Volume 8, Issue 2
Having been director of the Michigan Retirement Research Center for almost five years now, I am constantly impressed with the rapid pace of change within SSA and the Office of Policy. The most recent quarter has, however, been exceptional even relative to this standard: a new Commissioner, Michael Astrue, assumed office, and Andrew Biggs joined him as the new Deputy Commissioner. Michael Astrue combines a very distinguished record of public service, including an interval as Counselor to the Social Security Commissioner, with private-sector executive experience. Dr. Biggs steps in from five year’s tenure in the Office of Policy, where he most recently was Acting Deputy Commissioner. The MRRC will endeavor to continue to work to contribute to the scholarly foundation for policy deliberations, and we look forward to participating with the exciting new leadership team.
The MRRC recently held a Researcher Workshop in Ann Arbor for its scholars. This year we welcomed over 50 researchers. The Workshop stresses wide participation both in terms of numbers of presenters and equal time for floor discussion and the presentation on each paper. This year we scheduled two special panel discussions, one on recent findings on private saving behavior and another on the outlook for Social Security reform. For the latter, in addition to Kent Smetters (Wharton) and John Karl Scholz (University of Wisconsin), we were delighted to have John Sabelhaus from the Congressional Budget Office (CBO) as a speaker. John Sabelhaus’ interesting discussion included insights and information from work at CBO, and this Newsletter includes an article adapted from his talk.
John Laitner, Director MRRC
Mark Your Calendar for the 9th Annual Meeting of the Retirement Research Consortium August 9-10, 2007 at the National Press Club, Washington, D.C.
Information about this conference, including registration information, will be provided and updated as needed on the MRRC website: www.mrrc.isr.umich.edu.
The conference is being organized by the MRRC in cooperation with the Retirement Research Center of NBER , the CRR and the SSA Office of Policy.
In this issue Robert J. Willis discusses his MRRC-supported research.
Q: Talk about what motivates your MRRC-supported research.
Over the years, various colleagues and I have been supported by the MRRC and NIA to pursue a line of inquiry in the area of cognitive economics, or the role of cognition in economic decision-making. It is important to understand this link for several reasons. First, with increasing longevity, the elderly are a fast growing segment of the population. Older Americans face increasingly complex financial decisions. For example, there is increased scope for choice due to the decline of defined benefit pensions and the growth of 401(k) plans. Portfolio decisions concerning savings and wealth management, decisions about when and or whether to annuitize, choices about health care and medical insurance plans as well as understanding rules around Medicare benefits are all cognitively demanding. Furthermore, the cognitive abilities of older Americans are highly heterogeneous and changing as they age. Decisions faced by older individuals balancing risks and benefits of alternative financial and health care choices are genuinely difficult. The field of cognitive economics seeks to develop innovations in economic theory and measurement to address questions like: What are the limitations in knowledge, memory, reasoning, calculation? What is the role of emotion, social context, conscious vs. unconscious judgments and decisions? What is the role of health as determinant, outcome and context for economic activity, decisions and well being? What is the connection between economic welfare and measures of well being? Some of my work for MRRC begins to address some of these issues. In particular, I have focused on subjective expected probability beliefs, which are available in the Health and Retirement Study (HRS).
Q: How do you measure subjective beliefs?
The probability questions use a format pioneered by Tom Juster and Chuck Manski. An example question asks, “Using a number from 0 to 100, what do you think are the chances that you will live to be at least X?” where X = 80 for persons 50 to 70 and increases to 85, 90, 95, 100 for each five year increase in age. On average, researchers using these questions have found that they make sense. For example, survival probabilities conform to life tables and are predictive of actual mortality. However, we also find that individual probabilities are very noisy with heaping on the focal values of “0”, “50-50” and “100.”
Q: How do subjective expected probabilities relate to financial decisions?
Subjective Expected Utility (SEU) theory assumes that individuals decide on a given course of action by choosing that action which yields the highest expected utility. The SEU model presumes that individuals competently perform some extremely demanding tasks before making any given decision. The SEU model assumes that the individual has a coherent set of beliefs about the probabilities that each state will occur. There is a distinction between “risk” and “uncertainty” where risk refers to a situation in which a specific probability can be attached to a given outcome while uncertainty refers to a situation in which a probability cannot be specified. What Ellsberg identified in the famous “Ellsberg Paradox” and subsequent experimental research has verified, is that most people prefer a known risk to an uncertain one of equal expected value. This is known as uncertainty aversion. In an early MRRC paper, Lee Lillard and I devised a model which assumes that most individuals form their preferences among uncertain prospects in a context of real world choices. Examples include the returns from investments in human or physical capital, the choice of a marriage partner, or the returns from a stock held over time. We show that uncertainty aversion is simply a consequence of risk aversion. Lee and I began to look at the pattern of responses to the probability questions in the HRS as indicators of the degree to which they indicate people’s capacity to think clearly about subjective probability beliefs. We explored the idea that focal answers of “0”, “50” and “100” were perhaps indicators of less coherent or well-formed beliefs than non-focal (or “exact”) answers. We treated the probability questions like a psychological battery and constructed an empirical propensity to give non-focal answers. We hypothesized that people who give more focal answers are more uncertain about the true value of probabilities. If the uncertainty is about a repeated risk, such as the return to a stock portfolio held over time, we show that people who have more imprecise probability beliefs (i.e. are more uncertain about the “true” probability) will behave more risk aversely. We found that people who had a higher propensity to give exact answers tended to have higher wealth, had riskier portfolios, and achieved higher rates of return, controlling for conventional economic and demographic variables.
Q: How does this affect portfolio choice?
In another MRRC paper, Gabor Kezdi and I continue this line of investigation in order to attempt to explain why some people hold stocks and others do not. In addition to the measures of precision from Lillard and Willis, we construct a measure of optimism using questions on expectations about weather. We find that stock ownership and the probability of becoming a stockholder are strongly positively correlated with the index of optimism and the propensity to make precise (non-focal) probability expectations. The optimism index may reflect cognitive bias as well as optimistic beliefs that are justified by the individual's private information. Nonetheless, we find strong evidence that people who give more optimistic answers to probability questions in general are healthier, wealthier, and more educated.
Q: What about subjective survival probabilities?
Adeline Delavande and I have begun examining the relationship between how long people expect to live and their probability of claiming Social Security benefits early, at age 62. We find that subjective survival probabilities capture meaningful behavioral responses to incentives for early Social Security claiming when they are purged of measurement error using risk factors as instruments. Among people who are still working at age 62, those who expect to live longer are likely to delay claiming of Social Security benefits to a degree that is both statistically and economically significant. For example, an increase of 5 percentage points in the subjective probability of survival to age 75 of each person leads to a 1.9 percentage point decline in the proportion who claim before age 64, from 29.6 percent to 27.7 percent.
Q: Where are you headed with this research?
Delavande and I are interested in deepening our understanding of how people make complicated financial decisions such as when to claim Social Security benefits by blending findings from the literature pointing out behavioral mistakes, and our own findings on the role of survival expectations. We are especially interested in understanding financial decision-making in the context of scarce cognitive resources. We think that even thinking about decision such as whether to delay Social Security claiming or not is costly when cognitive resources are scarce. If you think of an individual who is fairly uncertain about his life expectancy, when reaching age 62, he is faced with the following decision: either claim early because this sounds beneficial when he computes his average life expectancy, or invest cognitive resources to find out more information about his mortality risks. Many people who claim early may just do so because they are fairly uncertain about their life expectancy, and think it will be costly and maybe useless to devote cognitive resources to evaluate whether delaying would be better for them.
Robert J. Willis is Professor of Economics and Research Professor in the Survey Research Center and the Population Studies Center of the Institute for Social Research. He is the Principal Investigator for the Health and Retirement Study, a longitudinal survey of over 22,000 persons over age fifty in the United States, which is supported by the National Institute on Aging. Professor Willis received his Ph.D. from the University of Washington in 1974. He joined the University of Michigan in 1995 and holds joint appointments with the Survey Research Center and the Population Studies Center at the ISR. Before coming to Michigan, Professor Willis held appointments at the University of Chicago, SUNY at Stony Brook, and Stanford University. He has served on a number of advisory boards including, currently, the Census Advisory Board, and the Board of Visitors of the Rockefeller Center for the Social Sciences at Dartmouth College.
Parts of this article are based on Willis’ Presidential Address to the Society of Labor Economics, May, 2007.
Projecting Behavioral Responses to the Next Generation of Retirement Policies by Alan L. Gustman and Thomas L. Steinmeier WP 2007-153.
Money in Motion: Dynamic Portfolio Choice in Retirement by Wolfram J. Horneff, Raimond H. Maurer, Olivia S. Mitchell and Michael Z. Stamos WP 2007-152.
Enhancing the Quality of Data on the Measurement of Income and Wealth by F. Thomas Juster, Honggao Cao, Mick Couper, Daniel H. Hill, Michael Hurd, Joseph P. Lupton, Michael Perry and James P. Smith WP 2007-151.
The Responsiveness of Private Savings to Medicaid Long Term CarePolicies by Purvi Sevak and Lina Walker WP 2007-150.
By John Sabelhaus | Congressional Budget Office
The Congressional Budget Office (CBO) produces a variety of publications related to Social Security. Two recent documents are particularly germane to the question posed to a panel at the recent MRRC research workshop: What is likely to change Social Security in the future? What changes are needed? This article briefly summarizes some of the issues discussed and alternatives offered.
As part of CBO testimony in May 2005 before the Senate Committee on Finance, the Long-term Modeling Unit prepared a detailed “menu” of options for changing scheduled benefits or revenue. At the Finance Committee Chairman’s request, the group addressed the budgetary and distributional implications of various options for slowing the growth of Social Security benefits. The report examines the outlook for Social Security using the same long-term economic and demographic assumptions used in the March 2005 report of the Social Security trustees.
CBO projects that scheduled Social Security outlays (those implied by the current benefit formula) will rise from 4.3 percent of gross domestic product (GDP) this year to 6.5 percent in 2050. Revenue, however, is scheduled to remain at 4.9 percent of GDP. Various policy options for closing the predicted gap between revenues and expenditures follow. It is important to note that they are considered independent of each other.
Indexing of Benefits
Under current law, average scheduled benefits for newly entitled Social Security beneficiaries are wage-indexed, meaning that they grow with average economy-wide earnings. Because average earnings typically grow faster than prices do, the purchasing power of those benefits will be higher than that of benefits paid today. Price indexing would link benefits to price levels rather than to earnings levels, so that average benefits would grow at the same rate as prices. If benefits were indexed to prices instead of to wages beginning in 2012, the date of trust fund exhaustion would be delayed indefinitely. However, because of concerns about the erosion in living standards of lower income workers, an alternative is to index benefits to prices only for the top 70% of workers in terms of earnings, so-called progressive price indexing. This measure alone would close 100 percent of the gap between revenues and outlays in 2050.
CBO also examined the potential to adjust the benefit formula to offset increases in life expectancy in order to ensure that total lifetime benefits did not grow as life spans increased. This type of longevity indexing would close 31 percent of the gap between revenues and outlays.
Cost of Living
The Social Security cost-of-living adjustment (COLA) adjusts benefits for inflation so that the value of benefits is preserved over time. The COLA is linked to the consumer price index (CPI). Another option for controlling the growth of benefits is to lower the COLA beginning in 2012. There is some concern among policy makers that the existing CPI overstates the actual change in cost of living, and some have proposed changing the index. This alternative would slow the growth of benefit increases by about 0.4 percent per year. This option closes 19 percent of the gap.
The normal retirement age (NRA) is the age of eligibility for full Social Security retirement benefits. Under current law, the NRA is slowly being increased from 65 to 67. It will reach 67 in 2027 for workers born in 1960. CBO considered the option of raising this another year to 68. This closes 15 percent of the gap.
In 1983, Congress set the taxable maximum at 90 percent of covered earnings. Under current law, the earnings cap rises with average wages and is set at $97,500 (it was $90,000 in 2005 when these simulations were prepared). Earning above this amount are not taxed. Over time, incomes of the Nation’s highest earners have risen faster than the average with the result that the maximum in effect now covers only 83% of earnings. Raising this taxable maximum to cover 90 percent of earnings closes about 10 percent of the gap.
In April of 2006, the CBO prepared an analysis of the long-term effects of S. 2427, the Sustainable Solvency First for Social Security Act of 2006 which was introduced by Senator Robert Bennett. The proposed legislation would reduce benefits through a variety of measures including decreasing benefits.
S.2427 would implement progressive price indexing, which, as mentioned, would slow the growth of benefits for the highest-earning beneficiaries. Benefits for the lowest earners would continue to grow with average earnings, and benefits for other groups would grow at a weighted average of those two rates. It would also introduce longevity indexing. The longevity indexing provision would reduce the replacement factors to reflect changes in life expectancy at age 67. Beginning in 2018, the provision would multiply the factors by a ratio that captures the increase in life expectancy at age 67. By 2100, that provision alone would result in a 20 percent reduction in replacement factors. Together these two provisions put the Social Security system well into surplus by 2080.
The proposed legislation also includes a faster increase in the normal retirement age. Under current law, the normal retirement age rose to 66 for people who turned 62 in 2005. After 12 years, it will again climb in two-month steps for six years, reaching 67 for people who turn 62 in 2022 or later. The proposal would shorten the 12- year hiatus by five years. The increase of two months per birth year would start with the 1950 birth cohort, and the normal retirement age would reach age 67 for the 1955 cohort.
The table above shows the benefit to tax ratios for cohorts born in the 1950s and in the 1990s under current law and under the proposed legislation. It compares the lowest, middle, and highest earnings quintiles on this outcome. It shows that with the reforms made in the Sustainable Solvency First legislation, the lowest quintile earners born in the 1950s will see no change in their benefit to tax ratio. Whereas those born in the 1990s would see a considerable decrease in their benefit to tax ration, under the Senator Bennett’s proposal, they would stay even with those born in the 1950s. Those in the highest quintile born in the 1990s would see a relative drop in their benefit to tax ratio under the proposed legislation whereas those born in the 1950s in the highest quintile will see only a modest decrease.
John Sabelhaus is Unit Chief of the Long-term Modeling Group at the Congressional Budget Office. This article is based on his presentation at the 2007 MRRC Researcher Workshop.
CBO Publications cited:
Menu of Social Security Options (May 2005).
For the third consecutive year, MRRC researchers gathered at the Institute for Social Research in Ann Arbor for a day and a half of collegial exchange about their MRRC-supported research. As the workshop increases in popularity, more than 50 researchers were willing to take their chances with April in Michigan.
John Laitner was pleased to welcome Sharmila Choudhury and Irena Dushi from the SSA Office of Policy. The workshop began with a panel on the topic of Health/Health Insurance and Retirement. John Bound presented Estimating the Health Effects of Changes in Retirement Age; John Bailey Jones presented joint work with Eric French on The Effects of Health Insurance and Self-Insurance on Retirement Behavior; Helen Levy reported on work with David Weir entitled Extra Help: Take-up of the Social Security Administration’s Low-Income Subsidy Program for Part D of Medicare in 2006; and Lina Walker discussed The Responsiveness of Private Savings to Medicaid Long-Term Care Policies, joint work with Purvi Sevak.
Purvi Sevak, Irena Dushi, Sarmila Chaudhury, Margorie Honig, Martin Farnham
Alan Gustman next chaired a panel on Labor Supply and Retirement. Presentations in that panel included What Explains Trends in the Labor Force Behavior of Older Men in the US? by David Blau; Decisions of When to Retire by John Laitner and Dan Silverman Discouraged Workers? Job Search Outcomes of Older Workers by Nicole Maestas; The Effect of Subjective Survival Probabilities on Retirement and Wealth in the US by David Bloom, David Canning, Michael Moore and Younghwan Song; and Determinants of Retirement and Annuitization Choice in Personal Account Plans: Evidence from a new retrospective data set in Chile by Estelle James and Alejandra Edwards.
Bob Willis continued as chair of Labor Supply and Retirement, part II. Ananth Seshadri presented work with John Karl Scholz on Children and Household Wealth; Emma Aguila presented Pension Reform and Saving; Michael Hurd presented joint work with Susann Rohwedder on Alternative Measures of replacement rates; Elizabeth Powers and Todd Elder presented on A Longitudinal Analysis of Entry of the Low-Income Elderly into the SSI Program; and Erik Hurst presented Consumption in Retirement: Recent Developments.
The day ended with a panel discussion which posed the question: Are many people over-prepared for retirement? Laurence Kotlikoff, Susann Rohwedder, and Erik Hurst offered their thoughts, and a lively discussion ensued.
Pierre-Carl Michaud, Raquel Fonseca, David Blau, Emma Aquila
The second day opened with a continuation of the panels on Labor Supply and Retirement with Eric French serving as Chair. Panel presentations included A Cross-National Comparison of Self-Employment Dynamics at Older Ages by Nicole Maestas; Trends in Labor Force Participation of Married Women by Christopher House, John Laitner and Dmitriy Stolyarov; and Housing Wealth and Retirement Timing by Martin Farnham and Purvi Sevak.
Olivia Mitchell chaired a session on Planning and Financial Literacy in which Alan Gustman presented Imperfect Knowledge of Pension Plan Type, joint with Tom Steinmeier and Nahid Tabatabai; Annamaria Lusardi presented Measuring Financial Literacy; Daniel Benjamin discussed Who Is Behavioral? Cognitive Ability and Anomalous Preferences; and Jonathan Zinman discussed his work entitled Cognitive Skills and Wealth Accumulation: Evidence from the NLSY.
Michael Hurd chaired a lunchtime panel discussion entitled: What is likely to change Social Security in the future? What changes are needed? John Sabelhaus, John Karl Scholz, and Kent Smetters served as panelists, sparking active discussion throughout.
Next, John Laitner chaired a panel on Pensions and Retirement with presentations on Trading Behavior in Personal Accounts: Lessons from 401(k) Pension Participants by Olivia Mitchell and Takeshi Yamaguchi (presenter); Money in Motion: Dynamic Portfolio Choice in Retirement by Wolfram Horneff and Raimond Maurer (presenter); and Public Pensions and State Budgets by Leslie Papke.
Irena Dushi, Marjorie Honig, Lina Walker
The final session was chaired by Charlie Brown and included A Dynamic Model of Retirement and Social Security Reform Expectations: A Solution to the New Early Retirement Puzzle by Hugo Benitez-Silva; How Did the Elimination of the Earnings Test Above the Normal Retirement Age Affect Retirement Expectations? by Pierre-Carl Michaud; Self-Assessed Retirement Outcomes: Determinants and Pathways by Susann Rohwedder; and Errors in Subjective Survival Probabilities: Prevalence and Behavioral Responses by Todd Elder.
The Federal Interagency Forum on Aging-Related Statistics (Forum) will hold a free, half day workshop to take place on Wednesday, August 8, 2007, 1:00-5:30pm, at the Bureau of Labor Statistics’ Conference Center in Washington, D.C.
Howard Iams of the Social Security Administration will make an introduction, and John Phillips of the National Institute of Aging will serve as moderator. Presentations will include Projecting 401(k) Accumulations and Developing Replacement Rates by Sarah Holden and Peter Brady of the Investment Company Institute; Measuring Defined Contribution Pension Wealth with Matched Administrative Data by Anil Kumar of the Federal Reserve Bank of Dallas; New Estimates of the Future Path of 401(k) Assets by Steve Venti of Dartmouth College; Approaches to Estimating Pension Wealth in the HRS by David Weir of the University of Michigan; and Estimating Pension Wealth in MINT by Eric Toder and Karen Smith of The Urban Institute.
Presentations will briefly report on the methodology that the researcher(s) use in estimating/projecting lifetime pension wealth for pre-retirees and/or recent retirees, specifically what problems were faced and how were they solved. After all the talks, speakers will participate in a general discussion with the audience. The Forum intends to put the presentations on its website (www.agingstats.gov) for easy reference by the research community. Documentation of alternative methodologies may lead to improvements in the estimates as ideas are shared. In addition, researchers using the measure or projections would benefit from better documentation identifying strengths and weaknesses.
An important issue in the retirement literature is expected pension wealth formed by the stream of all pension payments over a lifetime. Economists assume that people consider this expected stream of pension wealth in making decisions about leaving the career job, leaving the labor force, and taking up Social Security retirement benefits. Many studies of retirement behavior attempt to measure pension wealth incentives to claim benefits and/or leave the labor force because economic theory predicts outcomes will vary based on the magnitude of the financial incentives. Incentive measures typically focus on the marginal value of delaying benefit claims. On the other hand, adequacy studies focus on the level of income that is produced from retirement savings in order to calculate measures of well-being. Some include pension wealth in measuring pre-retirement wealth and economic well-being. Others focus on the extent that the projected income flow of a person/couple is “adequate” in retirement and pension benefits form one of the pillars of retirement income. They use projected lifetime pension wealth to estimate future retirement income from pensions. Thus, lifetime pension wealth is important to the analysis of retirement risks and economic well-being of current and future older persons. Though both of these constructs are valuable for the analysis of retirement broadly, they represent different concepts of pension wealth. The Forum on Aging Related Statistics is interested primarily in estimating lifetime pension wealth for the pre-retired or newly retired because of its importance to economic well-being of the aged and future retirees.
The Forum membership includes 14 Federal agencies that produce or use statistics on aging. It provides these agencies with a setting to discuss data issues and concerns that cut across agency boundaries; facilitates in development of new databases; improves mechanisms currently used to disseminate information on aging-related data; invites researchers to report on cutting-edge analyses of data; and encourages international collaboration.
The visitor's entrance for the Postal Square Building is located on First Street, N.E., across from Union Station. There are two entrances to the Postal Square Building on First Street. The visitor's entrance is located near the intersection of First Street and G Street; the other entrance is for the Postal Museum and is near the intersection of First Street and Massachusetts Ave. Contact Sherry Dockery to register for the workshop. Her email address is firstname.lastname@example.org and her phone number is 301-458-4187.
University of Michigan
Institute for Social Research
426 Thompson Street
Ann Arbor, MI 48106-1248