| July 2005
Volume 6, Issue 2
To help maintain and amplify its long tradition of research, the Social Security Administration created the Retirement Research Consortium in 1998, a national and international network of scholars who gather data and develop analytic tools to build and continuously modernize the knowledge base on Social Security, pension, and retirement policy in this country. The annual RRC Conference in Washington, D.C. is a key link in the design of this research network: each year, the SSA provides funding for researchers in the consortium to come to Washington to share their most recent research findings with each other and, importantly, with staff and administrators of the Social Security Administration, as well as other policymakers and analysts. This newsletter includes the Agenda for this year’s RRC Conference, along with information about how to register to attend. The SSA Office of Policy also recognizes the importance of smaller meetings that can stimulate scholarly discussion and interactions within the research Centers that compose the RRC. In this vein, the Social Security Administration recently provided support for a MRRC Researcher Workshop, which was held at the University of Michigan in Ann Arbor in May. Approximately 40 MRRC research scholars attended the workshop, where they had an opportunity to share findings and new ideas in a relatively informal setting. Susan Grad represented the Office of Policy and was the lunchtime speaker. As the MRRC continues to grow, it is important for our scholars to maintain close familiarity with each other’s work, including that of new members; to actively discuss research at all stages of its development; and, to identify gaps in our knowledge that merit future attention. The workshop was quite successful, I believe, and I hope that all who attended benefited from the exchanges. A write-up of the event is included in this newsletter. Looking forward to seeing you in D.C.
John Laitner, Director MRRC
August 11-12, 2005 National Press Club, Washington, DC
Information about this conference, including registration information and the agenda, will be provided and updated as needed on the MRRC website: http://www.mrrc.isr.umich.edu/news/events/
The 7th Annual Conference is being organized by the Boston Center for Retirement Research, in cooperation with the Michigan Retirement Research Center, the Retirement Research Center of NBER, and the Social Security Administration.
Below is a selection of abstracts from recently released MRRC working papers. Visit our website for full papers and other current papers. http://www.mrrc.isr.umich.edu/publications/papers/.
John Laitner and Daniel Silverman
Using pseudo-panel data, we estimate the structural parameters of a life-cycle consumption model with discrete labor supply choice. A focus of our analysis is the abrupt drop in consumption upon retirement for a typical household. The literature sometimes refers to the drop, which in the U.S. Consumer Expenditure Survey we estimate to be approximately 16%, as the “retirement-consumption puzzle.” Although a downward step in consumption at retirement contradicts predictions from life-cycle models with additively separable consumption and leisure, or with continuous work-hour options, a consumption jump is consistent with a setup having non-separable preferences over consumption and leisure and requiring discrete work choices. This paper specifies a life-cycle model with these latter two elements, and it uses the empirical magnitude of the drop in consumption at retirement to provide an advantageous method of identifying structural parameters–most importantly, the inter-temporal elasticity of substitution.
Estelle James and Alejandra Cox Edwards
Postponing retirement will become increasingly important as a means to increase labor force, its output and old age security, as populations age. Recent research has focused on incentives stemming from the social security system that influence the worker’s decision to retire. Defined benefit systems (both public and private) often contain penalties for postponing access to pensions or continuing to work while receiving a pension. In contrast, the tight link between contributions and accumulations and actuarial conversion of accumulations into pensions in privately managed defined contribution systems may lead workers to postpone pensions or to continue working withdrawals begin. The experience of Chile, which implemented its new system offers an opportunity to test if the change in incentives has indeed produced the expected change in retirement behavior. Using probit analysis of household survey data from 1960 to 2002, we estimate the impact of the pension reform on the probability of 1) becoming a pensioner and 2) dropping out of the labor force, for older workers. We find strong effects of the new system on both propensities, in the aggregate and at the individual level after controlling for individual and macro-economic variables. In particular, restricted access to early pensions and the exemption of pensioners from the pension payroll tax appear to exert a powerful effect on labor force participation rates.
John Laitner, Christopher House and Dmitriy Stolyarov
Economists’ principal tool for studying household behavioral responses to changes in tax and other government policies, and the magnitude and determinants of private saving, is the life—cycle model. The purpose of this paper is to attempt to incorporate into that model one of the most conspicuous changes in the U.S. economy in the last 50 years, the rise in labor market participation for married women. The increased presence of married women in the labor force has obvious benefits: women now earn much more income than they did in the past. On the other hand, working women presumably spend less time doing housework and other types of home production, and the forgone value of time at home reduces the net benefit of their work in the market. Conventional accounts do not provide measurements of the costs of lost home production, but we attempt to use comparisons of household net worth at retirement to deduce valuations indirectly. This paper modifies a standard life—cycle model to include women’s labor supply decisions, estimates key parameters of the new specification, and attempts to assess the significance of rising female labor market participation for aggregate national saving in the U.S. Using panel data from the Health and Retirement Study, we find that the difference between measured labor market earnings for married women and earnings net of the value of lost home production seems moderately small – about 30 percent – and that the corresponding long—run effect on the overall rate of private saving is minor.
Michael Hurd and Susann Rohwedder
The simple one-good model of life-cycle consumption requires “consumption smoothing.” According to previous results based on partial spending and on synthetic panels, British and U.S. households apparently reduce consumption at retirement. The reduction cannot be explained by the simple one-good life-cycle model, so it has been referred to as the retirement-consumption puzzle. An interpretation is that at retirement individuals discover they have fewer economic resources than they had anticipated prior to retirement, and as a consequence reduce consumption. This interpretation challenges the life-cycle model where consumers are assumed to be forward looking. Using panel data on anticipated consumption changes at retirement and on recollected consumption changes following retirement, we find that the median recollected change in spending at retirement is zero and that the recollections are broadly consistent with anticipations. Based on a measure of total spending in true panel we find that the actual mean and median changes are slightly positive. Therefore, we find no retirement-consumption puzzle.
This paper examines the consumption response to the 1972 Social Security benefit increase. Nominal benefits were increased by 20 percent while annual cost of living adjustments (COLAs) were contemporaneously implemented and scheduled to begin in less than three years. Taken in isolation, this benefit increase could be viewed as a large and permanent increase in real Social Security benefits. However, the prevailing high rates of inflation that were the impetus for the COLA legislation may have caused households to view the permanent real benefit increase to be substantially less than 20 percent. Using data from the 1972-73 Survey of Consumer Expenditures, the results provide a mixed picture of the consumption impact of the benefit increase. Strictly nondurable consumption increases significantly at the time of the benefit increase. However, this increase does not persist. Furthermore, the likelihood of making any purchases from an array of durable good categories does not change throughout this period.
History of the Social Security Administration
The Social Security Administration provides a large volume of material relating to the historical development of social insurance and to the Social Security Act in particular. In the context of potential reform, it is interesting to consider the history of Social Security. A brief review is provided at http://www.ssa.gov/history/brief.html. This short essay suggests that as the U.S. began its evolution from an agrarian to an industrial society, the need to care for vulnerable citizens began to emerge as a policy issue: “… both the States and the Federal Government had begun to recognize that certain risks in an increasingly industrialized economy could best be met through a social insurance approach to public welfare. That is, the contributory financing of social insurance programs would ensure that protection was available as a matter of right as contrasted with a public assistance approach whereby only those persons in need would be eligible for benefits.” Text of the original Social Security Act and other supporting documentation is available as well at http://www.ssa.gov/history/law.html.
The MRRC recently asked researchers a series of questions about their work. In this issue Alan Gustman and Tom Steinmeier discussed their MRRC-related work conducted over the years.
How would you describe your MRRC-supported work?
For the past decade, we have used the data in the Health and Retirement Study (HRS) to examine how potential changes in Social Security and pensions may change retirement and saving patterns. A distinguishing feature of our research is the use of a model which allows individuals to have differing degrees of impatience regarding current vs. future consumption.
Thanks to the continuing support of MRRC over the past 7 years, we have been able to pursue a coherent research agenda that is most helpful and informative for policy makers and has allowed us to systematically develop and apply models required to fully understand how Social Security affects retirement and saving.
Our work has demonstrated the importance of having models that jointly explain saving and retirement, that allow for differences within the population in taste for saving -- thereby explaining the spike in retirement activity at age 62, that distinguish partial from full retirement, that allow interactions in the retirement decisions of couples and the reluctance of husbands to retire before their wives or if their wife is sick.
With such models—which can only be estimated because of the availability of the HRS—it is possible to simulate retirement effects of provisions under current law. We can also evaluate the likely effects of changes in Social Security policies, such as changes in the early retirement age, earnings test, delayed retirement credit, introduction of personal accounts, allowance for lump sum settlements, etc…
We also have been able to better understand the distributional implications of various policies, especially since SSA allowed Social Security records to be matched with HRS respondents who gave their permission. Providing this information to researchers on a confidential basis protects the respondent.
What are some highlights of your findings that are of greatest policy relevance for SSA right now?
First, changes in the early retirement age are likely to have significantly more impact on retirement than would changes in the normal retirement age. Currently about 15 to 20 percent of each cohort of individuals retires during their 62nd year. Some individuals simply have not saved enough to retire before Social Security eligibility. Others value highly the benefits that would be lost by working and tend to ignore the future benefit increases which would occur as a result, even though the future increases are roughly actuarially fair. In short, they view the lost benefits more as a tax, which discourages work. Our model indicates that, of those who currently retire at age 62, roughly two-fifths would delay retirement if the early retirement age were increased.
If the early retirement age were increased, it is possible that more individuals would apply for disability. However, the number of SSDI male recipients who enter the program at age 60 is only about 18,000, with similar numbers at other ages in the 56-61 age range. Even if the numbers at ages 62 and 63 were to rise to similar levels because of an increase in the early retirement age, it would simply be too small to change the results appreciably.
Second, eliminating the current earnings test would also reduce retirement, though by a smaller amount than increasing the early retirement age. Much of the impact would be that fewer individuals would be working part time and more would be working full time if the earnings test were repealed. However, many more individuals would be collecting benefits at ages 62 through 64, which would be a drain on the finances of the system. This would also reduce the eventual level of benefits that these individuals could collect during their retirement.
Third, increases in the early retirement reduction would also have significant effects. Changing the reduction to the level specified in the President’s Commission report would increase the percentage of 62-65 year olds working full-time by around three percentage points, which is a little less than eliminating the earnings test but nonetheless substantial. Proposals to increase the delayed retirement credit would have much less of an effect, primarily because the earnings test does not apply to individuals above the normal retirement age.
Fourth, potential lump sums can have a large effect on retirement. If lump sums are tied to retirement, individuals may retire earlier in order to gain access to the lump sums, especially if the lump sums are made available at age 62. If the lump sums are not tied to retirement, we can expect a nontrivial fraction of the population to withdraw and spend the lump sums, leaving less income and assets available to support retirement later on. To make matters worse, these are likely to be individuals who have accumulated few other assets for retirement.
Our related work has generated other punch lines relevant to the design of Social Security policies. Despite the progressive appearance of the benefit formula, there is little redistribution among families with different potential earnings levels. Also, high-income immigrant families with short tenures in the U.S. get exceptionally high returns from Social Security, but there is an easy fix for this perhaps unintentional result. Regarding the proposals of the President’s Commission, the method used to calculate benefits implies if there were another round of benefit reductions after having adopted the proposals, total benefit reductions would be the same whether a person had chosen a personal account or not. Finally, simulations of the total effects of the Commission proposals would be to reduce the percentage of retired 62 year olds by about four percentage points.
“Offsetting the Principal in the New Social Security Accounts.” Tax Notes. (April 2005).
“Retirement Effects of Proposals by the President’s Commission to Strengthen Social Security.” National Tax Journal. Vol. 58 No. 1 (March, 2005), pp. 27-49.
“The Social Security Early Retirement Age In A Structural Model of Retirement and Wealth.” Journal of Public Economics. Vol. 89, Issues 2-3 (February 2005), Pages 441-463.
“The Social Security Retirement Earnings Test, Retirement and Benefit Claiming.” National Bureau of Economic Research Working Paper 10905 (November 2004).
“Personal Accounts and Family Retirement.” National Bureau of Economics Working Paper 10305 (February 2004).
“How Effective Is Redistribution Under the Social Security Benefits Formula?” Journal of Public Economics. Vol. 82, No. 1 (October 2001), pp. 1-28.
“Social Security Benefits of Immigrants and U.S. Born.” In George Borjas, editor, Issues in the Economics of Immigration, University of Chicago Press (2000), pp. 309-350.
A new version of the RAND HRS Data File will be released in July, 2005. The RAND HRS Data file is a cleaned and easy-to-use version of the Health and Retirement Study (HRS) with derived variables covering a broad range of measures and named consistently across waves. The file was developed by the RAND Center for the Study of Aging with funding from the National Institute on Aging (NIA) and Social Security Administration (SSA). The Office of Research, Evaluation, and Statistics at SSA provided important research direction in the design of this data file.
As of spring 2005, eight HRS waves are available for study. The complete HRS includes four entry cohorts, as does the RAND file. The current version of the RAND HRS Data file (Version D) is based on 1992, 1993, 1994, 1995, 1996, 1998, and 2000 final releases and the 2002 early release. Version E will incorporate the final release of 2002 data, corrections to known problems, and several new variables, including an expansion of insurance indicators, and the addition of pension, retirement expectation, and poverty measures.
RAND HRS development is based on the RAND-enhanced Fat Files. These files store most raw HRS variables with household level data merged to the respondent level. They are easily merged with the RAND HRS Data File. The distribution set includes one file of raw variables for each HRS interview year and files for each year that include detailed components of income and asset imputations that are incorporated in the RAND HRS. Analysts who need to construct measures not included in the RAND HRS may find the Fat Files useful. NIA and SSA have also funded the development of these files. The Fat File for final 2002 data will be available along with those for prior interview years when RAND HRS Version E is released.
With NIA funding, RAND is also developing a longitudinal file of HRS children and grandchildren. A preliminary version of this file will be available by the end of June. This initial release links children (and grandchildren) of core respondents across interview years using the Longitudinal Other Person Number (LOPN) developed by HRS. A respondent-child file and a sub-household-child file will be distributed. These will provide limited information about the child and flags indicating the quality of the links, e.g., whether the sex, birth year, and relationship are consistent over time and whether the child is consistently present when this would be expected. They will also include the identification variables needed to merge with HRS child level files. In the initial distribution relatively few discrepant links are corrected but RAND expects to continue development of this file by improving the links.
Information on the distribution of these files will be posted on the Data Products page of the RAND Center for the Study of Aging website when the files are released.
With the weather finally warming to comfortable spring temperatures, 40 MRRC researchers gathered in Ann Arbor for a day and a half of lively intellectual exchange on May 6th and 7th.
Meeting in the gothic surrounds of the University League, the scholars
were invited to present past and present research findings as well as
ideas for future work. John Laitner and Susan Grad, Deputy Associate Commissioner
in the Office of Research, Evaluation, and Statistics in the SSA Office
of Policy, chaired a session on consumption, well-being, and retirement
that included presentations by Michael Hurd, Steve Haider, Kathleen McGarry,
Nicole Maestas, Chris House, Dmitri Stolyarov, Dan Silverman, Debra Dwyer,
Mel Stephens, and Frank Heiland.
MRRC Executive Committee member David Weir chaired a session on disability
and health that included presentations by Richard Burkhauser, Dean Lillard,
Robert Weathers, Todd Elder, and Arie Kapetyn.
Next there was a session on education and financial literacy that included presentations by Susann Rohwedder and Annamaria Lusardi. Susan Grad graciously agreed to be the lunchtime speaker, sharing with those in attendance information about data that SSA will soon be making available to the research community for the first time. John Laitner and Kent Smetters led off the afternoon session with their research on efficiency gains and losses in the context of Social Security reform.
The day ended with a session on data use and enhancement with presentations by Susan Rohwedder on the RAND HRS, Robert Willis on new HRS data, Frank Stafford on the PSID, Tom Juster and Dan Hill on the topic of correcting longitudinal data inconsistencies, and Mike Nolte on the current status of the MRRC’s attempt to create a virtual private network for scholars to gain remote access to SSA administrative data through the secure data facility at the University of Michigan.
University of Michigan
Institute for Social Research
426 Thompson Street
Ann Arbor, MI 48106-1248