Journal Personal Finance Volume 16 Issue 1 by IARFC

Volume 16 Issue 1 2017 www.journalofpersonalfinance.com

Journal of Personal Finance

Techniques, Strategies and Research for Consumers, Educators and Professional Financial Consultants

IARFC INTERNATIONAL ASSOCIATION OF REGISTERED FINANCIAL CONSULTANTS

National Financial Plan Competition 2018 Be Part of the future

Molly Funk, student competitor and 2016 Winner from Bryant University connects to the group (a.k.a pretend case study family) on a more personal and level playing field.

The IARFC has extended its invitation to students to participate in the 2018 National Financial Plan Competition. The finalists and their faculty advisor will present their comprehensive financial plans to a live audience. We invite the IARFC members to join the competition and become part of the judging process. The National Financial Plan Competition is organized through the IARFC for undergraduate students in a curriculum related to Financial Services. Participants are given a case study of a fictional family with an overview of their financial picture. A financial plan is then produced with recommendations for current and future action. These plans are sent to the IARFC headquarters for an initial round of judging. From this group of entries, six are chosen to be presented electronically with the three semi-final winners going forward to present their plans in person at an IARFC event.

IARFC INTERNATIONAL ASSOCIATION OF REGISTERED FINANCIAL CONSULTANTS

Expenses incurred by participating in the judging of the Financial Plan Competition are not the responsibility of the IARFC.

Become a 2018 Sponsor Sponsorship levels

Silver

Gold

Platinum

Diamond

Individual

$50

$100

$250

$500

Corporate

$1,000

$5,000

$7,500

$10,000

Send more information on Corporate Sponsorship. Interested in judging the Financial Plan Competition. Please print or type information below. Mr. First Name

Ms.

Mrs. Last Name

Firm or B/D Street Address City

State

Zip Code

Phone E-mail

Payment Information: Credit Card

Check made payable to IARFC

CC # (Discover, MasterCard, Visa, AmEx) / Exp. date / CCV code

Signature

PLEASE MAIL OR FAX THIS COMPLETED FORM TO IARFC

International Association of Registered Financial Consultants P.O. Box 42506 Middletown, OH 45042-0506

Fax: 513.345.9479 Phone: 800.532.9060 Email: info@iarfc.org

Journal of Personal Finance

Volume 16, Issue 1 2017 The Official Journal of the International Association of Registered Financial Consultants ÂŠ2017, IARFC. All rights of reproduction in any form reserved.

Journal of Personal Finance

Contents Editors’ Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 The Home as a Risky Asset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 David Blanchett, PhD, CFA, CFP , Morningstar Investment Management LLC Despite being one of the most valuable assets in the world and one of the largest assets on many household balance sheets, real estate—in the form of homeownership—presents considerable risks that are generally poorly understood. Through an analysis, we find significant homeownership risks that are primarily idiosyncratic (i.e., not market-related), driven largely by the illiquid nature of owning a single home. We find the risk of homes is approximately double that of city-specific home price indexes (e.g., the S&P Case-Shiller Home Price Indices), with an annual standard deviation of 12%, which is approximately equivalent to the historical volatility of a portfolio invested in 60% stocks and 40% Treasury bills. While the return on house price indexes has exceeded inflation historically (with a real return of approximately ~1%), the actual real return realized by homeowners, after considering the various costs associated with owning and selling a home, has likely been negative in real terms. Renting is often a better option for many households, especially those households with lower marginal tax rates (i.e., households that do not itemize deductions) and have shorter expected housing durations. We note significant differences in the returns, volatility, and market risk of homes and REITs; these differences suggest REITs are a relatively poor proxy for residential real estate from an investment perspective. We also identify the factors, such as home price, county unemployment rate, housing turnover, home size, and even average annual temperature, that can differ by region and are strongly related to the returns, volatility and market risk of homeownership. Many households may use this factor information to better approximate the risk of their homes. Overall, the impact of owning a home on the optimal total wealth financial portfolio is likely to vary significantly by household, based on the unique risks associated with the home, household wealth, and other non-financial household assets. The Impact of Rates of Return on Roth Conversion Decisions and Retiree Savings Wealth . . . . . . . . . . . . . . 29 Lewis W. Coopersmith, Ph. D., Professor Emeritus of Management Sciences, Rider University, Lawrenceville, NJ, and founder and Chief Research Officer of VestaEdge, Inc. Alan R. Sumutka, MBA, CPA, CGMA, Associate Professor of Accounting at Rider University, Lawrenceville, NJ. A tax-optimal retirement savings withdrawal model, implemented as a linear programming application, is used to compare savings wealth growth when Roth conversions are permitted (RC) and when they are not (NoRC). Evaluations are made for combinations of percentage rates of return (ROR) for taxable, tax-deferred, and tax-free savings. PctDiff, the difference between tax-free and tax-deferred (TD) account RORs, is an important conversion consideration. When investment strategies target PctDiffs at two percent or greater, RC provides substantial benefits. As PctDiff increases, the percentage of initial TD savings that should be converted rises, the time to recover savings wealth lost to conversion-related taxes declines, and savings wealth growth surges. When PctDiff is less than two percent, savings wealth growth is small and savings wealth loss due to conversion-generated taxes persists for more than 13 years; retiree health and prospects of living long enough to realize savings wealth gains becomes a vital concern. Conversions are best made relatively early in retirement and at varying annual amounts. Conversion of all initial tax-deferred savings in the first year of retirement rarely results in maximum savings wealth growth. Do Financial Advisors Follow Their Own Advice? Evidence from 2008-2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Dominique Gehy Outlaw, PhD , Hofstra University Jesse Outlaw, CPA, Outlaw, Bruno, and Associates, Family Office Consistent with previous literature, we find that financial advisors’ trade recommendations do not outperform the trades that are independently initiated by their clients. At best, their recommendations outperform their cli©2017, IARFC. All rights of reproduction in any form reserved.

Volume 16, Issue 1

ents’ selections in the short-term, but still underperform the market. What is unknown in the literature is whether it is misaligned sales incentives that cause advisors to give their clients suboptimal advice. Using transaction-level data from a U.S. brokerage firm, we compare financial advisors’ personal trades to their clients. We find that financial advisors do not outperform their clients, suggesting that advisors are giving clients their best advice which they, too, follow. Expected vs. Actual Retirement Savings Behavior of Highly Educated Individuals . . . . . . . . . . . . . . . . . . . . . 51 Kristine Beck, Ph.D., Assistant Professor, California State University, Northridge Inga Chira, Ph.D., CFP®, Assistant Professor, California State University, Northridge Using a unique sample of 318 respondents, we design a custom survey to examine savings understanding and behavior with respect to demographic attributes, long-term financial goals, and the level of financial knowledge of highly educated individuals. We find that savings expectations differ from actual savings behavior with regard to demographics and individuals’ articulation of personal financial goals. However, we find a strong relationship between the level of financial knowledge and savings behavior. Financial knowledge is measured using awareness of the tax benefits of retirement savings, stock market performance, specifics of financial instruments, and self-reported financial savvy. Risk Tolerance and Goals-based Savings Behavior of Households: The Role of Financial Literacy . . . . . . . . 66 Swarn Chatterjee, University of Georgia, Athens, GA Lu Fan, University of Georgia, Athens, GA Ben Jacobs, University of Georgia, Athens, GA Robin Haas, University of Georgia, Athens, GA This study uses a national dataset to examine the association among risk tolerance, financial literacy, and goalsbased savings behavior of households. The results indicate that three out of five households do not have any emergency funds set aside, and about half the households have not calculated how much money they will need for retirement. However, both financial literacy and risk tolerance are associated with goals-based savings behavior, such as saving for emergencies, and planning for retirement among households. Although risk tolerance appears to be an important factor in the savings and investment decisions of households, the findings of this study provide further evidence regarding the role of financial literacy in improving household financial capability. Implications for policy makers, scholars, and researchers in the area of behavioral economics and household finance are included. Preventing Financial Elder Abuse. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Kenn Beam Tacchino, JD, LLM, RICP, Professor of Taxation and Financial Planning and Boettner Endowed Professor in Financial Planning, Widener University, Chester PA Clients may not only need a financial planner’s help and advice when it comes to their finances— they may also need their protection as well. Because billions of dollars are financially exploited each year, planners and financial service institutions are increasingly being called upon to walk the tightrope between a client’s autonomy and a client’s need to be safeguarded. In order for the financial planner to better serve his clients we first look at the financial elder abuse problem and share examples, which highlight the scope of the issue. We then point out some red flags that will alert financial planners to a potential problem. We examine the legal requirements and protections that are germane to understanding the financial planner’s responsibility in recognizing and preventing financial elder abuse. We review actions the financial planner can take to cope with or prevent financial elder abuse. And finally, we discuss a systematic response that an organization should take to create a business culture that seriously addresses financial elder abuse.

Journal of Personal Finance

Call for Papers Journal of Personal Finance (www.JournalofPersonalFinance.com) Overview The Journal of Personal Finance is seeking high quality submissions that add to the growing literature in personal finance. The editors are looking for original research that uncovers new insights— research that will have an impact on advice provided to individuals. The Journal of Personal Finance is committed to providing high quality article reviews in a single-reviewer format within 60 days of submission. Potential topics include: •

Household portfolio choice

•

Retirement planning and income distribution

Investment research relevant to individual portfolios

•

Household credit use

•

Individual financial decision-making

•

Household risk management

Professional financial advice and its regulation

•

Life-cycle consumption and asset allocation

•

Behavioral factors related to financial decisions

•

Financial education and literacy

Please check the “Submission Guidelines” on the Journal’s website (www.JournalofPersonalFinance. com) for more details about submitting manuscripts for consideration.

Contact Wade Pfau and Walt Woerheide, Co-Editors Email: jpfeditor@gmail.com www.JournalofPersonalFinance.com

Volume 16, Issue 1

Journal of Personal Finance Volume 16, Issue 1 2017 Co-Editors Wade Pfau, Ph.D., The American College Walt Woerheide, Ph.D., ChFC, CFP™, RFC®, The American College

Editorial Board Benjamin F. Cummings, Ph.D., Saint Joseph’s University

Ruth Lytton, Ph.D., Virginia Tech

Dale L. Domian, Ph.D., CFA, CFP™, York University

Lew Mandell, Ph.D., University of Washington

Michael S. Finke, Ph.D., CFP™, RFC® Texas Tech

Carolyn McClanahan, MD, CFP™, Life Planning Partners

Joseph W. Goetz, Ph.D., University of Georgia

Yoko Mimura, Ph.D., California State University, Northridge

Michael A. Guillemette, Ph.D., University of Missouri

Robert W. Moreschi, Ph.D., RFC®, Virginia Military Institute

Clinton Gudmunson, Ph.D., Iowa State University

Ed Morrow, CLU, ChFC, RFC®, IARFC

Sherman Hanna, Ph.D., The Ohio State University

David Nanigian, Ph.D., Mihaylo College at Cal State Fullerton

George W. Haynes, Ph.D., Montana State University Douglas A. Hershey, Ph.D., Oklahoma State University

Barbara M. O’Neill, Ph.D., CFP™, CRPC, CHC, CFCS, AFCPE, Rutgers

Karen Eilers Lahey, Ph.D., The University of Akron

James Taggert, Ph.D., Taggert Consulting

Douglas Lamdin, Ph.D., University of Maryland Baltimore County

Jing Jian Xioa, Ph.D., University of Rhode Island

Jean M. Lown, Ph.D., Utah State University

Tansel Yilmazer, Ph.D., CFP™, The Ohio State University

Angela C. Lyons, Ph.D., University of Illinois

Yoonkyung Yuh, Ewha Womans University Seoul, Korea

Mailing Address: IARFC Journal of Personal Finance 1046 Summit Drive, P.O. Box 506 Middletown, OH 45042

Postmaster: Send address changes to IARFC Journal of Personal Finance, 1046 Summit Drive, P.O. Box 506 Middletown, OH 45042

Permissions: Requests for permission to make copies or to obtain copyright permissions should be directed to the Co-Editors. Certification Inquiries: Inquiries about or requests for information pertaining to the Registered Financial Consultant or Registered Financial Associate certifications should be made to IARFC, 1046 Summit Drive, P.O. Box 506 Middletown, OH 45042.

Rui Yao, Ph.D., CFP™, University of Missouri

Disclaimer: The Journal of Personal Finance is intended to present timely, accurate, and authoritative information. The editorial staff of the Journal is not engaged in providing investment, legal, accounting, financial, retirement, or other financial planning advice or service. Before implementing any recommendation presented in this Journal readers are encouraged to consult with a competent professional. While the information, data analysis methodology, and author recommendations have been reviewed through a peer evaluation process, some material presented in the Journal may be affected by changes in tax laws, court findings, or future interpretations of rules and regulations. As such, the accuracy and completeness of information, data, and opinions provided in the Journal are in no way guaranteed. The Editor, Editorial Advisory Board, the Institute of Personal Financial Planning, and the Board of the International Association of Registered Financial Consultants specifically disclaim any personal, joint, or corporate (profit or nonprofit) liability for loss or risk incurred as a consequence of the content of the Journal.

General Editorial Policy: It is the editorial policy of this Journal to only publish content that is original, exclusive, and not previously copyrighted. Subscription requests should be addressed to: IARFC Journal of Personal Finance 1046 Summit Drive, P.O. Box 506 Middletown, OH 45042 Info@iarfc.org 1-800-532-9060

Journal of Personal Finance

Editors’ Notes Welcome to the Spring 2017 issue of the Journal of Personal Finance. We are excited to bring you six articles representing cutting edge research in the field of personal finance. Our first article by David Blanchett, Ph.D., CFA, CFP, explores the home as a risky asset. This is part of a series David has written to explore how household assets aside from a financial portfolio behave and interact with the financial portfolio. David finds that the risks of homeownership are considerable and are poorly understood. In particular, he finds that the risk of individual homes is approximately double that of city-specific home price indexes, with an annual standard deviation of 12%. This is approximately the same as the historical volatility of a portfolio invested in 60% stocks and 40% Treasury bills. David provides further analysis about historical returns on home ownership, the ability for REITs to proxy home ownership, and factors that are correlated with home prices. He seeks to provide details needed for homeowners to better determine how their home relates to their financial portfolio, with implications for asset allocation. In our second article, Lewis W. Coopersmith, Ph. D., and Alan R. Sumutka, MBA, CPA, CGMA, use linear programming to explore the potential benefits of using Roth conversions as part of a tax-optimal retirement savings withdrawal model. Optimal Roth conversions are explored for a variety of situations and a key driver of the outcomes is the difference in expected returns for the different retirement accounts. The tradeoffs regard the amount of time required to make up early losses from paying taxes on the converted amount. They find that conversions are best made relatively early in retirement and at varying annual amounts. Next, Dominique Gehy Outlaw, Ph.D., and Jesse Outlaw, CPA, find that financial advisors’ trade recommendations do not outperform the trades that are independently initiated by their clients. Using data from a brokerage firm, they also learn that financial advisors do not outperform their clients with personal trades. At least this suggests that

advisors are not influenced by misaligned sales incentives to provide suboptimal investment recommendations to their clients. Our fourth article is by Kristine Beck, Ph.D., and Inga Chira, Ph.D., CFP®. They design a survey that demonstrates a strong relationship between the level of financial knowledge and savings behavior. They link these variables to demographic attributes and long-term financial goals, finding that savings expectations differ from actual savings behavior with regard to demographics and individuals’ articulation of personal financial goals. Fifth, Swarn Chatterjee, Lu Fan, Ben Jacobs, and Robin Haas, use a national dataset to examine the association among risk tolerance, financial literacy, and goals-based savings behavior of households. They find that both financial literacy and risk tolerance are associated with goals-based savings behavior, such as saving for emergencies, and planning for retirement among households. Their findings provide additional evidence about how financial literacy plays and important role in improving the financial capabilities of households. Finally, Kenn Beam Tacchino, JD, LLM, RICP, provides a detailed assessment of how advisors can better protect their clients from potential financial elder abuse. He notes that planners and financial service institutions are increasingly being called upon to walk the tightrope between a client’s autonomy and a client’s need to be safeguarded. He highlights the scope of the financial elder abuse issue, points out red flags that can alert planners to a potential problem, examines the legal requirements and protections that are germane to understanding the financial planner’s responsibility in recognizing and preventing financial elder abuse, and reviews actions the financial planner can take to cope with or prevent financial elder abuse. We hope you enjoy the current issue of the Journal of Personal Finance.

— Wade Pfau — Walt Woerheide

Volume 16, Issue 1

The Home as a Risky Asset

David Blanchett, PhD, CFA, CFP, Head of Retirement Research Morningstar Investment Management LLC

Abstract Despite being one of the most valuable assets in the world and one of the largest assets on many household balance sheets, real estateâ&#x20AC;&#x201D;in the form of homeownershipâ&#x20AC;&#x201D;presents considerable risks that are generally poorly understood. Through an analysis, we find significant homeownership risks that are primarily idiosyncratic (i.e., not market-related), driven largely by the illiquid nature of owning a single home. We find the risk of homes is approximately double that of city-specific home price indexes (e.g., the S&P Case-Shiller Home Price Indices), with an annual standard deviation of 12%, which is approximately equivalent to the historical volatility of a portfolio invested in 60% stocks and 40% Treasury bills. While the return on house price indexes has exceeded inflation historically (with a real return of approximately ~1%), the actual real return realized by homeowners, after considering the various costs associated with owning and selling a home, has likely been negative in real terms. Renting is often a better option for many households, especially those households with lower marginal tax rates (i.e., households that do not itemize deductions) and have shorter expected housing durations. We note significant differences in the returns, volatility, and market risk of homes and REITs; these differences suggest REITs are a relatively poor proxy for residential real estate from an investment perspective. We also identify the factors, such as home price, county unemployment rate, housing turnover, home size, and even average annual temperature, that can differ by region and are strongly related to the returns, volatility and market risk of homeownership. Many households may use this factor information to better approximate the risk of their homes. Overall, the impact of owning a home on the optimal total wealth financial portfolio is likely to vary significantly by household, based on the unique risks associated with the home, household wealth, and other non-financial household assets.

Journal of Personal Finance

Introduction The total value of global real estate is approximately $217 trillion, according to London-based real estate services provider Savills.1 This is more than the value of other common financial assets, such as equities (~$62 trillion2) or bonds (~$122 trillion3), combined. Similarly, home value typically dominates the household balance sheet,4 represents the greatest household liability (the mortgage), and accounts for the largest household expenditure (e.g., 33% of expenditures for U.S. households5). Listings website Zillow 6 estimates the total value of residential real estate in the United States to be $28.4 trillion, which is consistent with data from the Federal Reserve,7 which places the value at $25.3 trillion. Given the key role real estate plays for investors, especially households, it is important to understand the attributes of homes from an investment perspective and how they fit onto a comprehensive balance sheet in conjunction with other household assets, primarily financial assets (i.e., owning a home should affect how a household invests its financial portfolio). Residential real estate (or a home) is a unique asset for households since it is both an investment and consumption good. A home is an investment good such that it allows the household to accumulate equity over the duration of homeownership, and it is a consumption good since it provides shelter. When assessing the value of homeownership, it is important to understand both aspects. That is, we should consider both the risks associated with owning a home and the best financial option to provide shelter (i.e., buying versus renting). Both of these perspectives will be explored at some length in this piece. Unlike stock portfolios, which can easily be diversified through the purchase of other stocks or some type of collective investment vehicle (e.g., a mutual fund or ETF), a homeowner is subject to significant idiosyncratic risk that cannot be easily (and is rarely) diversified away. While a homeowner can insure against types of risk, such as

fire or theft, most location-specific risk is virtually impossible to hedge away given the relatively illiquid market of homes today. Therefore, common proxies used to gauge the risk (and return) of residential real estate, such as the S&P/Case-Shiller Home Price Indexes, significantly understate the risks associated with owning a home since they represent a geographically diverse portfolio of thousands of homes (similar to how the risk of a single stock is much greater than that of a broad-market index, such as the S&P 500). Through our analysis, we find that the volatility of individual homes is likely more than double that of CaseShiller Indexes, with an annual standard deviation of at least 12%, which is equivalent to the historical volatility of a portfolio invested in 60% stocks and 40% Treasury bills (or 50% stocks and 50% bonds). Note, the 12% estimate is for a non-leveraged home. Leverage creates a significant potential risk for new (or first-time) homebuyers, and home equity is likely the riskiest asset in most homeownersâ&#x20AC;&#x2122; portfolio, significantly exceeding the risk associated with financial assets. In addition to idiosyncratic risk, homeownership comes with a lot of costs: at purchase, during ownership, and at sale. While the historical returns for many residential real estate indexes have exceeded inflation (generally by ~1%), the true return of owning a home is likely lower than inflation, and potentially negative (in nominal terms), after all the costs associated with homeownership are factored in (i.e., the cost of carry), such as insurance, taxes, maintenance, transaction costs, etc. Although many of the costs associated with homeownership are explicitly paid by owners (such as real estate taxes, sales commissions, etc.), these same costs are implicitly paid by renters (i.e., the costs are factored into the total rent), and therefore the relative attractiveness of home ownership (versus renting) is driven by a variety of factors, such as the prevailing rentto-buy price ratio, expected duration of homeownership, the tax deductibility of the various taxes and costs (e.g., mortgage interest) associated with home ownership, etc. Because of the idiosyncratic nature of homeownership, itâ&#x20AC;&#x2122;s difficult to model its expected risk and reward. Real

1. 2.

$162 trillion is residential, $29 trillion is commercial, and $26 trillion is land. http://data.worldbank.org/indicator/CM.MKT.LCAP.CD?end=2015&start=1975&view=chart&year=2015

3. 4. 5. 6. 7.

http://www.bis.org/statistics/c1.pdf That is, excluding intangible assets such as human capital and pensions. http://www.bls.gov/news.release/cesan.nr0.htm http://www.zillow.com/research/total-housing-value-2015-11535/ https://www.federalreserve.gov/releases/z1/current/z1.pdf

ÂŠ2017, IARFC. All rights of reproduction in any form reserved.

Volume 16, Issue 1

estate investment trusts would seemingly have similar risk and return characteristics as homes, since they are real estate-focused investments. However, REIT returns differ significantly from single-home investment returns, both domestically and abroad, suggesting that REITs are a poor proxy for homes. When building a total wealth portfolio, our optimization routine suggests significant allocations to home ownership when related costs are ignored. However, the optimizer gives low homeownership allocations when costs are included, suggesting that homes are not an efficient asset when viewed entirely from an investment perspective or even when jointly considering them as a consumption good (i.e., owning versus renting). However, again, much depends on the specific characteristics of the home and its owners. To provide some perspective on which region-specific qualities have been associated with historical returns, volatility, and the market risk of homes, we run a series of multivariate ordinary least squares regressions and find that certain zip code attributes, such as average home price, unemployment rate, population, education level, percent born in the U.S., percentage of English-language-only households, zip code size, housing turnover, average regional temperature, and home size all tend to have statistically significant coefficients that vary by dependent variable. Overall, though, there does appear to be some factors that at least historically have been positively and negatively associated with the return, volatility, and market risks of homeownership across regions that could be useful to homeowners when assessing the risk of their home. Finally, we provide guidance on the impact this research could have on an investor’s portfolio. While the impact of other assets, such as pension benefits, which are relatively bond-like, is much more straightforward, the unique nature and risks associated with homes make them a unique asset for households. Households that have a significant amount of leverage in the home (e.g., a mortgage over 80% of the value of the home) should likely be more conservative in other financial assets to balance the overall home equity risk; however, the impact of owning a home on the optimal total wealth financial portfolio is likely to vary significantly 8. 9. 10.

https://research.stlouisfed.org/fred2/series/RHORUSQ156N http://www.federalreserve.gov/pubs/bulletin/2014/pdf/scf14.pdf http://www.census.gov/housing/hvs/data/rates.html

by household, based on the unique risks associated with the home, the household’s wealth, and their other non-financial assets.

Homeownership Real estate is a unique asset in that it can be both an investment good and a consumption good. The income stream that owners could save from buying a home instead of renting is commonly termed “imputed rent.” Despite the potential benefits of renting (more on this later), homeownership is generally encouraged versus renting by national governments. The main economic argument for subsidizing homeownership is that ownership may give rise to positive spillovers for society, such as wealth accumulation, better outcomes for children, community engagement, etc., although the evidence of these benefits is mixed (Andrews and Sánchez, 2011). Countries use a variety of methods to encourage homeownership, such as tax policies (e.g., allowing the deduction of mortgage interest costs is common) and interventions in the financial markets (e.g., government guarantees in the housing finance market). Researchers have defined the term “homeownership” in two ways. The U.S. Census Bureau uses a housing-unitbased definition: the number of owner-occupied units in a region divided by the total number of units. Using this formula, the Census Bureau tracked homeownership in the U.S. since 1965, during which time it drifted between 63% and 69%. It is currently about 64%.8 Other agencies and organizations that study homeownership use a different, people-based definition—namely, the percentage of people (or families) who own their primary residence. Using this definition, we see significant variations in homeownership rates in the U.S. by income and region. For example, homeownership rates in 2013 were only 49.2% for the bottom half of the income distribution, but 93.5% for the top 10%.9 In terms of regions, homeownership rates are highest in New Hampshire (at 75%) and lowest in New York (at 51%).10 There is an even greater dispersion at the city level. Similar to regional differences in homeownership in the U.S., there are even more pronounced differences across

Journal of Personal Finance

countries, peaking in Romania at 96.6% and bottoming at 43.8% in Switzerland.11 At 64%, the U.S. homeownership rate is relatively low compared to other developed nations. Most countries in the OECD have seen an increase in homeownership over the past few decades, although the underlying factors driving the changes vary. Household characteristics and aging (since older people are more likely to be homeowners) explain a significant amount of changes (Andrews and Sánchez, 2011). Most individuals who purchase a home typically use a mortgage, since relatively few homeowners (especially first-time homeowners) have the necessary savings to purchase the home outright. Mortgage features also vary significantly by region, with government interventions often playing a significant role. While fixed rates on mortgages are most common, there are some countries (e.g., the United Kingdom) that have predominately floating-rate mortgages. According to Andrews and Sánchez (2011), terms vary from as low as 10 years in the Netherlands to 30 years in the U.S. and Denmark. Most countries require some type of down payment, typically in the range of 5%−30% of the value of the home. Additionally, while there is generally no penalty for prepaying a mortgage (e.g., in the U.S.), the lenders in some countries (such as in Germany and Japan) may be entitled to compensation for lost income.

The Home as a Consumption Good While many homeowners may view their home as an investment, a home’s primary function for households is as shelter (i.e., a consumption good). In this section we introduce some general concepts related to owning a home to put the role of the home into more of a consumption framework as a means to consider whether homeownership makes sense for shelter provision. While aspects such as potential appreciation and risk are central to the rentor-buy decision, other variables, such as expected duration of being in (or owning) the home, marginal tax rates, etc., can also be very important. Unlike assets such as mutual funds, which are relatively inexpensive to buy and maintain, homeownership comes 11. 12. 13.

with a significant number of costs, both transaction-related and ongoing. The costs of homeownership are similar to the costs associated with owning other physical commodities, commonly referred to as the cost of carry. For example, if you buy gold there are a variety of storage and insurance costs associated with owning the asset, which makes the price of gold for spot delivery lower than the price of gold for future delivery under normal market conditions. Similarly, it is important to understand the costs homeowners pay to continue to own homes. One common expense associated with owning a home is homeowner’s insurance. While homeowner’s insurance is not technically required for all homeowners (e.g., those who have paid off their mortgage), individual homes are subject to a significant amount idiosyncratic risk and homeowner’s insurance provides a guarantee (subject to certain provisions) that if something should happen to a home (e.g., the home were to burn) the owner would be indemnified. According to the Insurance Information Institute,12 the cost of homeowner’s insurance in the U.S. is approximately 0.5% of the value of the home. Approximately 80% of the premium insures the home’s value, while the remaining 20% covers the value of the contents of the home (i.e., is comparable to what a renter would pay to insure his or her goods inside the home but not the structure itself ). The costs of homeowner’s insurance vary significantly by region, based on the prevailing risks in the area. For example, homeowner’s insurance in Florida generally exceeds 1.0% of the home’s value, while homeowner’s insurance in California may be less than 0.25% of the home’s value. Additional types of insurance, such as flood or hurricane insurance, should also be considered by homeowners in higher risk areas and are generally priced based on the prevailing risk of claims. Property taxes can be another significant cost associated with homeownership. In the U.S., property taxes are generally less than 1% of the home’s value, although they ranged from 0.2% (Maui County, Hawaii) to 3.1% (Wayne County, New York) in 2011.13 Internationally there are countries with no property tax (e.g., Bahrain, Cayman Islands, Croatia, Liechtenstein, Israel, and Sri Lanka) while property taxes in Iceland (1.65%), France (3%), and the United Kingdom

http://www.pewresearch.org/fact-tank/2013/08/06/around-the-world-governments-promote-home-ownership/ http://www.iii.org/fact-statistic/homeowners-and-renters-insurance http://www.urban.org/sites/default/files/alfresco/publication-pdfs/412959-Residential-Property-Taxes-in-the-United-States.PDF

Volume 16, Issue 1

(4%) can be higher depending on the property value and location.14 Maintenance is another cost associated with a home. Anyone who has had to replace a roof or HVAC unit in a home has experience with the (potentially significant) costs associated with owning and maintaining a home. Maintenance costs are generally assumed to range between 1% and 3% of the home’s value annually (e.g., see Smith and Smith, 2006), and also vary significantly based on the size and location of the home. When thinking about maintenance costs, it’s important to distinguish between capital improvements (e.g., adding a deck or remodeling the kitchen), which increase the value of the home, and general maintenance (e.g., fixing a broken toilet), which is general upkeep. In reality, while most types of maintenance end up resulting in some type of increase to the home value, even with capital improvements the benefit is typically less than the amount spent on the project, and improvements may depreciate relatively quickly due to changing tastes and fashions.15 A problem with residential real estate indexes (e.g., CaseShiller), which typically track repeat sales of the same home, is that they do not (and really cannot) distinguish the changes in the value for a home that would have occurred if the home were in the exact condition between the two sales. In some (or potentially many) instances, the homeowner may have spent (significant) monies improving the home that increase the value (for index calculation purposes), but the expenses related to the improvements are not reflected in the repeat sales value. An extreme example of this would be a home purchased for the exclusive purpose of fixing up and reselling (i.e., flipping). There are costs to buying and selling a home—and they are significantly greater than selling other commonly held assets, especially when compared to financial assets (e.g., an ETF). For example, while many mutual funds can effectively be purchased for free, it may cost 10% or more of the home’s value to actually sell the home. In the U.S. it is common to use a real estate agent when purchasing a home (approximately 87% of all purchases are through an agent, according to a 2016 analysis by the National

14. 15. 16.

Association of Realtors 2016). Commissions of 6% of the sales price are commonly paid by the seller of a home (which is generally split between the seller’s and buyer’s agent). Buying a home typically incurs such costs as home inspection costs, taxes, appraisal fees, lender origination fees, moving costs, etc., that can easily exceed 2% of the home’s value. While a round-trip cost of 8% to buy and sell a home in the U.S. may seem excessive, it can be significantly worse in other countries. For example, round-trip costs can be as high as 25% in Russia, 20% in Monaco, and 16% in France, but as low as 2.4% in Lithuania, 4% in Estonia, and 5% in the U.K.16 National governments encourage homeownership in a number of ways, including through public policies that benefit homeowners. For example, in the U.S. a capital gain of $250,000 from the sale of your primary residence ($500,000 if married) can be excluded from taxation if the home has been owned by the individual and used as the primary residence for a period aggregating at least two years out of the five years prior to its date of sale. Additionally, real estate taxes and mortgage interest can be deductible for tax purposes if the individual claims the itemized deduction for tax purposes. As previously discussed, there are also a number of intangible benefits associated with homeownership, such as providing the most stable tenure arrangement to satisfy basic household needs and promoting a more active and informed citizenry (DiPasquale and Glaeser, 1999). Households that own homes also tend to accrue greater levels of wealth, which we’ll address in the next section.

Renting While there are a variety of explicit costs associated with owning a home (e.g., paying real estate taxes), it is important to realize that some costs will be paid regardless of whether the individual owns or rents. For example, while a homeowner pays real estate taxes explicitly, a renter will pay these taxes implicitly since these costs will be factored into the total rent. There are certain expenses, such as broker’s commissions, that may vary based on expected duration of ownership (e.g., fees are less of a burden on a longterm owner than on a short-term owner not only because

http://www.pwc.com/gx/en/tax/corporate-tax/worldwide-tax-summaries/assets/pwc-worldwide-tax-summaries-corporate-2015-16.pdf Eighty percent or less is a common recoup rate according to remodeling.hw.net (for rates see http://www.remodeling.hw.net/cost-vs-value/2016/). http://www.globalpropertyguide.com/faq/guide-transaction-costs

Journal of Personal Finance

Figure 1: Historical Rent-to-Price Ratios

they’re more frequent, but also because the property has had less time to appreciate, something compounded by mortgage repayment schedules, which favor interest over principal in early repayment years). One common metric used to determine whether it makes economic sense for a household to buy or rent is the “rent-to-price ratio,” which is the cost of renting relative to buying a comparable home. Szapiro (2014) explores this topic at length and notes that many homeowners would be better off renting than buying. Despite this and similar research, most Americans (84%) believe that owning a home is a good financial decision.17 This is not that surprising when put in the context of Beracha and Johnson (2012), who note that most homebuyers mostly ignore renting in favor of buying and do not consider the true cost of ownership. It is nevertheless worthwhile to understand the relative costs of each to make a more informed decision. Figure 1 provides some perspective about the historical relative costs of renting versus buying (i.e., the rent-toprice ratio) for various regions in the U.S. since 1988.18 There has been significant variation in the rent-to-price ratio historically in the U.S., especially by region. When the rent-to-price ratio is low (e.g., 4%) the cost of purchasing a home is relatively expensive (because rents are cheap), and vice versa. The actual potential benefit of renting versus 17. 18. 19.

buying is contingent on a number of variables, though, that extend beyond just rent-to-price ratio. To provide some general guidance on what is likely to result in the best outcome for a household (from a pure economic perspective), we developed a model and performed an analysis. The analysis is based on quarterly data on rent-to-price data from the U.S. Census (displayed in Figure 1). Actual historical values are smoothed over three quarters to minimize quarter-to-quarter jumps. The model assumes a household will either purchase a house or rent in that region for some period, ranging from one to 15 years. The household is assumed to use a 20% down payment to purchase the home. The remainder of the purchase (i.e., 80% of the value) is financed via a mortgage with an assumed 30-year duration based on prevailing rates from the Primary Mortgage Market Survey data provided by Freddie Mac.19 The mortgage is assumed to be refinanced if at any time during the homeownership period rates drop by 1% from the current mortgage rate (i.e., it is possible for multiple refinancings during the ownership period). The refinancing is assumed to be costless and each new refinancing is assumed to be for a new 30-year term. Purchase cost is assumed at 2% of the value, and selling cost is 6% of the sale value. Annual maintenance costs are

http://www.realtor.org/sites/default/files/reports/2015/national-pulse-report-2015-10-13.pdf https://www.census.gov/housing/hvs/data/histtab11.xls http://www.freddiemac.com/pmms/pmms_archives.html

Volume 16, Issue 1

Figure 2: Differences in Expected and Actual Duration of Homeownership

assumed to be 2%, which are assumed to include homeowner’s insurance. Real estate taxes are assumed be 1% of the value of home and are assessed annually. Mortgage interest and real estate taxes are assumed to be deductible and are credited to the account at the marginal tax rate assumed in the calculation. The amount of the interest payment is determined annually to capture the fact interest payments decline as a percentage of the total loan payment over time. Gains after selling the home are assumed to be tax-free regardless of the ownership duration or the amount of the gain. The assumed purchase and sale prices are based on the prevailing regional average cost of homes for that quarter. As noted in the previous section, using average sales prices likely overstates the actual future appreciation of the home because the maintenance costs associated with owning the home are not generally considered to be capital improvements. The analysis assumes an equalization of cash flows, where the down payment monies, along with any additional marginal savings (or costs), are invested in a portfolio invested in 50% stocks (represented by the S&P 500 Index) and 50% bonds (represented by the Ibbotson US Long-Term Government Bond Index). The portfolio is assumed to have a 1% annual fee. All gains are assumed to be realized annually from the portfolio account. Taxes paid are based on the marginal tax rate for the bond portion of the portfolio and half the marginal tax rate for the equity portion. All future cash flows are added or subtracted from this account (e.g., if the rent in the following year of purchase is cheaper than

the mortgage than this amount would be an assumed inflow). The value of the portfolio (i.e., side account) is compared to the monies the individual would realize after selling the home. Two key assumptions varied in the analysis are duration of homeownership and the individual’s marginal tax rate. Duration is an important consideration when thinking about homeownership. When researching the potential benefits of homeownership, a variety of values have been used, for example, Beracha and Johnson (2012) use eight years, Belsky, Retsinas, and Duda (2007) use durations of three, five, and seven years, and Rappaport (2010) uses 10 years. For the analysis we assume durations between one and 15 years to provide some perspective on how the potential benefits change over the ownership period. When thinking about the duration of homeownership, it is important to note the consistent disconnect between the actual and expected duration of homeownership in the U.S. According to the annual National Association of Realtors “Home Buyer and Sellers Trends” reports from 2007 to 2016 (the last 10 years) the actual length of homeownership has been significantly less than the expected duration of homeownership, especially for younger households. The average expectations and actual values over the 10-year test period are included in Figure 2. Given the relatively high transaction costs associated with selling (and buying) a home, these observed durations are notable since they suggest homeowners are likely experiencing more costs (e.g., seller’s commissions) than they may have expected when they first purchased the home.

Journal of Personal Finance

The other key assumption varied for the analysis is tax rates. While online calculators commonly assume that homeowners itemize and can therefore deduct various expenses associated with homeownership, such as real estate taxes and mortgage interest payments, only 30% of households in the U.S. itemize their deductions.20 This is even more important for renters, since the vast majority of renters saving to buy a home (92%) do not itemize their Federal taxes (Szapiro, 2014). Therefore, we assume five different levels of marginal tax rates, from 0% to 40% in 10% increments. While it is unlikely a homeowner would be in a 20% marginal tax bracket (since this bracket doesnâ&#x20AC;&#x2122;t exist currently in the U.S. at the Federal level), view these marginal tax rates as approximations, especially if the household crosses thresholds (e.g., if they rent they will not itemize, but if they buy they will be able to itemize half of the homeownership expenses) as well as given the more complex implications associated with the phase outs associated with credits and deductions. The results of the analysis are included in Figure 3. There are two panels. Panel A includes the median return frombuying for the respective scenario, and Panel B includes the probability of being better off buying (i.e., the percentage of scenarios in which the wealth was higher when the individual purchased a home versus rented). 20.

There are a few important takeaways from Figure 3. First, it is unlikely a household is going to be better off buying (again, from a purely economic perspective) if the expected homeownership duration is less than four years at the highest assumed marginal tax rate (40%) and 12 years at the 0% marginal tax rate. While the median real return of 4.38% for a household that stays in a home for 15 years with a 20% marginal tax rate may appear to be attractive, it is important to remember there is a significant deviation in the potential range of outcomes. It is clear, though, that the longer the duration of home ownership and the higher the marginal tax savings (i.e., rate) the higher the probability a household would be better off buying versus renting. These findings are consistent with research by others, such as Szapiro (2014) who notes the typical, median-income prospective homeowner today could generate over 50% more net wealth over the next 10 years by renting and investing instead of buying a home, and Beracha and Johnson (2012) who find that homeowners would have been better off renting and investing in 65% of the 30 years they analyzed. Also, this analysis assumes an average historical rent-to-price ratio between approximately 5.5% and 6.5% across the four regions. If the analysis were to be repeated using U.S. historical rent prices from Davis,

http://taxfoundation.org/blog/who-itemizes-deductions

Figure 3: Buying Versus Renting Panel A: Median Return From Buying

Panel B: Probability of Being Better Off Buying

ÂŠ2017, IARFC. All rights of reproduction in any form reserved.

Volume 16, Issue 1

Lehnert, and Martin (2008), where the average historical rent-to-price ratio is approximately 4.4% during the overlapping period (i.e., about 1.6 percentage points lower, on average), but going back to 1960, the potential benefits of home ownership look much worse, where the median real return even at the 15th year for the 30% marginal tax rate is only 1.02% and the probability of a being better off buying is still only 52.73%. In other words, the assumed rent-to-price ratio has significant implications on the relative attractiveness of owning a home. This is akin to value investing: Buying stocks priced high relative to earnings (or other measures) historically has led to lower future returns than buying underpriced stocks. One key assumption of this analysis is that all monies the investor who decides to rent would have spent on the mortgage are saved. While this may be a valid assumption for an individual who is completely rational from an economic perspective (i.e., homo ecnonomicus), it does not reflect actual household behavior. Homeownership is often described as a form of “forced savings” that results in a higher net worth for households. For example, based on the Federal Reserve’s 2013 Survey of Consumer Finances21, homeowners had on average a net worth of approximately $200,000 versus approximately $5,000 on average for

21. 22.

15 renters. One problem with these general statistics is that they overlook the significant differences in the cohorts that own homes versus rent (i.e., households that own homes tend to have higher incomes, which could potentially explain the entire difference in average net worths). Even after controlling for a household’s propensity to save and accounting for household characteristics, though, homeowners build more wealth than renters do, according to research by Di, Belsky, and Liu (2007), whose findings were mirrored by Boehm and Schlottman (2008), Turner and Luea (2009), and others. Therefore, while homeownership may appear to be less than a great “investment” for some from a pure rate-of-return perspective (when compared to renting), it does appear to be an effective “savings” vehicle, enabling households to accumulate wealth they may have spent elsewhere (so long as the household resides in the house for about five years or more).

Historical Risk and Return of Homes Understanding the role of homeownership within a household’s total wealth is important because it is generally the largest asset and liability. Figure 4 provides some perspective on the real returns of homes, compared to stocks, bonds, and bills, over 20-year rolling periods since 1900, with home prices from Robert Shiller’s website22 and stock,

http://www.federalreserve.gov/econresdata/scf/scfindex.htm http://www.econ.yale.edu/~shiller/data.htm

Figure 4: Stocks, Bonds, Bills... and Homes Panel A: Historical Real Returns: 1900–2015

Panel B: Historical Risk: 1900–2015

Journal of Personal Finance

Figure 5: Differences in Home Price Volatility by Region-Size

bond, and bill returns based on data from the Dimson, Marsh, and Staunton (2012) dataset. Figure 4 gives the impression that from 1900 to 2015 homes have had real returns and standard deviations similar to those of bills; however, it is unlikely the vast majority of homeowners have had this experience. While an investor could have purchased an investment that tracks the historical returns of stocks, bonds, and bills with relative ease,23 the home index tracks the historical performance of a broad index of homes that are regionally diversified, not a single home. Also, the home index is based on repeat sales values, so it likely overstates the true realized return for a homeowner since costs are ignored (as previously noted).

deviation) vary across regions of different sizes. Figure 5 contains the median standard deviation of home prices based on different regions: state, metro, zip code, and neighborhood, from 1996 to 2015 (based on year-end prices). Data for the analysis is from Zillow24 and is based on an annual frequency to minimize any potential implications of smoothing over shorter periods.

In reality there is a strong idiosyncratic component to the return from investing in an individual house; this significantly affects the risk and return attributes of owning a home. It is difficult to measure the volatility of home prices due the illiquid nature of the market. Also, given the fact most housing purchases are made with a consumption motive (rather than investment motive), as well as high transaction costs, heterogeneity in housing, and limits to arbitrage, the “true” value of homes is difficult to assess, which may lead to prolonged periods of “inefficient” pricing (Black, Fraser and Hoesli, 2006).

The median volatility of the home price index clearly increases as the size of the region decreases. The volatility for the entire country (note, the country-level standard deviation is not a median) is consistent with the historical Shiller data in Panel B of Figure 4 where the average rolling 20-year standard deviation was 5.35%. If we were to fit a simple regression model to the results in Figure 5 based on the approximate size of the respective regions, the estimated risk level for an individual home would be approximately 12%, which is approximately equivalent to the historical risk for a portfolio of 60% stocks and 40% bills, or 50% stocks and 50% bonds (using the same historical returns data to create Panel A in Figure 4). This 12% estimate is similar to, but slightly lower than, other estimates for the level of volatility for individual homes. For example, Flavin and Yamashita (2002) estimate the risk of individual homes to be approximately 14%, while Case and Shiller (1989) suggest 15%.

Figure 5 has been included to provide some perspective about how the levels of historical volatility (i.e., standard

Goetzmann (1993) finds that the risk of individual homes is about double the risk of regional portfolios, averaging

23. 24.

At least recently; it would have been more difficult, and more expensive, back in the early 1900s. http://www.zillow.com/research/data/

Volume 16, Issue 1

approximately 11.1% and 5.8%, respectively, in Atlanta, Chicago, Dallas, or San Francisco. These findings are consistent with the relation noted in Figure 5. Goetzmann also notes that regional diversification dominates local diversification, whereby investing in four homes in four different regions, an investor may achieve a reduction in risk roughly comparable to diversifying across thousands of homes in the same region. This is consistent with research by Glaeser and Gyourko (2007) who also note that unlike other financial assets, much of the variation in house prices is local, not national. Shiller makes available historical datasets on home prices for four cities: Atlanta, Chicago, Dallas, and Oakland from 1970 to 1986 on his website.25 While the house price data is somewhat dated, we can use it to estimate the potential distribution of real returns from owning a home over different durations. This is important since it is not clear to what extent the risk of homeownership changes over time. Panel A in Figure 6 includes the distribution of annualized real returns for individual home sales on a percentile basis, and Panel B includes information about the standard deviation of the returns by holding period.

25. 26.

Many potential homeowners may believe the risks associated with owning a home decline over time. A similar belief exists among many investors for equities, and is commonly referred to as “time diversification.” If an asset has negative autocorrelation, it may in fact become less risky over longer periods (versus if the returns are independent and identically distributed, or iid). Blanchett, Finke, and Pfau (2016) demonstrate that this “time diversification” effect has in fact existed historically for equities in 19 of 20 countries studied. Similarly, Panel B suggests (at least for the U.S. over the test period) that the benefit of time diversification also exists in the U.S., i.e., the annualized standard deviation declines by one over the square root of the investment period. This is consistent with the research of Belsky, Retsinas, and Duda (2007), who found that holding property for more time generally did not result in a greater chance of making the home a good investment. Figures 7 and 8 provide some perspective about the differences in real returns and volatilities of home indexes by U.S. city (Figure 7), based on Case-Shiller data, as well as by country (Figure 8), using data obtained from the Federal Reserve Bank of Dallas26 as described in Mack and Martínez-García (2011).

http://www.econ.yale.edu/~shiller/data.htm http://www.dallasfed.org/institute/houseprice/

Figure 6: Shiller Individual Home Sales Dataset: 1970–1986 Panel A: Individual Home Real Return Distribution

Panel B: Standard Deviation by Holding Period

Journal of Personal Finance

There have been significant differences in the real returns and volatility in the U.S. for city-specific Case-Shiller Indexes. For example, Portland has had a relatively high return with low risk (at 2.87% and 6.41%, respectively) while Las Vegas has had the opposite (at 0.20% and 13.24%, respectively). There is a positive relation between return and risk across cities, where cities with higher returns have had higher volatilities, although the strength of the relationship is relatively weak (with an R² of 14.66%).

has had less risk and a lower return than most countries. While the returns in Figure 8 may appear to be over a long historical period, it is only 41 years (from 1975 to 2015 inclusive). Using a sample of high-quality homes from Amsterdam from 1628 to 1974, Eichholtz (1997) finds that the nominal (real) return for homes in Amsterdam over the entire period was 1.8% (0.5%), which is lower than the allcountry real return average in Figure 8 (which is 1.84%)

Market Risk of Homes Similar to the U.S. cities, there have been significant differences in the historical real returns and volatiles across countries, although there is virtually no relation across real return and volatility metrics (with an R² of .28%). The U.S.

While homes have a relatively high level of idiosyncratic risk, their level of systematic risk is less clear. Ibbotson and Siegel (1984) compare the returns of residential real estate

Figure 7: Historical Real Return and Risks for Case-Shiller Indexes: 1988–2015

Figure 8: Historical Real Return and Risks for Country House Price Indexes: 1975–2015

Volume 16, Issue 1

to stocks and bonds from 1960 to 1982 and find that the stock beta of residential real estate is relatively low, at 0.05, and that real estate returns are not explained very well— either yearly or over the whole period studied—by market or inflation risk. They also note that smoothing techniques used may understate the true risk of residential real estate. Additional research by Goetzmann (1993), Flavin, and Yamashita (2002), and De Roon, Eichholtz, Koedijk (2012) all find low correlations between housing returns and other assets, which suggests that housing can be a potential diversifying asset from a total wealth perspective.

The final regression factor is liquidity. A liquid stock is one where it is possible to trade large quantities of the stock quickly, at low cost, without affecting the price. Research by Pastor and Stambaugh (2003), among others, demonstrates that illiquid securities tend to outperform more liquid stocks, which suggests the existence of a liquidity premium (LIQ). We use a monthly liquidity factor available on Pastor’s website.28 The five-factor regression formula is noted in equation 1.

To determine the market risk of homes, we used a factor regression approach. The approach begins with the threefactor model developed by Fama and French (1993), who demonstrate the existence of a small-cap premium (SMB) and value premium (HML) in addition to the market portfolio (RMkt). In addition to these three factors, a momentum (MOM) factor has also been included. Momentum is the effect reported by Jegadeesh and Titman (1993), among others, where stocks that have performed well (poorly) historically tend to continue to perform well (poorly). The risk-free (Rf ) return, SMB factor, HML factor, and MOM factor values are obtained from the data library on Kenneth French’s website.27

The analysis is based on data from Zillow for 4,561 neighborhoods from 1996 to 2015. Neighborhood is selected as the region (e.g., versus state or zip code) since neighborhood is the “smallest” region increment available and therefore is most likely to represent the idiosyncratic risk associated with a single home. To be included in the dataset, returns over the entire period must be available. The 5th, 25th, 50th, 75th, and 95th percentiles and average values (across all 4,561 regressions) are included in Figure 9. The actual values for the 50th percentile and average coefficients are included in Figure 9 as labels for reference purposes.

27. 28.

Rh– Rf = α + B1 (RMkt– Rf ) + B2 (SMB) + B3 (HML) + B4 (MOM) + B5 (LIQ) + ε [1]

The regression results suggest low systematic risk exists in single-home returns. We observed low coefficients of

http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html http://faculty.chicagobooth.edu/lubos.pastor/research/liq_data_1962_2015.txt

Figure 9: Average Market Factors for Various Domestic Home Price Indexes

Journal of Personal Finance

correlation between home returns and stock returns, as well as to the size, value, momentum and liquidity factors. Market coefficients were relatively small (averaging 0.05, but ranging between negative 0.07 and 0.23) with a slight value tilt (averaging 0.06 and ranging between negative 0.08 and 0.25).

Homes Are Not REITs REITs might seem to be a reasonable investment proxy or substitute for homes, since both are real estate-based investments, but this does not appear to be true. Cotter and Roll (2015) compare the returns of REITs to the S&P/CaseShiller Home Price Indexes and find that the Case-Shiller Indexes have a market beta that is approximately one-fifth the total volatility of REITs (0.58) and much higher levels of auto-correlation, especially at the monthly level. Waggle and Johnson (2004) model the risk of individual home standard deviation of 8.81% versus 14.98% for REITs. Table 1 shows the return, standard deviation, and market coefficient (from a single-factor model) for various REIT indexes (data obtained from NAREIT29) versus Case-Shiller indexes from 1994 to 2015. While both the REITs and Case-Shiller indexes are diversified indexes consisting of multiple properties (unlike individual homes), the historical return, standard deviation, and market risk for REITs have been significantly higher (more than double) those of respective home price indexes. It is important to note that the differences in returns of REITs and home prices are not just a domestic effect. Table 2 compares the returns, standard deviations, and one-factor betas for house price indexes using data obtained from the Federal Reserve Bank of Dallas30 as described in Mack and Martínez-García (2011), and REIT indexes from the FTSE. Similar to the domestic analysis, the return, risk, and market beta of international REITs have been significantly higher than those for domestic house price indexes. Perhaps even more interesting is that REITs have been a much less attractive investment internationally, with lower returns, higher risk, and a higher market beta when compared to U.S. REITs. While data on the risk of individual homes in other countries over this period is not readily available, even if the risk of individual homes was twice the house price index, the

29. 30.

risk (i.e., standard deviation) of homes would still be less than half that of REITs.

Allocating to Homes Up to this point, we have explored the risk of single-home investments in isolation from an investor’s or a household’s total wealth. It’s important, though, to consider homeownership in the context of total wealth, which includes both financial assets (such as retirement savings) and nonfinancial assets (such as human capital). Past research on appropriate housing allocations has delivered mixed findings. Englund, Hwang and Quigley (2002) study single-family housing returns in Stockholm, Sweden, from January 1981 to August 1993 and find that an efficient portfolio would include no housing for shorter periods, but for longer periods, low-risk portfolios would include between 15% and 50% housing. Goetzmann (1993) suggests an allocation of 50%, while de Roon, Eichholtz, and Koedijk (2002) note 30%. Assumptions around things like maintenance costs, taxes, expected duration of homeownership, etc., will have a significant impact on the analysis. The ideal analysis would likely contain different types of optimizations based on how the home fits into the investor’s or household’s total wealth and financial goals. For example, running some type of surplus optimization, where a liability is created to represent the effective short position the household has with respect to home consumption would be one potential approach; however, for simplicity purposes we use a more traditional optimization for this analysis. We use three main assumptions for the optimization: level of down payment, risk-aversion level, and whether or not the analysis includes home costs. The level of down payment can vary considerably by homeowner, so three levels are considered: 5%, 20%, and 100% (i.e., the home is purchased with cash). Additionally, three different levels of risk aversion are considered—low, moderate, and high—calibrated such that, in the absence of real estate as an available investment, the household would invest in a portfolio that is 80%, 50%, or 20% equities, respectively (which corresponds to risk-aversion coefficients of 1.2, 2.8, and 11.0, respectively, in the model).

https://www.reit.com/investing/index-data/monthly-index-values-returns http://www.dallasfed.org/institute/houseprice/

Volume 16, Issue 1

Table 1: A U.S. Perspective on Homes Versus REITs

Table 2: An International Perspective on Homes Versus REITs: 1990-2015

We consider three distinct costs: real estate taxes, general maintenance costs, and transaction costs. The analysis is based on the actual historical returns of house-price indexes for a given zip code, so the real estate taxes are based on the state-level tax rate based on values obtained from the Tax Foundation.31 The real estate tax is reduced by the assumed marginal tax rate of 25%, under the assumption the household can deduct the real estate taxes. 31.

http://taxfoundation.org/blog/how-high-are-property-taxes-your-state

Second, annual maintenance costs are assumed to be 2%. Third, total round-trip buy/sell costs of 8% are assumed to be amortized evenly over the holding period, which is assumed to be 10 years. When costs are included, the quarterly returns are reduced evenly by the total-cost value. We do not consider the costs associated with the mortgage nor the exclusion of capital gains taxes if a home is sold as a primary residence. While these are both important

Journal of Personal Finance

considerations for each household, financing costs must be paid regardless of when a leveraged purchase is made, and the potential exclusion of taxes only applies when the home is the primary residence of a household, and would not necessarily apply to investments held in certain account-types (e.g., a Roth IRA). Certain costs included would be paid whether the household rents or buys. If the household owns the home, the costs would be explicit (i.e., they would be paid out of pocket). In contrast, the costs would be factored into the cost of renting, since the landlord would be forced to pay these expenses and would therefore pass them on to the renter. Therefore, a blend of the “no costs” scenario” and “with costs” scenario is likely the best representation for a homeowner, while the “with costs” scenario is likely a reasonable approximation from the perspective of an individual looking to buy a home purely for speculative purposes who has no intent of staying in, or renting out, the property.

Third, the allocation to one of the 10 base investment asset classes cannot exceed 40% to ensure a portfolio of at least three asset classes. No maximum allocation constraint is imposed on the home allocation. An investment fee (i.e., expense ratio) of 20 basis points is assumed for the 10 investment asset classes to reflect reasonable costs associated with passive investing. Opposed to that used in a more traditional mean-variance optimization, our objective function is based on the Constant Real Risk Aversion (CRRA) utility function, where we target three risk-aversion levels (y) by maximizing equation 2, where t is the respective quarter, T is the total number of quarters, w is a vector of asset class weights, r is vector of returns (in percent), and c is some constant (which is set equal to 1). The risk-aversion levels for an aggressive, moderate, and conservative household are estimated to be 1.2, 2.8, and 11.0, respectively (these correspond to portfolios with equity allocations of 80%, 50%, and 20%, respectively). We use this objective function, versus a more traditional mean variance optimization approach, since we do not want to target a single return or standard deviation, but rather the portfolio that generates the optimal tradeoff between the two. χ ((‫ ) ޸ݎ ݓ‬+ ܿሻͽ·й σ ‫ݪ‬ ͽ·‫މ‬ ‫=ݐ‬0 1 – ‫ݕ‬ MAX 1–y [2] ܶ

The optimizations are based on quarterly changes in home prices for various zip codes from the third quarter of 1996 to fourth quarter of 2015. This period is selected since it is the range of available data from homes in the Zillow zip code database.32 Data must be available for the entire period for a zip code to be included. This limits the analysis to 10,368 specific regions (i.e., zip codes). We use quarterly returns (versus annual returns in previous studies) to allow for a greater number of return series in the optimization.

Table 4 contains the average allocations to home values for the different scenarios.

In addition to the “home,” which is considered a unique asset class in the optimization, 10 other asset classes are included—five equity and five fixed-income (see Table 3). We constrained the optimizer in three ways. First, no shorting (i.e., all weights must be greater than or equal to zero). Second, the sum of all weights must equal 100%.

The most notable difference in the scenarios is the impact of including home costs on the analysis, which significantly reduces the relative attractiveness of homes from an investment perspective. The average allocation with home costs across the nine other scenarios (by risk-aversion and down-payment level) is 2.0% versus 29.3% when excluding

32.

[((

http://www.zillow.com/research/data/

Table 3: Asset Classes Used

)

)]

Volume 16, Issue 1

Table 4: Average Home Allocation in Optimizations Panel A: Include Home Costs

home costs. It appears homes are relatively more attractive for investors with lower levels of risk aversion and moderate down payments (e.g., 20%). There are certain regions that receive much higher average allocations to homes. For example, if we group the zip codes by state, the highest average allocations to homes are California (39.0%), Hawaii (19.0%), and Massachusetts (15.6%). Not surprisingly these are the three states with the highest average quarterly returns over the test period (at 1.45%, 1.22%, and 1.12%, respectively). While risk (i.e., standard deviation) is also important, a multivariate regression where the dependent variable is the home allocation and the independent variables are normalized coefficients for returns, standard deviation, and market risk suggest that return is approximately five times more important than risk when explaining the variation in the home allocation among the portfolios, and approximately 20 times more important than market beta.

Regional Differences in Home Returns, Volatility, and Market Risk There are notable differences in the returns, risk, and market betas for home indexes in different regions. In this section we aim to provide some insight into some of the drivers of those differences so that the reader can more accurately assess the risks associated with homes by region. This analysis, again, is based on the Zillow home price indexes for zip codes from December 1996 to December 2015. We calculated the annual change in the home index, and from these values the real return, standard deviation, and market risk (using the five-factor model in equation 1) 33. 34.

http://www.bls.gov/lau/#tables https://www.census.gov/programs-surveys/acs/

Panel B: Exclude Home Costs

are determined. These three variables are the dependent variables for the OLS regressions. We included a variety of independent variables in the regression. It is important to note that while the dependent variables are based on some kind of variable throughout the period (from December 1996 to December 2015), the independent variables are mostly a single point in time (generally targeting the year 2010, based on availability). Therefore, when interpreting the coefficients, it’s going to be more important to consider the attributes of that zip code today, versus where it was historically, since many communities are changing over time. The independent variables included in the regression are: •

Price: Source: Zillow, prices as of December 2010

•

County Unemployment Rate: Source: Bureau of Labor Statistics 2010 Annual Averages33

•

Maximum Industry Exposure: Source: Custom calculations, primary data from the 2012 Economic Census of the United States

•

% of Households Family: Source: American Community Survey 201034

•

Population: Source: American Community Survey 2010

•

Education >= Bachelor’s: Source: American Community Survey 2010

•

%Same House as 1 Year Ago: Source: American Community Survey 2010

•

%Born in USA: Source: American Community Survey 2010

•

%HHs English Only: Source: American Community Survey 2010

•

Size of Zip Code (Square Miles): Source: American Community Survey 2010

Journal of Personal Finance

•

Housing Turnover: Source: Source: Zillow, turnover as of December 2010

•

Average Annual Temperature: Source: CDC35

•

Average Home Size: Zillow, prices as of December 2010

Descriptive statistics for the variables have been included in Table 5. We normalized all independent variables (by subtracting the mean and dividing by the standard deviation) to make it easier to compare the relative importance of the variables in the regression. For example, if the average home price in your zip code is the same as the national average ($219,619), the assumed value would be zero. If instead the home price in the area is $400,000, which is approximately one standard deviation above the average, the assumed value would be 1. Note that it would have been possible to increase the explanatory power of the model by transforming some of the independent variables (e.g., taking the natural log of the home Price); however, the raw values

Table 5: Descriptive Statistics for Regression Variables

are used so that the coefficients are easier to interpret by readers of varying levels of financial sophistication. Also, transforming the variables had only a minor impact on the results. The results of the three multivariate OLS regressions are included in Table 6. For returns, each of the coefficients were significant at the 1% level. The five most meaningful variables (based on the absolute value of the regression coefficient) were price, average home size, the percentage of households that speak only English, the percentage of occupants born in the U.S., and the average annual temperature, respectively. These five factors suggest that homes located in areas with other high-priced homes, that are relatively small (i.e., located in cities), with a diverse racial population, and warm temperature have experienced the highest average returns historically. For volatility, all of the coefficients were significant at the 1% level, except maximum industry exposure (not statistically significant) and percentage of households that are families (statistically significant at the 5% level). The five most meaningful variables (based on the absolute value of the regression coefficient) were the percentage of occupants born in the U.S., the county employment rate, percentage of households that speak only English, housing turnover, and percentage of households with at least college degrees, respectively. These five factors suggest that homes located in areas with a diverse racial population that is well educated, low unemployment, and low housing turnover have experienced the lowest levels of volatility historically (i.e., those that have thrived for the entire period of analysis). For market risk, all of the coefficients were significant at the 1% level, except average home size, which was statistically significant at the 5% level. The five most meaningful variables (based on the absolute value of the regression coefficient) were the same as volatility regression (with identical signs). The similar results are not that surprising given the relatively high correlation between volatility and market risk (0.818). Again, though, the regressions suggest that homes located in areas with a diverse racial population that is well-educated, has low unemployment and low housing turnover have experienced the lowest levels of market risk.

35.

http://wonder.cdc.gov/nasa-nldas.html

Volume 16, Issue 1

Table 6: The Drivers of Return, Risk, and Market Beta for House Price Zip Code Indexes

No Portfolio Is an Island: The Impact of the Home When building portfolios, investment professionals and financial planners often focus entirely on the available set of investible assets for the financial portfolio, such as stocks and bonds. In reality, the financial assets are only one of many assets owned by the client, and considering the risks of the “other” assets can be an essential aspect to building truly efficient portfolios when viewed from a total wealth perspective.36 For example, Cochrane (2007) contends that the optimal portfolio for an investor should deviate from the market portfolio to the extent that he or she is different from the average person. When thinking about the impact of the home on a portfolio, the first decision is whether to purchase a home at all. Our analysis suggests many households are likely better off renting, especially those who are likely to be in the home for less than six years and those paying a relatively low tax rate. However, to realize the potential benefit of renting, the 36. 37.

household must be disciplined and be willing to “save the difference” between renting and owning. It is worth noting that while the other potential benefits associated with owning a home may outweigh the investment considerations (e.g., the local school districts37 ) it is important to be aware of the additional potential costs associated with purchasing a home so that the household can plan accordingly. If an individual decides to purchase a home, the key risk to consider is the idiosyncratic risk associated with that home. The risks associated with owning a single home are very different than those for a diversified portfolio of homes (e.g., a Case-Shiller Index). While many people consider houses to be a “safe” investment, the historical volatility of individual homes has been approximately double that of city-specific home price indexes, with an annual standard deviation of 12%. This level of volatility has historically been associated with a portfolio that is 60% stocks and 40% bills. Therefore, a home is hardly a risk-free asset. The risks of home equity are further magnified by leverage. For example, a household that puts a 20% down payment on a home has

See “No Portfolio Is an Island” by Blanchett and Straehl (2015) for additional work on this topic. Or a spouse who is unwilling to be a “renter.”

Journal of Personal Finance

an asset with an equivalent level of volatility as large as a three-times leveraged stock index, easily the riskiest asset in the household portfolio. It is worth noting (again) that there is very little market risk for homes (the average market coefficient for the five-factor regressions was 0.18), and relatively little exposure to the size or value factors as well. Unlike other assets that have an obvious and somewhat intuitive impact on how a household should invest their financial assets, the impact of homeownership is less clear. For example, a household with a fully accrued pension benefit has an asset that is effectively bond-like, and can therefore take more risk in the remainder of their portfolio (versus the average household that may have fewer pension benefits). Defined-benefit pensions, for the most part, are all the same in terms of their impact on a household’s total wealth. In contrast, the potential impact of homeownership depends on things like the unique risks associated with the home (e.g., is it in an economically diverse area?), the extent of leverage used to purchase the home, the household’s other wealth and non-financial assets (e.g., human capital and pensions), and the household’s housing goals (e.g., does the household plan on staying in the house into the foreseeable future). Therefore, the true impact of a home is less clear and is going to be largely driven by that household’s unique facts and circumstances.

equivalent to the historical volatility of a portfolio invested in 60% (50%) stocks and 40% (50%) bills (bonds). While the return on house price indexes has exceeded inflation historically (a real return of approximately 1%), the actual real return realized by homeowners, after considering the various costs associated with owning and selling a home, is likely lower than inflation and potentially negative (in nominal terms). Therefore, renting is likely a better option for many households, especially those with lower marginal tax rates (i.e., households that do not itemize deductions) and those who plan to move again within a few years. We also note significant differences in the returns, volatility, and market risk of homes and REITs, which suggests REITs are a relatively poor proxy for residential real estate from an investment perspective. And we have identified some of the factors, such as home price, county unemployment rate, housing turnover, home size, and even average annual temperature, that can differ by region and are strongly related to the returns, volatility and market risk of homeownership. Households may use this factor information to better approximate the risk of their homes. Overall, the impact of owning a home on the optimal total wealth financial portfolio is likely to vary significantly by household, based on the unique risks associated with the home, household wealth, and other non-financial household assets.

Conclusions A home is a unique household asset since it is both an investment and a consumption good. A home is an investment good such that it allows the household to accumulate wealth (i.e., equity in the home) over the duration of homeownership, and a consumption good since it provides shelter. When assessing the value of homeownership, it is important to understand both aspects—that is, to understand the risks associated with owning a home and whether it is the best way of providing shelter (as opposed to renting). We explored both of these perspectives at some length. We found there are significant risks associated with homeownership. These risks are primarily idiosyncratic (i.e., not market-related), driven largely by the illiquid nature of owning a single home. We found the risk of homeownership is approximately double that of city-specific home price indexes (e.g., Case-Shiller indexes), with an annual standard deviation of 12%, which is approximately

References Andrews, Dan and Aida Caldera Sánchez. 2011. “The Evolution of Homeownership Rates in Selected OECD Countries: Demographic and Public Policy Influences.” OECD Journal: Economic Studies, vol. 2011/1. Barnes, Yolande. 2016. “Around the World in Dollars and Cents.” Savills. http://www.savills.co.uk/ research_articles/188294/198667-0 Belsky, Eric, Nicolas Retsinas, and Mark Duda. 2007. The Financial Returns to Low-income Homeownership. Ithaca and London: Cornell University Press. Beracha, Eli, and Ken H. Johnson. 2012. “Lessons from Over 30 Years of Buy versus Rent Decisions: Is the American Dream Always Wise?” Real Estate Economics, vol. 40, no. 2: 217-247.

Volume 16, Issue 1

Black, Angela, Patricia Fraser, and Martin Hoesli. 2006. “House Prices, Fundamentals and Bubbles.” Journal of Business Finance and Accounting, vol. 33, no. 9‐10: 1535-1555.

Englund, Peter, Min Hwang, and John M. Quigley. 2002. “Hedging Housing Risk.” The Journal of Real Estate Finance and Economics, vol. 24, no.1-2: 167-200.

Blanchett, David, Michael Finke, and Wade Pfau. 2016. “Optimal Portfolios for the Long Run.” Working Paper.

Fama, Eugene F., and Kenneth R. French. 1993. “Common Risk Factors in the Returns on Stocks and Bonds.” Journal of Financial Economics, vol. 33, no. 1: 3-56.

Blanchett, David, and Philip Straehl. 2015. “No Portfolio Is an Island.” Financial Analysts Journal, vol. 71, no. 3: 15-33. Boehm, Thomas P., and Alan Schlottmann. 2008. “Wealth Accumulation and Homeownership: Evidence for Low-Income Households.” Cityscape: A Journal of Policy Development and Research, vol. 10, no. 2: 225-256. Case, Karl E. and Robert J. Shiller. 1989. “The Efficiency of the Market for Single Family Homes.” American Economic Review, vol. 79, no. 1: 125-137. Cochrane, John H. 2007. “Portfolio Theory.” February 2007, http://faculty.chicagobooth.edu/john.cochrane/research/ papers/portfolio_text.pdf. Cotter, John, and Richard Roll. 2015. “A Comparative Anatomy of Residential REITs and Private Real Estate Markets: Returns, Risks and Distributional Characteristics.” Real Estate Economics, vol. 43, no. 1: 209-240. Davis, Morris A., Andreas Lehnert, and Robert F. Martin. 2008. “The Rent‐Price Ratio for the Aggregate Stock of Owner‐occupied Housing.” Review of Income and Wealth, vol. 54, no. 2: 279-284. De Roon, Frans, Piet Eichholtz, and Kees C.G. Koedijk. 2002. “The Portfolio Implications of Home Ownership.” Centre for Economic and Policy Research. No. 3501. DiPasquale, Denise, and Edward L. Glaeser. 1999. “Incentives and Social Capital: Are Homeowners Better Citizens?” Journal of Urban Economics, vol. 45, no. 2: 354-384. Di, Zhu Xiao, Eric Belsky, and Xiaodong Liu. 2007. “Do Homeowners Achieve More Household Wealth in the Long Run?” Journal of Housing Economics, vol. 16, no. 3: 274-290. Emrath, Paul. 1997. “A Long-Run House Price Index: The Herengracht Index, 1628-1973.” Real Estate Economics, vol. 25, no. 2: 175-192.

Flavin, Marjorie, and Takashi Yamashita. 2002. “Owner-occupied Housing and the Composition of the Household Portfolio.” The American Economic Review, vol. 92, no. 1: 345-362. Glaeser, Edward L., and Joseph Gyourko. 2006. “Housing Dynamics.” No. w12787. National Bureau of Economic Research. Goetzmann, William N. 1993. “The Single Family Home in the Investment Portfolio.” Journal of Real Estate Finance and Economics, vol. 6, no. 3: 201-22. Ibbotson, Roger and Laurence Siegel. 1984. “Real Estate Returns: A Comparison with Other Investments.” Journal of American Real Estate and Urban Economics Association, vol. 12, no. 3: 219-42 Jegadeesh, Narasimhan, and Sheridan Titman. 1993. “Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency.” The Journal of Finance, vol. 48, no. 1: 65-91. Mack, Adrienne, and Enrique Martínez-García. 2011. “A Cross-Country Quarterly Database of Real House Prices: A Methodological Note.” Globalization and Monetary Policy Institute Working Paper 99. National Association of Realtors. Home Buyer and Sellers Trends. Reports from 2007 to 2016. Available online. Pastor, Lubos and Robert F. Stambaugh. 2003. “Liquidity Risk and Expected Stock Returns.” The Journal of Political Economy, vol. 111, no. 3: 642-685. Rappaport, Jordan. 2010. “The Effectiveness of Homeownership in Building Household Wealth.” Federal Reserve Bank of Kansas City, Economic Review, vol. 95, no. 4: 35-65. Smith, Margaret Hwang, and Gary Smith. 2006. “Bubble, Bubble, Where’s the Housing Bubble?” Brookings Papers on Economic Activity 2006, no. 1: 1-67.

Journal of Personal Finance

Szapiro, Aron. 2014. “House of Cards: The Misunderstood Consumer Finance of Homeownership.” HelloWallet White Paper. Turner, Tracy M., and Heather Luea. 2009. “Homeownership, Wealth Accumulation and Income Status.” Journal of Housing Economics, vol. 18, no. 2: 104-114. Waggle, Doug, and Don Johnson. 2004. “Home Ownership and the Decision to Invest in REITs.” Journal of Real Estate Portfolio Management, vol. 10, no. 2: 129-144.

Volume 16, Issue 1

The Impact of Rates of Return on Roth Conversion Decisions and Retiree Savings Wealth

Lewis Coopersmith, Ph. D., Professor Emeritus of Management Sciences, Rider University, Lawrenceville, NJ, and founder and Chief Research Officer of VestaEdge, Inc. Alan R. Sumutka, MBA, CPA, CGMA, Associate Professor of Accounting at Rider University, Lawrenceville, NJ.

Abstract A tax-optimal retirement savings withdrawal model, implemented as a linear programming application, is used to compare savings wealth growth when Roth conversions are permitted (RC) and when they are not (NoRC). Evaluations are made for combinations of percentage rates of return (ROR) for taxable, tax-deferred, and tax-free savings. PctDiff, the difference between tax-free and tax-deferred (TD) account RORs, is an important conversion consideration. When investment strategies target PctDiffs at two percent or greater, RC provides substantial benefits. As PctDiff increases, the percentage of initial TD savings that should be converted rises, the time to recover savings wealth lost to conversion-related taxes declines, and savings wealth growth surges. When PctDiff is less than two percent, savings wealth growth is small and savings wealth loss due to conversion-generated taxes persists for more than 13 years; retiree health and prospects of living long enough to realize savings wealth gains becomes a vital concern. Conversions are best made relatively early in retirement and at varying annual amounts. Conversion of all initial tax-deferred savings in the first year of retirement rarely results in maximum savings wealth growth.

Journal of Personal Finance

Introduction Generally, Roth IRAs are coveted for their tax-free earnings and withdrawals. They are funded by nondeductible contributions from earned income (for example, salary) or from taxable Roth conversions from traditional IRAs. Since 2010, there has been no limit on the amount of traditional IRA savings that can be converted to Roth IRAs, which has led financial advisors to encourage their clients to execute Roth conversions. Ed Slott, a Roth conversion proponent, opines: “The Roth IRA is the single best gift Congress has ever presented to the American taxpayer. It allows us to build retirement accounts that, over the long haul, will grow to incredible size – and remain free of income tax forever”, Slott (2003). In 2010 “Slott…says he converted nearly all his six-figure IRA in January [of 2010]”, Saunders (2010). Because Roth conversions are taxable, they can cause immediate wealth reduction. However, the size and duration of this reduction is not well understood. Keebler (2010) states that Roth conversion planning is complex and advisors and CPAs must work together to determine which clients should convert and which clients should maintain their Traditional IRA. The factors that support some level of conversion generally outweigh the factors that favor no conversion. In an overgeneralization, the wealthier a person is, the greater the need to carefully study the Roth conversion question.

Literature Review Recent publications compare Roth conversion (RC) strategies to No Roth conversion (NoRC) strategies and focus on the merits of each strategy under varying marginal tax rate assumptions at the time of conversion and withdrawal. Geissler and Hulce (2014) conclude that traditional IRAs are not always preferable when a lower income tax rate is expected during retirement. Washer (2011) examines four strategies that differ by Roth conversion amounts. He determines the levels of marginal tax rates at which each strategy is preferred relative to wealth growth and longevity. Clayton, Davis and Fielding (2012) evaluate when to convert based on marginal tax rates but they do not

include taxable savings nor adequately address complicating factors such as required minimum distributions (RMDs) and alternative rates of return (RORs) for Roth and tax-deferred savings. They conclude that the decision to convert to a Roth IRA remains an individual one and may be influenced by factors other than the time to break even. Krishnan and Cumbie (2016) also evaluate the Roth conversion decision based on marginal tax rates and consider taxable savings. Cellucci (2014) defines specific strategies to convert an entire traditional IRA to a Roth IRA and uses Monte Carlo simulation to determine when to convert based on the tax impact at that time. All of these studies conclude that some amount of a Roth conversion provides greater benefits than no amount of Roth conversion when marginal tax rates are expected to be higher in retirement than at the time of conversion. However, the conversion amount is not easily forecasted and there is no assessment of the impact on total savings wealth. Welch (2016) uses a tax-optimal linear programming model to evaluate whether to execute Roth conversions based on the metric of disposable income. He concludes that there is no significant economic benefit to conversions. He observes that the conversion option pays less total tax than no conversion option and the patterns of tax payments are totally different. While there are many similarities between Welch and this research, the differences in the model, methodology and research objectives are contrasted throughout this paper. An early use of a tax-optimal withdrawal strategy is described in Ragsdale, Seila, and Little (1994). The linear programming formulation used in this study is based on the formulation described in Coopersmith and Sumutka (2011) and Coopersmith, Sumutka, and Arvesen (2009). Their formulation was enhanced to include Roth IRA accounts and IRA to Roth IRA conversions. Welch (2015) demonstrates the validity of the linear programming approach to tax minimization and the superiority of linear programming over a common practice simulator. Our results and methodology contribute to the body of literature demonstrating how linear programming is a useful platform for evaluating alternate strategies in retirement financial planning.

Volume 16, Issue 1

Research Design This research extends previous Roth conversion studies in two respects: (1) A tax-optimization model is used to determine withdrawals from three account types: taxable (TX), tax-deferred (TD), and tax-free Roth (TF) savings. The objective of the linear programming model used in this research is to maximize savings wealth at the end of the planning horizon, as measured by the final total savings of the three account types, referred to as Final Total Account Balance (FTAB). This is one of the essential differences between our methodology and Welch’s. We justify using maximization of FTAB as our optimization objective based on both our study of personal finance and personal experience with wealth managers in their practice of retirement planning. We have observed that retirement financial planning generally starts with setting a budget that includes pre-tax itemized living expenses based on lifestyle expectations. This is followed by planning in order to satisfy this budget plus taxes from income sources and retirement savings. We also observed that effective management of retirement savings and investments not only reduces the chance that savings will be exhausted during the planning horizon but could actually result in savings growth. Additional justification for our choice of optimization direction is that it allows an appropriate comparison of savings wealth growth between RC and NoRC when the set of model input data that includes pre-tax living expenses is fixed. We believe our optimization objective of maximizing FTAB is most appropriate for evaluating this paper’s key comparative measures of savings wealth reduction resulting from Roth conversion, as discussed below in how we extend Welch’s work. The importance of growing savings wealth in retirement is reflected in the quotes in the Introduction. Furthermore, all authors other than Welch in the Literature Review note the importance of achieving high levels of final savings wealth in their evaluation of Roth conversion strategies. (2) RC strategies are compared to NoRC strategies for various combinations of RORs by account type to analyze the impact of each strategy on retiree total savings wealth over a 30-year retirement planning horizon.

The tax-optimization model used in this research is tax-efficient according to the criteria stated in Sumutka, Sumutka, and Coopersmith (2012): •

“Consideration of more than one annual withdrawal strategy over a retirement horizon—each strategy differs by the sequence of withdrawals from different accounts, and each account has a different tax treatment

•

A realistic calculation of taxes for each strategy

•

Selection of the best strategy with respect to some performance measure (for example, final total account balance)”

A “tax-optimal” retirement savings withdrawal strategy results when the set of withdrawal strategies evaluated is all possible strategies. Tax optimization is achieved through the use of a mathematical programming model and its associated solution algorithm. An important reason for the use of tax-optimization is that for any given set of initial data, the tax-optimal strategy results in the mathematically maximum FTAB. In comparing RC to NoRC, annual tax-optimal withdrawals are determined for each strategy for the same set of model input data. For RC, the model also determines the optimal amount of annual Roth conversions that maximize the FTAB. The resulting Roth conversion amounts can vary from zero to the entire balance of the previous year’s TD savings. Thus, RC results could be: (a) the same as NoRC (the FTAB cannot be more than the NoRC strategy at any level of Roth conversion), (b) a partial Roth conversion (some Roth conversions in some years, but not in others), or (c) full Roth conversion (a one-time Roth conversion of all TD savings) Therefore, the annual withdrawal and/or conversion decisions are more effective than those prescribed by a rule-based strategy, such as the common rule which uses a predetermined sequencing of withdrawals (for example, exhaust sequentially TX, TD, and finally TF). We extend Welch’s work in these ways: •

Instead of annual spending we measure the improvement in FTAB for RC over NoRC.

Journal of Personal Finance

•

Welch uses the same RORs for account type while this study varies the ROR by account.

•

We evaluate the amount of savings wealth reduction: the greatest annual total savings wealth reduction of RC compared to NoRC early in the plan.

•

We report the number of years it takes before the total savings wealth of RC exceeds that of NoRC (i.e., the savings wealth break-even point).

•

We use a Monte Carlo Method to compute the difference in longevity risk between RC and NoRC. Longevity risk is defined as the chance of the plan failing before the end of the planning horizon.

The tax model incorporates the following for 2015: •

Federal tax brackets, specified itemized deductions (which always exceed standard deductions), and exemptions are increased annually by a 2 percent inflation rate.

•

Qualified dividend income (QDI) and long-term capital gains (LTCG) are taxed at their favorable rates of zero percent, 15 percent, or 20 percent.

•

All other income (for example, interest, non-qualified dividends, RMDs, etc.) is taxed at ordinary income tax rates, from 10 percent to 39.6 percent.

The study data assumes a 65-year-old retired couple with $2,000,000 in 2014 year-end savings, which is allocated among three accounts: TD = $1,400,000 (70%), TX = $400,000 (20%), and TF = $200,000 (10%). The following assumptions are fixed for all analyses: Pre-tax living expenses: Total 2014 pre-federal tax living expenses are $115,000, comprised of $34,500 in itemized deductions (as opposed to $12,600 standard deduction), and $80,500 in other living expenses, which are increased annually by a 2 percent inflation rate. Pre-federal tax living expenses are the same for the two strategies. Income sources: 2014 Social Security is $30,000 ($15,000 for each spouse) and increases annually by a 2% inflation rate. To satisfy the remaining $85,000 of cash needs ($115,000 pre-federal tax living expenses minus $30,000 of Social Security), the couple withdraws 4.25% from their 2014 year-end initial total savings of $2,000,000. RORs for TX accounts: For reasons described below, the ROR for taxable savings is 2 percent below the ROR for tax-free accounts.

Taxation of TX RORs: One-sixth of the taxable ROR is derived from fully taxable savings (for example, interest income), one-sixth is from favorably taxed QDI and LTCG, and two-thirds is from non-taxed capital appreciation. Taxation of TX account withdrawals: Fifty percent of withdrawn taxable savings is LTCG; the remainder is a return of non-taxable basis.

An Example to Illustrate Comparative Measures In this example, the RORs for TD savings (ROR.TD) and TX savings (ROR.TX) are 6 percent and the ROR for TF savings (ROR.TF) is 8 percent. ROR.TD and ROR.TX are set lower than ROR.TF due to the prudent financial reasoning that TD savings (used for Roth conversions and RMDs) and TX savings (used for tax-free basis, favorable tax rates, and tax bracket management) are likely to be withdrawn before TF savings (used for tax bracket management and bequests). Thus, to minimize the sequence of returns risk, TX and TD should have higher allocations of lower risk, lower return investments. In this example and for other combinations of RORs, we assume that accounts are rebalanced as funds are withdrawn or added to maintain specified ROR levels by account type. Obviously as funds are transferred between account types, the percentage of funds in a given account type to total savings may change resulting in changes to the ROR for total savings. This is discussed further in the section How Research Findings Can Be Used in Practice. By comparing a NoRC to RC strategy, Table 1 illustrates the first evaluation measure, “the amount of Roth conversion” which produces the greatest FTAB. Withdrawals for both strategies are displayed for only 15 years since key differences in withdrawals occur in these years. For NoRC, the model withdraws only from TX and TD accounts in each of the years. Taxes are low in the first four years because of lower-taxed withdrawals from TX. When RMDs start at age 70.5, RMDs and taxes increase annually. Note that the tax-optimal specification of TD withdrawals before the entire depletion of TX is contrary to the common rule of withdrawing all TX before TD. In contrast to NoRC, the tax-optimal RC strategy prescribes varying amounts of partial Roth conversions from ages 66−69 (which total $714,000 or 51% of the initial

Volume 16, Issue 1

Table 1: Comparison of Tax‐optimal Withdrawal Strategies Assumptions: ROR.TF=8%, ROR.TD=6%, ROR.TX=6% No Roth Conversion (NoRC) Roth Conversion (RC) (Entries in $000) (Entries in $000) Withdrawals Withdrawals TaxTaxAge Taxable deferred Federal Taxable deferred Tax-free (on Jan 1) (TX) (TD) RMD Taxes (TX) (TD) (TF) RMD 66 57 32 2 151 67 58 33 2 153 68 59 34 2 118 12 69 59 35 2 124 70 39 61 61 6 99 6 29 71 38 64 64 6 101 6 28 72 37 68 68 6 103 6 26 73 36 72 72 7 106 7 24 74 34 76 76 7 108 7 22 75 32 81 80 8 110 7 19 76 8 111 84 13 112 7 15 77 8 114 87 14 44 67 11 78 8 116 90 14 45 68 10 79 9 118 92 14 46 69 8 80 9 121 95 14 22 93 7

$1,400,000 TD savings) that increase taxes during those years, but favorably increase TF savings and reduce TD savings and future RMDs. When compared to NoRC, from ages 66−68, TX withdrawals increase and TD withdrawals are eliminated at ages 66 to 67. TX withdrawals from ages 66 to 68 exhaust initial TX savings (not shown) creating the need from ages 68 to 76 for TD and limited TF withdrawals;

Figure 1

Federal Taxes 63 64 39 31 11 11 12 12 12 12 13 2 2 2 2

Roth Conversion 245 251 163 55

the TF withdrawals help reduce taxes. After age 76, RMDs and TD withdrawals are reduced, TF withdrawals increase, and taxes decrease significantly. Figure 1 illustrates three additional evaluation measures. The RC strategy FTAB is $3,701,000 and the NoRC strategy FTAB is only $2,476,000, a difference of $1,225,000.

Journal of Personal Finance

Expressed as an evaluation measure, the “amount (percent) of savings wealth enhancement” using RC is $1,225,000 (49.5%). However, in early years RC generates higher taxes and lower annual total account balances (TABs).1 Therefore, the “greatest amount of savings wealth reduction” evaluation measure for any year is $176,000 or 8.3 percent less than NoRC and the “duration of savings wealth reduction” (or savings wealth breakeven point) evaluation measure is 15 years, to age 80.

(from 2 to 8 percent) were examined. For example, in the column labeled “4%,” a constant 4% PctDiff results if ROR. TF and ROR.TD are 8 and 4 percent, 9 and 5 percent, or 10 and 6 percent, respectively. For each iteration, ROR.TX was set at 2 percent below ROR.TF (for example, if ROR.TF = 8%, then ROR.TX = 6%). The average result for the three ROR combinations in a column is displayed in Figures 2, 4, and 5. The bolded pairs in Table 2 were used to compare longevity risk of RC versus NoRC through the use of Monte Carlo simulation (adapted for use with the tax-optimal method, as described in Appendix A).

Variable ROR Analysis Roth conversions cause a transfer of TD assets to the TF. Therefore, the RORs of each asset location impact the RC decision. For example, the conversion of TD assets with a 6 percent ROR to the TF with a 6 percent ROR may not be appropriate, since earning the same, albeit, tax-free ROR after a conversion may be insufficient to offset the tax cost of the conversion. However, if ROR.TF exceeds ROR.TD, the increased ROR.TF offsets some of the tax cost and the conversion becomes more viable. Therefore, the difference in percentages ROR.TF minus ROR.TD, referred to as “PctDiff,” influences the conversion decision.

Optimal Amount of Roth Conversion Figure 2 represents the percentage of initial TD savings that is optimally converted to Roth accounts based on PctDiffs. For example, at PctDiff = 2%, the optimal Roth conversion rate is 52 percent of initial TD assets. Results indicate that as the PctDiff increases from 0% to 6%, the optimal percentage of TD savings converted to TF savings increases. This is due to Roth conversion re-allocating assets from lower-yielding TD assets to higher-yielding TF assets while reducing future RMDs and any associated taxes. The trend shows Roth conversion percentages increase at a decelerating rate. When ROR. TF and ROR.TD are equal (PctDiff = 0%), only 15 percent of TD savings is converted. The conversion rates increase rapidly for PctDiffs of 1% and 2% and then level to under 70 percent when ROR.TF is significantly higher than ROR. TD (for example, at PctDiff = 5% or 6%). Figure 2 shows

As shown in Table 2, seven PctDiff amounts, varying between 0% and 6 %, were evaluated by examining three different ROR.TF and ROR.TD pairs2 for each PctDiff. In total, 21 pairs of ROR.TF (from 6 to 10 percent) and ROR.TD

1. 2.

Since there is a TAB for each age, age will be referenced when needed. For example, TAB(72) refers to the TAB at age 72. Welch uses the same ROR for all account types which is equivalent to a PctDiff = 0%.

Table 2: Variable ROR Analysis PctDiff1 0%

ROR.TF, ROR.TD Pairs 6%,6%

6%,5%

7%,7%2

7%,6%

7%,5%

7%,4%

8%,8%

8%,7%

8%,6%

8%,5%

8%,4%

8%,3%

8%,2%

9%,7%

9%,6%

9%,5%

9%,4%

9%,3%

10%,6%

10%,5%

10%,4%

Note: ROR.TX percentage value is set at 2% below ROR.TF. 1. PctDiff : Percentage points ROR.TF is greater than ROR.TD. 2. Bold entries are used to analyze longevity risk.

Volume 16, Issue 1

Figure 2

that the impact on the optimal level of Roth conversions diminishes as PctDiff increases above 2%. This implies that taxes generated by increased Roth conversions increasingly offset benefits of additional growth in the TF account. Interestingly, regardless of the PctDiff, some amount of Roth conversion maximizes the FTAB, but all conversion rates are below 100 percent, which implies only partial RCs are optimal.

Amount of Final Savings Wealth (FTAB) Enhancement The effect of the Figure 2 Roth conversions on FTABs is illustrated in Figure 3, which shows the percentage improvement in the FTAB of RC over NoRC when ROR.TF equals 8 percent for each PctDiff (see Table 2, third row of pairs). The percentage improvement of FTAB increases as the PctDiff increases from 0% to 6% as assets are increasingly re-allocated from lower-yielding TD to higher-yielding TF (as illustrated in Figure 2). Figure 3 shows that for PctDiff = 0% (both ROR.TF and ROR.TD are 8 percent), there is only a 5 percent improvement in the RC over NoRC strategy. However, as the PctDiff increases, the RC FTAB improvement rises steadily to 224 percent over the NoRC FTAB at a PctDiff of 6% (ROR.TF and ROR.TD are 8 and 2 percent, respectively). While not displayed, results for other ROR.TF values (in other rows of Table 2) show similar increasing trends in percentage improvement in FTAB with increases in PctDiff, but the trends are less steep as ROR.TF increases.

Figure 3

Greatest Amount of Savings Wealth Reduction3 Although Roth conversions increase final savings wealth, a potential deterrent to a conversion is that the conversion-caused increase in taxes reduces total savings immediately. Figure 4 illustrates the greatest annual percent TAB reduction over 30 years caused by RC4 is only 1.8 percent when PctDiff = 0%. However, when the PctDiff = 1% or 2%, the greatest savings wealth reduction increases to 6.5 percent and 8.5 percent, respectively. This precipitous drop in savings wealth for RC is due to a combination of: (a) Increasing savings wealth in the early years for the NoRC strategy caused by some transfers from TD to TX accounts without a major tax impact. (b) An early decrease in savings wealth in RC caused by higher taxes on TD distributions, which are offset only partially by slightly higher RORs earned in the TF account at such low PctDiffs. However, for PctDiff = 6%, RC transfers assets with lower ROR.TD to assets with a significantly higher ROR.TF so that the greatest savings wealth reduction is a more palatable 4.8 percent. Additionally, the moderation in savings wealth reduction is due to the fact that for PctDiffs greater than 2%, ROR.TX is greater than ROR.TD. For example, if ROR.TF = 7%, ROR.TD = 4% (PctDiff = 3%), then ROR.TX = 5% (since 3.

â&#x20AC;&#x153;Savings Wealth Reductionsâ&#x20AC;? are computed for each age as [TAB(age) for RC] minus [TAB(age) for NoRC]. The greatest reduction occurs for the age when the Savings Wealth Reduction is most negative. There is very little variation among results for the three cases contributing to each shown average value, generally within 0.4% from the average.

Journal of Personal Finance

Figure 4

Figure 5

it is set 2 percent below ROR.TF). Thus, for NoRC, the model optimally transfers balances out of the lower-yielding ROR. TD into the higher-yielding ROR.TX, increasing downstream TABs.

Duration of Savings Wealth Reduction

duration of time before the RC TAB exceeds the NoRC TAB (the savings wealth breakeven time) drops from 15 years to 5 years. For PctDiffs of 4% or more, the greatest savings wealth reduction is reached in only three years or less and the savings wealth breakeven time is relatively short at seven years or less.

As illustrated in Table 1 and Figure 1, RC can cause increased taxes and reduced TABs for many years. The dashed lines in the bars of Figure 5 show the average number of years to reach the greatest savings wealth reduction caused by RC for various PctDiffs. The bars of Figure 5 show the average number of years to recoup the lost savings wealth caused by the conversion (bars)5 (i.e., the savings wealth break-even period).

For PctDiffs of 0% to 6%, savings wealth breakeven ages corresponding to the number of years in Figure 5 are 85, 87, 80, 75, 72, 71 and 70, respectively. These results imply that the more ROR.TF exceeds ROR.TD, the greater the chance that retirees will live to realize the increased savings wealth benefits of Roth conversions.

Figure 5 supports a strong inverse relationship between PctDiff and these savings wealth reduction evaluation measures. As the PctDiff increases, the measures decrease. For example, Figure 2 shows that when PctDiff = 0% or 1%, Roth conversions create little additional savings wealth. In Figure 5, the dashed lines in the bars show that it takes just under 10 years for the savings wealth reduction to decline to its lowest point. Twenty and 22 years, respectively, elapse before RC TABs exceed NoRC TABS. However as PctDiff increases from 2% to 6% the time to reach the greatest savings wealth reduction drops from 4.3 to 2.0 years. The

Roth conversion benefits are unrealistic if it is unlikely that they would be realized over the entire planning horizon. Traditionally Monte Carlo methods are used to assess longevity risk, defined as the chance of outliving savings during the planning horizon. A Monte Carlo method of risk analysis has been adapted for use with the tax-optimal retirement planning method used in this research; it is explained briefly in Appendix A.

There is very little variation among results for the three cases contributing to each average value; with the exception of PctDiff = 0%, at most, one result was a year less than the average.

Longevity Risk

Table 3 shows longevity risk differences between NoRC and RC. Except for PctDiff = 0%, longevity risk is smaller for RC than for NoRC. For the data used in this paper, the chance of outliving oneâ&#x20AC;&#x2122;s savings using RC is low, between 8 percent and 15 percent. While only two differences from

ÂŠ2017, IARFC. All rights of reproduction in any form reserved.

Volume 16, Issue 1

Table 3: Longevity Risk1 Comparison: NoRC Versus RC PctDiff

NoRC

19%

15%

19%

24%

13%

22%

15%

12%

14%

10%

10%2

12%2

Difference NoRC%–RC%

1. Longevity risk is defined as the chance of outliving savings during the planning horizon. 2. Results for PctDiff = 4% and = 6% are statistically significant.

NoRC are statistically significant, the consistency of results implies RC is the less risky strategy. It appears that the higher total savings growth of RC most likely counteracts the greater risk that generally accompanies the potentially higher volatility associated with the higher ROR.TF of Roth savings.

when PctDiff equals ROR.TF minus ROR.TX. Reasons for this are given above in the section Greatest Amount of Savings Wealth Reduction where the greatest TAB reduction occurs at PctDiff = 2% and ROR.TF minus ROR.TX was also 2%.

Conclusions Additional Results To determine the effect of altering assumptions other than PctDiff, NoRC results were compared to RC results for (a) higher initial savings wealth amounts, (b) higher initial expenses of $118,450 (3% higher) and $121,900 (6% higher), and (c) ROR.TX values that were above and below the 4 percent to 8 percent range used in this paper. Interested readers can obtain results from the corresponding author on request. Patterns in evaluation measures for the additional results are almost identical to those in the figures, thus strengthening the implications related to PctDiff already discussed. Of note are the following additional implications: •

For higher initial total savings wealth of $3,000,000, the equivalent Figure 2 pattern of conversion rates is similar but higher, as it jumps to slightly over 40 percent at PctDiff = 0% and levels off at about 75 percent for the highest PctDiffs.

•

When initial expenses are higher than $110,000, the percentage Improvement in RC FTABs over NoRC FTABs is higher than line graph in Figure 3 for PctDiffs greater than 1%. This is due to lower FTABs when expenses are increased for both NoRC and RC plus the result that the FTABs for RC become relatively larger than the FTABs of NoRC as PctDiff increases.

•

In Figure 4, the ‘swoosh’ pattern is similar for other values of ROR.TX. However, the greatest TAB reduction of a RC strategy from a NoRC strategy occurs

To better understand the impact of Roth conversions on retiree total savings wealth, tax-optimal retirement withdrawal plans over a 30-year horizon were compared for two strategies: no Roth conversions (NoRC) versus model-determined Roth conversions (RC). Annual withdrawals and Roth conversion levels were determined using a tax-optimization model. When the tax-free ROR (ROR. TF) is higher than both the tax-deferred ROR (ROR.TD) and the taxable ROR (ROR.TX), RC increases total savings growth over NoRC. Using Roth conversions does not increase longevity risk; to the contrary, under a wide range of ROR scenarios for the various account types, they lower longevity risk. However, not all combinations of RORs for these different accounts justify the use of Roth conversions, as discussed below in “How Research Findings Can Be Used in Practice.” An important factor impacting the decision to convert is referred to as PctDiff, the difference between ROR.TF and ROR.TD. When PctDiff is below 2%, percentage gains in savings wealth growth for RC are small. When PctDiff is less than 3%, the savings asset balance reduction due to the need to pay federal taxes can persist for more than 13 years and retiree health and prospects of living long enough to realize savings wealth gains is a consideration in opting for Roth conversions. As PctDiff increases above 2%, the RC strategy provides the following increasing benefits:

Journal of Personal Finance

•

The percentage savings wealth growth of RC over NoRC increases.

•

The optimal percentage of initial TD savings converted to TF is greater.

•

The time to recover savings wealth lost due to Roth conversion related taxes is shorter.

For example, a PctDiff of 3% (ROR.TF = 7%, ROR.TD = 4%) provides greater benefits than a PctDiff of 2%. The relative nature of PctDiff is of critical importance, not the value of ROR.TF. Thus, for PctDiff = 3%, the benefits of RC over NoRC are similar for savings targeting returns of ROR.TF = 8% and ROR.TD = 5% as one targeting ROR.TF = 7% and ROR.TD = 4%. Roth conversion of all TD savings is never observed to be optimal; the highest optimal conversion rate equals about 75 percent of initial TD savings. Roth conversions should be made relatively early in retirement, with annual levels varying during the time interval from before, to not much later than, the age RMDs start.

How Research Findings Can Be Used In Practice Before opting for Roth conversions, financial planners should be confident that their use will create higher savings wealth for an acceptable number of retirement years. Based on this paper’s findings, substantially higher savings wealth can be anticipated only if the ROR for tax-free Roth savings is at least 2 percent higher than the ROR for tax-deferred savings. However, as highlighted above in Conclusions, benefits resulting from Roth conversions taper off for PctDiffs higher than 2%. This suggests a PctDiff of 2% as a practically attainable planning objective. Since tax-free savings are generally not needed until later in retirement, a higher tax-free ROR (ROR.TF) at the time of Roth conversion can be generated by reducing or eliminating less volatile, lower return investments, such as cash or short-term bonds from the tax-free portfolio. To assure that a retirement plan stays on target, the balance of investments by account type should be reviewed at least annually.

vulnerable to market volatility. This concern can be remedied to some extent by limiting the size of a Roth conversion in any year, for example, to not more than $50,000. While this limitation is suboptimal, we have observed in practice that this type of restriction significantly moderates the levels of wealth reduction, slightly increases the time interval of reduced savings wealth after Roth conversions, and reduces final total savings wealth by a small percentage. A key intangible benefit of Roth conversions is that eventually a large percentage of savings is in tax-free Roth accounts. Even when higher than optimal levels of Roth conversions can potentially lower savings wealth growth, retirees may feel that greater proportions of final savings wealth in tax-free accounts may be desirable and important to estate planning. The long-term benefits of Roth conversions can be significant, especially for taxpayers who do not need all of the cash flow generated by RMDs to satisfy living expenses. However, one must ensure that these benefits are not compromised by income-based surtaxes or tax phase-out provisions (such as the 3.8 percent surtax on investment income or itemized deduction and exemption phase outs), which can be triggered by the Roth conversion itself.

Areas of Future Research The tax-optimal method used in our research lends itself to comparing other strategies for differences in the impact on wealth growth. Of recent interest is the impact of Roth annuity plans on wealth growth and longevity risk. As referred to in “How Research Findings Can Be Used in Practice” it may be of interest to more thoroughly evaluate the impact of suboptimal use of Roth conversion strategies that take into account intangible factors such as retirees concern for market volatility. Other strategic questions besides those involving Roth savings could be evaluated. For example, tax-optimal methods can be used to evaluate the impact of alternative Social Security claiming strategies on total savings wealth growth, not just on total Social Security income.

Clients’ tolerances for fluctuations in total savings wealth differ. Some retirees may be uncomfortable with large reductions in wealth early in retirement due to large Roth conversions that they feel may make them more

Volume 16, Issue 1

Appendix A: Monte Carlo Longevity Risk Assessment for Tax-optimal Retirement Planning

Coopersmith, Lewis W. and Alan R. Sumutka (2011). “Tax-Efficient Retirement Withdrawal Planning Using a Linear Programming Model.” Journal of Financial Planning, 24 (9), 50−59. Coopersmith, Lewis W., Alan R. Sumutka, and James Arvesen

The main difference between the Monte Carlo (MC) method used in tax-optimal planning in this research and traditional MC methods is the use of a mathematical optimization model and solution algorithm to determine a withdrawal plan given a scenario of randomly selected annual RORs for each of the various types of savings accounts. The percentage of stocks, bonds, and cash are specified for each of the three savings types: taxable, tax-deferred, and tax-free. For each year of the planning horizon, a randomly selected ROR is determined separately for stocks and for bonds using statistical models that closely fit annual S&P 500 data for stocks and Shiller data on bond rates, both from 1951 to the present. A weighted average of these RORs is then used as the annual ROR for each savings type. Either an optimal withdrawal plan for the random scenario is determined or no plan is feasible. Results for a sample of size N (N=100 in this paper) of retirements are run and the number of feasible retirement plans, n, is determined. Longevity risk is computed as (1−n/N).

(2009).” Optimal Tax-Efficient Planning of Withdrawals from

Other differences include:

Saunders, Laura (2010). “What Experts Are Doing With Their Own

Retirement Accounts.” Academy of Financial Services Proceedings (October). Retrieved from www.academyfinancial.org/wp-content/uploads/2013/10/3B-Coopersmith-Sumutka-Arvesen.pdf Geissler, Greg and David Hulce (2014). “Traditional Versus Roth 401(k) Contributions: The Effect of Employer Matches.” Journal of Financial Planning 27(10), 54−60. Keebler, Robert, S. (2010). The Rebirth of Roth: A CPA’s Ultimate Guide for Client Care, American Institute of Certified Public Accountants, 2010, 21. Krishnan, V. Sivarama, and Julie Cumbie (2016). “Roth Conversion: An Analysis Using Breakeven Tax Rates, Breakeven Periods, and Random Returns.” Journal of Personal Finance, 11 (1), 37−46. Ragsdale, Cliff T., Andrew F. Seila, and Philip L. Little (1994). “An Optimization Model for Scheduling Withdrawals from Tax-Deferred Retirement Accounts.” Financial Services Review 3 (2), 93-−109.

Roth IRAs.” Wall Street Journal, Eastern Edition, August 14, 2010, B9.

1. Outcomes of the optimization model are determined in each sampled scenario using linear programming. These outcomes include the annual amounts withdrawn from each account type, the amount of federal taxes, and the annual amounts of Roth conversions.

Slott, Ed (2003). The Retirement Savings Time Bomb…and How to Defuse It. Penguin Group, 2003: 203. Sumutka, Alan R., Andrew M. Sumutka, and Lewis W. Coopersmith (2012). “Tax-efficient Retirement Withdrawal Planning Using a

2. Annual pre-federal tax expenses in this paper are fixed initially for all sampled scenarios at $115,000 followed by inflationary growth. However, for other studies and analyses, the pattern of expenses may be at the discretion of the user and is not limited to inflationary growth.

References Cellucci, Robert (2014). “When Is the Roth Conversion Optimal?” Journal of Financial Service Professionals, 68(9), 94−100.

Comprehensive Tax Model.” Journal of Financial Planning, 25 (4), 41−52. Washer, Kenneth (2011). “Partial Roth Conversions in Early Retirement: An Effective Strategy for Extending Portfolio Longevity.” Journal of Financial Service Professionals, 65(1,) 8−0-87. Welch, James S. Jr. (2015). “Mitigating the Impact of Personal Income Taxes on Retirement Savings Distributions.” Journal of Personal Finance, 14 (1), 17−37. Welch, James S. Jr. (2016). “Measuring the Financial Consequences

Clayton, Ronnie J., Lemuel W. Davis, and William Fielding (2012).

of IRA to Roth IRA Conversions.” Journal of Personal Finance, 15 (1),

“Converting a Traditional IRA to a Roth IRA: Breakeven Analysis.”

47−55.

Journal of Personal Finance, 11 (2), 10−35.

Journal of Personal Finance

Do Financial Advisers Follow Their Own Advice? Evidence from 2008–2011

Dominique Gehy Outlaw, PhD , Hofstra University Jesse Outlaw, CPA, Outlaw, Bruno, and Associates, Family Office

Abstract Consistent with previous literature, we find that financial advisors’ trade recommendations do not outperform the trades that are independently initiated by their clients. At best, their recommendations outperform their clients’ selections in the shortterm, but still underperform the market. What is unknown in the literature is whether it is misaligned sales incentives that cause advisors to give their clients suboptimal advice. Using transaction-level data from a U.S. brokerage firm, we compare financial advisors’ personal trades to their clients. We find that financial advisors do not outperform their clients, suggesting that advisors are giving clients their best advice which they, too, follow.

Volume 16, Issue 1

Most of the nation’s roughly 223,400 financial advisors will tell you you’re better off using their services. Do-it-yourselfers, meanwhile, shun hired advice as a waste of time and, yes, money.1 —Wall Street Journal

Introduction With the increase of information transmission and websites that assist individuals in managing their wealth, many individual investors are feeling more equipped to make their own investment decisions and question whether hiring a traditional financial advisor is still worth the cost. Since 2008, the number of households using online trading brokerage accounts has increased by 20.4%.2 However, the 2015 Certified Financial Planner survey finds that more individuals are turning to financial advisors, from 28% in 2010 to 40% in 2015.3 Taken together, the evidence suggests that there is an overall increase in the number of individual investors participating in the market. Some investors are managing their own portfolios, and many still look to financial advisors. The purpose of this study is to determine which is more valuable for retail investors. Several news articles continually give anecdotal suggestions to individual investors on whether it is still beneficial to hire a financial advisor. Due to a lack of data availability, there is limited empirical evidence that addresses this question. In the academic literature, the value of financial advisors is debatable. Several studies find mixed results on whether financial advisors help their clients obtain higher returns, mitigate behavioral biases, and achieve better risk-return profiles (Shapira and Venezia, 2001; Bluethgen, Gintschel, Hackethal, Muller, 2008; Kramer, 2012; Bhattacharya, Hackethal, Kaesler, Loos, and Meyer, 2012). Further, Inderst and Ottaviani (2009) model how sales incentives may actually disincentivize advisors from giving their clients optimal recommendations. This might result in unsuitable products that benefit the advisor, but not the client.

1. 2. 3.

http://www.wsj.com/articles/SB10001424052702304893404579529 663913965236 http://www.statista.com/statistics/228118/people-in-householdswith-an-online-investing-stock-trading-service-usa/ https://www.cfp.net/news-events/latest-news/2015/09/24/surveyamericans-use-of-financial-advisors-cfp-professionals-rises-agreeadvice-should-be-in-their-best-interest

In this study, we contribute to the debate on whether financial advisors select better stocks for their clients than the clients would have selected on their own. Specifically, we compare the performance of trades that were recommended by the advisor and trades that were independently initiated by the client. We examine this using a proprietary dataset containing transactions of approximately 25,000 individual investors from 2008–2011, a highly volatile period where financial advisors should be most valuable.4 We also determine whether financial advisors are intentionally giving their clients suboptimal advice by comparing the performance and characteristics of their own personal trades to those of their clients. Previous studies have documented that financial advisors do not add value to their clients’ portfolios, but with the exception of Foerster, Linnainmaa, Melzer, and Prevetiro (2015), to our knowledge, no other study determines whether the advisors are strategically giving suboptimal recommendations, or if advisors simply do not have superior stock selection abilities. Another contribution of this study is that the brokerage firm in our sample is located in the United States, where many agree is the origin of the global financial crisis. A majority of the papers in the financial advisor literature use datasets from European countries. Typically, the markets and investors in the U.S., when compared to those in European regions, have different trading cultures, best practices, and regulations. For instance, the financial advisors of the Dutch brokerage firm in Kramer (2012) and Kramer and Lensink (2012) do not receive commissions on trades. They are paid flat monthly fees regardless of the number of transactions. A majority of the financial advisors in our sample receive commissions on each trade. This makes it even more interesting to understand whether sales incentives and compensation structures influence the quality of financial advice and their clients’ performance. We hypothesize that since retail investors continue to hire financial advisors despite access to cheaper online self-directed trading platforms, financial advisors continue to add value to their investment decisions. We test for value 4.

It is during these panic periods that retail investors are known to irrationally sell their holdings and become afraid to buy, passing up some great investment opportunities. For instance, a 2015 article in Motley Fool advises investors to resist the temptation to sell during market corrections: http://www.fool.com/investing/general/2015/08/24/timesensitive-the-worst-investing-move-during-a-st. aspx.

Journal of Personal Finance

by comparing whether trades recommended by financial advisors are associated with higher returns. Specifically, we expect higher short- and long-term returns measured by cumulative abnormal returns (CARs) and alphas from calendar-time portfolio returns, respectively. Within the year of the trade, we find that financial advisors’ purchase recommendations are better than their clients’ selections, but still underperform the market. However, clients sell their stocks at a better time than financial advisors recommend. The long-run returns, which is the primary concern for a majority of the investors, are qualitatively similar for recommended and non-recommended trades. This now raises the question of whether advisors actually do have superior skill and expertise, but intentionally give suboptimal advice because it is advantageous in some cases (Inderst and Ottaviani, 2009). For instance, advisors may encourage clients to buy underperforming stocks and then recommend that they sell the same stock to increase their transaction commissions. To test this theory, we also compare the financial advisors’ and clients’ trades. A unique component of our dataset is that we also have the financial advisors’ personal trades. The brokerage firm requires that all financial advisors only use their brokerage to execute all personal transactions. If financial advisors are giving their best advice to their clients, we expect their personal trades to perform similarly to those that they recommend to their clients. We find that financial advisors’ personal trades do not outperform their clients. This suggests that financial advisors are giving their soundest advice. The results of this paper shed light on the ongoing issue of determining whether it is still worthwhile for investors to hire a financial advisor or to seek alternatives that better suit their personal investment needs. The remainder of the paper is organized as follows. Section II reviews the literature on the role of financial advisors. Section III describes the data and summary statistics. Section IV presents the results, and section V concludes with the implications of our study.

Literature Review Overall, the literature has yet to reach a consensus on whether financial advisors are value-adding for their clients. Some recent empirical studies cast doubt on the usefulness of financial advisors. Hoechle, Ruenzi, Schaub,

and Schmid (2016) find that advisors actually hurt trading performance, and it is even more pronounced if the advisor was the one to contact the investor before executing the trade. Their comprehensive Swiss dataset provides information on all contact between the financial advisor and the clients from mass emails to personal home visits. They codify recommended trades based on the time lapse between the communication and executed trade. Foerster, Linnainmaa, Melzer, and Previtero (2015) find that financial advisors are costlier rather than more beneficial. The financial advisors’ clients underperform the passive investment benchmarks and the top performing advisors in their Canadian dataset charge fees that exceed the alphas produced. Thus, over the long-term a significant portion of the client’s savings are spent on fees. A few studies also consider how investors’ financial literacy impacts the role of the financial advisor. Using investors’ transactions from a German brokerage firm, Bhattacharya, Hackethal, Kaesler, Loos, and Meyer (2012) document that investors who most need financial advice do not seek it. And even for the small percentage that do seek financial advice, they do not follow the advice, resulting in suboptimal portfolio efficiency. Karabulut (2013) reports that financial advice does not serve as a substitute for financial literacy because when less sophisticated investors consult an advisor before trading, they do not make better decisions as suggested by their lower returns. And in some cases, the advisor actually reinforces some of the behavioral biases (Mullainathan, Noth, and Schoar, 2012). However, there are a few studies that find that even though financial advisors do not increase returns for their clients, they help mitigate behavioral biases and unnecessary risk that can hurt portfolio performance. Shapira and Venezia (2001) find that the disposition effect is significantly less pronounced among professional investors than self-directed investors. Portfolios created with the help of financial advisors are better diversified and advisors have incentive to promote diversification. Bluethgen, Gintschel, Hackethal, Muller (2008) also find that financial advisors enhance portfolio diversification. Advisors make portfolios more consistent with predefined optimal model portfolios, although investors have higher fee expenses. Kramer (2012) suggests that financial advisors do add value to their clients. Using a Dutch dataset of individual investor transactions from 2003–2007, they find that self-directed trades do not have significantly better performance than advised

Volume 16, Issue 1

trades. However, the advised clients have better diversified portfolios and carry significantly less idiosyncratic risk which is also valuable to their clients. They also find that independent investors that switched to financial advisors during that period, also benefit from the financial advisors’ recommendations.

Data Our sample is comprised of individual investors’ transactions from 25,131 accounts from March 2008 to August 2011. Our proprietary transaction-level data is obtained from a United States full-service brokerage firm. Each account is assigned to a financial advisor. In some cases, the client chooses the specific advisor, and in some cases the client is assigned to the advisor by the brokerage firm. There are 65 advisors in the sample dispersed throughout the U.S.

Although all the clients are assigned to a financial advisor, all the trades are not recommended by the advisor. Each time the brokerage firm records a transaction, the financial advisor has to disclose whether or not they recommended the trade. For non-recommended trades, the client independently initiated the trade. The variable, Recommended, is a binary variable equal to one when the trade was suggested by the financial advisor, and zero when the client independently initiated the trade. 8.5% of the client trades are recommended by the advisor. Also, our dataset has a distinct advantage over most related studies because we also have access to the financial advisors’ personal stock transactions. The brokerage firm requires that advisors execute all their trades with the brokerage firm only so these transactions represent all the advisors’ trading activity. Only a small percentage of the trades are from the financial advisor (approximately 3.75%).

Table 1: Sample Description This table presents the descriptive statistics of the sample of 25,131 investors. All data is self-reported by the investor from a questionnaire that is required to be completed upon opening an account at the brokerage firm. Panel A reports the risk tolerance, primary investment objective, and investment time horizon. Panel B reports the annual income and net worth of all clients. Panel C reports the age of all the clients.

Panel A: Investor Investment Preferences Investment Objective

Risk Tolerance Low Medium High Unspecified

Annual Income ($) 0–19,999: 20,000–50,000: 50,001–100,000: 100,001–200,000: 200,001–500,000: 500,001–1,000,000: Over 1,000,000:

1.75% 74.82% 23.39% 0.05%

Capital Pres Speculation Growth Income Unspecified

0.57% 0.60% 88.19% 10.60% 0.05%

Time Horizon < 5 years 5–10 years >10 years Unspecified

Panel B: Investor Financial Characteristics Percentage Net Worth ($) 7.27% 25.21% 34.76% 22.30% 7.87% 1.60% 0.76%

0–19,999: 20,000–50,000: 50,001–100,000: 100,001–250,000: 250,001–500,000: 500,001–1,000,000: Over 1,000,000:

Panel C: Investor Age Age Percentage Younger than 18: 18-25: 26-40: 41-65: Older than 65:

2.00% 2.54% 6.52% 47.33% 41.61%

3.72% 14.36% 81.86% 0.05%

Percentage 3.11% 3.55% 7.54% 18.30% 22.67% 21.83% 22.90%

Journal of Personal Finance

Our data provides self-reported characteristics about the investors including age, relation to financial advisor, location, occupation, income, net worth, risk tolerance, and investment objective.5 In table 1, we report that 23.39% of the investors report their risk tolerance as high, 74.82% as medium, and 1.75% as low. Only a small percentage invest for capital preservation (0.57%) and speculation (0.6%). 88.2% of the investors report their primary investment objective as growth, while 10.6% report income. In our sample, a majority of the investors (57.06%) have an annual income of $50,000–$200,000. The remainder of the investors have an annual income of $0–$19,999 (7.3%), $20,000–50,000 (22.3%), $200,001–$500,000 (7.9%), 5.

There is an artificial account identifier to de-identify the subjects.

$500,001–$1,000,000 (1.6%), and over $1,000,000 (0.8%). 14.2% of the investors have a net worth less than $100,000. 18.3% of the investors have a net worth of $100,001– $250,000, 22.7% have $250,001–$500,000, 21.83% have $500,001–$1,000,000, and 22.9% have over $1,000,000. Only 11.06% of the sample is younger than 40 years old. The remaining 47.33% and 41.61% of the sample fall in the age group 41–65 and over 65, respectively. For each transaction, we have data on the date, ticker, transaction type, number of shares traded, and commissions. To obtain additional stock and accounting measures, we restrict the sample to firms with data available on CRSP and Compustat. Firms must be common stocks (CRSP share code 10 and 11) that are traded on the NYSE, AMEX, or NASDAQ. This leaves us with a sample of 213,278

Table 2: Stock Characteristics This table presents the stock summary statistics of all transactions in our sample from March 2008 to August 2011. Size is the natural log of the market capitalization of the stock. Market capitalization is the number of shares outstanding multiplied by the price per share. B/M is the ratio of book assets divided by market value at the end of the fiscal year ending as of December of the prior year. This variable remains the same from July of year t through June of year t–1. Returnt–1 is the stock return in the month before the trade. Momentum is the cumulative stock return over the prior six months. IVol is the idiosyncratic volatility, measured as the standard deviation of residual returns from regressions of daily returns during the month on the Fama and French (1993) three factors. Panel A reports the summary statistics for all transactions. Panel B and C report the statistics for all stocks purchased and sold, respectively.

Panel A: All Trades Variable

Mean

Std Dev

Min

Max

Size

213,278

15.630

1.821

8.148

20.023

B/M

213,278

1.820

17.665

0.000

1039.450

Returnt-1

213,278

0.003

0.137

-0.878

3.983

Momentum

213,278

0.032

0.409

-0.968

20.198

IVol

213,278

2.784

2.897

0.008

407.428

Panel B: Purchases Variable Size

Mean

Std Dev

109,092

15.600

1.832

Min

Max

8.148

20.023

B/M

109,092

1.537

10.647

0.000

1039.450

Returnt-1

109,092

-0.002

0.135

-0.748

3.983

Momentum

109,092

0.030

0.418

-0.952

20.198

IVol

109,092

2.766

2.714

0.069

84.177

Panel C: Sales Variable

Mean

Std Dev

Size

104,186

15.661

1.810

8.335

20.023

B/M

104,186

2.116

22.803

0.000

1039.450

Returnt-1

104,186

0.007

0.139

-0.878

2.597

Momentum

104,186

0.033

0.399

-0.968

11.048

IVol

104,186

2.802

3.078

0.008

407.428

Min

Max

Volume 16, Issue 1

transactions, which is approximately nine trades per investor. Table 2 presents the summary statistics of all the stocks in our sample. Individual investors trade stocks of relatively large, growth firms. In the month prior to trading, the stocks have a positive return of 0.3% and usually experience a 6-month stock price run-up (momentum) of 3.2%. In addition, the stocks display idiosyncratic, firm-specific risk of 2.8%. Barber and Odean (2008) show that investors display asymmetric biases when selling and purchasing stocks. Thus, in panels B and C (and throughout the rest of the paper), we present the summary statistics for the subsample of purchase and sale transactions. It is a fairly even split between the number of purchases (51.15%) and sales (48.85%). We find that investors purchase and sell stocks that are similar in size. According to the B/M values, they purchase more value stocks and sell more growth stocks. They tend to purchase (sell) stocks that experience a negative (positive) return in the previous month. There are no obvious differences in the 6-month momentum or idiosyncratic volatility for the stocks they purchase or sell. We do a more in-depth analyses in the next section to better understand the performance and characteristics of recommended and independent trades, and financial advisors’ personal trades.

Results The goal of our study is to compare the performance of the trades that are recommended by the financial advisors to the non-recommended trades that are initiated by the client. In addition, we compare the trades of financial advisors to the trades of their clients. We focus on the stocks’ short- and long-term abnormal returns and stock characteristics. For returns, we compute the one-month, 6-month, and one-year cumulative abnormal returns (CARs) using the market model. We measure the long-term abnormal returns using alphas of a simulated portfolio that traded the stocks and held them for three years using calendar-time portfolio regressions. We also compare additional stock characteristics such as size, growth vs. value, previous one-month return, and 6-month price momentum. To measure diversifiable risk, we calculate the idiosyncratic volatility of the stock returns (Brandt, Brav, Graham, Kumar, 2009; Kramer, 2012; Dorn and Weber, 2013).

Recommended vs. Independent Trades In table 3, we compare the returns and stock characteristics of trades recommended by advisors and trades that are independently initiated by the clients. Panel A reports the cumulative abnormal return (CARs) over the following daily intervals: (1, 20), (1, 126), and (1, 252). The independently-initiated purchases significantly underperform the market in all the short-term horizons (from –0.8% to –8.9%). The recommended purchases also underperform the market in the 126 and 252 trading days following the purchase (–2.5% and –5.4%), but to a lesser degree than the independent purchases. Thus, the recommended trades do only relatively better than the independent trades, but neither outperforms the market. Depending on the benchmark for good performance, this might imply bad advice from the financial advisors. Next, we examine the independently-initiated and recommended sales. Note that we should see negative returns following a sale suggesting it was a good time to sell. Here we find that independent sales have better short-term performance than the recommended sales. The independent sales have persistent significant negative performance for the 20, 126, and 252 trading days following the sale. The recommended sales have significantly positive abnormal returns in the 20 days following the sale, which suggests that the advisors recommend the sale a little too early. The recommended sales have negative performance in the 252 trading days following the trade (–2.4%), although it is less negative than the independently sold stocks (–4.7%). Overall, financial advisors make relatively better short-term purchase recommendations, but clients make better selections when selling their stocks. In Panel B, we compare the stock characteristics to determine if there are any notable differences in the stocks recommended by advisors and their clients. Clients tend to purchase larger stocks than the financial advisors recommend. This is not surprising since individual investors tend to trade large, attention-grabbing stocks (Barber and Odean, 2008). The difference in size for recommended and independent sales is statistically insignificant. Clients purchase and sell more growth stocks than advisors recommend (highly significant differences of 0.951 and 1.206, respectively). And consistent with table 2, high B/M stocks are sold most often, especially in independently initiated sales (2.214).

Journal of Personal Finance

Table 3: Recommended Trades vs. Client-Initiated Trades This table reports the mean cumulative abnormal returns (CARs) using the single-factor market model for trades that are recommended by the advisor (recommended) and those that are initiated by the client (independent). CAR (1,20), (1,126), and (1,252) represents 20, 126, and 252 trading days following the transaction. Panel B presents the stock characteristics of the stocks recommended by advisors or independently initiated by the client. Size is the natural log of the market capitalization of the stock. Market capitalization is the number of shares outstanding multiplied by the price per share. B/M is the ratio of book assets divided by market value at the end of the fiscal year ending as of December of the prior year. This variable remains the same from July of year t through June of year t-1. Returnt-1 is the stock return in the month before the trade. Momentum is the cumulative stock return over the prior six months. IVol is the idiosyncratic volatility, measured as the standard deviation of residual returns from regressions of daily returns during the month on the Fama and French (1993) three factors. t-statistics of the differences are reported in the parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Panel A: Cumulative Abnormal Returns Independent Recommended Diff t-stat

Independent Recommended Diff t-stat

CAR (1,20) –0.008*** (–21.87) –0.001 (–0.71) –0.007*** (–5.57)

Purchases CAR (1,126) –0.045*** (–44.34) –0.025*** (–7.75) –0.020*** (–5.95)

CAR (1,252) –0.089*** (–54.30) –0.054*** (–10.45) –0.035*** (–6.47)

CAR (1,20) –0.004*** (–10.57) 0.003*** (2.35) –0.007*** (–5.51)

Sales CAR (1,126) –0.013*** (–12.30) 0.004 (1.28) –0.018*** (–4.98)

CAR (1,252) –0.047*** (–27.18) –0.024*** (–4.27) –0.023*** (–3.98)

Panel B: Stock Characteristics N 95,548 95,548 95,548

Mean 15.626 1.620 -0.002

Purchases Recommended N Mean 8,951 15.405 8,951 0.669 8,951 0.006

95,548 95,548

0.033 2.762

8,951 8,951

0.003 2.793

0.030*** –0.031

(7.47) (–1.03)

N 92,083 92,083 92,083

Mean 15.666 2.214 0.007

Sales Recommended N Mean 8,577 15.697 8,577 1.008 8,577 0.003

Diff –0.032 1.206*** 0.004**

t-stat (–1.30) (9.02) (2.50)

92,083 92,083

0.033 2.797

8,577 8,577

–0.002 –0.066**

(–0.34) (–2.10)

Independent Size B/M Returnt-1 Momentum IVol

We also examine the returns prior to the trades to determine whether previous returns influence trading decisions. We find that clients independently purchase stocks that recently experience a negative return in the previous month and sell stocks that experience a positive return, more so than the recommended trades. The differences are significant at the 1% (purchases) and 5% (sales) level. Clients independently purchase stocks that have a significant 3% larger 6-month stock price run-up. It seems that clients

0.035 2.862

Diff 0.221*** 0.951*** –0.008***

t-stat (8.96) (17.74) (–5.30)

independently chase positive momentum, winner stocks, but make the purchase just after there is a negative dip in returns. Both independent and recommended sales display qualitatively similar momentum. We find no significant difference in idiosyncratic volatility for the purchases, but we do find that advisors recommend selling stocks with higher idiosyncratic volatility. 6 6.

Due to data limitations, we are unable to examine how the purchase or sale of stocks with significant idiosyncratic volatility impacts the

Volume 16, Issue 1

Table 4: Financial Advisor Trades vs. Client Trades This table reports the mean cumulative abnormal returns (CARs) using the single-factor market model for financial advisor trades and their client trades. CAR (1,20), (1,126), and (1,252) represents 20, 126, and 252 trading days following the transaction. Panel B presents the stock characteristics of the stocks recommended by advisors or independently initiated by the client. Size is the natural log of the market capitalization of the stock. Market capitalization is the number of shares outstanding multiplied by the price per share. B/M is the ratio of book assets divided by market value at the end of the fiscal year ending as of December of the prior year. This variable remains the same from July of year t through June of year t-1. Returnt-1 is the stock return in the month before the trade. Momentum is the cumulative stock return over the prior six months. IVol is the idiosyncratic volatility, measured as the standard deviation of residual returns from regressions of daily returns during the month on the Fama and French (1993) three factors. t-statistics of the differences are reported in the parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Panel A: Cumulative Abnormal Returns FA Client Diff t-stat

FA Client Diff t-stat

CAR (1,20) –0.006*** (–3.52) –0.007*** (–21.07) 0.001 (0.48)

Purchases CAR (1,126) –0.023*** (–4.53) –0.044*** (–44.70) 0.021*** (3.98)

CAR (1,252) –0.056*** (–6.95) –0.086*** (–55.02) 0.030*** (3.62)

CAR (1,20) –0.001 (–0.59) –0.004*** (–9.51) 0.002 (1.11)

Sales CAR (1,126) –0.010 (–1.69) –0.012*** (–11.45) –0.002 (–0.39)

CAR (1,252) –0.045*** (–4.95) –0.045*** (–27.26) 0.000 (–0.02)

Panel B: Stock Characteristics FA Trades Size B/M Returnt-1 Momentum IVol

N 4,593 4,593 4,593

Mean 15.437 1.513 -0.003

Client Trades N Mean 104,499 15.607 104,499 1.538 104,499 –0.002

4,593 4,593

0.021 2.793

104,499 104,499

t-stat (–5.76) (–0.14) (–0.44)

0.030 2.764

–0.010 0.028

(–1.53) (0.66)

Sales Client Trades

FA Trades Size B/M Returnt-1 Momentum IVol

Diff –0.170*** –0.025 –0.001

N 3,526 3,526 3,526

Mean 15.433 2.275 0.015

N 100,660 100,660 100,660

Mean 15.669 2.111 0.007

Diff –0.236*** 0.164 0.008***

t-stat (–7.29) (0.38) (2.82)

3,526 3,526

0.037 2.804

100,660 100,660

0.033 2.802

0.004 0.002

(0.64) (0.05)

FA Trades vs. Client Trades Next, we compare the returns and characteristics of stocks traded by financial advisors in their personal portfolio and those of their clients. Overall, we find that the shortterm returns are similar for the financial advisors and clients, with the exception of purchases for the 126 and overall portfolio performance, or why financial advisors make this recommendation.

252 trading day horizons. For these two horizons, we find that the financial advisors’ purchases perform better than their clients’, but both advisors and their clients experience statistically significant negative short-run returns. The difference in sales for the two groups is statistically insignificant at all horizons. This reaffirms the previous finding that financial advisors display some superior stock-picking ability relative to their clients, but their recommendations and personal purchases still underperform the market. In

Journal of Personal Finance

most cases, the difference in CARs are insignificant for the financial advisors and clients.

Long-run Performance

In Panel B, we find that clients trade relatively larger stocks than the advisors, which is probably due to large stocks’ popularity. There is no significant difference in the B/M of the firms they purchase or sell. Also, advisors and clients purchase stocks with similar returns in the previous month, but clients purchase winner stocks that experienced a significantly larger 6-month momentum. The difference in momentum is a highly significant 1.53%. Advisors sell stocks after a 0.8% larger positive return in the previous month, but both groups similarly sell stocks after a 3.3%– 3.7% run-up in the stock price over the previous 6 months. Lastly, we find no difference in the idiosyncratic risk of stocks that advisors and clients purchase and sell. Thus, in many cases it seems financial advisors trade similarly to their clients, suggesting honest, yet non-superior advice.

Although we examine the short- and long-term returns in this study, it is worth noting that the short-term returns are not the central focus since in this sample, a majority of the investors report a longer investment horizon (81.86% from Table 1). Thus, the results up to this point may be due to financial advisors and their clients not being overly concerned with the short-run returns and volatility. Therefore, we also compute the 3-year abnormal returns using calendar-time portfolio regressions which control for overall market performance, size, B/M, and momentum (Fama and French, 1993; Carhart, 1997). The calendar-time portfolios mimic the returns of portfolios that purchased and sold the stocks and held them for three years before falling out of the portfolio. The alphas are the abnormal returns adjusted for the market performance, size (SMB), book-market (HML), and momentum (UMD) systemic factors. These factors are available on Kenneth French’s website.7 7.

http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/

Table 5: Long-run Abnormal Returns This table presents three-year calendar-time portfolio abnormal returns for the clients’ recommended vs. independent trades (panel A) and advisors vs. clients’ trades (panel B). The one-factor alpha is the intercept on a regression of monthly returns from a rolling strategy controlling for the market return. The three-factor alpha is the intercept on a regression of monthly excess returns from the rolling strategy controlling for the Fama and French (1993) factors. The four-factor alpha is augmented with the Carhart (1997) momentum factor. Portfolios are rebalanced every calendar month to maintain value weights. Reported in the parentheses are the t-statistics. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. Panel A: Recommended vs Independent Trades Recommended One-factor alpha Three-factor alpha Four-factor alpha

Purchases

Sales

Diff

Purchases

Independent Sales

Diff

Rec–Indep Diff

0.021

0.000

0.027

0.010

0.017

–0.017

(1.25)

(1.34)

(–0.03)

(1.43)

(0.65)

(0.88)

(–0.66)

0.014

–0.001

0.017

0.001

0.016

0.017

(0.88)

(1.06)

(–0.04)

(1.02)

(0.09)

(0.86)

(0.65)

0.014

0.012

0.002

0.020

0.000

0.019

0.017

(0.94)

(0.91)

(0.14)

(1.17)

(0.02)

(1.05)

(0.65)

Panel B: FA vs Client Trades FA One-factor alpha Three-factor alpha Four-factor alpha

Client

Purchases

Sales

Diff

Purchases

Sales

Diff

FA–Client Diff

0.053**

0.015

0.038

0.024

0.016

0.008

0.030

(2.36)

(0.60)

(1.49)

(1.50)

(1.22)

(0.56)

(1.12)

0.045**

0.010

0.035

0.014

0.007

0.028

(2.09)

(0.42)

(1.36)

(1.03)

(0.71)

(0.52)

(1.04)

0.048**

0.008

0.040

0.016

0.005

0.010

0.030

(2.27)

(0.33)

(1.61)

(1.19)

(0.58)

(0.78)

(1.13)

Volume 16, Issue 1

We present the results in table 5. In Panel A, the 3-year abnormal returns, or alpha, show that there are no statistically significant differences in performance between the purchases and sales recommended by the advisors and initiated by the clients. Both groups yield no abnormal returns, whether we control for just the market performance (one-factor alpha), augment the model with size and B/M (three-factor alpha), or also include momentum (four-factor alpha). Thus, financial advisors do not seem to be value-adding as they do not recommend better stocks than what their clients independently select. Next, we compare the financial advisors’ long-run performance to their clients. Again, we find that the returns are qualitatively similar for the two groups. Only for the purchases, we find that financial advisors select stocks that display significantly positive long-run returns depending on which factors we control for, but the difference is insignificant. Thus, we are unable to conclude that financial advisors’ trades outperform their clients. The overall findings suggest that financial advisors do not make better trade recommendations to their clients, and it is not misaligned incentives that encourages them to do so because they do not exhibit superior stock-picking ability in their own portfolios.

Conclusion The literature has yet to reach a consensus on whether the popular practice of hiring a financial advisor is value-adding for retail investors. There are also questions on whether financial advisors’ compensation structures are aligned with their clients’ best interest. This is especially important as the number of individual investors in the markets continually increases. Using a unique dataset of individual investor transactions from a U.S. brokerage firm during the most recent financial crisis period, we add to this literature and confirm some of the previous findings that financial advisor recommendations do not result in higher returns. While a majority of the clients in this sample are more concerned with long-run performance, the results show that financial advisors do not help them achieve those goals better than they would on their own. Further, financial advisors do not show superior skill and expertise as their personal trades do not outperform the trades they data_library.html

recommend to their clients. For the most part, financial advisors’ recommendations are aligned with their own personal decisions, which suggests that they give honest advice. Therefore, it is up to the individual investor to determine whether there are additional benefits that are worth hiring a financial advisor.

References Barber, B. and Odean, T. (2008). All that glitters: The effect of attention and news on the buying behavior of individual and institutional investors. Review of Financial Studies, 21, 785–818. Bhattacharya, U., Hackethal, A., Kaesler, S., Loos, B., and Meyer, S. (2012). Is unbiased financial advice to retail investors sufficient? Answers from a large field study. Review of Financial Studies, 25, 975–1032. Bluethgen, R., Gintschel, A., Hackethal, A., and Muller, A. (2008). Financial advice and individual investors’ portfolios. Working Paper, EBS Universität für Wirtschaft und Recht and Goethe University Frankfurt. Brandt, M., Brav, A. Graham, J., and Kumar, A. (2010). The idiosyncratic volatility puzzle: Time trend or speculative episodes? Review of Financial Studies, 23, 863–899. Carhart, M. (1997). On persistence in mutual fund trading, Journal of Finance, 52, 57–82. Dorn, D. and Weber, M. (2013). Individual investors’ trading in times of crisis: Going it alone or giving up? Working Paper, Drexel University and Universitait Mannheim. Fama, E. and French, K. (1993). Common risk factors in the returns on stocks and bonds, Journal of Financial Economics, 33, 3–56. Foerster, S., Linnainmaa, J., Melzer, B., and Previtero, A. (2015). Retail financial advice: Does one size fit all? Journal of Finance, forthcoming. Hoechle, D., Ruenzi, S., Schaub, N., and Schmid, M., (2016). The impact of financial advice on trade performance and behavioral biases. Review of Finance, forthcoming. Inderst, R. and Ottaviani, M. (2009). Misselling through agents. American Economic Review 99, 883–908.

Journal of Personal Finance

Karabulut, Y. (2013). Financial advice: An improvement for worse? Working paper, Erasmus University. Kramer, M. (2012). Financial advice and individual investor portfolio performance. Financial Management, 41, 395–428. Kramer, M. and Lensink, R. (2012). The impact of financial advisors on the stock portfolios of retail investors. Working Paper, University of Groningen. Mullainathan, S., Nöth, M. and Schoar, A. (2012). The market for financial advice: An audit study. NBER Working Paper. Shapira, Z. and Venezia, I. (2001). Patterns of behavior of professionally managed and independent investors. Journal of Banking and Finance, 25, 1573–1587.

Volume 16, Issue 1

Expected vs. Actual Retirement Savings Behavior of Highly Educated Individuals Kristine Beck, Ph.D., Assistant Professor, California State University, Northridge Inga Chira, Ph.D., CFPÂŽ, Assistant Professor, California State University, Northridge

Abstract Using a unique sample of 318 respondents, we design a custom survey to examine savings understanding and behavior with respect to demographic attributes, long-term financial goals, and the level of financial knowledge of highly educated individuals. We find that savings expectations differ from actual savings behavior with regard to demographics and individualsâ&#x20AC;&#x2122; articulation of personal financial goals. However, we find a strong relationship between the level of financial knowledge and savings behavior. Financial knowledge is measured using awareness of the tax benefits of retirement savings, stock market performance, specifics of financial instruments, and self-reported financial savvy.

Journal of Personal Finance

Introduction and Motivation Economic modeling of saving and investment behavior usually treats individuals as rational, utility-maximizing decision makers. As a result, the focus is often on the optimal design of pension plans and on the increased need for financial education aimed at encouraging participation in retirement savings. In reality, an individualâ&#x20AC;&#x2122;s motivations differ from what a rational model would predict. Benartzi and Thaler (2007) study savings behavior in the context of increased participation of defined contribution plans. They point out that many, if not all, people do not solve optimization problems; instead, they use biases and heuristics and employ those to savings in the context of retirement. Similarly, Agnew (2006) identifies a number of behavioral biases that affect retirement planning. Generally, studies derived from secondary data focus on the final decision rather than on the process employed to decide. For example, we can observe that a particular employee made the decision not to enroll in a 401(k) plan, but we do not know how the individual arrived at that decision. Benartzi and Thaler (2007) take a step in this direction by examining the reasons employees make the decision to join a voluntary retirement plan in the first place, about how much to contribute to their defined contribution plans, and how to invest their contributions. The authors show that these decisions rely on rules of thumb, biases, and heuristics. For example, instead of consulting experts, participants tend to ask their spouses, coworkers and friends for retirement and investment advice [Benartzi and Thaler (1999), Duflo and Saez (2002)]. Benartzi and Thaler (1999), show that, generally, most people spend less than one hour making retirement decisions. Given the long-lasting impact of such decisions on wealth and retirement, it is important to understand why this does not appear to be important to the participants in the study. One possible explanation is that there is a specific segment of the population that is more prone to plan for retirement1 , and the segment can be identified and categorized by specific demographic characteristics. The remainder of the population spends considerably less time 1.

Even though the time spent on retirement planning may not necessarily indicate superior retirement planning capabilities and outcomes, spending less than one hour on analyzing a new plan comparing the investment choices and decisions on the asset allocation is probably not enough time to make optimal contribution and allocation decisions.

making retirement planning decisions. Another possibility is that individuals have other priorities/goals which makes the choice to spend so little time and effort on retirement planning a rational choice. A third possible explanation is that individuals realize that retirement savings are important but they do not have the knowledge or the tools to translate their understanding into actions. We collect primary data through a survey to study the drivers for decisions about individual savings by linking savings expectations and actual savings to (1) individual demographic characteristics, (2) personal goals, and (3) financial knowledge. Our main goal is to shift the discussion from how individuals save to why they make specific savings decisions. Within this context, we analyze specifically how highly educated individuals think about retirement savings and what they do to prepare for retirement. A National Center for Policy Analysis study from 2013 shows that increased levels of education are associated with higher retirement income outside of Social Security and the more years of education a person has, proportionally, the higher is the increase in retirement income. So-Hyun and Grable (2005) also show a link between higher education/higher income and the existence of workplace retirement savings. Concomitantly, a number of consumer surveys point out that only 20% of U.S. households use financial planners for help [for example, see Elmerick, Montalto, and Fox (2002), Hanna and Lindamood, (2010)]. But as the authors themselves point out, this picture is misleading as the use of planners is concentrated in a specific (wealthy and educated) strata of the population. Similarly, Kim and Hanna (2015) find college-educated households are more likely than high school-educated households to work with a financial planner and have adequate retirement savings. Chira (2016) finds that about 57% of highly educated individuals met with a planner. Given the prior literature findings about those who save for retirement and those who use a financial planner, we concentrate our sample on highly educated individuals. These individuals are more likely to actively make longterm financial savings decisions, as well as know about and employ financial planning services. Instead of using secondary surveys that contain vastly heterogeneous data, our sample includes individuals who are relatively more educated and wealthy compared to the median U.S. household. Taken together, our sample

ÂŠ2017, IARFC. All rights of reproduction in any form reserved.

Volume 16, Issue 1

represents individuals who are more likely to think about retirement, save for retirement, and employ professional services to do so. Our contribution is twofold. First, we categorize the reasons and explore the relative importance of each of the three broad categories that were previously shown to drive retirement savings behavior: individual demographic characteristics, specific financial goals, and financial knowledge. Second, unlike the studies that use a varied audience of respondents, we focus on a sample that is known to take an active role in retirement savings and to seek professional advice [for example, So-Hyun and Grable (2005), Elmerick et al. (2002), or, Hanna and Lindamood, (2010)].

Literature Review and Hypotheses Development Our theoretical framework builds on prospect theory. Developed by Kahneman and Tversky (1979) and extended by Bowman, Minehart and Rabin (1999) to incorporate the effects of consumption, the theory can be used to understand savings and consumption behavior of the U.S. population. Applying this theoretical framework to the 2008 financial crisis, Fredrickson (2013) shows that household savings patterns differ based on the losses or gains experienced between 2007 and 2009. Additionally, we integrate Shefrin and Thaler’s (1988) behavioral lifecycle savings theory to further examine our hypotheses and the inclusion of control variables in our models. Savings behavior is often studied along with financial knowledge and education by stratifying the population along different demographic characteristics by using the framework of a traditional lifecycle savings theory. Demographics have been shown to influence responses on financial literacy surveys and on the retirement planning process. For example, using the Survey of Consumer Finances (SCF), Hilgert and Hogarth (2002) find that education, race, and age correlate with financial literacy. And, using the Health and Retirement Study (HRS), Lusardi and Mitchell (2007a and 2007b) confirm the link between knowledge and race/gender. Specifically, these studies find that the more financially literate individuals are more educated, are white, and are neither young or old (between 36−65). Using structural equation modeling, Hershey and Mowen (2000) similarly conclude both “personality constructs and

financial knowledge are significant determinants of pre-retirement planning” (687). We examine the effects of setting goals on saving behavior within the context of Lopes’s (1987) SP/A theory. A psychological framework that examines decision making under uncertainty, the framework attempts to balance the feeling of security (S) with aspirational levels (A), or in our case, goals. The theory predicts that investors will try to match their savings motives with their aspirational level. Thus, the savings behavior of households will relate to their goals. Generally, the financial literacy literature points to a strong positive link between financial knowledge and savings behavior. Due to the availability of data, many financial literacy studies have a tendency to focus on high school/ college students [for example, see Lyons (2007), Xiao et al. (2010) or Tew and Tew (2014)]. Others (particularly those based on SCF) examine working adults with relatively little education/income compared to our sample or people in a specific age bracket (for example, the Lusardi and Mitchell work based on HRS studies). Consequently, the differences in questions asked in the administered tests/surveys are relatively small. Most of the questions asked focus on textbook economics and applied personal finance material. One of the main conclusions many of the studies offer is that individuals are not adequately prepared for retirement because they do not possess the financial knowledge or because they do not plan for retirement [see Lusardi and Mitchell (2006a), (2007a)]2. The authors illustrate that the U.S. savings rates are low and investment/retirement decisions are rarely based on time-tested financial principles. As a result, people approach retirement unprepared. This conclusion is relatively common in the research on financial literacy and savings. Other studies, such as Moore (2003) and Agnew and Szykman (2005), explore the direct understanding of financial instruments. These studies show that respondents have little understanding of how and where to invest or how mutual funds work. Ironically, these are examples of the types of financial decisions responders are expected to make in their own retirement planning. Lusardi and Mitchell (2007a) connect the demographic characteristics 2.

For example, Lusardi and Mitchell (2007a) provide a review of current financial education and literacy programs concluding that “many households are unfamiliar with the most basic economic concepts needed to make investment and savings decisions” (page 35).

Journal of Personal Finance

of the respondents to savings behavior, concluding that individuals who possess more math knowledge (such as lottery division) are more likely to have saved for retirement. Earlier work by Hilgert, Hogarth, and Beverly (2003) shows a similar link. Smith and Barboza (2014) find a positive relationship between self-reported overconfidence on financial knowledge and student debt. Although the recommendations that follow these studies differ, the lack of financial literacy usually calls for increased financial education as a solution to the lack of savings and retirement preparedness. As previously noted, there is a positive relationship between an increase in the education level, an increase in income, and an increase of using professional advice, such as the advice of financial planners. However, the access of lower income/lower education individuals to financial planners is below that of higher education/higher income individuals. For example, Hanna and Lindamood (2010) question the common 20% number seen in some studies by concluding that in the lower net worth households, only 10% may be using any professional financial advice, while in wealthier households, employment of professional services is higher than 20%. Lower income and net worth households may not even have access to professional financial help from financial planners, given the fees and minimum assets associated with such services. According to the 2013 US census report in the population age 25 and over, 58.33% of adults have some type of college education, 41.50% have at least an associate degree, and one third of the population have completed a baccalaureate degree. Comparing the financial literacy assessments to this data, one would expect the U.S. population (or at least the half that has been exposed to college) to be more prepared to understand retirement savings and investments. One of the problems we see with many of the surveys is the focus on â&#x20AC;&#x153;textbook materialâ&#x20AC;? that may not necessarily correlate with practical savings or investment behavior. For example, while solving math problems may be an excellent tool for predicting problem solving skills, it may not translate into a better understanding of retirement planning. Tokar Asaad (2012) reinforces this point by showing that both self-assessed knowledge and factual knowledge affect financial behavior. In many cases the self-assessed, or perceived, knowledge has a greater impact on behavior.

Many college educated individuals have been exposed to a basic course in economics and have a basic understanding of (or at least familiarity with) concepts such as interest rates, compounding, and financial markets. Understanding the link between what educated individuals know and their actual savings behavior could offer insights into the use and value of financial education for retirement savings for other segments of the population. Our formal hypotheses are stated as: Hypothesis 1a: Demographic characteristics contribute to explaining individual retirement savings behavior. Hypothesis 1b: Setting financial goals positively contributes to explaining individual retirement savings behavior. Hypothesis 1c: Finance-related knowledge positively contributes to explaining individual retirement savings behavior.

Survey Design and Participation Faculty and staff at Oregon State University were surveyed regarding factors relating to expected retirement savings, actual savings behavior, and financial planning. A survey was emailed to all employees who have completed at least their undergraduate education. Additionally, a random sample of employees and faculty at Florida Atlantic University, the University of Alabama, and the University of Georgia were contacted to expand the sample.3 We emailed different groups (colleges) at these universities to avoid skewing the sample to specific departments. For example, the colleges of Arts and Letters, Engineering and Computer Science and Education were emailed at Florida Atlantic University while the College of Communication and Information Science, College of Community Health Science and the School of Social Work were contacted at the University of Alabama. The survey was anonymous and

Given that the responses were voluntary, the possibility of self-selection bias is a concern. To alleviate this concern, a total of 50 faculty members at Oregon State were asked to participate in the survey as part of non-financial event. This was a workshop organized on campus for a completely unrelated event about teaching effectiveness. We do not have any reason to believe that the participants had any particular interest in financial literacy or retirement savings. After the event, all the participants were asked to voluntarily participate in the survey with no reward. The participation rate was 93%.

ÂŠ2017, IARFC. All rights of reproduction in any form reserved.

Volume 16, Issue 1

consisted of 59 randomized questions.4 Survey questions and categories are identified below. The participants (entire departments) were emailed during the summer of 2014; the email explained the purpose of the study and included an invitation to participate in the survey. Our response rate varied between 10-18%, depending on the college. The highest response rate was observed in the department of philosophy and the lowest in science departments (such as physics and chemistry).

Retirement Scenarios We present four different scenarios in which respondents select the employee percent contribution to a hypothetical 401(k) account. The four scenarios altered the annual compensation (either $50,000 or $100,000) and the employer contribution. Employer contributions ranged from 0% to 4%, presented either as dollar-for-dollar contribution or a percent per dollar. This exercise was designed to mimic a real scenario that can be found upon acceptance of employment; in fact, the most complicated scenario is presented exactly as it appears in a medium-sized company in the Los Angeles area. This particular group of questions was designed to gauge whether respondents can hypothetically make the minimal allocation required in order to take advantage of the employer contribution, and how this decision changes with a change in income. Additionally, we ask two other questions to assess how much the respondents know about retirement savings in general. Specifically, we ask what percent of the gross salary should be saved starting with age 25 (and then again, starting with age 35) in order to maintain the same standard of living in retirement as currently available. We ask the questions in sequence intentionally; a difference in results should point to a basic understanding of time value of money. The last question in this category assesses the degree of importance respondents place on starting savings as soon as possible after graduating from college. Combined, the seven questions serve as the basis for our assessment of the perceptions respondents have about retirement savings, i.e., whether they know the importance 4.

A number of questions in the survey were about investment allocation in a hypothetical retirement plan along with questions that assess respondentsâ&#x20AC;&#x2122; risk tolerance. Those questions are outside of the scope of this study. The full survey and collected data are available upon request.

55 of savings and the optimal savings under changing conditions.

Actual Savings Behavior The next group of questions changes the focus from expected savings behavior to the actual savings behavior of the specific respondent. Participants are asked if they are currently saving, what percent of their salary they are currently saving, if they are currently participating in a retirement plan, and if they have ever participated in a retirement plan.

Financial Goals A number of questions focus on the long term financial goals of the respondents. The participants are asked to identify their main long-term financial goals as well as to select what they consider to be the most important goal. The purpose for these questions is twofold. First, we explore how well-articulated long-term goals are both in quantity and precision. Second, we categorize the goals in subsequent analysis. As a result, we are able to count the number of goals as well as to sort them into five categories: general debt reduction (credit cards, student loans, and car payments), house/mortgage, retirement, education, and other.

Financial Knowledge Eight questions measure the financial knowledge dimension of savings. Most of the questions are factual. These questions are intended to be as precise as possible and to relate directly to retirement saving knowledge, although some reference general financial expertise. For example, in one question we ask the 2014 IRS limit for allowable deferred compensation. In another question, we require a time value of money calculation on a monthly installment loan. Some questions are more subjective. For example, we ask the respondents to assess how financially savvy they consider themselves to be, or to rate how well college prepared them for making financial decisions. We also tabulate demographic information that could increase the financial knowledge and add it to this category. We identify respondents who have formal education at both the undergraduate and graduate levels in economics or business, as they are more likely to know the correct answer to some of these questions.

Journal of Personal Finance

Demographic Information The demographic information includes education (level attained and field), place of birth, the time spent in the U.S., gender, age, marriage and children, income, student status, credit score, and state of residence.

Data Description Table 1 presents the descriptive statistics for the sample. The education level of survey respondents is quite advanced: 38% of respondents have a Ph.D., and another 39% have a Master’s or professional degree (such as a JD). The remaining 23% have an undergraduate degree. Additional breakdown of the sample shows that only 3% of the Ph.D.s in the sample are in Business; 16% and 13% of the respondents have Master’s and undergraduate degrees in business, respectively. The average age of survey respondents is 42 (log 3.69), 83% of the sample was born in the United States, and 58% are female. Sixty-six percent of the respondents are married, and 43% have children. In addition to a high education level and family focus, the household income is fairly high: Almost 75% of the

respondents have a household income of $50,000 or more, and one third have household income over $100,000. When we eliminate full time students, 92% of the sample earns more than $50,000. The credit scores reflect the ability of the respondents to repay their debts; 47% have a credit score of over 760. Despite the education level of the respondents, 22% do not know their credit score. Table 2, Panel A addresses savings expectations and knowledge. The questions assess whether the percentage of hypothetical saved income depends on presented salary and employer contribution and whether the respondents have a good grasp of the amount of savings necessary at different stages in life in order to replace income at retirement. A number of findings are noteworthy. First, the higher the hypothetical income in a given scenario, the higher the percentage of savings chosen. The mean savings is 4% higher at an income of $100,000 compared to an income of $50,000. Second, employer contribution does not seem to have a significant impact on the percentage saved. The mean savings selected, however, is above the threshold of maximum employer match in all scenarios. Third, participants realize that the older you are when you

Table 1. Sample Descriptive Statistics Education MS=1 if the person’s highest degree is a master’s degree or a professional degree such as JD or MD and 0 otherwise; Education PhD=1 if the person’s highest degree is a doctorate degree and 0 otherwise; US Born=1 if the person was born in the US and 0 otherwise; Gender =1 if the person is male and 0 otherwise; Age is calculated as the natural log of the actual age provided by the respondent; Relationship=1 if married and 0 otherwise; Children=1 if the person has children and 0 otherwise; household income is represented by a variable where 1=–<30,000 per year, 2= 30,001−50,000, 3=50,001−75000, 4=75,0001−100,000 and 5>100,0001; Student =1 if the person is currently a full time student and 0 otherwise; college prepared is represented by a scale from 1 (not at all) to 10 (very well) that represents how well college prepared the respondent for making financial decisions for herself; credit score is a variable where 1 represents less than 660, 2=660−720, 3=721−760, 4=>760 and 5 is represented by “I don’t know.” The sample size is represented by any participant who answered the specific question. Variable

Mean/Proportion

Standard Deviation

Min

Max

Education MS (N=312)

0.7756

0.4183

Education PhD (N=312)

0.3846

0.4873

US born (N=312)

0.8269

0.3789

Gender (N=312)

0.4283

0.4953

Age (N=295)

3.69

0.3149

3.04

4.45

Relationship (N=312)

0.6603

0.4744

Children (N=312)

0.4327

0.4962

House. Income (N=297)

4.141

1.827

Student (N=312)

0.2276

0.4199

College Prepared (N=271)

3.2029

2.5209

Credit Score (N=296)

3.7297

1.0356

Volume 16, Issue 1

start saving for retirement, the more you must save. The mean increases from 14.38% to 21.04% as the starting retirement age shifts from 25 to 35 years of age. These answers were collected in free form, without multiple choices.5 Overall, hypothetical savings behavior points to the fact that most people are aware of how much they should save and are willing to save a significant amount that would certainly put them on the right path to retirement, if implemented.

should save for retirement, but in reality they are saving less. Another concern is retirement participation where 12% of the respondents who work full time and have access to retirement plans at their place of employment do not currently participate in the plan. This trend will be analyzed further in the multivariate analysis. Table 3 presents the answers for the knowledge part of the survey. One of the most surprising answers is the respondents’ lack of knowledge about the maximum allowed IRS limit for tax-deferred contributions. Only 28.5% of the participants were able to identify the correct limit, with 62.5% admitting that they do not know. Given the relatively high income of those in the sample and the advantage of tax deferrals, one would expect this population to seek ways to minimize taxes and be aware of the yearly limit. Additionally, only 22% of the respondents knew the exact employer match for their respective pension plan [401(k) or 403(b)]. Many of the participants mentioned that their university employer offers a match/guarantee (depending on the hire date) but they do not know how much it is

The second part of Table 2, Panel B, focuses on actual savings rather than expected savings. We find that 73% of the respondents are currently saving consistently. Eliminating the 18% of the sample representing graduate students (who are mostly PhD students on low assistantships) shows that approximately 10% of the working respondents do not save at this time. Furthermore, the mean savings rate is under 5% of salary. This points to an apparent disconnect between intention and implementation. At least hypothetically, respondents seem to be aware about how much one 5.

Depending on the source, financial planners advise saving between 10 and 15% starting at age 25.

Table 2A: Savings Expectations The table presents hypothetical savings behavior. Columns 2−5 represents the percentage chosen to be saved in a defined contribution savings plan when (1) the employee salary is $50,000 and the employer is not offering a match, (2) the employee salary is $50,000 and the employer is offering a dollar for dollar match up to 4% of the salary, (3) the employee salary is $50,000 and the employer is offering a $0.25 per dollar match up to 4% of the salary, and (4) the employee salary is $100,000 and the employer is not offering a match. Columns 6 and 7 represent the percentage of salary that should be saved if an employee starts retirement savings at the ages of 25 and 35 respectively. Variable

50K No Match

Mean 11.12 Median 10 SD 8.46 N 294 Min 0 Max 60 Note: * Eliminated 3 answers of 100%.

50K $1/$1* 10.17 8 8.22 295 0 60

50K $0.25/$1

100K No Match

11.45 10 9.27 285 0 75

15.07 13 9.05 271 0 60

Savings 25

Savings 35

14.38 10 9.26 299 2 68

21.04 20 11.09 297 4 75

Table 2B: Actual Savings Behavior The table presents actual savings behavior. Column 2 represents the percentage of employee consistently savings now, column 3, the percent of the salary saved at this point, column 4, identifies the percentage of respondents currently participating in a retirement plan and column 5, the percentage of respondents that have ever participated in a retirement plan.

Variable Mean Median SD N Min Max

Percent Consistently Saving 73.04 – 297 – –

Percent of Salary Saved <5% 5−10 – 300 – –

Percent Current Retirement Participation 69.96

Percent Ever Retirement Participation 62.80

–

– 294 – –

– 293 – –

Journal of Personal Finance

Table 3: Knowledge The table represents knowledge-based information collected from participants. Column two identifies the correct or positive answer in percentage points, column 3 presents the percent of answers stating the lack of knowledge about the question being asked and column four identifies the most common answer/mean. The number of respondents who answered a particular question is displayed in the last column.

Variable

Correct/ YES

IRS Limit

28.53

ER Contribution Known1

I don’t know

Most Common

62.50

I don’t know

312

22.43

5-10

205

On my own (49.32%)2

296

Source Knowledge Financial Savvy

5.53

290

Stock Market Performance 2013

53.84

19.14

Very well/amazing

295

ETF versus MF

30.773

69.23

I don’t know

283

Have you ever had an Installment Loan

88.81

Yes

295

Can you calculate the PMT

91.49

Yes

294

Calculation PMT

73.08

Correct

182

1 (not at all)

271

How well did college prepare you?

3.20

(24)

Notes: 1. Only the participants who answered yes to a current retirement participation were asked about employer contribution. 2. 14.85% get their knowledge from a financial professional 3. 87 people specified something correct about the difference. 4. Mean is 3.20 and median is 2.

Table 4: Goals The table presents the four most common answers identifying participants’ financial goals. Mean (median)/Percent Number of Goals Listed

1.80 (2) (1−7)

Get out of debt

12.01

House/Mortgage

21.91

Retirement/Security

85.87

Education

17.67

Other (Charity, Legacy, Individual Goals)

28.27, 4.95, 1.41

283

because the contribution happens regardless of employee participation. When asked how financially savvy they are, the average response is 5.5 on a scale of 10. Forty-nine percent of respondents gather financial knowledge on their own. Those sampled believe college prepared them very poorly for making any kind of financial decisions (3.2 out of 10). To compare their self-assessed knowledge to actual financial knowledge, we asked about the general performance of the stock market in 2013 (53.8% have a good understanding that the stock market performed very well), and about the differences between ETFs and mutual funds

(only 30% of the respondents could identify one or more of the main differences). Finally, we ask whether the respondents would be able to calculate a monthly payment on a car loan and then, ask them to calculate the payment. While 91% of the respondents identified themselves as capable of performing the calculation, only 73% of the participants calculated the payment correctly. Table 4 presents a summary of the long-term financial goals identified by the survey respondents. The participants had no specific guidelines or categories to choose from; they could enter any text as a free response.

Volume 16, Issue 1

The number of goals ranged from 1 to 7 with a mean of 1.8 and a median of 2. The four most common categories mentioned are: (1) get out of debt, (2) buy a house/pay the mortgage, (3) retirement or security in old age and (4) further education for the family. Other goals were categorized under the heading ‘other,’ with the most common related to charity and legacy/estate. Overwhelmingly, the most important goal was retirement, mentioned by 85.8% of the respondents. Seventy-eight percent identified retirement as the most important goal.

Multivariate Analysis Table 5 presents the analysis of savings perceptions and behavior. Panel A shows the demographic factors impacting savings. Panel B presents goals, Panel C financial knowledge, and Panel D combines all three factors. Models 1 and 2 analyze savings expectations while Models 3 through 5 focus on actual savings behavior.6 We use OLS regressions in Models 1 and 2. The dependent variables are represented by the hypothetical percentage saved when the salary is $50,000 and the employer does not contribute to the retirement plan in Model 1, and the percentage of salary that should be saved for retirement if one starts savings at the age of 35 in Model 2. In Model 3 we employ logistic regressions, where the dependent variable is equal to one if the respondent identified herself as consistently saving at the present time and 0 otherwise. In model 4, we once again use an OLS regression to assess the relationship between our variable of interest, actual percentage of salary currently saved, and the different types of independent variables. Finally, in Model 5, we revert to the use of logistic regression to explore the association between current participation in employer provided retirement plans and the independent variables. Our dependent variable is equal to 1 if the participant is currently enrolled in a retirement plan at work and 0 otherwise. We find that demographic characteristics explain about 10% of savings expectations. Some limited evidence points to a difference in gender: males choose a higher percentage of savings than females. A consistent finding is the impact of children on savings perceptions. There is strong 6.

In a hypothetical scenario, participants were asked to select the percent of pension contribution if the employer did not offer any match and if the salary was $50,000 per year. In a second scenario, participants were asked how much they should save for retirement if they start saving around age 35.

negative relationship between how much participants think they should save and having children. Respondents with children choose a smaller percent contribution and a smaller percent of savings compared to respondents with no children. One possible explanation is that when people have children, they realize that savings patterns are influenced by the children and in turn, they adjust their savings expectations. Shifting the focus from ideas to reality, models 3 through 5 identify a number of factors that contribute to how much respondents currently save and their participation in retirement plans. We find that between 20−30% of actual savings behavior can be explained by demographic characteristics. Specifically, we find that household income and the existence of children consistently impacts savings behavior. The link between income and savings can also be explained within the prospect theory framework. These results are also consistent with Fisher and Montalto (2011), who find that “the resistance to lowering consumption in the face of lower-than-normal current income was greater than the resistance in increasing consumption in the face of higher-than-normal current income” (page 4). The higher the household income, the more likely respondents are to save, the higher the percentage saved, and the more likely participants are to participate in retirement plans. Again, the presence of children is negatively related to the existence of savings and the amount actually saved. Age and credit score also have an impact on the percent saved. Older respondents and participants with higher credit scores are more likely to save a higher percentage of their income. Finally, graduate students are less likely to participate in a retirement plan, which is not surprising given their full time student status. Panel B explores the impact of goal setting and types of goals on savings and retirement savings. Unlike the demographic characteristics that are found to have an impact on hypothetical savings behavior, personal goals do not impact the savings scenarios presented. Participants can likely dissociate themselves from their personal goals when they are making a hypothetical savings decision. In a way, they treat the hypothetical scenarios as problems that need to be solved with a specific correct answer. As such, they do not apply their personal goals to those scenarios. This is particularly interesting given the strong impact that children have on the same scenarios. Demographic

Journal of Personal Finance

Table 5: Multivariate Analysis Savings The table identifies the demographic (Panel A), goal (Panel B), and knowledge characteristics (Panel C) associated with the perceptions and the propensity of saving. Panel D combines all the variables into a comprehensive model. The dependent variables are: Model 1— the hypothetical percentage saved when the salary is $50,000 and the employer does not contribute to the retirement plan; Model 2— the percentage of salary that should be saved for retirement if one starts savings at the age of 35; Model 3—consistent current savings; Model 4—actual percentage of salary saved; and, Model 5—active current participation in a retirement plan. Models 1, 2, and 4 are based on OLS regressions; Models 3 and 5 employ logistic regression. The independent variables include: Education MS is equal to one if the respondent has a graduate degree and 0 otherwise; Education PhD is equal to one if the respondent has PhD and 0 otherwise; US born is equal to 1 if the respondent was born in the US and 0 otherwise; Gender is equal to 1 if male and 0 if female; Age is the current age (in years) of the respondent; Relationship is equal to 1 if the participant is married and 0 otherwise; Children is equal to 0 if the participant has children and 0 otherwise; Salary is presented by a range between 1 and 5 where 1 represents a total household salary less than $30,000 and 5, a household salary of more than $100,000; Student is equal to 1 if the respondent is a current full time student and 0 otherwise; College prepared ranges between 1 and 10 where 1 represents not at all and 10 represents very well and corresponds to how well college prepared participants for making financial decisions; C redit score ranges from 1 to 5 where 1 is less than 660, 4 is more than 760 and 5 is I don’t know. *, **, and *** represents significance at 10%, 5% and 1% respectively.

Panel A: Demographic Characteristics Variable

Constant (β) (P-value) Odds ratio Education MS

Education PhD

US born

Gender

Age

Relationship

Children

House. Income

Student

College Prepared

Credit Score

N Probability Model (Adj/Pseudo) R2

Model 1 How1 much to save 50k no match 1.868 (0.850) — –0.349 (0.833) — –1.231 (0.247) — -2.361* (0.099) -2.114* (0.057) — 2.346 (0.351) — 0.662 (0.606) — –3.430*** (0.001) — — — 2.814 (0.128) — –0.044 (0.825) — 0.677 (0.135) — 249 0.000*** OLS 0.1130

Model 2 Savings at 35 3.273*** (0.005) — –1.967 (0.332) — 0.938 (0.532) --1.179 (0.638) — 0.994 (0.518) — –3.430 (0.235) — –2.534 (0.192) — –4.523*** (0.000) — 0.582 (0.341) — 2.348 (0.416) — 0.192 (0.521) — 0.733 (0.231) — 262 0.005*** OLS 0.1046

Model 3 Consistently save

Model 4 Percent Saved

–4.324 (0.129) 0.029 –0.462 (0.299) 0.635 0.182 (0.639) 1.285 0.226 (0.616) 1.179 0.221 (0.505) 1.285 0.819 (0.297) 1.946 –0.292 (0.498) 0.738 –0.723* (0.071) 0.486 0.676*** (0.000) 1.966 0.369 (0.521) 1.447 0.030 (0.644) 1.031 0.037 (0.799) 1.038 263 0.000**** Logit 0.2057

–1.831 (0.182) — –0.062 (0.782) — 0.305 (0.112) --0.244 (0.271) — 0.160 (0.361) — 0.871** (0.020) — –0.373 (0.103) — –0.696*** (0.000) 0.270*** (0.000) — 0.197 (0.946) 0.028 (0.386) 0.139** (0.050) — 262 0.000*** OLS 0.2099

Note: 1. Verified the results with the two options for employer match. $1/$1 and $0.25/$1 with no significant differences.

Model 5 Current retirement plan –1.823 (0.499) 0.161 0.343 (0.494) 1.410 –0.170 (0.700) 0.843 0.398 (0.444) 1.489 0.175 (0.639) 1.192 0.238 (0.767) 1.269 –0.284 (0.567) –0.262 (0.526) 0.770 0.655*** (0.000) 1.927 –1.785*** (0.001) 0.168 –0.011 (0.883) 0.989 –0.164 (0.320) 0.849 263 0.000*** Logit 0.3480

Volume 16, Issue 1

Table 5: Multivariate Analysis Savings, continued Panel B: Goals The independent variables include: Number of goals identifies the total number of goals listed by the respondent; Debt reduction is equal to 1 if one of the goals identifies relates to debt reduction, including student loans, credit cards and car loans and 0 otherwise; Mortgage is equal to one if one of the goals identified relates to paying off the mortgage, savings for a house down payment or buying a house in the future and 0 otherwise; Retirement is equal to 1 if one of the goal identifies enough savings for retirement or security in retirement and 0 otherwise; and Education is equal to 1 if one of the goals relates to college payments for oneself, spouse or children and 0 otherwise. *, **, and *** represents significance at 10%, 5% and 1% respectively. Variable Constant (β) (P-value) Odds ratio Number Goals Debt reduction Mortgage Retirement Education N Probability Model (Adj/Pseudo) R2

Model 1 How much to save 50k no match 1.029*** (0.000) — 0.131 (0.856) –1.601 (0.142) — 0.180 (0.878) — 0.998 (0.476) — –0.604 (0.659) — 285 0.3989 OLS 0.0087

Model 2 Savings at 35 2.293** (0.000) — 0.086 (0.929) –1.027 (0.550) — 0.466 (0.801) — –1.839 (0.394) — –2.095 (0.178) — 292 0.6986 OLS 0.0084

Model 3 Consistently save –0.150 (0.659) 0.860 0.280 (0.201) –0.869** (0.019) 0.419 –0.229 (0.560) 0.795 0.879** (0.011) 2.402 0.236 (0.598) 1.266 303 0.000*** Logit 0.0569

Model 4 Percent Saved 2.091*** (0.000) — 0.310** (0.021) –0.943*** (0.000) — –0.315 (0.154) — 0.183 (0.434) — –0.428 (0.097) — 294 0.000*** OLS 0.0689

Model 5 Current retirement plan –0.393 (0.271) 0.675 0.010 (0.958) 0.121 (0.743) 1.129 –0.161 (0.675) 0.851 1.297*** (0.000) 3.661 0.486 (0.204) 1.627 303 0.004*** Logit 0.0468

Panel C: Knowledge The independent variables include: IRS limit is equal to 1 if the respondents correctly chose the 2014 limit for tax deferred retirement contributions and 0 otherwise; ER contribution represents the percentage of salary match contributed by employer to a retirement plan where 1 is “I don’t know” and 0 is a specific percent identified by the respondent; Source of knowledge is equal to 1 if the majority of financial knowledge is self-taught and 0 otherwise; Financially savvy ranges between 1 and 10 where 1 is not at all and 10 is very savvy and represents a self-identified number; Stock market equals to 1 if the participant answered very good/excellent when asked about the stock market performance in the US in 2013 and 0 otherwise; PMT calculation is equal to 1 if the participants calculated correctly the monthly payment on a car loan and 0 otherwise; ETF vs Mutual Funds is equal to 1 if the respondent identifies correctly one of the main differences between the two investments and 0 otherwise; and, Education MS B and Education BS B equals to 1 if the respondents has a graduate or undergraduate degree in economics or business and 0 otherwise. *, **, and *** represents significance at 10%, 5% and 1% respectively. Variable

Model 1 Model 2 How much to save 50k Savings at 35 no match Constant (β) 0.999*** 1.702*** (P-value) (0.000) (0.000) Odds ratio — — IRS Limit 2.605*** 4.551*** (0.008) (0.004) — — ER Contribution –1.900*** 0.813 (0.000) (0.492) — — Source of Knowledge –0.594 –1.051 (0.537) (0.400) — — Financial Savvy 0.008 0.465 (0.968) (0.101) — — Stock Market –0.476 –1.375 (0.616) (0.312) — — PMT Calculation –0.410 –0.93 (0.648) (0.885) — — ETF vs Mutual Funds 4.029*** 1.353 (0.003) (0.366) — — Education MS B 1.433 2.097 (0.259) (0.272) — — Education BS B –1.891 0.742 (0.112) (0.697) — — N 263 274 Probability 0.000*** 0.002*** Model OLS OLS 0.1423 0.0800 (Adj/Pseudo) R2 Note: We verified the results including state fixed effects

Model 3 Consistently save –0.206 (0.585) 0.813 0.801** (0.047) 2.228 0.563* (0.055) 1.757 –0.144 (0.657) 0.865 0.165** (0.033) 1.180 0.908*** (0.008) 2.481 –0.125 (0.691) 0.882 0.276 (0.471) 1.319 –1.452*** (0.001) 0.234 0.147 (0.767) 1.159 277 0.000*** Logit 0.1346

Model 4 Percent Saved 1.686*** (0.000) — 0.631*** (0.001) — 0.109 (0.353) — 0.057 (0.734) — 0.093** (0.013) — 0.418** (0.024) — –0.204 (0.215) — 0.256 (0.216) — –0.427* (0.069) — –0.224 (0.375) — 276 0.000*** OLS 0.1711

Model 5 Current retirement plan 0.207 (0.598) 1.231 0.778** (0.030) 2.178 0.107 (0.667) 1.113 0.397 (0.193) 1.489 0.122* (0.098) 1.129 0.219 (0.487) 1.245 –0.133 (0.654) 0.875 0.895** (0.010) 0.408 –1.089*** (0.006) 0.336 –0.045 (0.915) 0.955 277 0.006*** Logit 0.0780

Journal of Personal Finance

Table 5: Multivariate Analysis Savings, continued Panel D: Combined Model Variables Constant (β) (P-value) Odds ratio Education MS Education PhD US born Gender Age Relationship Children House. Income Student College Prepared Credit Score Number Goals Debt reduction Mortgage Retirement Education IRS Limit ER Contribution Source of Knowledge Financial Savvy Stock Market PMT Calculation ETF vs Mutual Funds Education MS B Education BS B N Model Probability (Adj/Pseudo) R2

Model 1 How much to save 50k no match 1.190 (0.566) — –0.858 (0.631) — –0.789 (0.474) — –3.013*** (0.035) — 1704 (0.101) — 1.067 (0.744) — 0.506 (0.757) — –2.616** (0.010) — — 3.429** (0.039) — –0.226 (0.330) — 0.774* (0.077) — –1.052 (0.142) — 0.611 (0.615) — 0.179 (0.881) — 2.238* (0.070) — 1.797 (0.137) — 2.016** (0.044) — –2.274*** (0.000) — –0.325 (0.745) — –0.115 (0.639) — –0.293 (0.814) — –0.016 (0.986) — 3.531** (0.012) — –0.445 (0.745) — –0.996 (0.401) — 239 OLS 0.000*** 0.2544

Model 2

Model 3

Model 4

Savings at 35 3.322** (0.016) — –2.400 (0.303) — 1.244 (0.425) — –2.109 (0.433) — 0.302 (0.847) — –5.383 (0.173) — –2.540 (0.216) — –3.561*** (0.008) — 0.425 (0.545)

Consistently save –4.893 (0.148) 0.007 –0.361 (0.475) 0.696 –0.145 (0.748) 0.867 0.365 (0.522) 1.441 –0.140 (0.744) 0.869 0.370 (0.673) 1.448 –0.333 (0.462) 0.716 –1.013** (0.028) 0.360 0.882*** (0.000)

2.694 (0.373) — –0.171 (0.653) — 0.598 (0.351) — –1.103 (0.365) — 0.582 (0.781) — 0.783 (0.698) — 0.673 (0.765) — –1.351 (0.438) — 3.439** (0.028) — 0.787 (0.421) — 0.865 (0.641) — 0.755** (0.037) — 0.865 (0.641) — –0.323 (0.827)) — 0.341 (0.848) — –0.013 (0.996) — 0.052 (0.982) — 250 OLS 0.001*** 0.1726

0.276 (0.696) 1.319 0.764 (0.437) 1.079 –0.080 (0.638) 0.922 0.434 (0.159) 1.544 –1.037** (0.027) 0.355 –0.076 (0.437) 0.926 0.111 (0.834) 1.118 0.418 (0.445) 0.520 0.900 (0.110) 2.461 0.512*** (0.005) 1.669 –0.205 (0.669) 0.814 0.226** (0.028) 1.254 0.204 (0.664) 0.227 –0.278 (0.494) 0.757 0.650 (0.189) 0.916 –1.826*** (0.002) 0.161 0.270 (0.643) 1.311 251 Logit 0.000**** 0.3525

Percent Saved –1.218 (0.411) — –0.069 (0.773) — 0.311* (0.098) — –0.213 (0.361) — –0.068 (0.724) — 0.527 (0.181) — –0.361 (0.110) — –0.619*** (0.001) — 0.287*** (0.000) 0.417 0.066 (0.806) — –0.000 (0.806) — 0.091 (0.221) — 0.188 (0.176) — –0.663*** (0.008) — –0.104 (0.631) — –0.110 (0.643) — –0.153 (0.536) — 0.547*** (0.005) — 0.129 (0.240) — 0.058 (0.741) — 0.107*** (0.007) — 0.030 (0.887) — –0.156 (0.334) — 0.258 (0.186) — –0.335 (0.159) — –0.067 (0.792) — 250 OLS 0.000*** 0.2091

Model 5 Current retirement plan –1.793** (0.031) 0.0000 0.821 (0.148) 2.227 –0.315 (0.552) 0.729 0.267 (0.671) 1.307 0.320 (0.480) 1.378 1.409 (0.140) 4.095 –0.162 (0.760) 0.849 –0.632 (0.186) 0.531 0.760*** (0.000) 2.141 –1.724*** (0.004) 0.178 0.069 (0.523) 1.072 –0.089 (0.653) 0.914 0.063 (0.837) 1.066 0.379 (0.490) 1.462 0.138 (0.797) 0.134 (0.827) 1.144 0.852 (0.140) 2.345 4.505** (0.024) 4.506 0.358 (0.121) 1.432 0.029 (0.947) 1.030 0.182* (0.075) 1.200 –1.321*** (0.007) 0.266 –0.412 (0.311) 0.662 0.542 (0.270) 0.581 –1.539*** (0.017) 0.214 0.122 (0.816) 1.130 251 Logit 0.000*** 0.4235

Volume 16, Issue 1

characteristics have a stronger impact on hypothetical scenarios compared to financial goals. Conversely, when it comes to actual savings behavior personal goals have a strong impact on whether the participants in the survey save and how much they save. There is an inverse relationship between the existence of savings and the type of personal financial goals. Respondents who identified non-mortgage debt reduction as a primary personal goal are less likely to save, and if they do save they put aside a smaller percentage of their salary. At the same time, participants who identified retirement and stability as their primary goal are more likely to save and are more likely to contribute to a current retirement plan. The relationship between savings and financial knowledge is presented in Panel C. It shows that between 8 and 17 percent of savings expectations and behavior can be explained by the knowledge component of the survey. We find a strong relationship between savings perceptions and the knowledge component. Not all knowledge, however, is created equal. Participants who are aware of the current limit imposed by the IRS on the maximum deferral of income for retirement savings are more likely to choose a higher savings percentage in the scenarios presented, are more likely to save, save for retirement, and save a higher percent of their income. We find this a very encouraging finding as it sheds some light on the type of education needed to encourage savings. An understanding of the tax savings associated with pension deferrals could be targeted to increase savings participation in employer retirement plans and savings on an individual level. Other financial knowledge information that has a positive effect on savings is the awareness of stock market performance, and to a lesser degree, the awareness of employer contributions to existing pension plans. Both have a positive relationship with the percentage saved. We also find that the degree of sophistication of financial knowledge possessed has an impact on participation in retirement savings. Participants who could explain basic differences between ETFs and mutual funds are more likely to currently participate in a retirement plan. One consistent factor that determines high savings behavior is the self-assessment of financial knowledge. Participants seem to have a good grasp on their financial knowledge; the higher they rate their financial knowledge, the more likely they are to save, save more, and save for retirement.

One of the surprising findings is the inverse relationship between the existence of a graduate degree in business/ economics and actual savings behavior. One would expect that the more business education one has, the more likely the person is to realize the importance of savings and savings early and as a result, act on that knowledge. We find no relationship between graduate business education and hypothetical savings behavior, which in itself is troublesome as these respondents are equipped with the knowledge to at least select the optimal targetâ&#x20AC;&#x201D;if not actualâ&#x20AC;&#x201D;savings behavior. Looking at the actual behavior, we find a consistent negative relationship. Graduate business education is associated with less saving and less retirement saving. One possible explanation is the expense associated with an MBA. It is possible that after graduation a higher student loan burden has to be relieved before savings behavior is modified. Panel D presents the combined model from Panels A through C. With the highest correlation coefficient of 0.429 and a VIF of 1.93, we include all the independent variables in a simultaneous model. The advantage of an overall model is potentially identifying variables that were found to serve as explanations for savings expectations and behavior that are proxies for other variables. Enforcing the individual results, we find that demographic characteristics, personal financial goals, and the degree and sophistication of financial knowledge possessed drives ideal savings behavior. Theoretical higher savings is associated with lack of children and knowledge about retirement tax incentives. Other important factors vary by the model and include whether one was born in the U.S. or abroad, student status, credit score, and the type of financial knowledge. When looking at the actual savings behavior, we confirm that the presence of children in a family is inversely associated with savings, while an increase in household income is positively associated with savings. Again, people who consider debt reduction as a primary goal are less likely to save while people who consider themselves and are measured as financially savvy are more likely to save, as well as specifically for retirement. Overall, the all-inclusive model explains between 20â&#x2C6;&#x2019;42 percent of actual savings behavior, from how much people save to their savings priorities.

Journal of Personal Finance

Conclusion Starting from a sample of highly educated individuals, we aim to identify characteristics that are associated with expected and actual savings behavior in the context of retirement planning. We find that overall, the level and depth of savings knowledge is adequate. The hypothetical scenarios present an attempt to assess whether individuals can calculate an appropriate savings rate given a specific scenario and whether they have the general knowledge of how much one should save for retirement given the age he/she starts saving. Theoretically, individuals are aware they should be saving and should choose an optimal portion of income to save when presented with multiple scenarios. Specifically, they realize that maximizing an employer 401(k) match is optimal savings behavior. They also realize that someone who starts saving for retirement at age 35 needs to save a higher percent of his/her income compared to someone who starts saving at age 25. In practice, participants save less than they say they should. The percentage saved is inversely associated with having children and having specific goals that consume a large percentage of income, such as paying off debt. Simultaneously, those who are more likely to save are wealthier individuals, those who think of themselves as financially savvy, and those who possess highly specific knowledge about savings and investments. Our findings relate to two separate strands of literature that explore retirement savings and expectations. First, similar to a number of other studies, we show a consistent stratification of the individuals who save for retirement and the ones who do not. Specifically, consistent with So-Hyun and Grable (2005) and Kim and Hanna (2015), we find that level of education is positively associated with retirement savings. Second, we explore the relative importance of the three major factors that have been shown to affect retirement savings: demographic characteristics, financial goals, and financial knowledge. Putting those three types of factors in a broader unified perspective, we show that when combined with other factors, demographic characteristics, the existence of goals, and actual financial knowledge all add explanatory power to actual savings behavior. These results are in line with prior studies focusing on only one of the three components, such as findings by Hershey and Mowen (2000), Hilgert, Hogarth and Beverly (2003) or Smith and Barboza (2014), among others.

One of the limitations of this study is the possibility of bias due to question order. For example, savings behavior is self-reported and measured after hypothetical savings scenarios. Respondents may under or over-estimate their actual savings when conditioned by hypothetical scenarios. Another limitation is the objective measurement of financial knowledge. Respondents took the survey unsupervised, with access to online resources. Some of the correct answers may have been the result of an online search rather than the actual participant’s measurement of financial knowledge. In addition, our survey may suffer from the typical shortcomings of survey work: participants may not be motivated or willing to provide honest answers that could showcase them in an unfavorable light. We hope that the anonymity of our survey mitigates this issue. This study has implications for those concerned about savings behavior in the United States, as increasing financial literacy does in fact increase actual savings for highly educated consumers. This finding can be used by financial planners and financial advisers who work with highly educated individuals to develop a niche, or specialization, and stand out from their competitors. Advisors who focus on highly educated individuals can use our findings to focus on client financial education. Although factors such as mortgage debt and having children decrease savings, knowledge regarding tax benefits, stock market performance, and financial instruments increase savings. Given that demographic attributes explain 11%−35% of variance in savings behavior in our sample, it is important to incorporate other variables when exploring U.S. households’ saving. We find that the articulation and presence of financial goals slightly increases our understanding of savings behavior, while the impact of knowledge related to savings behavior has a much larger impact.

References Agnew, Julie R. 2006. “Do Behavioral Biases Vary across Individuals? Evidence from Individual 401(k) Data.” Journal of Financial and Quantitative Analysis, 41(4): 939−62. Agnew, J., and L. Szykman. 2005. “Asset Allocation and Information Overload: The influence of Information Display, Asset Choice and Investor Experience.” Journal of Behavioral Finance, 6, 57−70.

Volume 16, Issue 1

Benartzi, S. and R. Thaler. 1999. “Risk Aversion or Myopia? Choices in Repeated Gambles and Retirement.” Management Science, 45(3): 364−381.

Lusardi, A., and O. Mitchell. 2006. “Financial Literacy and Planning: Implications for Retirement Wellbeing.” Pension Research Council Working Paper 1, The Wharton School.

Benartzi, S. and R. Thaler. 2007. “Heuristic Biases in Retirement Savings Behavior.” The Journal of Economic Perspectives, 21 (3): 81−104.

Lusardi, A., and O. Mitchell. 2007a. “Baby Boomers Retirement Security: The Role of Planning, Financial Literacy and Housing Wealth.” Journal of Monetary Economics, 54, 205−224.

Bowman, D., D. Minehart, and M. Rabin. 1999. “Loss Aversion in a Consumption-Savings Model.” Journal of Economic Behavior and Organizations, 38: 155−178. Chira, I. 2016. “Perceptions about the Financial Planning Profession.” Journal of Wealth Management, 18(4), 22−28 Duflo, E., and E. Saez. 2002. “Participation and Investment Decisions in a Retirement Plan: The Influence of Colleagues’ Choices.” Journal of Public Economics, 85: 121−148.

Lusardi, A., and O. Mitchell. 2007b. “Financial Literacy and Retirement Preparedness: Evidence and Implications for Financial Education.” Business Economics, 35−44. Lyons, A. 2007. “Credit Practices and Financial Education Needs of Midwest College Students, Working Paper.” Available at: http://www.usc.edu/dept/chepa/IDApays/publications/credit_practices.pdf.

Elmerick, S. A., C.P. Montalto, and J.J. Fox. 2002. Use of Financial Planners by U.S. Households. Financial Services Review, 11(3), 217−213.

Moore, D. 2003. “Survey of Financial Literacy in Washington State: Knowledge, Behavior, Attitudes and Experiences.” Technical report 03-39, Social and Economic Sciences Research Center, Washington State University.

Fredrickson, J. 2013. “Prospect Theory and Savings Behaviors During the Great Recession: 2009 SCF Panel Survey Results.” Journal of Behavioral Studies in Business, 6 (1).

Shefrin, H., and R. Thaler. 1988.” The Behavioral Life-Cycle Hypothesis.” Economic Inquiry, 26 (4): 609−643.

Hanna, S. D., and S. Lindamood. 2010. “Quantifying The Economic Benefits of Personal Financial Planning.” Financial Services Review, 19(2), 111−127.

Smith, C. and G. Barboza. 2014. The Role of Trans-Generational Financial Knowledge and Self-Reported Financial Literacy on Borrowing Practices and Debt Accumulation of College Students.” Journal of Personal Finance, 13(2): 28−50.

Hershey, D., & J. Mowen. 2000. “Psychological Determinants of Financial Preparedness for Retirement.” The Gerontologist, 40 (6): 687−97.

Tew, K., and P. Tew. 2014. “Financial Literacy and the College Student, Working Paper.” Available at http://papers.ssrn. com/sol3/papers.cfm?abstract_id=2430795.

Hilgert, J. and M. Hogarth, 2002, Financial Knowledge, Experience, and Learning Preferences: Preliminary Results from a New Survey on Financial Literacy, Consumer Interest Annual 48.

Tokar Asaad, C. 2012. “Perceived Financial Knowledge and Actual Financial Behavior: An International Assessment. Working Paper.” Kent State University.

Hilgert, M., J. Hogarth, and S. Beverly. 2003. “Household Financial Management: The Connection Between Knowledge and Behavior.” Federal Reserve Bulletin, 309−322.

Xiao, J., J. Serido, and S. Shim. 2011. “Financial Education, Financial Knowledge, and Risky Credit Behavior of College Students.” Chapter in Consumer Knowledge and Financial Decisions. New York: Springer.

Kahneman, D. and A. Tversky. 1979. “Prospect Theory.” Econometrica, 47 (2): 263−292. Kim, K. T., and S. Hanna. 2015. “Does Financial Sophistication Matter in Retirement Preparedness?” Journal of Personal Finance, 14 (2): 9−20.

Journal of Personal Finance

Risk Tolerance and Goals-Based Savings Behavior of Households: The Role of Financial Literacy

Swarn Chatterjee, University of Georgia, Athens, GA Lu Fan, University of Georgia, Athens, GA Ben Jacobs, University of Georgia, Athens, GA Robin Haas, University of Georgia, Athens, GA

Abstract This study uses a national dataset to examine the association among risk tolerance, financial literacy, and goals-based savings behavior of households. The results indicate that three out of five households do not have any emergency funds set aside, and about half the households have not calculated how much money they will need for retirement. However, both financial literacy and risk tolerance are associated with goals-based savings behavior, such as saving for emergencies, and planning for retirement among households. Although risk tolerance appears to be an important factor in the savings and investment decisions of households, the findings of this study provide further evidence regarding the role of financial literacy in improving household financial capability. Implications for policy makers, scholars, and researchers in the area of behavioral economics and household finance are included.

Keywords Risk tolerance, financial literacy, savings behavior, financial decision making, household finance, financial capability

ÂŠ2017, IARFC. All rights of reproduction in any form reserved.

Volume 16, Issue 1

Introduction The term financial risk tolerance is defined as investors’ willingness to lose a portion of their investment with the prospect of higher expected future returns (FINRA Suitability Standard, Rule 2111). Similarly, Grable (2008) defines risk tolerance as individuals’ willingness to act toward attaining a certain goal, where the outcome of their actions is uncertain and usually accompanied by the possibility of some loss. In general, risk tolerance as a concept is borrowed from the area of economic psychology and is approximately inversely associated with the economic concept of risk aversion (Kimball, Sahm, & Shapiro, 2008). A vast amount of research over the past half century has been devoted to understanding financial risk tolerance and factors that affect risk-averse behavior of individuals. Riley and Chow (1992) find that educational attainment of individuals was associated with their financial risk tolerance. Yao and Hanna (2005) find association between risk tolerance and gender difference, with single men being more risk tolerant than either married men and women and single women, and single women being more risk averse than either married or single men and married women. Sung and Hanna (1996) conclude that the differences in risk tolerance across gender, race, ethnicity, and income strata are primarily due to differences in understanding and interpretation of the nature of risk. Educational attainment is therefore a factor in risk tolerance because better educated individuals are better informed and hence have a better understanding of the nature of risk than individuals with lower levels of educational attainment. According to Faff, Mulino, and Chai (2008), financial risk tolerance is a combination of a person’s willingness and ability to take on financial risks. Therefore, to accurately measure one’s financial risk tolerance, the level of financial knowledge must be considered along with other psychological characteristics, time horizon, financial resources, and capacity to take risk. Similarly, Hallahan, Faff, and McKenzie (2004) find that educational attainment along with income and wealth is related to financial risk tolerance. Sages and Grable (2010) report that financial numeracy and household financial management skills of individuals affect their perception of financial risk tolerance. The authors conclude that individuals with low financial knowledge are unable to plan, prepare, or attain

their goals without further financial education or the help of a professional financial advisor. The evidence from earlier studies, as discussed in the previous paragraphs, suggests a positive association between knowledge or the ability to understand and interpret financial information and risk tolerance. Interestingly, according to the findings from a recent study (Huzdik, Beres, & Nemeth, 2014), many individuals have a mistaken belief that they are financially knowledgeable. The authors suggest that up to 30% of the respondents perceived themselves as being more financially knowledgeable than their measured financial literacy score would suggest. The role of financial knowledge or financial literacy is important because of its association with risk tolerance. Jacobs-Lawson and Hershey (2005) find that individuals with higher levels of financial knowledge who are risk tolerant are more likely to save for retirement than others, but individuals with lower levels of financial knowledge who have high risk tolerance are not. Financial literacy can be defined as an individual’s ability to use knowledge to manage financial resources effectively. Lusardi and Mitchell (2007) find that financial literacy is associated with retirement security of near-retirement households. Nicolini, Cude, and Chatterjee (2013) report that educational attainment and experience in managing various types of household assets and debt are positively associated with individuals’ financial literacy. According to Hanna, Waller, & Finke (2008) objective risk tolerance, even when measured with a valid risk tolerance questionnaire, may not provide sufficient information to enable the construction of an asset allocated portfolio that will match a client’s utility preferences optimally. This is because majority of the currently available risk tolerance scales are constructed based on responses to questions that implicitly correspond to the participants’ risk capacity and expectations. The authors, therefore, suggest that financial planners, when advising their clients, also need to examine their clients’ assets and other financial resources to separately understand the role of risk capacity, and provide financial education to manage their client’s expectations to more realistic levels. The authors caution that relying solely on the risk tolerance profiles based on “ad-hoc” risk tolerance questionnaires, when providing investment advice to clients, may lead to “inappropriate” portfolio recommendations. Similarly, Callan and Johnson (2002) suggest that in addition to

Journal of Personal Finance

measuring risk with a valid instrument, when advising their clients, the financial planners need to also examine their clients’ financial profile, better understand their resources or the clients’ capacity to take more risks, and conduct an interview with their clients to seek out more information about their risk preferences and expected returns, before they make any portfolio recommendation. Recent studies have found that financial literacy is positively associated with wealth accumulation and retirement savings of households across time after controlling for risk tolerance, among other factors (Van Rooij & Lusardi, 2012; Croy, Gerrans, & Speelman, 2010). However, with the exception of a few studies, such as Jacobs-Lawson and Hershey (2005), very few studies have looked at the interaction among financial literacy, risk tolerance, and its effect on the retirement savings behavior of individuals. The purpose of this study is to explore the relationships between financial literacy scores and assessments of risk tolerance, as well as their interactions and associations with individuals’ financial goals-based savings behavior. The goal is an improved understanding of the effects of the interaction between financial literacy and risk tolerance on the financial planning and savings behavior of households. The findings from this study will be useful for policy makers interested in improving the retirement security of households. The findings will also be useful for financial planners in better understanding their clients’ financial capability when providing them with financial advice. Moreover, this study will fill an important gap in the literature by examining the nuanced effects of financial literacy and risk tolerance on individuals’ goals-based financial decisions.

field of behavioral economics, individuals make often predictable, psychological mistakes when making financial decisions under uncertainty. According to prospect theory (Kahneman & Tversky, 1979), people are risk averse in the domain of gains and risk seeking in the domain of losses. Other behavioral economic studies have found that instead of integrating their financial decisions, individuals tend to look at financial decisions in a more task-specific way that is independent from each other (Statman, 1999). This is known as mental accounting. These mistakes are magnified when individuals have insufficient knowledge and inability to understand the risks associated with certain financial decisions. When individuals do not have sufficient knowledge or do not have the ability to estimate the expected outcome of a specific financial decision, they tend to perceive such decision from the domain of preserving gains and, hence, demonstrate a preference for the status quo and become more risk averse. Previous studies have found that individuals’ aversion to risk regarding investments drops with better ability to process and understand financial information and the risk environment (Riley & Chow, 1992). As a result, risk tolerance is positively associated with educational attainment and perhaps, more specifically, financial literacy. Figure 1 illustrates the association among financial literacy, risk tolerance, and goals-based household savings decisions. Figure 1. Financial literacy, risk tolerance, and goalsbased savings

Conceptual Framework Every individual has a unique perception of the uncertainty or risk they face in the outcome of investment choices or financial decisions they make in everyday life. Therefore, it is difficult to develop a universally optimal portfolio allocation for goals-based savings strategies, such as saving for retirement, saving for emergencies, or planning for children’s education. As a result, household financial decisions may differ and often appear irrational when compared with the optimal allocation for a specific goal that rational economic theories suggest (Thaler & Benartzi, 2004). However, this difference is explained by the different levels of risk tolerance that households possess (Jacobs-Lawson & Hershey, 2005). According to findings from the burgeoning

Financial literacy and educational attainment improve individuals’ capacity to better estimate their risks and comprehend information regarding the financial markets in which they participate (Van Rooij & Lusardi, 2012). Greater knowledge has also been positively associated with risk tolerance, as discussed in the previous literature. Given this association between knowledge and risk tolerance (Riley & Chow, 1992; Sung & Hanna, 2005; Grable, 2008), and the association between financial knowledge and educational attainment with wealth accumulation and planning behavior (Lusardi & Mitchell, 2007; Chatterjee &

Volume 16, Issue 1

Zahirovic-Herbert, 2010) in prior literature, we hypothesize that: H1: Individuals with higher financial literacy and knowledge have higher risk tolerance scores than those with lower financial literacy scores after controlling for other socioeconomic and demographic variables associated with risk tolerance in previous literature. H2: Individuals with higher financial literacy and greater risk tolerance are more likely to plan for goals-based savings behavior after controlling for other socioeconomic and demographic variables associated with wealth accumulation and retirement planning in previous literature.

Methods Data Data for this analysis use the 2012 wave of the Financial Industry Regulatory Authority (FINRA) Investor Education Foundation National Financial Capability Study (NFCS). The NFCS is a national dataset that includes information on financial literacy, risk tolerance, retirement planning, and financial market participation of households in the United States. The dataset also includes extensive information on the socioeconomic and demographic characteristics of the households participating in the survey; 25,509 respondents were included in the 2012 wave of the NFCS dataset. In the analyses of this study, we have restricted our sample to the 15,233 non-retired households after excluding the missing and non-responses because some of the retirement planning and financial market participation-related

questions were more appropriate for the retired respondents in the dataset.

Dependent Variables Risk Tolerance The first dependent variable for our analysis is a measure of risk tolerance. The participants were asked the following questions: “When thinking of your financial investments, how willing are you to take risks?” The respondents were then asked to enter their responses on a scale of 1 to 10, with “very willing to take risk” being 10, and “not at all willing to take risk” being 1. We constructed three binary variables based on the responses to the scale shown in Table 1. Those respondents in approximately the bottom third of this scale (risk tolerance score of 3 or lower) were coded as low risk tolerance (1=YES; 0=NO). The respondents in middle third of this scale (risk tolerance score 4−6) were coded as moderate risk tolerance (1=YES; 0=NO). In an earlier study on risk tolerance, Hanna and Lindamood (2004) had estimated that a response of substantial risk in the SCF might correspond to a relative risk aversion level of roughly 2, which is not “risk seeking.” Consistent with this we have avoided the term “risk seeking” in this study, and the respondents in the top third of the scale (risk tolerance score 7−10) in our study were coded as high risk tolerance (1=YES; 0=NO). Goals-based Savings Decisions Have Emergency Funds: A binary variable was constructed based on the participants’ responses to the following

Table 1: Distribution of the Responses to the Risk Tolerance Measure Risk Tolerance

Frequency

1,858

1,111

1,615

1,485

9.75

39.84

1,972

12.95

52.79

1,966

12.91

65.69

2,126

13.96

79.65

1,586

10.41

90.06

648

4.25

94.31

866 15,233

Percentage 12.2 7.29 10.6

5.69 100

Cumulative 12.2 19.49 30.09

100

Journal of Personal Finance

question: “Have you set aside emergency or rainy day funds that would cover your expenses in case of sickness, job loss, economic downturn, or other emergency?” If the respondents answered “YES” to this question, the response was coded as 1; if they answered “NO.” it was coded as 0. Have Planned for Retirement: A binary variable was constructed based on the participants’ responses to the following question: “Have you ever tried to figure out how much you need to save for retirement?” If the respondents answered “YES” to this question, the response was coded as 1 and if they answered “NO,” it was coded as 0. Setting Aside Money for Child’s Education: A binary variable was constructed based on the participants’ responses to the following question: “Are you setting aside any money for your children’s college education?” If the respondents answered “YES,” the response was coded as 1 and if they answered “NO,” it was coded as 0.

Independent Variables Measured financial literacy was a variable of interest in this study. Construction of the financial literacy measure is described in Lusardi, Mitchell, and Curto (2010). The respondents were asked questions related to personal finance that everyone needs to apply in everyday life. These questions were as follows: Interest rate question: Suppose you have $100 in a savings account and the interest rate is 2% per year. After 5 years, how much do you think you would have in the account if you left the money to grow? The multiple choice options were: (1) More than $102, (2) Exactly $102, and (3) Less than $102. Inflation question: Imagine that the interest rate on your savings account is 1% per year and inflation is 2% per year. After 1 year, how much would you be able to buy with the money in this account? The respondents had to select from the following options: (1) More than today, (2) Exactly the same, and (3) Less than today. Bond question: If interest rates rise, what will typically happen to bond prices? The multiple-choice options were: (1) They will rise, (2) They will fall, (3) They will stay the same, and (4)

There is no relationship between bond prices and interest rates. Mortgage question: A 15-year mortgage typically requires higher monthly payments than a 30-year mortgage, but the total interest paid over the life of the loan is less. The respondents had to choose between: (1) True and (2) False. Stock question: Buying a single company’s stock usually provides a safer return than a stock mutual fund. The respondents had to choose between: (1) True and (2) False. The responses to each of the above questions were coded as 1 if correct and 0 if incorrect. The responses were then summed to create a composite financial literacy score ranging from 5 if the respondents answered all five questions correctly to 0 if they did not answer any of the questions correctly. The respondents who answered “prefer not to say” were dropped from this study. Other control variables included in this study were based on findings from previous literature (Hallahan, Faff, & McKenzie, 2004; Yao, Gutter, & Hanna, 2005; Roszkowski & Grable, 2005; Grable, 2008; Gilliam, Chatterjee, & Zhu, 2010). These variables are age, gender, marital status, presence of children in the household, educational attainment, employment status, income, and wealth.

Analyses The dependent variables for this study were binary. Empirical analyses in two previous studies on risk tolerance have used the cumulative logit models (Yao, Hanna, & Lindamood, 2004; Yao, Gutter, & Hanna, 2005). Following Wooldridge (2012), we ran probit models to estimate the determinants of being low, moderate, and high risk tolerance among the respondent. The remaining three dependent variables were emergency funds, planned for retirement, and planned for children’s education. These were coded as binary variables. Hence, we have also used probit models for the analyses of these three variables.

Results Table 2 shows the descriptive statistics for this study. The descriptive statistics indicate that 30% of the respondents had low risk tolerance, 36% had moderate risk tolerance, and 34% had high risk tolerance. We also see that 40%

Volume 16, Issue 1

Table 2: Descriptive Statistics Standard N=11,720 Mean Deviation Low risk tolerance 30% Moderate risk tolerance 36% High risk tolerance 34% Emergency funds 40% Save for childrenâ&#x20AC;&#x2122;s education 41% Planned retirement 47% Financial literacy score 3.479 1.697 Age Age 25-34 17% Age 35-44 17% Age 45-54 20% Age 55-64 19% Age 65 plus 17% Female 55% Race White 69% Black 12% Hispanic 10% Asian 5% Other 4% Marital Status Married 56% Single 27% Divorced/separated 13% Widowed 4% Have children 43% Number of children 0.729 1.092 Educational Attainment Less than high school 7% High school 26% Some college 33% College 21% Graduate 13% Income Income <15k 13% From 15,000 to 24,999 12% From 25,000 to 34,999 11% From 35,000 to 49,999 15% From 50,000 to 74,999 19% From 75,000 to 99,999 12% From 100,000 to 149,999 11% Income>149,999 7% Employment status Self-employed 12% Employed 66% Not employed 22% All variables are dichotomous (0 or 1) except as otherwise noted

reported having an emergency fund, 41% of the parents had saved for their childrenâ&#x20AC;&#x2122;s education, and 47% reported having calculated their retirement needs. Approximately 55% of the sample was women, and 69% of the sample was non-Hispanic white. Additionally, 12% were black, 10% were Hispanic, 5% were Asian, and the remaining 4% belonged to other races. Married respondents comprised 56% of the sample. The results also indicate that 34% of the

Min

Max

respondents had completed college or higher, 66% were employed, and 12% were self-employed. Determinants of Risk Tolerance and the Role of Financial Literacy The results of the probit estimation from Table 3 indicate that financial literacy is negatively associated with low risk

Journal of Personal Finance

Table 3: Ordered Probit Estimates of Financial Literacy and Risk Tolerance Low Risk Tolerance

Moderate Risk Tolerance

High Risk Tolerance

Coef

MFX

Sig

Coef

MFX

Sig

Coef

MFX

-0.019

0.011

-0.004

0.037

0.010

0.014

***

0.020

0.008

Age 18-24

-0.872

0.089

-0.124

***

0.107

0.076

0.040

0.596

0.078

0.234

***

Age 25-34

-0.622

0.079

-0.106

***

-0.087

0.066

-0.031

0.594

0.068

0.233

***

Age 35-44

-0.462

0.076

-0.085

***

-0.009

0.064

-0.003

0.371

0.066

0.145

***

Age 45-54

-0.269

0.070

-0.053

***

-0.019

0.060

-0.007

0.284

0.063

0.111

***

Age 55-64

-0.110

0.070

-0.023

0.009

0.060

0.003

0.133

0.063

0.052

0.325

0.035

0.071

0.151

0.029

0.055

***

-0.465

0.029

-0.177

***

Black

-0.009

0.054

-0.002

-0.091

0.047

-0.032

0.206

0.047

0.081

***

Hispanic

-0.041

0.059

-0.009

-0.062

0.048

-0.022

0.163

0.048

0.064

***

Asian

-0.170

0.090

-0.033

-0.183

0.164

-0.064

0.319

0.262

0.126

Other

-0.044

0.087

-0.009

0.097

0.073

0.036

-0.037

0.076

-0.014

Married

0.027

0.047

0.006

0.076

0.039

0.028

-0.131

0.039

-0.050

Widowed

-0.193

0.117

-0.037

0.087

0.103

0.032

0.103

0.109

0.040

Divorced or separated

0.026

0.057

0.006

0.047

0.051

0.017

-0.088

0.052

-0.034

-0.043

0.017

-0.009

-0.020

0.014

-0.007

0.056

0.014

0.022

***

-0.111

0.063

-0.023

-0.038

0.063

-0.014

0.213

0.067

0.083

***

Financial literacy score

Sig

Age (Ref: over 64)

Female

***

Race/ethnic (Ref: White)

Marital Status (Ref: Single)

Number of children

***

Educational Attainment (Ref:<HS) High school Some college

-0.273

0.065

-0.055

***

-0.030

0.063

-0.011

0.288

0.067

0.112

***

College

-0.368

0.071

-0.071

***

-0.011

0.067

-0.004

0.369

0.071

0.144

***

Graduate

-0.465

0.082

-0.083

***

-0.001

0.072

0.000

0.363

0.075

0.142

***

From 15,000 to 24,999

-0.136

0.062

-0.027

-0.013

0.062

-0.005

0.139

0.064

0.054

From 25,000 to 34,999

-0.183

0.064

-0.036

***

0.088

0.062

0.033

0.108

0.065

0.042

From 35,000 to 49,999

-0.363

0.065

-0.066

***

0.094

0.060

0.035

0.276

0.062

0.108

***

From 50,000 to 74,999

-0.457

0.064

-0.083

***

0.085

0.059

0.031

0.342

0.060

0.134

***

From 75,000 to 99,999

-0.659

0.074

-0.107

***

0.139

0.163

0.052

0.457

0.065

0.180

***

From 100,000 to 149,999

-0.876

0.080

-0.129

***

-0.013

0.065

-0.005

0.639

0.066

0.251

***

Income>149,999

-0.920

0.091

-0.127

***

0.008

0.071

0.003

0.727

0.072

0.284

***

-0.442

0.062

-0.078

***

0.076

0.050

0.028

0.300

0.051

0.118

***

-0.275

0.210

-0.061

0.039

0.036

0.014

0.208

0.236

0.080

-0.334

0.038

***

-0.693

0.092

-1.137

0.097

Income (Ref: <$15,000)

Employment status (Ref: Unemployed) Self-employed Employed Intercept

***

*** p<.001; **p<.01; *p<.05 MFX=marginal effect.

tolerance, but is positively associated with moderate risk tolerance. The marginal effects of independent variables are shown in the columns for MFX. Respondents younger than 55 were significantly less risk tolerant than the reference group of individuals older than 64. Conversely, the

respondents younger than 64 were more likely to have high risk tolerance than individuals older than 64. The women were significantly more likely to have low or moderate risk tolerance than men, but were 18% less likely than men to have high risk tolerance. The black respondents

ÂŠ2017, IARFC. All rights of reproduction in any form reserved.

Volume 16, Issue 1

Table 4: Probit Estimates of the Role of Financial Literacy and Risk Tolerance on Goals-Based Savings Decisions

Risk tolerance (Ref: Low risk tol.) Moderate risk tolerance High risk tolerance Financial literacy score Female Education (Ref: <HS) High school Some college College Graduate Marital Status (Ref: Single) Married Widowed Divorced or separated Racial/ethnic (Ref: White) Black Hispanic Asian Other Number of children Income (Ref: <$15,000) From 15,000 to 24,999 From 25,000 to 34,999 From 35,000 to 49,999 From 50,000 to 74,999 From 75,000 to 99,999 From 100,000 to 149,999 Income> 149,999 Employment status (Ref: Unemployed) Self-employed Employed Age (Ref: Over 64) Age 18-24 Age 25-34 Age 35-44 Age 45-54 Age 55-64 Intercept

Have Emergency Funds Coef St. Error MFX Sig

Setting Aside Money for Childâ&#x20AC;&#x2122;s Education Coef St. Error MFX Sig

Have Planned for Retirement Coef St. Error MFX Sig

0.351 0.536 0.034 0.037

0.061 0.059 0.010 0.030

0.138 0.207 0.014 0.015

*** *** ***

0.362 0.563 0.001 0.022

0.139 0.450 0.016 0.047

0.129 0.222 0.001 0.009

0.304 0.514 0.101 0.028

0.056 0.058 0.010 0.030

0.131 0.198 0.040 0.011

0.243 0.249 0.474 0.461

0.072 0.072 0.075 0.080

0.097 0.099 0.188 0.182

*** *** *** ***

0.155 0.264 0.490 0.621

0.107 0.108 0.112 0.120

0.061 0.103 0.192 0.244

0.107 0.334 0.447 0.531

0.074 0.073 0.076 0.081

0.043 0.133 0.177 0.208

-0.016 -0.080 -0.195

0.040 0.109 0.054

-0.006 -0.031 -0.076

0.070 0.200 0.090

0.062 -0.032 -0.068

0.168 0.166 0.099

0.040 0.108 0.053

0.067 0.066 0.040

***

0.159 -0.083 -0.179

-0.134 -0.053 0.209 -0.168 -0.092

0.048 0.049 0.165 0.078 0.014

-0.053 -0.021 0.083 -0.066 -0.037

***

0.072 0.067 0.092 0.118 0.023

0.008 0.000 0.024 0.002 -0.013

-0.210 -0.032 0.078 0.019 -0.002

0.064 0.050 0.048 0.078 0.014

-0.083 -0.013 0.031 0.007 -0.001

***

** ***

0.020 0.000 0.062 0.005 -0.033

0.189 0.227 0.328 0.550 0.764 0.955 1.239

0.067 0.067 0.064 0.063 0.067 0.069 0.077

0.075 0.091 0.130 0.216 0.294 0.359 0.437

*** *** *** *** *** *** ***

0.232 0.393 0.563 0.757 0.872 1.059 1.438

0.222 0.117 0.114 0.112 0.117 0.120 0.129

0.092 0.155 0.222 0.295 0.336 0.401 0.506

*** *** *** *** *** ***

0.056 0.203 0.194 0.400 0.496 0.636 0.716

0.068 0.067 0.064 0.063 0.068 0.070 0.077

0.022 0.081 0.077 0.158 0.194 0.246 0.272

*** *** *** *** *** ***

0.276 0.154

0.052 0.037

0.110 0.061

*** ***

0.294 0.130

0.081 0.058

0.116 0.050

*** ***

0.263 0.250

0.052 0.037

0.106 0.099

*** ***

-0.411 -0.537 -0.609 0.631 0.390 -1.025 N=11,720

0.081 0.071 0.168 0.165 0.165 0.106

-0.156 -0.203 -0.229 0.238 0.150

*** ***

0.894 0.674 0.171 0.170 0.177 0.224

0.398 0.277 0.234 0.189 0.076

-0.558 0.082 -0.325 0.070 *** 0.355 0.267 *** 0.241 0.233 0.097 0.064 *** -1.619 0.106 N=11,720

-0.213 -0.128 0.139 0.095 0.038

*** ***

1.072 0.709 0.598 0.482 0.193 *** -2.334 N=4,805

** *** *** ** **

*** *** ***

***

*** p<.001; **p<.01; *p<.05 MFX=Marginal effect

were less likely than the white respondents to have moderate risk tolerance, but the black and Hispanic respondents were more likely than the white respondents to have high risk tolerance. Compared to the single respondents, the married respondents were more likely to be moderate

risk tolerance, and conversely were 5% less likely to have high risk tolerance. Number of children reduced the likelihood of having low risk tolerance, and increased the likelihood of having high risk tolerance. When compared with respondents who did not complete high school,

Journal of Personal Finance

respondents who had completed high school or higher were less likely to have low risk tolerance, but were more likely to have high risk tolerance. Also, as observed in Table 3, the probability of this association between risk tolerance and education increased with the level of educational attainment. Compared to the reference group of respondents with income less than $15,000, respondents with income greater than $15,000 were less likely to have low risk tolerance, and more likely to have high risk tolerance. Similarly, being self-employed decreased the probability of having low risk tolerance by 8%, and increased the probability of having high risk tolerance by 12%. Determinants of the Role of Risk Tolerance and Financial Literacy on Goals-based Savings The results from Table 4 suggest that higher levels of risk tolerance and financial literacy are positively and significantly associated with having emergency savings and planning for retirement. In our analyses, every unit increase in the financial literacy score resulted in a 1.4% increase in the probability of having an emergency fund, and a 4% increase in the probability of having planned for retirement. Additionally, moderate risk tolerance is also associated with saving for children’s education. There is a significant association between educational attainment and goals-based savings behavior. Respondents with educational attainment high school or higher were more likely to have an emergency fund, and those with an educational attainment of some college or higher were more likely to plan for retirement and, if they had children, were more likely to save for children’s education. The probability of goals-based savings behavior increased with educational attainment. When compared with respondents with educational attainment of less than high school, the respondents with educational attainment of college were 18% more likely to save for an emergency and plan for retirement, and were 19% more likely to save for their children’s education. The probability of having an emergency fund (marginal effects= 18%), saving for children’s education (marginal effects= 24%) and planning for retirement (marginal effects= 21%) were even higher for those who had attained graduate level education. Compared to the single respondents, the married respondents were more likely to save for college (marginal effects= 6%), and plan for their retirement (marginal effects= 7%). Conversely, the divorced or separated respondents were less likely to have

an emergency fund, save for college, or plan for their retirement. Compared to the reference group of white respondents, the black respondents were less likely to have an emergency fund or have planned for their retirement. The other races variable was also negatively associated with having an emergency fund. The likelihood of having an emergency fund also decreased with number of children. Compared to respondents with income less than $15,000, those with an income of $15,000 or more were significantly more likely to save for an emergency, and additionally, the respondents with an income of $25,000 or more were significantly more likely to have planned for retirement, and have started saving for children’s college education. The likelihood of having an emergency fund, saving for children’s college education, and having planned for retirement increased with being self-employed or employed. When compared with individuals older than 64, individuals younger than 35 were less likely to have an emergency fund or have planned for their retirement. But individuals between 35−54 years of age were more likely to have saved for their children’s education. Post-estimation Determination of the Interaction of Risk Tolerance and Financial Literacy in Goals-based Savings Behavior Table 5 shows the interaction effect of having “moderate or high” risk tolerance and financial literacy on having each of three saving behaviors, after controlling for the demographic variables, income, and employment status variables. In Table 5, the variable “risk tolerance” indicates the respondent had high or moderate risk tolerance, with the reference category being low risk tolerance. The likelihood of having an emergency fund and planning for retirement was higher for those with moderate or high risk tolerance than for those with low risk tolerance, and the effect of having moderate or high risk tolerance on saving for an emergency fund and planning for retirement increased with financial literacy. Similarly, the interaction of moderate to high risk tolerance and financial literacy was also positively associated with planning for retirement. Figure 2 shows the interaction of the effects of risk tolerance and financial literacy on the likelihood of having emergency funds; the patterns for the interaction effects on the likelihood of having planned for retirement were somewhat similar.

Volume 16, Issue 1

Table 5: Probit of the Interaction of Financial Literacy, and Risk Tolerance on Goals-Based Savings Decisions Setting Aside Money for Childâ&#x20AC;&#x2122;s Education

Have Emergency Funds Coefficients

Standard Error

Moderate or high risk tolerance

0.378

Financial literacy Moderate or high risk tolerance* financial literacy Intercept

Have Planned for Retirement

Sig

Coefficients

Standard Error

Sig

Coefficients

Standard Error

Sig

0.039

***

0.465

0.059

***

0.518

0.033

***

0.202

0.059

***

0.329

0.081

***

0.599

0.051

***

0.167

0.069

0.173

0.094

0.101

0.047

-0.586

0.027

***

-0.724

0.044

-0.531

0.027

***

*** p<.001; **p<.01; *p<.05 The other independent variables used in Table 4 were also controlled.

Figure 2: Likelihood Have Emergency Funds (based on Table 5)

Conclusion The results from this study reveal several insightful nuances in the relationship among financial literacy, risk tolerance, and their association with the goals-based financial decisions of households. The findings confirm the hypothesis that financial literacy is associated with individual risk tolerance. Although financial literacy is negatively associated with being risk averse, it is positively associated with being moderately risk tolerant. Confirming the second

hypothesis, both financial literacy and risk tolerance are significant and positively associated with having emergency funds, and retirement planning behavior among households. Additionally, moderate risk tolerance is also associated with saving for childrenâ&#x20AC;&#x2122;s college education. We also find in our additional analyses that risk tolerance when interacted with financial literacy is associated with having emergency savings and retirement planning behavior among households. The findings of this study underscore

Journal of Personal Finance

the importance of both risk tolerance and financial literacy in the goals-based savings behavior of households. Although risk tolerance has been associated with greater participation in financial markets and wealth accumulation across time (Finke & Huston, 2003), very high levels of risk tolerance has been associated in previous studies with overconfidence and irrational household investment behavior (Barber & Odean, 2001). The findings from this study indicate that financial literacy plays an important role in helping individuals with higher risk tolerance make normative financial decisions. One limitation of this study has been that the risk tolerance scale used in our empirical analysis is based on the respondents’ answers to questions that implicitly correspond to their risk capacity and expectations, whereas the economic concept of risk aversion is independent of those factors. Therefore, although this scale is a proxy for perceived risk tolerance of the respondents, it does not correspond exactly to the economic concept of risk aversion. Further research is needed to fully understand the interaction between financial literacy and risk tolerance. Among other findings, the fact that less than half of the parents had saved for their children’s education and only about a half of the households had calculated their expected retirement needs was alarming. It was also interesting that 60% of the households did not have any emergency funds. The silver lining in this story is that the likelihood of all three of these goals-based savings behaviors increased with moderate risk tolerance. That the likelihood of having emergency funds decreased with the number of dependent children in the household is also concerning and needs greater research, as families with more dependent children have a greater need for emergency savings. Overall, the findings on the association among risk tolerance, financial literacy, and other factors corroborate the findings from previous studies (Riley & Chow, 1992; Gilliam, Chatterjee, & Zhu, 2010). The findings from this study also suggest to policy makers that individuals’ emergency savings and other goals-based long-term savings behavior are positively associated with financial literacy and risk tolerance. Furthermore, financial literacy is also positively associated with an increase in the perception of financial risk tolerance among households. These findings indicate that financial literacy plays a role in enhancing the financial capability of households, thus providing further validation for the work of the Financial Literacy and Education Commission in developing and implementing a national

strategy (Knoll & Houts, 2012) to improve the overall financial understanding and financial capability of households across the country.

References Barber, B. M., and T. Odean. 2001. “Boys Will Be Boys: Gender, Overconfidence, and Common Stock Investment.” The Quarterly Journal of Economics, 116(1), 261−292. Callan, V. J., and M. Johnson. 2002. “Some Guidelines for Financial Planners in Measuring and Advising Clients about Their Levels of Risk Tolerance.” Journal of Personal Finance, 1(1), 31−44. Chatterjee, S., and V. Zahirovic-Herbert. 2010. “Retirement Planning of Younger Baby-Boomers: Who Wants Financial Advice.” Financial Decisions, 22(2), 1−12. Croy, G., Gerrans, P. and C. Speelman. 2010. “The Role and Relevance of Domain Knowledge, Perceptions of Planning Importance, and Risk Tolerance in Predicting Savings Intentions.” Journal of Economic Psychology 31(6), 860−871. Faff, R., Mulino, D., and D. Chai. 2008. “On the Linkage Between Financial Risk Tolerance and Risk Aversion.” Journal of Financial Research, 31(1), 1−23. Finke, M. S., and S.J. Huston 2003. “The Brighter Side of Financial Risk: Financial Risk Tolerance and Wealth.” Journal of Family and Economic Issues, 24(3), 233−256. FINRA Manual: NASD Conduct R. 2320. 2011, available at http://finra.complinet.com/en/display/display_main. html?rbid=2403&element_id=3643 Gilliam, J., Chatterjee, S., and D. Zhu. 2010. “Determinants of Risk Tolerance in the Baby Boomer Cohort.” Journal of Business & Economics Research, 8(5). Grable, J. E. 2008. “Risk Tolerance.” In Handbook of Consumer Finance Research (pp. 3−19). Springer New York. Hallahan, T.A., Faff, R.W., and M.D. McKenzie. 2004. “An Empirical Investigation of Personal Financial Risk Tolerance.” Financial Services Review, 13, 57−78. Hanna, S. D., and S. Lindamood. 2004. “An Improved Measure of Risk Aversion.” Journal of Financial Counseling and Planning, 15(2), 27−45.

Volume 16, Issue 1

Hanna, S. D., and W. Waller 2008. “The Concept of Risk Tolerance in Personal Financial Planning.” Journal of Personal Finance, 7(1), 96−108. Huzdik, K., Beres, D. and E. Nemeth. 2014. “An Empirical Study of Financial Literacy Versus Risk Tolerance among Higher Education Students.” Public Finance Quarterly, 4, 444−456. Jacobs-Lawsona, J. M., and D.A. Hershey. 2005. “Influence of future Time Perspective, Financial Knowledge, and Financial Risk Tolerance on Retirement Saving Behaviors.” Financial Services Review, 14, 331−344. Kahneman, D., and A. Tversky. 1979. “Prospect Theory: An Analysis of Decision Under Risk.” Econometrica: Journal of the Econometric Society, 263−291. Kimball, M. S., C.R. Sahm, and M.D. Shapiro. 2008. “Imputing Risk Tolerance from Survey Responses.” Journal of the American Statistical Association, 103(483), 1028−1038. Knoll, M. A., and C.R. Houts. 2012. “The Financial Knowledge Scale: An Application of Item Response Theory to the Assessment of Financial Literacy.” Journal of Consumer Affairs, 46(3), 381−410. Lusardi, A., and O.S. Mitchell. 2007. “Baby Boomer Retirement Security: The Roles of Planning, Financial Literacy, and Housing Wealth.” Journal of Monetary Economics, 54(1), 205−224. Lusardi, A., O.S. Mitchell, and V. Curto. 2010. “Financial Literacy Among the Young.” Journal of Consumer Affairs, 44(2), 358−380. Nicolini, G., B.J. Cude, and S. Chatterjee. 2013. “Financial Literacy: A Comparative Study Across Four Countries.” International Journal of Consumer Studies, 37(6), 689−705. Riley Jr, W. B., and K.V. Chow. 1992. “Asset Allocation and Individual Risk Aversion.” Financial Analysts Journal, 48(6), 32−37.

77 Roszkowski, M. J., and J.E. Grable. 2005. “Estimating Risk Tolerance: The Degree of Accuracy and the Paramorphic Representations of the Estimate.” Journal of Financial Counseling and Planning, 16(2). Sages, R. A., and J.E. Grable. 2010. “Financial Numeracy, Net Worth, and Financial Management Skills: Client Characteristics that Differ Based on Financial Risk Tolerance.” Journal of Financial Service Professionals, 57−65. Statman, M. (1999). “Behaviorial Finance: Past Battles and Future Engagements.” Financial Analysts Journal, 55(6), 18−27. Sung, J., and S. Hanna. 1996. “Factors Related to Risk Tolerance.” Financial Counseling and Planning, 7(1), 11−20. Thaler, R. H., and S. Benartzi. 2004. “Save More Tomorrow™: Using Behavioral Economics to Increase Employee Saving.” Journal of Political Economy, 112(S1), S164−S187. Van Rooij, M. C., A. Lusardi, and R.J. Alessie. 2012. “Financial Literacy, Retirement Planning and Household Wealth.” The Economic Journal, 122(560), 449−478. Wooldridge, J. (2012). Introductory Econometrics: A Modern Approach. Boston: Cengage Learning. Yao, R., Hanna, S. D., and S. Lindamood. 2004. “Changes in Financial Risk Tolerance, 1983−2001.” Financial Services Review, 13(4), 249−266. Yao, R., Gutter, M. S., and S.D. Hanna. 2005. “The Financial Risk Tolerance of Blacks, Hispanics and Whites.” Financial Counseling and Planning, 16(1), 51−62. Yao, R., and S.D. Hanna. 2005. “The Effect of Gender and Marital Status on Financial Risk Tolerance.” Journal of Personal Finance, 4(1), 66.

Journal of Personal Finance

Preventing Financial Elder Abuse

Kenn Beam Tacchino, JD, LLM, RICP, Professor of Taxation and Financial Planning and Boettner Endowed Professor in Financial Planning, Widener University, Chester PA

Abstract Clients may not only need a financial planner’s help and advice when it comes to their finances—they may also need their protection as well. Because billions of dollars are financially exploited each year, planners and financial service institutions are increasingly being called upon to walk the tightrope between a client’s autonomy and a client’s need to be safeguarded. In order for the financial planner to better serve his clients we first look at the financial elder abuse problem and share examples, which highlight the scope of the issue. We then point out some red flags that will alert financial planners to a potential problem. We examine the legal requirements and protections that are germane to understanding the financial planner’s responsibility in recognizing and preventing financial elder abuse. We review actions the financial planner can take to cope with or prevent financial elder abuse. And finally, we discuss a systematic response that an organization should take to create a business culture that seriously addresses financial elder abuse.

Key Words Financial elder abuse, financial planning, FINRA, mandatory reporter, trusted third party, pausing distributions

Volume 16, Issue 1

Introduction There is no doubt that the morally repugnant crime of financial elder abuse is currently underrecognized, underreported, and underprosecuted. The financial services world has an important role to play in changing this disturbing matter. However, many financial planners may be justifiably confused regarding their role in the process. It is difficult to be vigilant about something that lurks in the shadows and is not easily identified. It is not always clear whether relationships a client has are supportive and wellmeaning or are controlling and criminal. Suspicious actions may evoke concern; but to what extent are financial planners to undermine the autonomy of the client? And to what extent must they protect the client’s right to privacy? What’s worse, the law pertaining to these matters is rapidly evolving on both the state and federal level and this makes it hard for the planner to know what standards to apply. Nevertheless, financial planners owe a duty to clients to take a hard look at what is happening and act appropriately when the signs of financial elder abuse are present. The societal importance of this topic has led to recent federal and state protections for senior clients as well as some useful resources for banks and credit unions to play a role in stopping financial elder abuse. Much of the regulatory focal point has been on what the financial service institution can and should do. However, the target of this paper is not on the role financial institutions can play, but it is on the role the financial planner can play. We will focus on the small insurance agency or mom and pop investment advisory firm and the financial planners who work for them. We will raise awareness of the financial elder abuse problem so that planners can better recognize the problem. We will provide insights concerning the role a financial planner can play in recognizing abuse and preventing abuse by describing the major laws pertaining to financial elder abuse and suggesting action steps the planner can take. Finally, we will present best practices the planner’s firm can use to create an environment that responds to the financial abuse threat.

from the Consumer Financial Protection Bureau (CFPB) (issued in March 2016, and 61 pages long). We have adapted many great ideas from this publication and scaled them to meet the needs of “kitchen table” financial and insurance planning firms and agencies. We consider the CFPB monograph to be the gold standard on the issue of financial elder abuse and we consider this article as a logical follow-up for a different audience.

What Is Financial Elder Abuse? The National Center on Elder Abuse defines financial elder abuse as “the illegal or improper use of an elder’s funds, property, or assets.”1 Alternatively, Stopfraud.gov defines elder fraud as “an act targeting older adults in which attempts are made to deceive them with promises of goods, services, or financial benefits that do not exist, were never intended to be provided, or were misrepresented.”2 Other key regulatory players also define the term differently as do the various state governments and this can complicate the financial planner’s role in helping to prevent financial elder abuse. These varied definitions emanate from the focus of the organizations involved. Some are primarily concerned with acts of fraud and deception, some with the sub-topic of identity-theft, while others look to crimes which emphasize manipulation of personal relationships and prey on the dementia which may impact older clients. There is common ground, however. In all cases, each organization recognizes that an older client’s larger net worth and mental and physical vulnerability can cause predators such as salespeople, telephone solicitors, lawyers, financial planners, caregivers, home repair contractors, con artists, people with power of attorney or those who are guardians of property, and perhaps most shockingly, family members, to prey on the elderly.3 Regardless of how it is defined the impact of financial elder abuse is substantial. Several studies estimate that 20% of elderly people have fallen victim to financial exploitation.4 Even worse, 1.

For an outstanding review of what home office, larger firms, and banks and credit unions can do to combat financial elder abuse we strongly recommend reading “Recommendations and report for financial institutions on preventing and responding to elder financial exploitation”

2. 3.

See Http://www.ncea.aoa.gov /FAQ/Type_Abuse/index. aspx#financial Http://www.stopfraud.gov/protect-yourself.html. According to the National Committee for the Prevention of Elder Abuse, persons over the age of 50 control 70 percent of the nation’s wealth, making them tempting targets. See “Prevalence and Correlates of Emotional, Physical, Sexual, and

Journal of Personal Finance

experts believe that for every known case of financial abuse four to five cases go unreported.5 A MetLife study analyzed that known financial abuse costs $2.9 billion annually; other studies have put it as high as $36 billion.6 The same MetLife Study found that women were nearly two times more likely to be victims of financial elder abuse than men; the most common age range for victims is 80−89; and not surprisingly, the victims are often socially isolated.7 Anothercommon theme is the presence of cognitive impairment, physical decline, or bereavement. However, it is not only vulnerable seniors who are victimized. Recent research has found that some typical victims of financial elder abuse are self-reliant, financially savvy, and college educated.8 In any case your clients not only are faced with the financial ramifications of financial elder abuse, but they are also subject to the substantial emotional scars that accompany the crime. After being exploited, a client’s physical or emotional health may be impacted; they may lose their independence and even die younger.9

7. 8. 9.

Financial Abuse and Potential Neglect in the United States: The National Elder Mistreatment Study,” American Journal of Public Health pp. 292−297, 2010. See also Crescente, Fernanda, “Senior Citizens Lose Billions, Fear to Report Financial Exploitation,” USA Today, June 15, 2016. According to Sandra Timmermann, “Protecting the Most Vulnerable from Financial Abuse: What Should we Know?”, Journal of Financial Service Professionals, May 2009, pages 23 and 24, elder abuse goes unreported because 1) the elder person may feel responsible for what happened, 2) the elder person may feel they will be placed in a nursing home if they speak out, 3) the elder person may not want their family member to go to jail, and 4) the elder person may feel the issue can be resolved in the family. The MetLife Study of Elder Financial Abuse: Crimes of Occasion, Desperation, and Predation Against Elders (2011) page 2. A Consumer Finance Protection Bureau study estimates annual losses at up to $36.48 billion. See Kieler, Ashlee, “Banks are the Key to Stopping Scammers That Target One in Five Older Americans,” consumerist. com, March 23, 2016. The MetLife Study of Elder Financial Abuse: Crimes of Occasion, Desperation, and Predation Against Elders (2011) page 3. McNeil, Donald “Are You Potential Fraud Bait,” MarketWatch, January 15, 2016. “Recommendation and report for financial institutions on preventing and responding to elder financial exploitation,” Consumer Finance Protection Bureau (CFPB), March 2016, page 10

Examples of Financial Elder Abuse Perhaps the best way for a financial planner to recognize financial elder abuse is to look at some examples of how it has manifested.10 Reviewing these examples may be the best way to understand the unsavory world of financial elder abuse and types of activities that financial planners should help to prevent. These acts not only hoodwinked the senior out of money—they also robbed them in many cases of their dignity and quality of life. •

A 29-year-old pled guilty to stealing more than $100,000 from vulnerable elders with whom she had romantic relations. She even married one of her victims in order to steal from him.

•

A 76-year-old son was charged with grand larceny. He got access to the parent’s financial information when his 98-year-old mother broke her hip and committed forgeries and cashed out and redeemed several CDs.

•

Two water purification salesmen were arrested for taking more than $37,000 from an 88-year-old women for filtration equipment valued at no more than a few hundred dollars.

•

So-called “grandparent schemes” exist in which the senior is persuaded to wire money to bail out “grandchildren” or pay their expenses. The recipient of the funds was not a grandchild.

•

When an older woman was under the weather she gave a “nice neighbor” her ATM card to shop for her. The neighbor used the card for her own purposes.

•

A caregiver stole $200,000 from a man who had hired her to take care of his wife who had dementia.

•

A pastor persuaded a man with Parkinson’s disease to allow him to manage his finances and over 130 withdrawals later he had stolen $25,000.

•

An elderly gentleman received a phone call from a woman identifying herself as from the bank. She explained the bank was having a computer problem and persuaded the man to reveal all types of personal information including his Social

10.

Some examples adapted from The MetLife Study of Elder Financial Abuse: Crimes of Occasion, Desperation, and Predation Against Elders (2011) pages 5−24.

Volume 16, Issue 1

Table 1: Examples of Forms of Elder Financial Exploitation, by Type of Perpetrator Perpetrators Strangers

Forms • Lottery, mail, telephone or internet scams in which the person creates trust and then a sense of urgency • Door-to-door home repair scams • Identity Theft Financial service providers • Sale of fraudulent investments (Ponzi or pyramid schemes) – Unregistered products being sold – Unlicensed representatives holding themselves out as experts – Oil and gas and real estate scams • Sale of financial products or services unsuitable for an older adult’s circumstances • Theft of cash or other valuables Family members, friends, homecare providers, and legal • Withdrawals from bank accounts or use of credit cards • Transfer of deeds guardians • Misuse of an older adult’s power-of-attorney • Misappropriation of an incapacitated older adult’s income or assets – Inability to say no to relatives, even if “abuse” is not illegal • Identity theft Source: GAO analysis of published research with additions. (GAO-13-6265)

Security number. His accounts were emptied within a few days. •

•

Two elderly women were beaten to death with a crowbar by the handyman who took and pawned off all their valuables. New York socialite Anthony Marshall conspired with lawyer Francis Morrissey to amend his mother’s will. The mother is Brooke Astor. The son took millions without the mother’s consent and also took artwork from the mother’s home.

Even when the abuse is prosecuted, the money is often impossible to recover because it is either spent or overseas. Table 1 summarizes some other examples. It is taken in part from testimony before Congress on the issue.11

Red Flags Financial planners should be on the lookout for indications that financial elder abuse is affecting their clients. When warning signs are detected the planner may be in a good position to address the situation and help to prevent financial elder abuse before it happens. Financial elder abuse falls into six distinct 11.

Testimony Before the Subcommittee on Commerce, Manufacturing, and Trade, Committee on Energy and Commerce, House of Representatives, ELDER JUSTICE, Statement of Kay Brown, Director Education, Workforce, and Income Security, May 16, 2013.

categories—transactions a client may make, identity theft, coercion, life style changes a client is undergoing, changes in the client’s personal relationships, and dementia-related events. Transaction indicators focus on actions that a client may take and they include things like expensive gifts being made to a caregiver, unusually large withdrawals from accounts, changes in banking patterns, checks written out to cash, new spending patterns following the addition of a new authorized user, daily maximum currency withdrawals from an ATM, uncharacteristic attempts to wire large sums of money, inquiries about international wire transfers, significant changes in spending patterns, and sudden and abrupt changes regarding financial management. It seems apparent that many of these activities may first be detected by a bank or credit union. However, financial planners should be the first to notice abrupt changes regarding financial management. This would include among other things unusual spending, allowing a policy to lapse, and changing the beneficiary of a policy. While these changes may be benign, it is important for the planner to take a deeper look. Financial planners need to ask the important questions: Can you tell me why you need to take a large withdrawal so we can look at other options that are more tax sensitive? Is it really a good idea to drop insurance coverage? What will you be doing with the money that was formerly used for premiums? Can we take a closer look at all your beneficiary designations since changing one beneficiary

Journal of Personal Finance

may have a domino effect on your entire estate plan? If the answers seem suspicious, preventative action (discussed below) may be required. Identity theft indicators include mentions of lottery or sweepstakes winnings, a new third party speaking for the older adult, financial statements or documents that no longer come to the client’s home, and requests to send account statements to a third party’s address. Financial planners should let their clients know that if a financial transaction seems too good to be true it probably is! Coercion may be suspected if the client gives power of attorney to an unlikely person, or retitles property to include an unlikely person.12 Coercion is often the result of undue influence. From the financial planner’s perspective, a new third party speaking for the client and a change in power of attorney and property retitling should raise concern. A deeper inquiry may be necessary to ascertain the legitimacy of these sudden changes. Lifestyle clues for which the planner should be diligently watching include missed appointments, uncharacteristic nonpayment of services, withdrawn behavior or disheveled appearance, undue anxiety about finances, or lack of knowledge about financial status. Obviously the financial planner is not in a position to judge mental competency or depression; however, the planner needs to make sure that dealings are with a client who is in control and full possession of their mental properties. Action must be taken (see below) if you suspect the client’s behavior is abnormal. Personal relationship sore points include social isolation, an unusual degree of fear or submissiveness to a caregiver, missing belongings, family presence of alcohol, substance abuse, and/or gambling, and signs of intimidation and threats by another. Be particularly vigilant if your client has a new “best friend” and their financial behaviors suddenly change. Since clients are most vulnerable when dementia is developing, it is always a good idea when the financial planner meets with the client to make note of whether the client was confused about instructions, has difficulty in following 12.

It may be prudent to suggest that a client have specific powers of attorney rather than general powers of attorney. With the former the agent is limited to what he or she can do. With the latter the agent is given broad powers from which elder-financial-abuse may be possible.

directions, forgets past decisions, or has trouble handling paperwork. We cannot overstress the import link between financial abuse and mental incapacity. Extra caution is warranted if the client seems mentally vulnerable. Also it is a good idea to involve trusted family members to make sure mental capacity is still present and to videotape the execution of changes to key documents. In addition, planners should discuss with clients financial incapacity and financial elder abuse well in advance of cognitive changes when possible. They should catalogue the name of attorneys and trusted parties and obtain permission to contact them when warranted. However, financial planners must be aware of the ethical tight rope they must walk when a family member is involved in a diminished capacity situation: •

The planner is required to advance the interests of the client over and above the interests of others; however, multiple representation assumes that the interests of all family members are aligned and remain aligned. (This is not always the case!)

•

The planner is required to spell out the conditions of engagement. He or she must indicate there may be areas from which joint representation must be withdrawn and indicate there may be areas in which confidentiality must be maintained.

Legal Requirements and Protections A financial planner’s state law will play a big part in determining the specific legal requirements they face and the actions that they can take without infringing on the client’s privacy or breaching duty to the client.13 For example, a financial planner might be considered a “mandatory reporter” but exactly what that means will depend on where they live. In some situations, the failure to report suspicions of financial elder abuse under state law may lead to administrative or civil action against the financial planner, and in some instances criminal penalties will apply! Reports should go to the local office of Adult Protective Services (APS) who have in place a system for investigating elder abuse. Adult Protective Services will

13.

For a discussion of some state’s laws see Petrasic, Sachs, and Kowalski, “Elder Financial Exploitation: An Increasing Compliance Concern,” Journal of Taxation and Regulation of Financial Institutions, 28J.Tax’n F. Inst. 31, 2015

Volume 16, Issue 1

make an initial determination whether action is required and intervene to counteract financial elder abuse when necessary. Since these units differ from state to state, the exact nature of their organization and operating procedures varies. A discussion of this is beyond the scope of this paper. Federal law also comes into play. Financial institutions must report actual or suspected financial abuse by filing a suspicious activity report (SAR). These can be filed with the Financial Crimes Enforcement Network (FinCEN). Elder financial abuse is also reported at the local level with district attorney’s offices and police departments. What follows is a review of the financial planner’s responsibilities and required action under a series of pertinent regulations and laws.14

Gramm-Leach-Bliley Act (GLBA) Astute planners will recall that privacy issues for clients are protected under federal law under something called the opt-out rule. The opt-out process is what institutions must do to protect client privacy. Essentially under the Gramm-Leach-Bliley Act (GLBA), the opt-out rule requires the financial institution to notify clients and give them an opportunity to opt-out before providing nonpublic personal information to a third party. However, regulators15 have indicated that financial institutions and financial planners can report financial elder abuse to authorities without concern for violating the opt-out process required by federal law.16 For example, 14.

15.

16.

We note that the Dodd Frank Act and the Elder Justice Act were the genesis for several of the regulatory initiatives discussed in the following section. However, we chose to omit a discussion of these laws from the list below in an effort to focus on the specific requirements and protections that are germane to financial planners and this discussion. We also note that The Elder Justice Act was passed as part of the Affordable Care Act. It is not fully appropriated, but some monies have been appropriated to expand the system and it has created more awareness of financial elder abuse. The Securities and Exchange Commission (SEC), Consumer Financial Protection Bureau (CFPB), Federal Trade Commission (FTC), Federal Deposit Insurance Commission (FDIC), the Board of Governors of the Federal Reserve System, the office of the Controller of the Currency, and the National Credit Union Administration. See Koco, Linda. “Regulators Weigh in on Elder Financial Abuse,” insurancenewsnet.com. September 30, 2013.

the financial planner can report incidents of taking an older adult’s fund without their consent or they can also report someone being asked to sign over assets through misrepresentation. They can spot and report irregular transactions, account activity, or behavior that signals financial abuse without fear of violating the privacy protections under the opt-out rule. Financial planners can also comply with regulatory investigations or respond to subpoenas. 17 Since the GLBA allows disclosure of personal information when a consumer consents, it would be a best practice for the financial planner to establish a procedure for offering clients the opportunity to consent to share information to third parties under specified circumstances.18 Financial planners should take comfort in the exceptions under the GLBA which allow planners and their institutions to: •

comply with federal or state laws (including mandatory reporting)

•

comply with a civil or criminal investigation

•

prevent fraud and unauthorized transactions

Common Law Under common law the doctrine of undue influence protects against overreaching by a wrongdoer seeking to take unfair advantage of a person who is susceptible to such wrongdoing on account of the donor’s age, physical or mental weaknesses, or other factors. Financial planners should also be aware that there is a tort of “intentional interference with expected inheritance.” Legal standards have also been established under which a testator cannot make a valid will unless lucid and free from the power of others. Financial planners should be aware, however, that diminished capacity does not always mean that the client lacks testamentary capacity.

17. 18.

See section 502 of the GLB Act (15 U.S.C 6802(e). “Recommendations and report for financial institutions on preventing and responding to elder financial exploitation.” Consumer Financial Protection Bureau, March 2016, page 45.

Journal of Personal Finance

FINRA Dispute Resolution and Unsuitability: Addressing Planner Abuse One of the perpetrators of financial elder abuse may sometimes be a financial planner. FINRA protections exist for clients if their financial planner has crossed the line. Almost all customer agreements with a brokerage firm contain a mandatory binding arbitration clause. Because of this, client disputes against planners are resolved through this forum by using the FINRA Dispute Resolution System (arbitration or mediation). There is even an extended statute of limitation (a six-year eligibility period). FINRA also has a phone hotline (1-844574-3577) on which clients can raise concerns and get assistance. Planners should also be aware of the insurance rules for unsuitability. However, a discussion of this topic goes beyond the scope of this article.

North American Securities Administrators Association (NASAA) Model Law In January 2016, a model law was adopted by the NASAA to protect vulnerable adults (generally anyone 65 or older) from financial exploitation.19 The requirements of the model law (which has been adopted by several states) are in stark contrast to the privacy requirements that financial planners have been accountable to for most of their careers. For one thing the model law mandates reporting to state securities regulators and to state adult protective services agencies when financial elder abuse is suspected. It also requires the sharing of financial records with state adult protective services (APS).20 If a financial planner has a reasonable belief that financial abuse is going on, he or she, like doctors and social workers who are mandatory reporters for things like physical and child abuse, must act under law to protect their client.21 In addition to making the planner a 19.

20. 21.

Specifically a person 65 years of age or older or a person subject to state adult services definitions. NASAA Model Legislation or Regulation to Protect Vulnerable Adults from Financial Exploitation, adopted 1/22/16. Remarks of the NASAA President Judith Shaw at the ARM 2016 Annual Educational Conference, March 8, 2016, nasaa.org. The model law refers specifically to investment advisor, agents, and broker–dealers as defined by state law. NASAA Model

mandatory reporter, the NASAA model law also allows the planner to set up a system under which his or her client designates a trusted third party for the planner to notify if the financial planner suspects financial elder abuse. The model law also allows financial planners to delay disbursements in order to prevent the financial abuse from being consummated. The triggering events under the model law include: •

wrongful or unauthorized taking, withholding, appropriation, or use of money, assets, or property

•

converting money, assets, or property in a way that deprives the client ownership, use, benefit, or possession of these resources

•

unseemly acts or omissions taken by persons holding the power of attorney, guardianship, or conservatorship rights22

The model law provides liability protection for financial planners for reporting what they believe in good faith to be financial elder abuse (to the state, adult protective services, and the designated third party). The planner is also given immunity for delaying financial distributions in order to prevent the financial elder abuse. Several states have already adopted the law (albeit with minor modifications).23

Financial Industry Regulatory Authority (FINRA) Proposed Rules FINRA has suggested similar federal rules to the NASAA state rules that would also empower financial planners to combat financial exploitation.24 FINRA is proposing that financial planners ask for the name and contact

22.

23.

24.

Legislation or Regulation to Protect Vulnerable Adults from Financial Exploitation, adopted 1/22/16. Unseemly includes obtaining control through deception, intimidation or undue influence. NASAA Model Legislation or Regulation to Protect Vulnerable Adults from Financial Exploitation, adopted 1/22/16. See Schoeff, Mark, Financial-Abuse Reporting Mandatory Starting Friday, Investment News, June 30, 2016. The article references Alabama, Indiana, Louisiana, and Vermont as adopting the model law and Delaware, Missouri, and Washington State as already adopting similar laws. Financial Exploitation of Senior and Other Vulnerable Adults, FINRA Regulatory Notice 15−37, October 2015.

Volume 16, Issue 1

information for a “trusted person” at the opening or updating of a client’s account. However, the financial planner will be allowed to open a new account without establishing a trusted person contact as long as they asked the client to provide the information. This FINRA rule will require that the financial planner disclose to the client that the firm is authorized to contact the trusted person and find out information about the client’s account to confirm the specifics of the clients current contact information, health status, and the identity of legal guardian, executor, trustee, and holder of power of attorney.25 It is FINRA’s hope that the trusted contact person will be able to become a resource for the financial planning firm to prevent financial exploitation. A second proposed change enables action when financial elder abuse is suspected. When there is a reasonable basis to believe that financial exploitation of a vulnerable adult (anyone 65 or older) is occurring, the financial planning firm can place a temporary hold on disbursements of funds or securities from a client’s account and the planner can notify the client’s “trusted person” contact (or if unavailable an immediate family member).26 Basically, the temporary hold would expire no later than 15 business days after the date that the firm first placed the hold on the account. This rule does not make it obligatory for financial planners to place holds if they suspect financial exploitation.27 When the planner decides to place the temporary hold, however, FINRA requires the firm to conduct an internal investigation of the financial exploitation and requires the firm to retain records related to the case.28 In the event that the financial planner suspects that the trusted contact person is the perpetrator in the financial exploitation, the disclosure would allow the firm to contact the next of kin.

25. 26. 27. 28.

Financial Exploitation of Senior and Other Vulnerable Adults, FINRA Regulatory Notice 15−37, October 2015, pp. 3. Financial Exploitation of Senior and Other Vulnerable Adults, FINRA Regulatory Notice 15−37, October 2015, pp. 4−5. Financial Exploitation of Senior and other Vulnerable Adults, FINRA Regulatory Notice 15−-37, October 2015, pp. 4−5. Financial Exploitation of Senior and Other Vulnerable Adults, FINRA Regulatory Notice 15−37, October 2015, pp. 4−5.

Electronic Funds Transfer Act (EFTA) and Regulation E Although this law needs to be understood better by the broader financial institution than it does by the financial planner, financial planners must nonetheless be aware of it. EFTA and Regulation E provide consumers with a variety of rights: •

The client has no liability for a timely reported unauthorized transfer and limited liability when the unauthorized transfer involves a lost or stolen access device.29

•

The consumer or her representative may notify a financial institution of an unauthorized transaction in person, by telephone, or in writing.30

Summary of Financial Planner Actions If the planner suspects elder abuse, her first action should be to contact the compliance department of the home office, in-house legal counsel, or another designated party from the organization to seek advice concerning her course of action. Planners should document their concerns and ask for documented instructions on how to proceed from the appropriate authority within the company. After receiving direction from the company, planners most likely will be asked to do one of three things: to contact a trusted third party and ask for their help; to contact adult protective services, an ombudsman, or local law enforcement; or to delay disbursement of funds in order to prevent the client from being subject to fraud. If the planner acts as a mandatory reporter, she can alert adult protective services by going to the National Adult Protective Services Association website (http://www. napsa-now.org/get-help/help-in-your-area/) and using the get help tab. Since adult protective services regulations vary from state to state, the planner will need to select the appropriate state from the map provided to get the contact information needed for their situation. Adult protective services will follow up and take action when warranted. Planners should take comfort in the

29. 30.

12CFRSections 1005.6 and 1005.11. Regulation E.

Journal of Personal Finance

fact that adult protective services will maintain their confidentiality. In other words, in the vast majority of cases, the client will not know that it was the planner who reported the situation. If the planner is dealing with a client in a nursing home, she should contact the National Long-Term Care Ombudsman Resource Center (http://theconsumervoice. org/get_help) and use the “get help” tab. Once again since these services vary from state to state, the planner will need to select the appropriate state from the map provided to get the contact information needed for her situation. Long-term care ombudsmen are advocates for residents of nursing homes, board and care homes and assisted living facilities. Ombudsmen provide information about how to find a facility and what to do to get quality care. Ombudsman are trained to resolve problems like financial elder abuse. If the planner sees a financial crime, he should assist his parent company in filing a suspicious activity report to the Financial Crimes Enforcement Network (discussed earlier). Planners should also do the following: •

Comply with regulatory investigations and/or respond to subpoenas.

•

Initiate the identification of “trusted third parties” with the client and bring them into the conversation when warranted.

•

Advise clients to seek legal advice regarding undue influence and intentional interference with expected inheritance.

•

Ensure the client has testamentary capacity to make a valid will.

•

Advise clients who have been wronged by third party planners to seek legal advice about the FINRA dispute resolution program and state insurance suitability laws.

•

Delay disbursement of funds when permissible to prevent financial elder abuse from being consummated.

•

Alert a client that he or she may have no liability for timely reported unauthorized transfers.

•

Network with (and build alliances with) community resources that serve the aging population and with other professionals who work with older people. This would include elder law attorneys and geriatric care managers.

Financial planners should also become familiar with the symptoms and red flags of dementia in order to protect vulnerable clients from financial elder abuse. The Alzheimer’s Association is a good resource for this (http:// www.alz.org/). Other helpful resources planners should consult regarding the financial elder abuse issue include: •

The National Association of Adult Protective Services (NAPSA) (http://www.napsa-now.org/)

•

The National Center on Elder Abuse (https://ncea. acl.gov/)

•

The National Committee for the Prevention of Elder Abuse (http://www.preventelderabuse.org/).

•

The Women’s Institute for a Secure Retirement http://www.wiserwomen.org/?id=661

As a general rule, planners who take action after receiving direction from their organization will not face any repercussions for acting in good faith and will have the full support and resources of their company to combat actions by an overly litigious predator, even if the planner’s suspicions were unfounded. In addition a proposed law called the Senior Safe Act extends a shield against liability for advisors and broker dealers who report incidents to adult protective services, law enforcement, and state and federal regulators. However, this legislation has not become a federal law at the time of this writing.

A Systematic Response for a Financial Planning Firm: Ten Best Practices to Consider A small financial planning firm or insurance agency should create a business culture of taking financial elder abuse seriously. From the firm’s perspective a strong business culture to prevent financial elder abuse requires planners to document everything! From the standpoint of protecting clients, the ten best practices discussed below should be a starting point. If small financial firms

Volume 16, Issue 1

haven’t already, they should consider taking the following steps to combat financial elder abuse: 1. Build elder financial abuse screening into the factfinding process. Financial planners must have the ability to recognize the relevant facts in the situation in order to recognize and prevent elder financial abuse. They need to collect a broader spectrum of information that addresses the material raised in this paper, construct a narrative placing emphasis on the previously discussed red flags, and understand what the situation is and how it occurred. 2. Establish procedures so clients can provide advanced consent to sharing account information with a designated and trusted third party (when the planner suspects that the client is at risk for financial elder abuse.31 One way to get the trusted third party involved is to consider adding language to the investment policy statement (IPS) that calls for a trusted and named third party to be contacted if the financial planner deems circumstances warrant it (be sure to include contingent third parties in case the original predeceases the client). The checks and balances of having children and other financial professionals involved could be the primary way in which a financial planner can prevent financial elder abuse before it happens. 3. Have a process for making sure proper legal documents are executed. Financial planners need to discuss durable power of attorney, powers of attorney for healthcare, and revocable living trusts with their clients. Use of these documents is crucial. More importantly, misuse of these documents can directly result in financial elder abuse. Financial planners should be extra vigilant and make sure the appropriate parties are named in these documents. 4. Offer opt-in account features such as cash withdrawal limits, geographic transaction limits, alerts for specified account activity or changes in beneficiary designations. Also, it may be prudent to assign a cotrustee or to require double signatures for certain

31.

Recommendations and report for financial institutions on preventing and responding to elder financial exploitation” Consumer Financial Protection Bureau, March 2016, page 44.

dollar amounts.32 Putting reasonable restrictions on what the client can do may not catch all shady activities but it may help to limit the damage. 5. Educate the client about common scams. The financial planning firms website may want to contain alerts about financial scams that are in the news, or relevant to the situation (e.g., scams that occur at Christmas time). Give clients a copy of the “Financial Self-Defense Guide for Seniors” (a consumer guide from the Certified Financial Planner Board of Standards). Let clients know about the AARP foundation’s fraud hotline. Consumer tips to pass along include the following: –

–

Shred personal information.

–

Be careful on social media and do not post anything that could reveal your location or absences from home.

–

Only keep required information in your purse or wallet.

–

Never conduct financial transactions on a site that only has an “http” in the URL; it should have an “https” and a yellow lock icon.

–

Check bank and credit card statements for suspicious activity.

–

Get second opinions.

–

Use the FINRA “broker check.”

–

Ask for written verification about a charity.

6. Have the compliance department review the local and federal requirements as well as best practices. Communications from the compliance department can include workshops, on-line tests, and lunch-and-learns. Financial planners should be well informed about their obligations as a so-called “mandatory reporter.” They should know under what circumstances that reporting is required, to whom to report, and their role in any follow-up activity and/or their role with the client after a report is made. Planners also should be well aware 32.

“Elder Fraud: Nine Tips to Protect Clients,” Financial Planning, June 30, 2015., www.financialplanning .com

Journal of Personal Finance

of the circumstances under which they can put a temporary hold on a disbursement of funds. 7. Connect with local services and hire someone as a consultant to put together materials for the firm. Consider contacting adult protective services, state’s attorneys’ general (the fraud and consumer protection units), long-term care ombudsman, and attorneys who can talk about guardianship and civil commitment. 8. Hire, train, and bond daily money managers who can work with a client to pay bills and keep track of financial records. Financial planners may not have the time or ability to “hand hold” their elderly client’s everyday financial affairs; however, the profession of daily money managers can help the client with the little things.33 Financial planners should let clients know that they have trusted and trained colleagues who can pay bills, write checks, keep tax records, and help manage the daily budget. 9. Report all cases of suspected exploitation to relevant authorities. For example, Adult Protective Services will assess whether the client is eligible for services and whether the reported information meets the legal definition of abuse, neglect, or exploitation.

assure their client that embarrassment should not get in the way of justice. And remind them that without action on their part there may be other victims. When it is a family member committing the abuse, the client may feel they will be placed in a nursing home (or otherwise give up independent living) if they speak out. The victim may not want a family member to go to jail. They feel issues can be resolved in the family. If this is the case, the financial planner can make the client aware of his/her options and act as a sounding board as the client weighs difficult decisions. Finally, professional ethics require financial planners to be vigilant about and to intervene in instances where elder financial abuse is occurring. The financial planner needs to become front-line fighters against the plague of financial elder abuse that is taking place in our society. Their clients not only need their help and advice when it comes to their finances, they may also need their protection as well.

10. When the client resides in a caregiving facility alert the long-term care ombudsman to the financial elder abuse. The ombudsman can protect the client in ways that the financial planner cannot.34

Conclusion In many instances financial planners are in a position to notice elder abuse more than other professionals such as lawyers and doctors because they have a long-standing relationship with the client and they understand family dynamics. They can help to prevent the so-called crime of the 21st century. Financial planners must understand that financial elder abuse may go unreported unless they offer to assist the client. In some cases, the elderly client feels responsible for what happened and they are too embarrassed to make a complaint. Financial planners should 33.

34.

For more information about Daily Money Managers see the website for the American Association of Daily Money Managers (http://www. aadmm.com/). For more information about Ombudsman see the website 1tcombudsman.org.

Ethics Approved

IARFC Code of Ethics Brand Your Ethics Approved Status Set yourself apart from other consultants • Affix your Ethics Approved Seal to your framed RFC® or RFA® Certificate • Send an IARFC Ethics Approved media release to your contacts • Order business cards with the Ethics Approved Seal • Place digitized Seal on your website in a prominent position • Mention this program in client newsletters • Order additional Ethics Approved Seals as a visual reminder • Display the IARFC Code of Ethics plaque in office • Put a link to the IARFC Code of Ethics on your website Visit the IARFC store for these valuable branding tools www.IARFC.org/Store or contact 800.532.9060, info@iarfc.org

IARFC INTERNATIONAL ASSOCIATION OF REGISTERED FINANCIAL CONSULTANTS

Journal of Personal Finance International Association of Registered Financial Consultants - IARFC 1046 Summit Drive P.O. Box 42506 Middletown, Ohio 45042

www.journalofpersonalfinance.com

EXPERIENCE EDUCATION INTEGRITY BECOME AN RFCÂ® Registered Financial Consultant call 800.532.9060 or visit www.IARFC.org

IARFC INTERNATIONAL ASSOCIATION OF REGISTERED FINANCIAL CONSULTANTS