution is required interpreting these mean energy usage statistics, due to the number of low populations in some categories, and the annual variations in weather patterns.

The explanatory variables that emerge from an extensive modeling process are a one-period lag of WHSQFTDAY, a 0,1 dummy to reflect the month of ECM installation by participant (RETROFIT), and the number of cooling (CDD) and heating (HDD) degree days per period.

TABLE 14: COMMERCIAL FIXED EFFECTS MODEL Source: Seidman Research Institute analysis

The commercial fixed effects model has excellent explanatory power, non-troublesome errors, and all of the included explanatory variables are significant.

The coefficient values suggest the initial effect of a retrofit is a reduction of 3.24 watt hours per square foot per day for commercial participants. In the long-run equilibrium, the reduction is estimated at 10.79 watt hours per square foot per day.

TABLE 13: COMMERCIAL PRE-/POST-ECM ENERGY USAGE ANALYSIS BY ECM

Using the mean value of the dependent variable as a base, this suggests the short-run effect of a retrofit on commercial participants is a 5% reduction in energy usage immediately after the implementation of one or more ECMs, and a long-run reduction of approximately 17%.

Source: Seidman Research Institute analysis

Single-Family Residential – Inferential Testing To ascertain the existence of a statistically-significant reduction in single-family residential energy usage, statistical tests are performed in a similar fashion to the commercial Energy Efficiency on an Urban Scale

102

Global Institute of Sustainability

Back to Main Table of Contents

sector on residential participants’ mean energy usage preand post-ECM implementation. Mean billed kilowatt-hours per day by period11 by participant are calculated, matched in pairs across the sample for months 1-3, 4-6 and 1-6 after ECM implementation, for comparisons with the same time horizon in the prior year. The results are reported in Table 15.

extensive examination of alternatives, read kilowatt-hours per day per period (READKWHDAY ) are used as the dependent variable,12 and a variety of model specifications examined in terms of explanatory variables. The model that statistically performs best is one that subsumes the cross-sectional elements within a series of intercept dummies through a fixed effects panel least squares estimation of the form outlined in Table 16.

Omitting all projects with erroneous or incomplete read data and projects with less than 6 months of pre- or post-ECM installation data reduces the sample size to 138 participants.

These 138 participants are used for all three inferential tests.

The explanatory variables emerging from an extensive modeling process are a one-period lag of READKWHDAY, a 0,1 dummy to reflect the month of ECM installation by participant (RETROFIT ), and the number of cooling (CDD ) and heating (HDD ) degree days per period.

The mean difference between mean energy usage during the first 3 months after the ECM installation and the corresponding 3 months during the previous 12 months is 2.60 kWh per day. A p-value of 0.01 suggests that this is significantly different from zero. The difference between the second 3 months and the corresponding time periods one year prior does not produce a statistically significant mean difference between the two periods.

The model has excellent explanatory power, non-troublesome errors, and all of the included explanatory variables are highly significant.

The coefficient values suggest the initial effect of the installation of one or more ECMs is a reduction of 2.57 kWh per day, and a long- reduction of 4.72 kWh per day.

TABLE 15: SINGLE-FAMILY RESIDENTIAL PRE-/POST-ECM ENERGY USAGE ANALYSIS

TABLE 16: SINGLE-FAMILY RESIDENTIAL FIXED EFFECTS MODEL Source: Seidman Research Institute analysis

Using the mean value of the dependent variable as a base, the short-run effect of the installation of one or more ECMs is therefore estimated at a 6.7% energy usage reduction in the immediate aftermath of the installation of ECM(s), and a long-run reduction of approximately 12% for residential participants.

Source: Seidman Research Institute analysis

A comparison of months 1-6 pre- and post-ECM installation yields a statistically significant mean energy usage decrease of 1.91 kWh per day.

The small sample size makes inferences difficult to draw.

However, the data suggests there are some dynamics associated with the post-ECM installation period for the single-family residential sector.

5. COST PER ACTUAL KWH AND EMISSIONS SAVED The econometric analysis of the commercial and residential panels yields long-run estimates of watt hours per square foot per day for commercial energy savings, and daily kilowatthour energy savings by residential participant. These are used to estimate projected annualized energy savings across all commercial and single-family residential participants.

Single-Family Residential – Econometric Analysis To investigate in greater detail the effects of ECMs on energy usage reductions in the single-family residential sector, panel data econometric methods are used to estimate the short- and long-run effects of the ECM(s), controlling for the heterogeneity among participants and seasonal variations in weather conditions.

The long run energy savings per commercial participant are 10.79 watt hours per square foot per day. The total estimated commercial square footage covered by the Energize Phoenix program through September 2013 is 34,105,096 square feet.

There is usable cross sectional data for 217 participants for 64 monthly read cycles between January 2008 and March 2013 – a maximum of 13,888 observations. However this is an unbalanced panel because there is missing data for some cross-sectional variables and time series energy usage. After Energy Efficiency on an Urban Scale

Buildings in the residential sector do not exhibit as much variation in their energy usage due to size. Thus, there is no need to standardize energy usage by dividing it by the area of the building. However, the variation in the number of days in the month is taken into account by dividing kWh by the number of days in each calendar month.

11

12

103

There is no accurate data on the square footage encompassed within each ECM.

Global Institute of Sustainability

Back to Main Table of Contents

Thus, the projected annualized savings across all commercial participants in the long-run is estimated to amount to approximately 134.32 GWh per year.

Estimating the ROI for Energize Program is difficult without an appropriate discount rate, forecasts for energy prices, or estimates of the length of time over which the assessed energy savings accrue. Energize Phoenix’s annualized dollar savings can be calculated assuming residential prices of 10.73c/kWh and commercial prices of 9.35c/kWh.17 These are also shown in Table 17. The program is estimated to save over $12.6MM in energy costs per annum across both the residential and commercial sectors for the participants.

The long-run energy savings per residential participant are 4.72 kWh per day. There are an estimated 400 residential participants through the end of the program. Thus, the annualized energy savings across all participants in the longrun is estimated to be approximately 689.12 MWh per year.

There is an immediate short-run energy reduction effect when an ECM is implemented which increases over time until the long-run effect is reached, which takes into account the effect from all previous time periods. For this reason, the first 12 months after ECMs are installed yield the slightly smaller energy savings estimates in the commercial and residential sectors of 118.13 GWh per year and 688.66 MWh per year, respectively.

Any assessment of the energy savings and their value needs to be examined in the context of the cost of the Energize Phoenix program. With this in mind and looking at one year’s worth of energy savings in the commercial sector, a total of $0.35 per kWh was invested by all parties combined. For single-family residential participants, the investment was $3.18. The EP program incentive costs per kWh saved in a year (that is, only taking into account the amount spent by the Energize Phoenix program on commercial and residential ECMs) are $0.12 and $1.42, respectively18. Taking into account all program expenses, including Administrative, Commodities & Training for both commercial and residential programs, the total cost per one year of kWh savings is $0.18.

One possible explanation in the commercial sector for the difference between the short-run and long-run effects of having an ECM installed is that it may require a few months to properly calibrate new and old equipment in the post-ECM environment. In the residential sector, the increase in the effect of ECMs through time may be explained by behavioral changes. More research is required to validate the change in energy savings between the short-run and long-run effects of the model.

Including just annual energy savings generated as the return from the program and using a discount rate of zero, the payback periods for total investments in the single-family residential and commercial sectors are estimated at 29.6 years for one year of energy savings for the single-family residential sector, and 3.8 years for the commercial sector.

The payback period for program incentives is estimated at 13.2 years for the residential sector and 1.3 years for the commercial sector. Taking into account all program costs, the payback period is estimated at 1.9 years. Comparativelyspeaking, the commercial part of the Energize Phoenix Program therefore appears to be more successful.

Table 17 illustrates details the annualized energy savings, costs and CO2e reduction associated with the Energize Phoenix Program.

TABLE 17: ENERGIZE PHOENIX PROGRAM ANNUAL ENERGY SAVINGS, COST PER KWH SAVED AND CO2E EMISSIONS

Table 17 also calculates the carbon emissions saved. This initially applies APS’ annual coal and natural gas power plant emissions described on the EPA Flight website (2013), alongside19 lifecycle GHG/CO2e estimates for nuclear and 13

Commodities and training include office supplies, BPI training costs and legal expenses.

This is assessed using energy prices of 9.35c per kWh for the commercial sector and 10.73c per kWh for the residential sector. Source: U.S. Energy Information Administration, Form EIA-861, “Annual Electric Power Industry Report.”

14

Source: Seidman Research Institute analysis

Energy Efficiency on an Urban Scale

104

15

This assumes a discount rate of 0%.

16

This assumes a discount rate of 0%.

17

These are average energy costs for Arizona provided by the EIA.

18

These figures do not take into account Administrative, Commodities & Training expenses.

19

Source: http://ghgdata.epa.gov/ghgp/main.do

Global Institute of Sustainability

Back to Main Table of Contents

renewables from the NEA (2012) and IEA (2008), to calculate APS’s direct GHG emissions per year at 10,911,301 CO2e metric tons.20 Using the coefficients from the two regression models for commercial and single-family residential participants, the GHG emission reductions attributable to the Energize Phoenix program are therefore estimated at 486 metric tons for the residential program, and 94,769 metric tons for the commercial program.

A combined 95,256 metric tons of CO2e saved is equivalent to: • Annual greenhouse gas emissions from 19,845 passenger vehicles; • CO2 emissions from 1,256 tanker trucks’ worth of gasoline; and • CO2 emissions from the electricity use of 14,260 homes for one year.

Sources: Nuclear Energy Agency (NEA), (2012), The Role of Nuclear Energy in a LowCarbon Energy Future, OECD/NEA 6887, Paris, France; and IEA (2008), Solar PACES Annual Report for 2008, edited by C. Richter, IEA Solar Power and Chemical Energy Systems, available at: http://www.solarpaces.org/Library/AnnualReports/docs/ATR2008.pdf.

20

Energy Efficiency on an Urban Scale

105

Global Institute of Sustainability

Back to Main Table of Contents

INTRODUCTION

APPENDIX J ANALYSIS OF THE ENERGIZE PHOENIX FINANCE PROGRAMS

This report, undertaken in response to a request for additional research from Energize Phoenix (EP) program staff, analyzes the Energize Phoenix Finance Programs with a primary focus on the Commercial Loan Program and a secondary emphasis on the Energy Assist 60/40 residential loan program. The goals of the analyses were to determine potential reasons for unexpectedly low participation rates, as well as lessons learned and recommendations that might assist other organizations that may contemplate establishing energy efficiency finance programs.

TABLE OF CONTENTS INTRODUCTION 106 KEY RESULTS – COMMERCIAL LOAN PROGRAM

106

KEY RESULTS – ENERGY ASSIST 60/40 LOAN PROGRAM

107

BACKGROUND 107 COMMERCIAL LOAN PROGRAM CHRONOLOGY

108

KEY RESULTS – COMMERCIAL LOAN PROGRAM

PROGRAM DESIGN AND DEVELOPMENT

108

RECESSION LENDING MARKET

108

• The EP Commercial Loan Program placed six loans totaling $707,055 – four to auto dealerships and two to manufacturers. Zero loans were in default as of October 16, 2013. As a function of the structure of its banking arrangement with National Bank of Arizona (NBAZ), the City of Phoenix (CoP) provided $210,172 in loan funds (30% of the total loan value). Further, the CoP posted $106,058 in a Loan Loss Reserve fund (15% of the total loan value) and covered $30,000 in bank charges (over and above the $731K in loans). Thus, the total CoP outlay for the EP program was $346,230, of which $316,230 plus earned interest is to become available for future loans upon successful repayment of the initial six loans.

This level of loan activity and associated dollar volume, either in conjunction with NBAZ or independently, is likely insufficient to create and sustain the revolving loan program that is a major EP goal.

MARKETING

109

PROGRAM ADJUSTMENTS

109

ACTIVITY AND RE-ALLOCATION

110

THE LOAN SURVIVORS

110

BORROWERS SPEAK

111

DECISION FACTORS FOR NONPARTICIPATING COMMERCIAL ENTITIES

113

COMMERCIAL CONTRACTORS’ PERSPECTIVES

113

CONCLUSIONS

113

LESSONS LEARNED

114

RECOMMENDATIONS

114

ENERGY ASSIST 60/40 LOAN PROGRAM

114

BORROWERS SPEAK

115

A RESIDENTIAL CONTRACTOR’S PERSPECTIVE

116

LESSONS LEARNED

116

RECOMMENDATIONS

116

SUMMARY CONCLUSION

116

• Four of the six borrowers – representatives of three auto dealerships and one manufacturer – opted to participate in the follow-up interview process.

• In all but one case among the four interviewees, the borrowers had the ability and willingness to self-fund their energy upgrade projects. For those three borrowers, the EP loans were consummated in order to gain access to EP rebates that otherwise were not accessible by the borrowers because of geographic program restrictions.

In two cases, the borrowers posted cash collateral for the loans in order to avoid intrusive aspects of the underwriting process.

• The Commercial Loan Program was de-funded incrementally beginning Fall, 2012, due to lack of market uptake, and all but $357,211 of the remaining funds had been re-allocated to commercial rebates by the time of this analysis. NBAZ reported that the insufficient loan volume also created a loss to the bank.

Energy Efficiency on an Urban Scale

106

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

• The possibility of potentially expensive and intrusive federal auditing that accompanies federal funding can serve as an impediment to private/public sector cooperation in federally-sponsored programs. Program staff reported that delays in concluding CoP agreements with both APS and NBAZ were caused by the need to define mechanisms to assure that neither APS nor NBAZ would come to have control over any federal funds.

relative participation of EP homeowner grant recipients (thirty) vs. only three homeowner loan participants.

BACKGROUND In 2010, the City of Phoenix (CoP), in conjunction with Arizona Public Service (APS) and Arizona State University (ASU) applied for a $75 million award from the U.S. Department of Energy (DOE) for an urban-scale energy efficiency upgrade program dubbed “Energize Phoenix” (“EP”). The $25 million eventually awarded was allocated among several different city departments and ASU for the purposes of managing a variety of subprograms, as well as marketing, administrative costs and overall program evaluation costs. Consistent with APS’ desires, none of the federal funds were allocated to APS. Separate APS energy efficiency rebates, which played an integral role in the EP subprograms, were funded by an Arizona Corporation Commission-approved set aside of ratepayer funds.

• In a period of financial turmoil and economic uncertainty, prospective program participants’ behavior indicates a strong preference to receive rebates over the obligations attendant to taking out a loan, even if offered under favorable terms. This was demonstrated conclusively by the relative participation of over 500 commercial rebate recipients vs. only six borrowers.

• Funding cycles for federally-funded programs do not necessarily coincide with the capital improvement review processes of private businesses, lowering the pool of potential participants.

The importance of the EP Commercial Loan Program to DOE and City of Phoenix is reflected in the stated objectives of the DOE’s Funding Opportunity Announcement:

• Assuring consistent local staffing to respond to federal programs of relatively short duration is a challenge for local grant recipients.

• Energy savings: Deliver verified energy savings through energy efficiency upgrade projects

KEY RESULTS – ENERGY ASSIST 60/40 LOAN PROGRAM

• Increased participation: Achieve broad market participation from a variety of residential, commercial, industrial, and public customers

• The Energy Assist 60/40 Loan Program placed three loans totaling $33,808 to homeowners within the 7th Street to 7th Avenue EP Corridor. All loans were outstanding as of October 16, 2013. Zero loans were in default. As a function of its servicing arrangement with Neighborhood Housing Services, Inc., the CoP provided all of the loan funds. Thus, the CoP total outlay for the program was $33,808, of which $33,808 plus earned interest would be available for future loans upon successful repayment of these three. This level of loan activity and associated dollar volume is likely insufficient to create and sustain a revolving loan program that is a major CoP goal for the overall EP program.

• Economies of scale: Demonstrate the benefits of gaining economies of scale • Enhanced resources: Enhance the resources available to support energy efficiency upgrades by effectively leveraging award funding • Financial sustainability: Design a viable strategy for program sustainability beyond the award period • Replicable pilot programs: Exemplify comprehensive community-scale energy efficiency approaches that could be replicated in other communities across the country as well as in the goals of the Energize Phoenix proposal:

• The main beneficiary of the limited participation in the 60/40 loan program, as well as the 60/40 grant program, was the multi-family program, to which under-utilized funds were re-allocated in support of its goals.

• Upgrade 1,700 residential units for greater energy efficiency and reduce energy consumption for residential participants by 30%

• In a period of financial turmoil, economic uncertainty, and substantial mortgage obligations, prospective homeowner program participants’ behavior indicates a strong preference to receive rebates over the obligations attendant to taking out a loan, even if offered under favorable terms. As was the case with the EP Commercial Loan Program, this preference was evidenced by the Energy Efficiency on an Urban Scale

• Upgrade 30 million square feet of office and industrial space for greater energy efficiency and reduce energy use for commercial participants by 18% • Cut carbon emissions by as much as 50,000 metric tons per year 107

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

• Leverage federal funds 5:1 with other investment

for nonresidential customers who wanted to finance energy efficiency projects for existing buildings. The minimum loan size for appropriate upgrade projects, after subtracting APS and Energize Phoenix incentives, was $50,000. Participants could obtain low, fixed–interest rate loans for 12 months to 120 months. Collateral was sometimes required, depending upon loan size, term, and underwriting requirements.

• Create 1,000 direct and indirect jobs • Create a sustainable revolving loan fund to perpetuate the program beyond the three-year award period As the funding for the program was derived from the American Recovery and Reinvestment Act of 2009 (also known as the “Stimulus” program), an additional federal goal was to expend all funds as soon as practicable in order to quickly stimulate the local economy.

Energize Phoenix and National Bank of Arizona jointly funded the program. 70% of each “participation loan” was funded by private bank capital and 30% by EP capital. The program was further aided by a cumulative loan-loss reserve that supported both lenders in case of a default. The cumulative loan-loss reserve was funded by EP at a rate equal to 15% of each loan issued.

The City of Phoenix planned to design and implement a commercial loan program that could operate on a “revolving loan” basis. That is to say, the loan program would be of sufficient size that it could generate repeated instances of economic stimulus, as first-generation loans were repaid and then loaned to subsequent generations of borrowers. It was anticipated that such a successful revolving loan program could extend the economic benefit to the community beyond the one-time impact that rebates represent. Additionally, financing allows owners to overcome the challenge of upfront capital investment, instead more closely aligning monthly payments with resulting monthly utility savings.

Recession Lending Market The core purpose of the Energize Phoenix Finance Program was lending money in order to facilitate the installation or application of energy saving measures that otherwise would not be implemented. The loan program was launched and operated at the same time that the economy was in a serious and deep recession. This had several implications for the program.

At the time the program launched, both residential and commercial lending was highly restrained by very strict underwriting criteria and by a general attitude in the banking industry to only lend to borrowers who had pristine credit. Lenders also sought more than sufficient collateral and increased debt service coverage requirements (ability to service a loan). This lending mentality prevailed at the same time that individuals and companies were impacted by reductions in their cash flow, which created further credit problems. The result was that very few loans were made and those potential borrowers who could meet the stricter underwriting guidelines were less likely to subject themselves to the rigors of the underwriting process.

COMMERCIAL LOAN PROGRAM CHRONOLOGY Program Design and Development While the City of Phoenix initially hoped to roll out both the EP commercial rebate program and the commercial loan program (which was linked to both APS and CoP rebate components) simultaneously, this did not occur. The commercial rebate program commenced in February, 2011, following joint program design and the execution of a Memorandum of Understanding with APS. A Request for Proposal (RFP) for commercial loan services was issued by the CoP Economic and Community Development (ECD) department in October, 2010 and National Bank of Arizona was the only RFP respondent. Negotiations, documentation and planning over a period of approximately 14 months resulted in an agreement and then the launch of the program. The range of issues explored in the negotiations between the parties is not clear but the process was reportedly slowed down by determination of the appropriate size of a loan-loss reserve, the segregation of funding sources to avoid potentially intrusive federal audits, lengthy document turnaround times, and other issues.

The recession and ensuing economic recovery also took place when there was, and remains, very significant amounts of cash held by many large and small corporations. In contrast, other more recent recessions were cash-less. In fact, the existence of significant cash reserves has created continued issues with recovery, because it has kept downward pressure on debt pricing due to competition for the best assets at returns that are historically low.

This environment persisted during the entire time period of the EP program. Offsetting these negative lending conditions were very low interest rates. Those who could borrow had access to very inexpensive capital. This fact did little to motivate

The loan program was structured as follows: The Commercial Loan Program provided a revolving loan fund Energy Efficiency on an Urban Scale

108

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

borrowers. Those who were financially healed either did not need the capital, the process for accessing it was more trouble than they felt it was worth or they did not want to incur debt and increase risk due to the uncertainty of the economic future. Those who needed the loans were damaged by the recession and could not meet stricter underwriting criteria. So, regardless of access to inexpensive capital, those who needed it could not qualify and those who qualified did not need it.

Therefore, the loan program was not, by itself, a motivator to participate in the upgrades as offered by the EP program. This applied to both commercial and residential borrowers.

In December 2011, upon completion of the lending agreement with NBAZ, ECD transmitted the first of several email blasts to real estate brokers, property owners and tenants within the original EP target geographical area. Staff reported that the email blast generated no activity/interest at all from those who were contacted. However, ECD staff did receive anecdotal feedback to the effect that “It looked like one hand (the loan program) didn’t know what the other hand (the earlier commencing rebate program) was doing.”

Also in December 2011, the geographic boundaries for both the commercial and residential EP programs were expanded to include 7th Street to 7th Avenue between Jefferson and Missouri, and from Airlane on the south to the S.R. 202 freeway on the north, extending east to S.R. 143. The expansion of the target area was predominantly driven by a desire to increase the number of single family homes eligible to participate in EP programs. An additional impetus was the desire to include the entirety of the neighborhoods located east and west of Central Avenue, which had previously been subdivided when the boundaries only encompassed the area between 3rd Street and 3rd Avenue.

Regardless of the interest rate, access to borrowed capital requires scrutiny imposed by underwriting criteria, some applied by individual lenders and most imposed by Federal regulatory agencies. There is no escaping the fact that commercial lending requires a sometimes long and arduous process of documentation and review. This process is time consuming and frustrating, especially in an environment when underwriting criteria require near perfect compliance.

Debt also increases risk. At a time when there is great uncertainty with the global and local economies, increasing debt meant increasing risk. Thus, the uncertain economy meant that those who could qualify for a loan likely would hesitate to increase risk. In many cases, those who qualified did not need the capital and therefore would not undergo the time and scrutiny associated with underwriting.

ECD staff transmitted another email blast in March 2012 that also evoked no response. The EP team thinking evolved to the point of considering further expansion of the geographic eligibility boundaries.

Of the forty mentions of Energize Phoenix in press publications, eight related directly to the EP Commercial Loan Program as a result of reporting related to the press releases issued by the program. The original Energize Phoenix Commercial Finance announcement and the EP Corridor boundary expansion announcement press release were sent out to eleven media outlets and sixteen community and professional organizations on December 11, 2011. The second announcement, related to program modifications described below, was issued on May 2, 2012.

Access to “cheap capital” has not been a motivating factor for those who could qualify. Those who could qualify likely already had access to cheap capital from other sources with whom they had existing relationships, or they had cash available to accomplish installation or application of energy saving devices. Others could not take advantage of the cheap capital because of underwriting requirements, i.e. they could not satisfy securitization requirements (due to a lack of tenants or due to existing loans encumbering their property) or would not realize financial benefits sufficient to warrant the underwriting process and perceived risks of borrowing in uncertain times.

Program Adjustments In May 2012, EP program staff adjusted program rules to allow for an increase in rebates up to $500,000 if a business applicant were to agree to take out a loan in the amount of $50,000 or more. The existing rebate limit of $200,000 was retained as the cap for rebate-only applicants. A third email blast was transmitted, this time to the expanded EP Corridor area, to announce the new program ground rules, and EP staff started to receive some interest from potential participants.

Marketing EP program staff placed primary responsibility for marketing both the commercial rebate and commercial loan programs with the energy contractors involved. A training session on the planned EP Finance Program for commercial contractors was held on June 20, 2011. Additionally, ECD staff contacted some of the large participating Energy Services Companies directly to encourage them to discuss the program with their prospective clients.

Energy Efficiency on an Urban Scale

The limited sentiment gleaned by staff from the market was: “Why take out a loan when there is free money available?”

109

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

EP staff perceived competition between the rebate-only and the loan-plus-rebate programs. Further, some EP staff felt that the loan program would have performed better had the EP boundaries been expanded earlier, as the rebate program had already marketed itself thoroughly in the original target area. As one market participant observed: “There was no way that an owner or landlord would have taken the loan financing without having a tenant [in his building]”, particularly if rebate funding were readily available.

expressed by businesses at/near the airport, several hospitals and a large financial services company.

As of September 30, 2012, EP required that interested parties submit real letters of commitment, including energy checkups.

The parties were also required to provide a letter from APS, indicating APS willingness to provide appropriate rebates. This yielded a total of eleven prospects, which in November were turned over by ECD staff to CoP’s Public Works Department for further processing/action. Six of those prospects followed through on the application process to then participate in the commercial loan program.

In June 2012, the decision was made to expand the EP Commercial Loan Program boundaries to include all of the City of Phoenix that falls within the APS utility service area. Though this was not communicated in a press release, it generated the first serious interest by the marketplace. The decision to expand citywide was spurred in part by a spring, 2012 site visit by Department of Energy officials. Based upon their then current assessment of the EP Commercial Loan Program, DOE suggested that EP take more aggressive measures to expend federal dollars by dramatically broadening the target area and by reallocating commercial loan funds to the commercial rebate program.

The Loan Survivors Of the six firms that did proceed to close on an EP commercial loan (and thereby secure the associated rebates), four were auto-related businesses, while two were manufacturing firms.

All six became geographically program-eligible only upon the loan program boundaries being expanded beyond the original and revised target areas to encompass all of APS service territory with the city.

Taking the loan in order to access capital that could not be obtained elsewhere was a reality for only one of the four firms interviewed. For the other three, borrowing the money was an unnecessary but required step in order to tap into the pool of rebates provided by the EP program. Each of the three had the financial capacity to self-fund their energy upgrades without outside financing.

Activity and Re-Allocation

By September, 2012, the EP team decided that the reallocation of loan dollars into rebates would commence at the end of September if no project applications materialized by that time. In the meanwhile, preliminary interest was

TABLE 1: ENERGIZE PHOENIX COMMERCIAL FINANCE PROGRAM PROJECTS

Source: City of Phoenix

Energy Efficiency on an Urban Scale

110

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

By late 2012, loan terms might include a 1- to 5-year payment period at a 3.99% interest rate with possible collateral or guarantee required (depending upon the financial profile of the borrower). So, hypothetically, if a borrower who could otherwise self-fund a $200,000 upgrade package were to contract for a $50,000 minimum loan for twelve months (with accompanying $100,000 in rebates) and then repay it when due, the borrower’s net cost to access $150,000 in capital would be $1,495.00 per year. Within the borrower pool, some chose to repay the loans even earlier than obligated, thus further reducing their net costs.

ECD staff had heard from another source of the dealership’s plans to undertake an energy upgrade project.

Borrowers Speak

Initially, the energy services company that both advised the dealership and was, in effect, the general contractor was not APS-certified. However, in order to participate in the EP program, it became APS-certified and EP-certified.

According to the Controller, ownership had already approved the project by tapping internal funding and was “ready to go” [forward] with it. The scope of work for the project included replacing all of the interior and exterior lights on a 16-acre, six building complex. Further, Dealership A placed controls on all HVAC units in order to lessen compressor run times, as well as voltage controls on all major electrical boxes to minimize electrical fluctuations.

A number of interviews were conducted in August, 2013, with key representatives of the borrowers – three auto dealerships and one manufacturer – who were participating in the commercial loan aspect of the Energize Phoenix program.

One auto dealership and one manufacturer borrower did not participate in the interviews. The interviews were based upon a seventeen-question survey. All respondents were very amenable to sharing their experiences though not all questions were uniformly answered.

The upgrade construction took place between August and October of 2012 at a total cost of $358,784. EP provided a loan to Dealership A in the amount of $50,000 for a 1-year term, at an interest rate of 3.99% (interest-only, CD secured).

APS and EP provided rebates in the amounts of $70,506 and $67,968, respectively, leaving Dealership A with an owner contribution in the project of $170,310.

In response to the question “How did you first hear about the Energize Phoenix Commercial Loan Program?”, two of the respondents indicated that they had heard of the program through electrical service companies, while the other two were made aware from phone calls from ECD staff.

The majority of the improvements consisted of replacing parking lot and building interior lighting. On the parking lots, 1,000W HID bulbs were replaced by a locally-manufactured fluorescent tube. The new tubes were each projected to have a 40,000 hour life span (with a current annual usage of 3,500 hours) and to use one third of the power consumed by an HID lamp. The building interiors saw a changeover to LEDs and the installation of updated fluorescent tubes.

While one of the four had a vision of creating a “green building” as a part of a rehabilitation effort, the other three were already in differing degrees of taking action on energy upgrades. In one case, the upgrade project had already been approved to use internal funding and was ready to go at the time it received the initial ECD staff phone call. Another was initially intending to only address more efficient lighting for an exterior area, as it had already installed a new control system there. However, once made aware of the benefits of the combination of loan and rebate aspects of the program, the business expanded the scope of upgrade to include its interior areas as well.

Pre-upgrade, the Controller indicated that the dealership’s, total monthly energy costs had ranged between $25,000 and $30,000 per month, while the post-upgrade total monthly cost comprises 50 – 55% of that amount, or $12,500 at the low end vs. $16,500 at the high end.

The contractor handled the EP application process on behalf of Dealership A and assured its compliance with the DavisBacon requirements.

The latter two firms cited above and one other had the internal financial capacity to fund the upgrades, yet saw an opportunity to leverage more rebate funding by taking out a loan. However, rather than undergo the personal credit review required by the lending process, two of the three owners opened certificate of deposit accounts at NBAZ for amounts equivalent to their respective loan values so as to fully collateralize the loans.

Dealership C

Dealership A

The Director of Facilities for Dealership C first heard about the EP Commercial Loan Program during a phone call from ECD staff. He indicated that Dealership C was already considering a changeover of its parking lot lights to LEDs. The Director summarized its transaction with Energize Phoenix stating: “It worked out really well [for the dealership].”

The Controller of Dealership A first heard of the EP Commercial Loan Program during a phone call from ECD staff. Evidently,

The upgrade project took place between March and May of 2013. The Director described the scope of the upgrade as

Energy Efficiency on an Urban Scale

111

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

replacing 43 acres of parking lot lighting (type 875 metal halide lamps) with LED lamps and a control system. The total project cost was $373,202, which was partially financed by a $50,000 EP loan for a 1-year term at an interest rate of 3.99% (interest-only). APS and EP provided rebates in the amount of $88,652 and $88,652 respectively, leaving Dealership C with an owner contribution to the project of $145,898.

$45,067 and $48,584, respectively, leaving Dealership D with an owner contribution to the project of $43,420.

The Manager estimated that the package of upgrades had yielded monthly savings of $4,400 per month. Further, under the ongoing maintenance provision of the lighting installation contract, the manufacturer will provide parts replacement for a 3-year period. The manager estimates that this will save Dealership D an additional $1,000 to $1,500 per month.

According to the Director, the economics of the transaction with EP (including rebates) yields a 1.8-year payback on Dealership C’s investment. Upon installation of the control system and new lighting, the energy cost savings were substantial when compared to the same time period in 2012.

For example, he stated the May through July 2012 energy costs approximated $45,000 for the lots, while similar period 2013 costs approximated $16,000. He expected even greater savings as the year progressed to the 2013 fall and winter months, with the accompanying longer nighttime utilization of the more efficient lighting.

Manufacturer A Manufacturer A’s owner selected an older building as its new home that required a broad range of physical upgrades. In the back of his mind, he held a vision of wanting to create a “green building” as a part of that rehabilitation program.

With that as one of his goals, his mechanical contractor made him aware that EP had funds available for energy efficiency upgrades.

The bulk of the upgrade scope of work focused upon replacing numerous rooftop HVAC units. The controller also recalled the installation of a number of “solar devices” as within the scope.

The Director indicated that the EP loan was not needed by the dealership to undertake the upgrade, but that it had to accept it in order to qualify for the rebates. As the owner was unwilling to have his personal income tax returns reviewed by the EP lender, the dealership secured the $50,000 loan by depositing $50,000 in an NBAZ account for the loan’s duration.

Completed during the month of November, 2012, the project cost for the energy upgrade totaled $424,494. EP provided a loan to Manufacturer A of $368,746 with a 5-year term at 3.99% (interest-only, UCC blanket lien1). The loan was funded in March, 2013. APS and EP provided rebates in matching amounts of $6,898, leaving Manufacturer A with an owner contribution in the project of $41,952.

He concluded by saying that if there had not been a rebound in car sales, their decision to participate in EP may have been different, possibly leading them to delay the work and self-finance the improvements. As it was, the dealership compressed its pre-construction activities in order to accommodate the EP completion timeline and avail itself of the rebates.

The Controller characterized the application process as a nightmare, as early on the contractor mistakenly represented to the company owner that the EP loan interest rate was only 1.99%. Even though the Controller found the application relatively easy with a simple, 1-page form, she recalled that NBAZ initially came back with a 4.75% rate quote, compounding the already uncomfortable situation. The rate was finally set at the correct 3.99% level.

Dealership D The Operations Manager of Dealership D became aware of the EP Commercial Loan Program in September, 2012, from the owner of an energy services company. His original intention was to solely address more energy efficient lighting upgrades within his 7-acre parking lot, as Dealership D had already installed a lighting control system there. The scope and location of work was expanded to include the replacement of metal halide fixtures with T-5 fluorescents in the showroom and service area.

Manufacturer A did not investigate any other financing methods, either before or after completion of the improvements. “We didn’t shop the financing deal around.”

However, the Controller thought that Manufacturer A’s total upgrade project could have ranged as high as $500,000 if there had been a 1.99% interest rate for the loan. The interest rate misunderstanding resulted in the installation of fewer efficiency upgrades.

The upgrade construction took place from mid-October to mid-November, 2012 at a total project cost of $204,923. EP provided a loan to Dealership D in the amount of $67,852 for a 3-year term at an interest rate of 3.99% (interest-only, unsecured). APS and EP provided rebates in the amounts of Energy Efficiency on an Urban Scale

U.C.C. Blanket Lien — A blanket lien under the Uniform Commercial Code is a reference to the lender’s perfecting a security interest in all of the personal property assets of a debtor. Those assets can include receivables, equipment, machinery, contract rights or the like. All of those items are personal property.

1

112

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

The Controller indicated that the size of the rebates ($13,796 in rebates out of the $424,500 in total cost) “were not a huge inducement” to participate in the program. Further, she stated that Manufacturer A’s major inducement to undertake the upgrade was the cost savings associated with “bringing the HVAC units up to an efficient standard and a high degree of reliability”. Another benefit was the company’s income tax savings associated with the energy upgrades.

funds that would have targeted energy upgrades will be released and dedicated for other purposes, either on-site or elsewhere within the company. For the national ownership to have taken on a loan obligation in the current year here would have necessitated reducing current funding for another purpose elsewhere in the portfolio. This was simply not possible, given the corporate review schedule and the company’s budgeting philosophy.

Other EP Commercial Loan Borrowers

In the case of the medical facility, the multi-facility corporate ownership maintains a very sizeable, self-funded, Capital Improvements Fund containing tens of millions of dollars for ongoing maintenance and infrastructure improvements for its several facilities. The proposed upgrade project focused solely on installing variable frequency drives on several chillers at the medical facility. Following a review of the project economics, its corporate finance department made a determination that it would rather finance its energy improvements internally, supplemented solely by an APS rebate, rather than take out an EP loan in order to obtain the additional EP rebate.

Neither Dealership B nor Manufacturer B responded to multiple requests for interviews. The description of their loan activity below is derived from Table 1.

Completed during the month of January, 2013, the total project cost for Dealership B’s energy upgrades totaled $232,415. EP provided a loan to Dealership B of $79,964 with a 3-year term at 3.99% (interest-only, unsecured). APS and EP provided rebates in the amounts of $57,975 and $57,607, respectively.

This left Dealership B with an owner contribution in the project of $3,833.

Completed during the month of November, 2012, the project cost for Manufacturer B’s energy upgrades totaled $109,417.

EP provided a loan to Manufacturer B of $57,457 with a 4-year term at 3.99% (interest-only, unsecured but with a personal guarantee provided). APS and EP provided rebates in the amounts of $20,699 and $17,508, respectively. This left Manufacturer B with an owner contribution in the project of $13,753.

Commercial Contractors’ Perspectives During a presentation of overall EP research findings to participating contractors, one commercial contractor voiced the observation that many capital projects are planned and budgeted far in advance and that a loan program needs a few years of lead time to gain consideration in those decisionmaking and budgeting cycles. The contractor also expressed that businesses had no interest in taking on additional debt in a down economy and preferred to only fund projects out of existing resources. Another contractor whose large firm offers in-house financing expressed similar observations and noted that his firm had been unsuccessful in convincing clients to use its financing over the same time frame. APS indicated that its finance program also saw very low activity levels during the 2011 – 2012 timeframe.

Decision Factors for Nonparticipating Commercial Entities Interviews were conducted with representatives of two different commercial entities that chose not to proceed with the EP commercial loan program. One entity was a mixed-use, multi-building complex and the other a medical facility. One was included in the list of eleven interested parties for a commercial loan and the other was not (rebate participant only).

Conclusions While the commercial loan program did have six takers for its loan product, the funding mechanism of choice for undertaking commercial energy upgrades was clearly the commercial rebate program. On the one hand, the preference for using a rebate was simple business economics, influenced by a policy goal that led to monetary savings. The transaction/bargain was that we (the federal government through EP) will grant you (the building owner) this money (rebate) for a specific purpose (the policy goal) that will also lower your operating costs (monetary savings).

In the case of the mixed-use facility, the ownership is a nation-wide owner and manager of office buildings. As such, it conducts five-year capital improvement budgeting for its entire portfolio, which it updates on an annual basis in the fall. The owner undertook a local project that was able to participate in the commercial rebate program because the monetary grant funds enabled it to undertake energy improvements in advance of the year when they were originally scheduled to occur. The $100,000+ rebate enabled ownership to upgrade to energy efficient lighting in two of its buildings.

The alternative transaction/bargain was that we (the federal government through EP) will loan you (the building owner)

In turn, a like amount of the project’s future year corporate Energy Efficiency on an Urban Scale

113

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

this money (a loan which you will need to repay, albeit under favorable terms, during the uncertainty of the great recession when your building may be only partially leased and your business may be financially stressed) for a specific purpose (the policy goal) that will also lower your operating costs (monetary savings that may or may not exceed your cost of capital).

design changes resulted in borrowing as an unnecessary but required step for organizations that did not need financing and were not interested in it.

• Finance programs are not a silver bullet for scaling energy efficiency. They are one tool in a suite of programs and policies and their success will fluctuate with market and program conditions.

The rebate transaction/bargain prevailed for those who already had resources and an interest in pursuing energy efficiency upgrades. With one known exception, those who did not have the resources but had the interest appeared to either not participate or found other sources of financing.

Recommendations • As there appears to be insufficient funding to proceed with a revolving loan fund of any significant scope, consideration should be given to re-allocating loaned funds as they are repaid by borrowers and use them as rebates for commercial energy upgrades or as matching grants for energy efficiency upgrades for local non-profit organizations.

Lessons Learned • Due to broader lending market and economic conditions, it is reasonable to conclude that the loan programs, themselves, were not a motivating factor to encourage or cause participation in the EP commercial energy upgrade programs.

• An alternative option in line with the original program goals of creating a sustainable finance program would be to re-allocate repaid funds into policy work aimed at development and passage of Property Assessed Clean Energy (PACE) enabling legislation and then investment into the implementation infrastructure needed to administer such a program. A similar alternative would be to apply resources toward work with the utilities to design an on-utility bill financing program.

• In retrospect, both EP and NBAZ felt that it would have benefitted the commercial loan program if they had taken more active roles in its marketing. They postulated (though it has not been corroborated) that the recessiondiminished contracting community was suspicious that NBAZ (as a member of the weakened banking community) would not actually be able to provide the funds to borrowers at the necessary time. Postulating thus, it was felt that the contractors did not bring sufficient marketing enthusiasm to the table. Financing would have been seen by contractors as one more element that could potentially impede a deal.

• Policymakers and program managers should maintain focus on the primary goal of generating building energy upgrades, using financing as one tool to enable that goal rather than letting financing become a goal unto itself.

ENERGY ASSIST 60/40 LOAN PROGRAM The Energy Assist 60/40 Residential Loan program was one of multiple programs that received funding from EP and were administered by the CoP’s Neighborhood Services Department (NSD). The income-qualified program derives its name from the upgrade contribution split between the program and the participant: After netting out APS rebates, the program covered 60% of the remaining project costs through a grant. The participant was responsible for the remaining 40%, which could be financed by the program at a low interest rate. An RFP was issued in the fall of 2010 and National Bank of Arizona (NBAZ) was selected as the 60/40 program lender. NBAZ decided in spring, 2011, to concentrate on EP’s commercial lending program and withdrew from the 60/40 program. NSD (with 20% of CoP funding) subsequently became the lender. Neighborhood Housing Services, Inc. was contracted as the servicer.

• Finance programs for energy efficiency projects need substantial time to succeed. Between program design, bid, negotiation, contractual agreements, marketing, lead times for participant capital budgeting cycles, and establishing critical mass, three years is a relatively short window to succeed. Despite the urgency of a compressed time frame, it took almost eighteen months to launch the EP Finance program, leaving only eighteen months to market and succeed in establishing a borrower base during a down economy while simultaneously weighing alternatives for deploying stimulus dollars before a hard deadline.

• The pressure to loan out funds led to program design changes that encouraged unintended behavior contrary to the broader goals of the EP Finance program. Although the EP Commercial Loan Program was intended to reduce or eliminate upfront capital costs and thereby enable organizations to make energy efficiency upgrades that they could not otherwise undertake, the evolutionary program Energy Efficiency on an Urban Scale

The 60/40 program commenced operation in May 2011. Leads for potential participants in the 60/40 program were primarily 114

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

identified from among the 300 target area homeowners who had undergone energy checkups. Contractors passed along referrals to EP program staff based upon knowledge they had acquired of an individual homeowner’s income level. The income limitation for the 60/40 program equated to 400% of the Phoenix MSA poverty level (less than $89,000 in annual gross income for a family of four).

while, but worked out well.”

The scope of improvements funded by the program included the replacement of one (of two total) air conditioning units, ductwork cleaning, re-insulation of most of the house, and door (“but not window”) weatherization. The total value of the program-funded work approximated $13,500, of which $1,000 was covered by APS rebates. 60% of the remainder was covered by an EP grant and 40% was funded by a no money down, 2%, 5-year loan from the program.

Out of a pool of forty homeowners who initially explored participating in 60/40, thirty actually undertook home upgrades. The remaining ten were characterized by NSD staff as having discontinued the process at some point in the application process. Total grant dollars expended for the sixty percent component of the 60/40 program came in at $220,129 for the thirty homes that were upgraded.

Resident A

Total Upgrade Cost $13,500

Less: APS Rebates -$1,000

Subtotal $12,500

60% EP Grant -$ 7,500

40% Owner Contribution $5,000

Plus Associated Tax $800

Total Owner Out-of-Pocket Cost $5,800

5-year EP Loan at 2% -$5,800

Net Owner Out-of-Pocket Cost $0

Among the thirty participating homeowners, twenty-seven paid for their participant’s contribution using their own resources.

Only three chose to participate in the finance component of the 60/40 program. This component enabled each of them to finance their contribution with a 2% EP loan for a 5-year term.

The experiences of two homeowner participants and one contractor participant in the 60/40 finance program are described below.

Construction was completed in the fall of 2012, and Resident A commented that she “would not have been able to make the improvements to the full extent without the loan.”

Borrowers Speak

The only negative element of the experience was that she felt that the contractor did not follow through sufficiently in finding the lost application file and that there was a disconnect somewhere among the contractor’s sales person, the contractor, the city, and the homeowner.

Resident A Resident A first heard about the EP program through her local church, which had already been a participant in the program.

Her home is located within the original EP Corridor boundaries.

Even though two contractors had knocked on her door directly to make her aware of their availability to assist her, she had previously made contact with one of the two through CoPprovided information.

Resident B Having purchased a home and moved into the EP target area in April, 2012, Resident B first heard of the EP residential program in the summer of 2012 through the newspaper. In addition, a neighbor had previously participated and made him aware of EP’s merits.

She established contact with NSD staff, who evaluated her housing and income characteristics and “ran the numbers”, and then advised Resident A that the 60/40 program would best suit her needs.

Resident B contacted NSD staff, who analyzed his program eligibility and then recommended the 60/40 option. Resident B found the application process to be “no big deal”, largely because he had previously assembled much of the information during the purchase process for his new home.

The chosen contractor conducted a home energy checkup on the house then submitted the application to NSD for processing. Although Resident A stated that her application had been misplaced for some period of time, she could not recall if it was a contractor or a city issue. Once the application was located, NSD staff came to the house and specified the amount the program could provide for its share of the scope of the energy improvements, beyond rebates to be provided by APS. NPS staff mentioned the loan aspect of the EP program and Resident A decided to proceed with it. She related that the loan application process was very easy. Her overall summary of the process was that: “The process took a Energy Efficiency on an Urban Scale

Based upon an energy checkup, Resident B interviewed two contractors before choosing one for the upgrade. The scope of work included upgrading one heat pump from 2.5 to 3.0 tons (with a much higher SEER rating) and adding 12 – 14” of insulation. The total cost of the work was $9,920, of which $520 was covered by APS rebates. $5,640 was funded by an EP grant, while $4,360 was financed by a four-year, 2% loan.

The work was completed in March 2013.

115

Global Institute of Sustainability

Back to Main Table of Contents

Appendix J Table of Contents

Resident B

Further, he described an instance in which, when unpacking his equipment to undertake an energy checkup, a homeowner refused to proceed with even the preliminary home energy evaluation. She cited that filling out the various application forms, including personal income information, was “simply too much for me.” He also cited an ASU behavioral survey as being a part of the excessive paperwork.

Total Upgrade Cost $9,920

Less: APS Rebates -$520

Subtotal $9,400

60% EP Grant -5640

40% Owner Contribution $3,760

Plus Associated Tax $600

Total Owner Out-of-Pocket Cost $4,360

5-year EP Loan at 2% -$4,360

Net Owner Out-of-Pocket Cost $0

Lessons Learned • Moderate income residents took advantage of the grant portion of the Energy Assist 60/40 program but 10:1 were not sufficiently attracted to the financing portion to follow through with borrowing.

Resident B considered the alternative of self-financing the $4,360 portion of the work, but given the favorable terms of the 60/40 loan program, he found it financially advantageous to accept the EP loan.

• One of the potential factors was the combined amount of program participation paperwork resulting from federal requirements, local government requirements, ASU research, income qualification and underwriting.

Resident B was pleased with the overall 60/40 financing process, stating: “Everyone in the process was so friendly and informative. I felt very at ease with the Neighborhood Service Department staff.”

Recommendations • As there appears to be insufficient funding to proceed with a revolving loan fund of any significant scope, consideration should be given to re-allocating loaned funds as they are repaid by borrowers and using them as grants for residential energy upgrades elsewhere in the target corridor, perhaps as supplementary funding for the Weatherization Assistance Program.

A Residential Contractor’s Perspective An interview was conducted with the owner of a company that performs both energy checkups and energy upgrades.

During the course of the Energy Assist 60/40 home program, the company performed approximately twenty checkups that subsequently led to ten-to-twelve upgrade projects. The typical construction job was valued at approximately $5,000 and included duct sealing and the installation of roof insulation.

Air conditioning units were replaced as warranted.

• More research should be conducted to understand why and how the remaining twenty-seven moderate income residents financed their 40% contributions to their Energize Phoenix upgrades. At a 2% interest rate, it would be challenging to find a lower cost of borrowing in the formal financial markets. It is possible that the transactional costs of securing the EP financing in terms of the time investment to compile the necessary paperwork and the related stress may have tipped the balance for most participants toward paying their contribution out of savings or financing their contribution in other ways.

The company found out about the program through email blasts that were transmitted by EP staff and the company’s staff underwent EP contractor training.

The company’s favorite aspect of the program was the flow of customer leads that it received through EP, based upon the apparently broad awareness of the program in the target area. According to the owner, with the attractiveness of the available rebates and loans, it did not take much convincing for homeowners to choose to move forward with the upgrades.

SUMMARY CONCLUSION Analyses and interviews suggest that a confluence of uncertain economic conditions, restrictive underwriting conditions, short timelines, lengthy procurement and negotiation processes, program design and marketing decisions, program modification responses, and individual participant behavior led to the unsustainable results experienced by the EP finance programs. While finance programs are a valuable tool to enable and scale energy efficiency, they are hard-pressed to overcome negative broader market conditions and they must be designed and executed with streamlined processes and with the broader goal of motivating energy efficiency upgrades always clearly in sight.

When asked if he had done any of the door-to-door marketing that other service providers had undertaken in the EP target area, he said it was not necessary, due to the flow of EP leads.

When questioned as to what improvements he would recommend for any future programs, he cited the need to reduce the overall volume of paperwork required. As the application facilitator on behalf of his homeowner clients, he cited instances where the program required electronic submittal of reports, the size of which could not be accommodated by the city’s email server. As a result, he was required to submit a printed version.

Energy Efficiency on an Urban Scale

116

Global Institute of Sustainability

Back to Main Table of Contents

APPENDIX K ENERGIZE PHOENIX, 2010-2013: AN ECONOMIC IMPACT ANALYSIS

EXECUTIVE SUMMARY

TABLE OF CONTENTS

The expenditure inputs provided by the City of Phoenix for the analysis, totaling approximately $24.4 million, are illustrated in the table below. The inputs provided by the City of Phoenix are based on actual expenditures for 2010-2012, and a combination of actual and projected expenditure for 2013, as of July 19, 2013. These inputs also only represent the federal investment in the program.

This report estimates the economic impact of the Energize Phoenix program principally in Maricopa County but also more generally in the State of Arizona for the period 2010-2013.

1. INTRODUCTION

118

2. STUDY METHOD AND INPUTS

118

2.1. STUDY METHOD

118

2.2. INPUTS

119

3. SIMULATION RESULTS

119

4. CONCLUSIONS

120

ENERGIZE PHOENIX ACTUAL AND PROJECTED FEDERAL PROGRAM PAYMENTS EXPENDITURE

REFERENCES 121 APPENDIX 121

A.1. INPUTS PROVIDED BY CLIENT

121

A.2. T HE REMI MODEL

121

A.3. EFFECTS NOT INCORPORATED

INTO THE ANALYSIS

Source: City of Phoenix 07.19.13

121

A robust economic impact study focuses on the flow of new money in an economy, rather than the redistribution of existing resources present. Utility and customer contributions to the upgrade projects, estimated at an additional $31.8 million, are therefore excluded from the inputs in the table as they represent a redistribution of existing money already present in the State of Arizona economy.1 Based on the City’s expenditure data, the annual and cumulative economic impacts of the Energize Phoenix program for Maricopa County and the State of Arizona are estimated as follows: ECONOMIC IMPACTS OF ENERGIZE PHOENIX, 2010-2013

The utility and customer payments consist of an estimated $1.2 million for the residential and $30.6 million for the commercial participants in the Energize Phoenix program.

1

These are direct, indirect and induced jobs for all industries and sectors, including public employees and farm workers.

2

3

Source: Authors’ calculations

A job year is equivalent to one person having a full-time job for exactly one year.

These are direct, indirect and induced jobs for all industries and sectors, apart from public employees and farm workers.

4

5

A job year is equivalent to one person having a full-time job for exactly one year.

Energy Efficiency on an Urban Scale

117

Global Institute of Sustainability

Back to Main Table of Contents

Appendix K Table of Contents

1. INTRODUCTION

2. STUDY METHOD AND INPUTS

When a program such as Energize Phoenix is launched in the State of Arizona, it has three distinct types of impact on the local economy. These are the direct, indirect and induced impacts.

2.1. Study Method This study makes use of an Arizona-specific version of the REMI regional forecasting model (PI+ version 1.3.5) updated at the Seidman Research Institute, and currently used for all projects commissioned by the State of Arizona.

• Direct impacts include the initial capital investment when the program is launched, and the people directly employed to deliver the Energize Phoenix program’s upgrades.

REMI is especially useful when examining the economic impact of a business or program expanding or relocating to a particular region, state or country. Through its dynamic modeling, REMI takes account of variations in the economic impact of a business or program as it moves from the establishment to operations phase, and also shows how estimates can vary through time for the project. These estimated impacts are the difference between the baseline economy and the baseline economy augmented with the new program or enterprise. As a result, this analysis measures the Maricopa County and State of Arizona’s economy up to 2013 with and without the existence of the Energize Phoenix program.

• Indirect impacts are the economic effects resulting from inter-industry transactions or supplier purchases.

• Induced impacts occur when the workers either directly or indirectly associated with the Energize Phoenix program spend their incomes in the local economy, when suppliers place upstream demands on other producers, and when state and local governments spend new tax revenues.

The indirect and induced economic impacts are second order expenditures and jobs created as a result of the initial “injection” of expenditure and direct jobs. For example, a City of Phoenix employee hired to implement the Energize Phoenix program represents a direct job. A worker employed by an upstream manufacturer to produce the energy efficient light bulbs represents an indirect impact; and the income that the program’s employees spend in the local economy will in turn create revenues/income for a variety of other businesses, which is an example of induced effects.

The use of a county level model also enables a more detailed disaggregation of results to occur, estimating the “leakage” of economic impacts outside the host county, Maricopa.

Finally, given its overall flexibility, REMI allows for the examination of a whole host of different scenarios – different businesses and/or different establishment and operations phases – while simultaneously providing estimates that are consistent across projects.

The rounds of expenditures are not self-perpetuating in equal measure. Through time, they dissipate as more of the income/ expenditures “leak” out of the local economy.6 The cumulative impact of these rounds of expenditures or “ripple effects” is also known as the multiplier effect.

The method for estimating the economic impacts involves four fundamental steps: 1. Prepare a baseline forecast for the state economy: This baseline scenario provides a forecast of the future path of the Maricopa County and State of Arizona economies based on a combination of the extrapolation of historic economic conditions and an exogenous forecast of relevant national economic variables. This is often referred to as the Business as Usual (BAU) case.

Importantly, there is no one single multiplier for every conceivable scenario. Due to the inter-linked nature of the State of Arizona’s economy and its links to the rest of the U.S. (and the world), the eventual ripple effects depend on numerous factors.7 A full understanding of the total impact of the Energize Phoenix program for the State of Arizona economy is therefore rather more complex than just an extrapolation of direct impacts.

2. Develop policy scenario: This policy scenario describes the direct impacts that the Energize Phoenix program will generate in Maricopa County and the State of Arizona.

Section 2 describes the approach taken to model the full impact of the Energize Phoenix program, and summarizes the inputs provided by the City of Phoenix. Simulation results are presented in Section 3, and conclusions in Section 4.

Energy Efficiency on an Urban Scale

For example, in the form of savings, or payments for goods and services produced outside the State of Arizona.

6

In very simple terms, what matters is the size of the direct impact, where it occurs (that is, in which county and which sector of the economy), consumers’ propensity to purchase different commodities and the duration of the impacts.

7

118

Global Institute of Sustainability

Back to Main Table of Contents

Appendix K Table of Contents

3. Compare the baseline and policy scenario forecasts.

2.2. Inputs

4. Produce delta results: The differences among the future values of each variable in the forecast results estimate the economic magnitude of the Energize Phoenix program at a county and state level, relative to the baseline.

The inputs provided by the City of Phoenix are based on actual expenditures as of July 19, 2013, and projected expenditures through to the end of all program operations, planned for December, 2013. These are detailed in Table 1.

TABLE 1: ENERGIZE PHOENIX FEDERAL PROGRAM PAYMENTS

The economic impacts measured in this study are: • Total Employment: An estimate of the total number of full-time (or equivalent) jobs in Maricopa County or the State of Arizona, encompassing every sector and industry, including public employees and farm workers. Total employment therefore includes employees, sole proprietors and active partners, but excludes unpaid family workers and volunteers.

Source: City of Phoenix

3. SIMULATION RESULTS The study period is 2010-2013 inclusive. Results are presented for Maricopa County and the State of Arizona.

• Total Private Non-Farm Employment: An estimate of the total number of full-time (or equivalent) jobs in Maricopa County or the State of Arizona, encompassing all sectors and industries but excluding public employees and farm workers. This again includes employees, sole proprietors and active partners, but excludes unpaid family workers and volunteers.

Using REMI, the results incorporate the direct economic impacts associated with the establishment and operation of Energize Phoenix, as well as any potential indirect and induced impacts that may occur due to the increased economic activity associated with the newly-established program. It is important to note that all figures presented below are set against the Business as Usual (BAU) case.8

• Gross State Product: This is the market value of goods and services produced by labor and property in Maricopa County or the State of Arizona. It represents the dollar value of all goods and services produced for the state or county’s final demand, but excludes the value of intermediate goods and services purchased as inputs to final production. It can also be defined as the sum of employee compensation (wages, salaries and benefits, including employer contributions to health insurance and retirement pensions), proprietor income, property income, and indirect business taxes.

Table 2 illustrates the total employment and total private non-farm employment job impacts. The distinction is important. Total employment refers to any job in the public or private sector created as a result of the Energize Phoenix program, including public employees. Total private non-farm employment simply refers to the private sector, and therefore excludes public employees and also any agricultural jobs. All figures are expressed in job years.9 TABLE 2: ENERGIZE PHOENIX’S EMPLOYMENT IMPACTS, 2010-2013

• Real Disposable Personal Income: This is an estimate of the total after-tax income received by any person residing in Maricopa County or the State of Arizona, deflated by the Personal Consumption Expenditure (PCE)-Price Index, but available for spending or saving. Technically speaking, real disposable personal income is the sum of wage and salary disbursements, supplements to wages and salaries, proprietors’ income, rental income of persons, personal dividend income, personal interest income, and personal current transfer receipts, less personal taxes and contributions for government social insurance.

Source: Authors’ Calculations

If gross state product is estimated to be x dollars higher than the baseline case, this does not mean it is x dollars higher than what gross state product is today. It means that it is x dollars higher than the gross state product forecast for that given year if the program had not located in Arizona.

8

9

Energy Efficiency on an Urban Scale

119

A job year is equivalent to one person having a full-time job for exactly one year.

Global Institute of Sustainability

Back to Main Table of Contents

Appendix K Table of Contents

Looking at Table 2, we estimate that Arizona-wide employment in all sectors is approximately 12 full-time (or equivalent) jobs higher relative to the baseline in 2010; 55 full-time (or equivalent) jobs higher relative to the baseline in 2011; 128 full-time (or equivalent) jobs higher relative to the baseline in 2012; and 219 full-time (or equivalent) jobs higher relative to the baseline in 2013.10 At least 98% of these employment impacts are estimated to occur in the host county, Maricopa.

In aggregate terms, during the study period 2010-2013, and based on the Energize Phoenix federally-funded program expenditure data supplied by the City of Phoenix, total GSP is estimated to be cumulatively higher by $30.9 million (2012$), and total real disposable personal income is estimated to be cumulatively higher by $18.2 million (2012$) in the State of Arizona. The majority of the impacts are again estimated to occur in Maricopa County.

For total private non-farm employment, the Arizona-wide impacts are estimated at 9 full-time (or equivalent) jobs in 2010; 43 full-time (or equivalent) jobs in 2011; 109 full-time (or equivalent) jobs higher relative to the baseline in 2012, and 198 full-time (or equivalent) jobs higher relative to the baseline in 2013. Again, at least 98% of these employment impacts are estimated to occur in the host county, Maricopa.

TABLE 4: ENERGIZE PHOENIX’S NON-EMPLOYMENT IMPACTS, 2010-2013

TABLE 3: PRIVATE NON-FARM EMPLOYMENT ACROSS SECTORS IN ARIZONA, 2010-2013 (JOBS PER YEAR)11 Source: Authors’ Calculations

In addition to the impacts identified in Tables 2-4, small lagged effects associated with the Energize Phoenix program are forecast in 2014. This equates to 4 private non-farm jobs (job years) in the State of Arizona in 2014, $309,000 GSP (2012 $), and $869,000 real disposable personal income (2012 $).

4. CONCLUSIONS The goal of this study is to assess the impact of the Energize Phoenix program on economic activity in the State of Arizona, with a particular focus on Maricopa County (the host county).

This is assessed by an estimate of the employment, GSP, and real personal disposable income impacts associated with the federally-funded Energize Phoenix program payments alone.

Source: Authors’ Calculations

Table 3 looks specifically at the private non-farm employment sectors benefitting the most from the Energize Phoenix program. This table estimates, as expected, that a large proportion of the jobs created (38.7%) are in the construction sector. Other smaller but significant impacts are also estimated to occur in educational services (14.2%), retail trade (8.3%), health care and social services (6.4%), and professional and technical services (6.1%).

From an employment perspective, using the City of Phoenix’s historical data, Energize Phoenix is estimated to account for 12 full-time (or equivalent) jobs in 2010 across all sectors, 55 full-time (or equivalent) jobs in 2011 across all sectors, and 128 full-time (or equivalent) jobs in 2012 across all sectors.

The 2013 employment impact, based on actual and projected expenditure, is estimated at 219 jobs.

Table 4 estimates the gross state product (GSP) and real disposable personal income impacts. All dollar amounts are measured in 2012 dollars (2012$).

A cumulative total employment impact for the entire program period is not appropriate, as these employment impacts are expressed in job years. A “job year” is defined as one person holding a full-time job for exactly one year. This means, for example, that a City of Phoenix employee working on the

Please note: the report always states the full-time (or equivalent) jobs associated with Energize Phoenix for a specific year, such as 2012. These job years should not be summed across years.

10

11

A job year is equivalent to one person having a full-time job for exactly one year.

12

Thus, we are looking at real, rather than nominal, dollars to be able to compare across years.

Energy Efficiency on an Urban Scale

120

Global Institute of Sustainability

Back to Main Table of Contents

Appendix K Table of Contents

Energize Phoenix program throughout 2010-2013 will account for 4 job years, but represent only 1 job.

and accounts for dynamic feedbacks among its economic and demographic variables. The REMI model is also an “open” model in that it explicitly accounts for trade and migration flows in and out of the state. A complete explanation of the model and discussion of the empirical estimation of the parameters/equations can be found at www.remi.com.

Examining only private non-farm sector employment impacts, these are estimated at 9 jobs in 2010, 43 jobs in 2011, 109 jobs in 2012, and 198 jobs in 2013.

Energize Phoenix is also estimated to add during the entire period (2010-2013) approximately:

The operation of the REMI model has been developed to facilitate the simulation of policy changes, such as a tax increase for example, or many other types of events – anything from the opening of a new business to closure of a military base to a natural disaster. The model’s construction includes a large set of policy variables that are under the control of the model’s operators. To simulate the impact of a policy change or other event, a change in one or more of the policy variables is entered into the model and a new forecast is generated.

The REMI model then automatically produces a detailed set of simulation results showing the differences in the values of each economic variable between the control and the alternative forecast.

• $30.92 million (2012$) in gross state product.

• $18.17 million (2012$) in real disposable personal income.

In addition, some small lagged effects associated with the program are forecast in 2014 – specifically, 4 private non-farm employment jobs in the State of Arizona, $309,000 GSP (2012 $), and $869,000 real disposable personal income (2012 $).

REFERENCES Arizona Department of Economic Security. www.azdes.gov U. S. Bureau of Census (2012) State Government Finance Database 2010. www.census.gov

The specific REMI model used for this analysis was Policy Insight Model Version PI+ version 1.3.5 of the Arizona economy (at the county level) leased from Regional Economic Models Inc. by a consortium of State agencies, including Arizona State University, for economic forecasting and policy analysis.

U. S. Bureau of Economic Analysis (2012) State Personal Income and Employment Database. www.bea.gov U. S. Bureau of Labor Statistics (2012) Quarterly Census of Employment and Wages Database. www.bls.gov/cew/ U. S. Bureau of Labor Statistics (2012) Occupational Employment Statistics Database. www.bls.gov/oes/

REMI is currently used by the Seidman Research Institute for all projects commissioned by the State of Arizona.

APPENDIX

A.3. Effects Not Incorporated into the Analysis

A.1. Inputs Provided by Client

The analysis focuses exclusively on the federally-funded Energize Phoenix program payments. It does not take into account any utility or customer payments, as this money is already present in the State, and would therefore in all probability have been predominantly spent on other State-based activities in the absence of the Energize Phoenix program.

Source: City of Phoenix

It also excludes approximately $438,000 in estimated finance associated with the program in 2013.

A.2. The REMI Model REMI is an economic-demographic forecasting and simulation model developed by Regional Economic Models, Inc. REMI is designed to forecast the impact of public policies and external events on an economy and its population. The REMI model is recognized by the business and academic community as the leading regional forecast/simulation tool available.

Unlike most other regional economic impact models, REMI is a dynamic model that produces integrated multiyear forecasts Energy Efficiency on an Urban Scale

121

Global Institute of Sustainability