Volume 14 Issue 2 - May 2014
MRRC Convenes Tenth Annual Workshop
MRRC researchers gathered in Ann Arbor, Michigan, for the 10th annual workshop on April 11-12, 2014, at the University of Michigan’s Ross Executive Education Center. During the fast-paced two-day meeting, 34 speakers presented research papers. A collegial exchange of ideas characterized the workshop, with plenty of questions and discussion both within sessions and during breaks.
John Laitner opened the day with a warm welcome to participants: “We are really excited about the talks we have lined up, and are glad that you could join us.”
Claire Green and Joel Feinleib from the Social Security Advisory Board (SSAB) attended this year. According to the website www.ssab.gov, “The SSAB is an independent, bipartisan board created by Congress and appointed by the President and the Congress to advise the President, the Congress, and the Commissioner of Social Security on matters related to the Social Security and Supplemental Security Income programs.”
Panel I. Fiscal Reform
The first session on Friday, April 11 dealt with Fiscal Reform and was chaired by John Laitner. R. Anton Braun presented The Implications of a Graying Japan for Government Policy, coauthored with Douglas Joines. Rapid aging of the Japanese population will put considerable pressure on Japan’s budget in future years. Under current policy, government expenditures on health in Japan will rise from 8% to 14% of GDP between 2014 and 2050. Public pension outlays will account for another 14% of GDP in 2050. The authors develop a model to examine the following: Is current fiscal policy sustainable? How will taxes have to evolve in future years to cover current government commitments to the old? How does the fiscal situation change if current commitments are reduced?
Selahattin Imrohoroglu presented the paper Increasing the Social Security Payroll Tax Base in General Equilibrium, coauthored with Shinichi Nishiyama. According to one projection by Social Security, in 2088 the dependency ratio will rise to 44% (from 24% in 2013). Using a macroeconomic model, the authors aim to determine how to finance future Social Security expenditures by taking into account the size and consequences of demographic shifts due to an aging population. They explore various ways to bring actuarial balance to Social Security.
Karen Kopecky discussed Old, Sick, Alone and Poor: A Welfare Analysis of Old-Age Social Insurance Programs, cowritten by R. Anton Braun and Tatyana Koreshkova. The authors assess the welfare effects of means-tested social insurance programs (MTSI) for retirees using a quantitative model of the U.S. economy. They find that these programs are highly valued, and find that, if anything, the scale of the current U.S. old-age social insurance system is too small. Finally, they show that there are important interactions between MTSI and Social Security. For example, if Social Security was removed, MTSI take-up rates by the elderly would approximately triple.
Shinichi Nishiyama presented preliminary findings from The Generational Impact of Closing the Fiscal Gap, coauthored by Felix Reichling. In the paper, the authors introduce stylized fiscal policy changes — an increase in progressive income tax rates, a cut in the government’s transfer spending, an increase in a flat consumption tax rate, and others — to close the fiscal gap to a heterogeneous-agent overlapping-generations economy. Finally, they analyze the effect of each change on the macroeconomy and welfare levels by age.
Panel II. Life-Cycle Portfolio Choices
The second session was chaired by Dmitriy Stolyarov. Matthew Shapiro discussed Resolving the Annuity Puzzle: Estimating Life-Cycle Models without (and with) Behavioral Data, cowritten by John Ameriks, Joseph Briggs, Andrew Caplin, and Christopher Tonetti. The paper includes preliminary findings from the Michigan-NYU-Vanguard Project (MINYVan) and the Vanguard Research Initiative (VRI), which are supported by National Institutes of Health and the Vanguard Group Inc. These projects seek to provide innovations in wealth measurement to better understand well-being in retirement and determine why retirees are reluctant annuitize.
Jeffrey Wenger talked about Income Diversification as Retirement Self-Insurance, written with coauthors Christian Weller and Benyamin Lichtenstein. According to Survey of Consumer Finance data, self-employment of households age 50 and over has ticked upward over the last two decades. The percentage of households who are entrepreneurs increased from around 7% to about 11%, while the percentage of independent contractors has stayed at around 6%. Twelve percent of entrepreneur households had Social Security plus other retirement income. The authors examine to what extent retirees seek other income streams to self-insure.
Olivia S. Mitchell presented How Family Status and Social Security Claiming Options Shape Optimal Life Cycle Portfolios, coauthored by Andreas Hubener and Raimond Maurer. The authors have developed a life-cycle model that generates optimal work hours and Social Security claiming ages for men and women, as well as decisions about stocks, bonds, and life insurance. Their model also integrates key details of Social Security benefit options. The authors simulate hours of work, saving and investment, asset allocation, and benefit claiming decisions by singles and couples. Consistent with time-use surveys, husbands specialize in paid work, while married women do more home production. In addition, wives are shown to optimally claim Social Security benefits early, while husbands claim late. When the authors compare their predictions with actual claiming behavior in the Health and Retirement Study, they show that the model does a good job matching work hours and retirement age profiles. Nor surprisingly, singles do not demand much life insurance. Among couples, by contrast, there is a large demand for life insurance by husbands, and only a small demand by wives.
Victoria Prowse discussed Life-Cycle Inequality and the Actuarial Unfairness of Defined-Benefit Pension Systems, cowritten with Peter Haan and Daniel Kemptner. Current pension schemes cause redistribution towards those with high life expectancy. While education impacts positively on both earnings and life expectancy, redistribution is typically regressive because life-cycle earnings are correlated with life expectancy. Furthermore, non-actuarial fairness may work to increase life-cycle consumption inequality. The authors use a dynamic life-cycle model of employment and savings behavior to explore the effect of introducing actuarial fairness into a defined-benefit pension system. They aim to quantify how actuarial fairness with respect to life expectancy decreases life-cycle consumption inequality: preliminary results show increases in life-cycle consumption of 2% for low educated individuals, and decreases of 7% for the highly educated.
Panel III. Data and Measurement
The third session was chaired by Michael Hurd. Ananth Seshadri presented the paper Real Interest Rates and Economic Growth, coauthored with Bruce Hansen and John Karl Scholz. When constructing long-range (LR) projections for the OASDI Trustees Report, is it reasonable to assume no relationship between LR real interest rates and LR economic growth? Using data from 1901-2011, the authors estimate the correlation between three measures of economic growth (economic productivity, real wage growth, and real GDP growth) and real interest rates. They find a moderately negative correlation between economic growth and real interest rates. They then examine the implications of their estimates for Social Security Trust Fund accumulations.
T. Lynn Fisher discussed findings from the paper Is the 2010 Affordable Care Act Minimum Standard to Identify Disability in All National Datasets Good Enough for Policy Purposes? coauthored by Richard V. Burkhauser, Andrew J. Houtenville, and Jennifer R. Tennant. They find that a work limitation (WL) question identifies a greater proportion of Disability Insurance/Supplemental Security Income recipients than the 6 Question Sequence (6QS) currently used by the Current Population Survey. The WL question identifies about 2/3 of the beneficiaries that the 6QS misses. However, the 6QS identifies about half of the beneficiaries that the WL question misses. Therefore, they propose adding the WL to the 6QS.
Frank Heiland and Na Yin discussed Anchoring Vignettes in the Health and Retirement Study: How Medical Professionals and Disability Recipients Characterize the Severity of Work Limitations. They analyze systematic differences in the characterization of identical health and disability cases. They compare how the following types of respondents describe the severity of work limitations (1) health professionals with medical training, (2) disability recipients who have first-hand experience with work limitations and how severity is officially determined, and (3) non-disabled, nonmedical respondents. They aim to develop objective response scales in conjunction with anchoring vignettes to adjust self-reported measures of health and disability.
V. Joseph Hotz presented The Add Health Parent Study: An Overview. This new intergenerational dataset is important for investigating the health and well-being of family members and families. The Add Health Parent Study (AHPS) interviews the parents of adult children enrolled in the National Longitudinal Study of Adolescent Health (Add Health) which itself is a nationally representative, cohort sample that contains rich biological, genetic and social longitudinal data. The AHPS will collect data on three domains of interest: health and health behaviors; cognition, personality and preferences; and intergenerational relationships — ties, exchanges and support, along with detailed measures of economic capacity of parents and adult children.
Daniel Silverman discussed the paper Using Naturally Occurring Data to Measure Major Financial Changes Late in Life coauthored by Michael Gelman, Shachar Kariv, Matthew Shapiro, and Steve Tadelis. Silverman explained how the researchers are investigating using a mobile banking application to track and measure income and expenditures. Using such a mobile app could increase the frequency, accuracy, and comprehensiveness of recording daily financial transactions of existing users. While there are some drawbacks to using this method, it provides high quality measurement at low cost and could augment survey research in the future.
Panel IV. Program Interactions, Transfers, and Responses
The fourth session was chaired by Daniel Silverman. David Blau presented The Effect of Inheritances on Labor Supply of Older Married Couples: Estimates and a New Test of Commitment, coauthored with Ryan Goodstein. The authors estimate the impact of inheritance receipt by an individual on own and spouse Labor Force Participation (LFP) among older married couples. Analyzing inheritance receipt in the Health and Retirement Study, they find that the effect of an inheritance shock on LFP is larger for the recipient than for the spouse.
Ethan M.J. Lieber discussed the paper Costs and Benefits of In-Kind Transfers: The Case of Medicaid Home Care Benefits cowritten by Lee Lockwood. The authors assess the welfare costs and benefits of in-kind transfers within the context of Medicaid home care for the elderly. They analyze the costs using data from the Cash and Counseling experiments and the targeting benefits using data from the Health and Retirement Study. Their estimates indicate that both the costs and benefits of in kind provision are large, suggesting that increases in policy efficiency could yield significant welfare gains.
Stefan Staubli presented Extended Unemployment Benefits and Early Retirement: Program Complementarity and Program Substitution, written with Lukas Inderbitzin and Josef Zweimüller. The authors investigate how changes in the potential duration of unemployment (UI) benefits affect (1) the incidence of early retirement of the older unemployed and (2) the pathway to early retirement of the older unemployed. They estimate the impact of introducing and abolishing the Regional Extended Benefits Program that offered targeted extension of UI benefits to unemployed Austrian workers in the late 1980s and early 1990s. They examine transitions to early retirement, disability pensions, and to old-age pensions.
Gustavo Ventura presented the paper, Childcare Subsidies and Household Labor Supply, coauthored with Nezih Guner and Remzi Kaygusuz. The literature has shown that availability and cost of childcare is a central determinant of female labor supply. The authors ask, what are the labor supply, output, and welfare effects of subsidizing child care for the US economy? Because female labor supply is quite elastic, child-care subsidies could have potentially large effects. The authors develop a life-cycle model with heterogeneous married and single households and costly childbearing to quantify the effects of child care subsidies. They find large effects on labor supply of married females, and larger responses for those with lower levels of education.
Panel V. Well-Being in Retirement
The fifth session was chaired by Kathleen McGarry. Alan Gustman presented the paper The Social Security Windfall Elimination and Government Pension Offset Provisions for Public Employees in the Health and Retirement Study, cowritten with Thomas L. Steinmeier and Nahid Tabatabai. About 3.5% of households in the Health and Retirement Study (HRS) are subject to either the Windfall Elimination Provision (WEP) or Government Pension Offset (GPO) provision features of the Social Security benefit determination process that limit the Social Security benefits of employees who worked in and earned a pension on jobs not covered by Social Security, but who are also are eligible for Social Security benefits. The present value of the Social Security benefits of affected households is reduced by roughly 1/5, amounting to 5 to 6% of the total wealth affected households accumulate before retirement. Households affected by both WEP and GPO lose about 1/3 of their benefit. Those with the least reductions to their Social Security benefits are those who benefit from the progressive Social Security benefit formula, having worked in both covered and uncovered employment, and are entitled to Social Security benefits, but with little or no pension from uncovered work.
David Love discussed Immigrants, Social Security, and the Adequacy of Wealth, coauthored by Lucie Schmidt and Purvi Sevak. Little is known about immigrants’ preparedness for retirement; they may be especially vulnerable. The authors plan to calculate measures of comprehensive wealth for immigrants and natives using 1998-2010 Health and Retirement Study (HRS) data. They will examine median profiles of annualized comprehensive wealth over the retirement period. Using a life-cycle framework to try to understand observed trajectories of annualized wealth, they aim to identify the role of uncertain longevity, bequest motives, precautionary savings, and housing.
Gema Zamarro presented Is Working Longer Good for You? A Simultaneous Equations Approach, cowritten with Raquel Fonseca, Arie Kapteyn, and Jinkook Lee. Continued improvements in life expectancy and fiscal insolvency of public pensions have led to an increase in pension entitlement ages in several countries, but the consequences for subjective well-being are largely unknown. The authors examine the effect of retirement on subjective well-being within 12 countries, using panel data for the 50+ population from the U.S. Health and Retirement Study (HRS) and the Survey of Health, Ageing, and Retirement in Europe (SHARE). They seek to understand the paths through which retirement induced through Social Security pension eligibility affects subjective well-being.
Panel VI. Behavioral Economics and Health
John Laitner chaired the sixth session of the workshop. Daniel Benjamin discussed Proxy-Phenotype Method Identifies Common Genetic Variants Associated with Cognitive Performance coauthored with D. Cesarini, C. Chabris, T. Esko, M. Johanesson, P. Koellinger, D. Laibson, S. Medland, N. Rietveld, O. Rostapshova, P. Turley, P. Visscher. In HRS, across specifications, a “polygenic score” (an index composed of many genetic variants) predicts 2-4% of variance in cognitive measures. While explanatory power is not large, it provides sufficient power for some analyses in the HRS, such as heterogeneity of effects of policy. As available sample sizes grow, more specific genetic variants associated with cognitive performance will be identified, and researchers will be able to construct a polygenic score with greater explanatory power.
Erik Meijer presented Labor Force Transitions at Older Ages: the Roles of Work Environment and Personality, on behalf of Marco Angrisani and other coauthors Michael Hurd, Andrew Parker and Susann Rohwedder. They study the relationship between retirement decisions, work environment and personality. They find that the following dimensions are strongly predictive of labor force transitions to part-time work or full retirement: health insurance, wages, physical demands, use of computer, age discrimination, whether work interferes with personal life, and relations with coworkers or supervisor. However, they do not find a significant direct effect of personality on labor force transitions, except for openness to experience, which is associated with an increase in the likelihood of retirement.
Meghan Skira discussed the paper Missing Work is a Pain: The Effect of Cox-2 Inhibitors on Sickness Absence, cowritten with Aline Bütikofer. The authors analyze how the availability of Cox-2 inhibitors, a class of pharmaceuticals used for treating pain and inflammation, affected the sickness absence of individuals suffering from chronic joint pain. They exploit the market entry of the Cox-2 inhibitor Vioxx and its sudden market withdrawal as exogenous sources of variation in Cox-2 inhibitor use. They use administrative data from Norway on sickness leave and merge it with survey data on health. Preliminary results suggest Vioxx’s removal increased sickness days by 10 to 14% and increased the probability an individual with joint pain receives disability benefits.
Panel VII. Shocks and Household Responses
Alan Gustman chaired the seventh session. Michael Hurd presented, Changes in Spending during the Great Recession Following Stock and Housing Wealth Losses, coauthored with Marco Angrisani and Susann Rohwedder. The authors examined the effects of wealth loss in Great Recession on household spending, housing wealth and stock market wealth. Using data from the Consumption and Activities Mail Survey of the Health and Retirement Study in three different years pre- and post-recession, they analyze 36 spending categories and total spending levels. Incorporating housing price data from the Federal Housing Finance Administration index by state for all 50 states into their analysis, they find large declines in housing wealth, and great variation in housing prices by location. Their analysis indicates that the housing price decline contributed to a decline in consumption. The stock market decline may have contributed to a decline in spending also, but its impact was smaller since it is a less important asset for most.
Brooke Helppie McFall discussed the paper The Impact of Hypothetical Wealth Shocks on Retirement Timing, cowritten with Matthew Shapiro. The authors ask, to what extent do individuals think hypothetical wealth losses or gains would impact their retirement plans? How do reactions to hypothetical losses compare with reactions to actual losses such as those in the Great Recession? Using 2013 data from the American Life Panel, they model hypothetical gains/losses on each individual, to see how it impacts their spending levels and whether they expect to work full-time longer as a result. She compares the hypothetical results to her 2011 paper based on data from the Health and Retirement Study that measured actual behavior during the Great Recession. They find that the amount that individuals expect to extend full-time work after age 65 is correlated with the size of wealth shocks.
Frank Stafford presented results from Family Responses to Housing Consumption Commitments, coauthored with Bing Chen. Many middle wealth and income Americans became home owners up to 2007. However, the financial crisis that began in 2007 impacted income and housing. Many small stake stock account holders who were experiencing mortgage distress sold off their shares, bearing capital losses. The housing crisis was characterized by a debt-deflation cycle. Mortgages taken out closer to the end of the boom were more likely in foreclosure by 2011. This had a disproportionate impact on African-Americans. With the stock market crash and mortgage distress, many Americans sold off stocks (held directly and in pensions) during 2007-2011. Higher stock wealth owners held their ground and were the net buyers in the fire sale on directly held stocks.
Panel VIII. Planning and Life-Cycle Behavior
Ananth Seshadri chaired the eighth session. Susann Rohwedder presented the paper, Expectations and Household Spending, cowritten with Michael Hurd. The authors estimate the effect of expectations about unemployment on household spending using high-frequency panel data from the RAND American Life Panel. The data were collected during the Great Recession and its aftermath, a time of great economic uncertainty. Using monthly data both on total household spending and on subcategories of spending, they find that changes in total spending made in response to changes in the chances of becoming unemployed are difficult to detect empirically. This is because many categories of spending, such as rent, utilities, and car payments, tend to be fixed from month to month. Nevertheless, when studying subcategories of spending that are more easily adjusted in the short-term they find significant effects. For example, in response to an increase from 0 to 1 in the probability of becoming unemployed, we estimate that households reduce spending on clothing by about 14%, dining out and other entertainment by 11%, and personal care by 12%.
Elena Gouskova discussed On the Dynamics of Household Health Care Costs: Are Medical Expenditure Risks Overestimated? The author examines forward telescoping -- a type of memory error -- wherein respondents recall events as happening in more recent time periods than actually occurred. This is a potentially serious reporting problem. Gouskova asks, how are the observed dynamics of medical costs reports different from that of the true medical costs if respondents over-report? To answer this question, she examines health care expenses from the Health and Retirement Study during 2002-2010, simulating the magnitude of telescoping using different assumptions. She finds that forward telescoping increases the level and variance of medical cost estimates, which may explain the differences in medical cost estimates across surveys observed by other authors.
John Sabelhaus discussed Job Change or Job Shock: What Drives Early Withdrawals from Retirement Accounts? cowrittten with Victoria Bryant. The authors seek to quantify pre-retirement plan “leakage” from defined-contribution plans across groups over time. They aim to determine the effects of income shocks and marital dissolution and the effect of job change, with or without shocks. They link data from 2004-2010 on early retirement plan withdrawals to job changes. Their preliminary results show (1) no surge in withdrawals during the Great Recession, (2) income shocks double the withdrawal probability, (3) job change is strongly correlated with early withdrawals, whether shocks occurred or not, (4) early withdrawal rates are higher at lower incomes, which is partially explained by higher shocks at low incomes; and (5) early withdrawals are found to be substantial at all income levels.
Giovanni Gallipoli presented Skill Complementarity and the Geography of Intergenerational Mobility, coauthored with Brant Abbott. The economics literature documents very persistent differences in the correlation of intergenerational earnings of parents and children, both across countries and across regions within the United States. The authors hypothesize that the way things are produced (i.e., the degree of skill complementarity in production) shapes outcomes like intergenerational mobility, inequality, and the prevalence of redistributive education policies. Using OECD data about industry composition together with proxies for the degree of skill complementarity by industry based on O*NET data, they find that the intensity of skill complementarity is significantly correlated with the degree of intergenerational earnings persistence and economic mobility. They estimate a rich model that fits the data at both the aggregate and industry level. They also find that industry composition and patterns of specialization are consistently associated with observed patterns of earnings taxation and shares of public spending on education. According to their estimates, the measured differences in skill complementarity account for 1/5 of international variation in earnings persistence (and up to 1/3, if indirect effects due to public policy are included).
Panel IX. Retirement Choices / Work Environment
John Laitner chaired the final session of the workshop. Kathleen Mullen discussed Suitable Work Conditions and Employment of Older Workers, cowritten with Nicole Maestas, David Powell, Till von Wachter and Jeffrey Wenger. Mullen described the American Working Condition Project, a new Internet Survey of individuals ages 50-70 that is being fielded in 2015. Funded by the Sloan Foundation Working Longer Program, it will be a representative panel of 2,000 U.S. workers. Researchers plan to harmonize the questionnaire with the Sixth European Working Conditions Survey (2015). EWCS is survey of 28 EU countries, plus in 2015 Brazil, Korea, and the United States. They will investigate issues such as, what job conditions make work sustainable over a longer work life? Survey questions will focus on several broad areas: (1) employment arrangements and compensation, (2) task complexity, autonomy, and environmental exposures, and (3) how do workers reconcile working time with other demands on their time? They have developed their questions with an eye toward how job characteristics were defined and elicited in other national studies.
Lauren Hersch Nicholas presented Lifetime Job Characteristics and Social Security Claiming Decisions, coauthored with Nicolae Done and Susan Yeh. Work is increasingly appreciated as a determinant of long-term health outcomes and mortality, though little is known about how the cumulative health effects of occupations influence workers’ health in retirement or ability to continue working and delay Social Security benefit receipt. This project will link all lifetime jobs reported by Health and Retirement Study respondents to a comprehensive set of job and worker characteristics collected by the Department of Labor’s Occupational Information Network. The authors’ preliminary analysis looked at the relationship between several measures of job physical and cognitive demands and indicators for whether a worker ever received SSDI benefits and the age Social Security benefits were first claimed. A one standard deviation increase in non-routine cognitive job demands was associated with a 2.3 percentage point decline in the probability of applying for and receiving DI benefits and a 0.33 year increase in age at initial benefit claiming. In contrast, a one standard deviation increase in routine manual demands increased probability of DI application by 1.5 percentage points and receipt by 1.6 percentage points.
David Powell discussed The Effect of Local Labor Demand Conditions on the Labor Supply Outcomes of Older Americans, cowritten with Nicole Maestas and Kathleen Mullen. The authors find that a large fraction of the growth in labor force participation of older individuals in the United States since 1990 can be attributed to labor demand increases in professional service industries. Analyzing Census and HRS data, they find evidence that this relationship is causal. Areas with large exogenous shocks to labor demand for professional services experienced faster growth in labor force participation for older people. They find that professional service industry jobs (1) require less physical effort, (2) require less heavy lifting, (3) require less stooping/kneeling/crouching, and (4) are more likely to allow employees to reduce hours during a regular work schedule.