bR I E f # 35 juLy 2012 Program on Retirement Policy www.urban.org InSIdE THIS ISSuE •If labor force participation increases more than projected, shortfalls in systems like Social Security will be smaller. •Almost all formal models predicting labor force participation among older workers fail to account for the importance of labor demand. •Older workers are the largest underused source of labor and human capital in the economy. Correcting Labor Supply Projections for Older Workers Could Help Social Security and Economic Reform C. Eugene Steuerle and Caleb Quakenbush Changes in life expectancies, birth rates, and health care (among other conditions) affect the employment of older workers. Older workers’ employment, in turn, has wide-ranging implications for the sustainability of the nation’s entitlement programs and the broader federal budget, as well as the future economic growth of both the elderly and the nation. Will future workers continue to spend longer portions of their lives in retirement? Or will the recent shift from declining to increasing labor force participation among older workers continue, or even accelerate—and not just during recessions? T he answer has enormous implications for how policy reforms can or should be designed. More years of work by the population produces more income, higher private saving as people draw down their assets over fewer years, fewer years of dependence on government, and higher Social Security, Medicare, and income tax revenue that can support higher annual and lifetime benefits at any tax rate (Butrica, Smith, and Steuerle 2006). Yet, the Congressional Budget Office (2012) has only recently started to show how policy changes, such as increases in the retirement age, could increase personal income and revenues outside Social Security.1 Whatever the cause, if labor force participation increases more than projected, shortfalls in systems like Social Security will be smaller. Policies that lean with rather than against the wind could accelerate or enhance that trend. This brief first points out some fundamental flaws in both theory and empirical work surrounding labor supply. Although it is theoretically possible for workers to supply more or less work over time, these flaws lead to biased estimates that over-predict the share of life spent in retirement and underpredict the future labor force participation of older workers. Next, the brief reviews some recent trends in both actual and predicted labor force participation by the Social Security Administration. Finally, it highlights how estimates associated with Social Security reform proposals traditionally have accounted inadequately for employment Both public and private policy must change to allow increased employment among older workers to blossom further. Correcting Labor Supply Projections for Older Workers Could Help Social Security and Economic Reform effects (regardless of their size). Avoiding these shortcomings would mean that any balanced system reform would produce higher benefit rates at any tax rate or lower tax rates at any benefit rate. All in all, there is a good reason to be optimistic—at least on this front—about the future of the economy. However, both public and private policy must change to allow increased employment among older workers to blossom further. Leisure and Retirement years: not the Same Thing Economists often view leisure as a superior good: that is, as the economy grows richer, people demand more time in activities outside the formal labor market. Economists then apply this view to retirement, using it to help explain the large growth in retirement years in countries across the globe in the 20th century. As real incomes rise, they argue, these larger lifetime incomes (the income effect) tend to dominate the higher wages made possible by working later in life (the substitution effect), and workers tend to retire earlier. Costa (1999) is among the best known of those who reviewed this evidence. According to her, increased incomes from private and public pensions, decreased travel and communications costs, and a burgeoning market targeted at retirees increased workers’ real wealth. It established retirement as a social norm, making withdrawal from the labor force appealing at increasingly younger ages. Costa cautioned that the small uptick in labor force participation rates just gaining notice in the 1990s could be a temporary deviation from trend, not a reversal of retirement-age declines. In other words, the jury was still out on whether the reversal would be permanent. Costa was not alone. At about the same time, the Social Security actuaries projected continually lower or stagnating labor partici- pation rates of older workers far into the future. Some technical panels of economists, demographers, actuaries, and other professionals advising Social Security actuaries on their long-range projection methodologies also shared this point of view. The 2003 Technical Panel noted that, “given the projection of a 127 percent increase in income [over the projection horizon], it seems inadvisable to assume increases in labor force participation rates for older persons. Such a reversal in the trend toward earlier retirement has never been seen in the historical record of any country.” In March 2001, Steuerle and Carasso predicted that just the opposite would happen. In their view, the classic labor-leisure distinction in economics oversimplifies how workers approach their decisions to participate in the labor force. In today’s economy, where work for most is less physically strenuous than it was for previous generations, workers may actually find benefits in work similar to those hypothesized for leisure (Steuerle, Spiro, and Johnson 1999). At the same time, the demand for a better life can play out in different ways than increasing retirement years from, say, 20 to 25. There may be increased demand for part-time or bridge jobs. Rising real incomes and shifting attitudes about work-life balance could bring about demand for shorter workweeks, longer vacations, or more parental leave. The limitations of the simple leisure theory derive from economic simplifications that are often useful but sometimes fail to capture real-world nuances. In truth, leisure is often put in economic models to complete them mathematically. Work is assumed—not proven—to provide negative value or utility, so people trade off the “bad” from working for the “good” from working—a paycheck—until those bad consequences are marginally more costly in absolute terms than the pay people receive (Steuerle 2007). Sociologists and psychologists would scoff at this type of simplification, trusting more in what a person says gives him or her a sense of well-being. They would be much more likely to recognize that labor force participation can be a source of social and intellectual fulfillment, particularly if older workers have more flexibility in when and where they work. The simplified economic theory is still useful if work is defined as activity outside the home that provides decreasing and, eventually, negative utility at some margin (for example, increasing work from 8 to 9 hours a day). However, it is unnecessary to assume that increasing labor force participation from 0 to 4 hours a day provides negative value (absent pay) to everyone or almost everyone. If that were true, for instance, there would be no voluntary labor for charitable organizations. Yet, many older individuals volunteer or continue to volunteer as they enter retirement (Zedlewski 2007). Workers as One-Hoss Shays The notion of complete retirement from the workforce is actually a very modern one. Until recently, most of the population worldwide was primarily agrarian. Work life revolved around the farm; the hours were long, and little if anything was available in the way of financial saving for retirement. Older farm workers did not sit back while the rest of the family did all the work. Rather, they pitched in with what they could, taking account of their depreciated skills. With the growth of an industrial economy putting huge stress on manual labor, many people wore out and needed to retire from those strenuous jobs. This led to the development of retirement plans that often treated workers like “one-hoss shays” (horse-drawn buggies) that worked perfectly, then suddenly collapsed and depreciated to uselessness. Social Security started and continues with 2. Correcting Labor Supply Projections for Older Workers Could Help Social Security and Economic Reform figure 1. Labor force Participation Rates by Age Group (percentage of age group population, both sexes) 80 Ages 16–24 70 60 Ages 55–64 50 40 30 20 Ages 65+ 10 0 1948 1958 1968 1978 1988 1998 2008 Source: Bureau of Labor Statistics. this model of human capital depreciation in mind. For instance, it does not offer partial retirement options, though some people might effectively take them indirectly.. The one-hoss shay stereotype never represented the diversity of work and family experiences, but today it relates even more poorly to how individuals’ skills and human capital evolve. To make matters even more complicated, most workers still retire at age 63 or 64. If they were to retire today for the same number of years as when Social Security benefits were first paid, they would be retiring at age 75, and in another 50 years or so at age 80. Workers in 1940 and 1950 retired at age 68 on average; today the combination of earlier retirement and longer lives gives workers more than 10 additional years of retirement than workers in 1940. That any demand for leisure—no matter how superior a good— would evolve this way makes little sense unless the availability of Social Security benefits at age 65, then 62, is taken into account. Labor demand: Missing from the Predictive Models While no one yet knows how people will adjust to increasingly higher lifetime earnings, almost all formal models predicting labor force participation among older workers— including those used by the Social Security Administration—contain a fundamental flaw: they fail to account for the importance of labor demand on labor force participation. Employment at a point in time measures not labor supply, but the intersection of labor supply and labor demand. The models, however, ignore demand and project future labor supply of different age groups by following their employment trends over time, almost as if each group were totally independent of the others. 3. Correcting Labor Supply Projections for Older Workers Could Help Social Security and Economic Reform Older workers are to the first half of the 21st century what women were to the last half of the 20th: the largest underused source of labor and human capital in the economy. As an extreme example of how such projection fails to account for labor demand, consider a population where the birth rate falls to zero and everyone will eventually be age 65 or older. In this case, it is unlikely that the future labor force participation of those older than 65 would in any way parallel that of older workers in an earlier population with many younger people. As long as citizens demand goods and services, businesses will seek out the labor necessary to satisfy that underlying demand. This example shows the danger of extrapolating labor force participation rates of different age groups without considering the number of potential workers relative to the population. Take an extreme case where labor demand is inelastic—that is, we need about the same proportion of the population to provide us with the goods and services we want. When the decline in work among older workers can no longer be offset by an increase among younger workers, the changed age distribution will dramatically affect the demand for workers at different ages and the trend lines of specific age groups.2 What allowed that demand to be filled in the last half of the 20th century, even while longer-term workers retired earlier and earlier? Mainly the introduction of women and baby boomers into the workforce. In fact, the share of adults working increased almost every non-recessionary year over that halfcentury. Thus, where labor demand existed, it could easily be met by the many additional productive, more highly educated entrants. As the entrance of new women and young workers slows, employers are likely to train their hiring sights on older workers and the vast pool of talent they represent. Adjusting the labor force of different age groups by education levels produces yet another explanation for why past trends have slowed or reversed themselves: new labor force entrants no longer demonstrate average skill levels well above those at or near potential retirement ages. College graduation rates of those now entering the workforce are no longer significantly above those of the workers they replace. These changes, too, increase the relative demand for older workers. Today’s older workers possess qualities that employers will increasingly seek out: their education level matches that of younger workers, they have lifetimes of experience, and they are healthier than their predecessors were at the same age (Munnell 2006). Trends in Actual and Predicted Labor force Participation So far we have mainly discussed theory. Now let’s look at the data. Overall labor force participation increased over the last half of the 20th century. While it has been trending downward recently as the economy suffered, it is by no means at record lows. Who has been driving this recent decline? Younger workers, who are now more likely to remain out of the labor force while pursuing education or to take more time to engage the workforce more fully. In today’s unfriendly job environment, more young adults may feel forced into or opt for more schooling or other non-labor alternatives in the face of poor employment prospects. Meanwhile, among citizens age 55 and older, the trend of declining participation has largely reversed. Participation among this group has increased from its low in the early ’90s and now slightly surpasses its 1980s level (figure 1). This is not just the result of a continued increase in participation among women: among men age 55 and older, labor force participation fell from 45.6 percent in 1980 to 37.7 percent in 1993, but it has since climbed to 46.3 percent in 2011 (BLS, not shown). Additionally, while the Great Recession saw a short uptick in new Social Security retirement claims among eligible workers, retirement benefit uptake has since fallen to its lowest level since 1976, and a decade-long trend in delaying retirement benefits may have resumed (Johnson 2012). Social Security’s Predictions of Labor force Participation by Cohort In developing its annual OASDI Trustees Reports, the Social Security actuary attempts to project future labor force participation rates among the working-age population. While specific methodology is refined over time, employment statistics used in the 4. Correcting Labor Supply Projections for Older Workers Could Help Social Security and Economic Reform figure 2. Trustees’ Assumptions versus Actual Participation Rates (men, age 65–69) 45 2012 SSA projections by year 2009 40 2005 35 BLS historical data 30 2000 25 1992 1986 20 1987 15 1985 1995 2005 2015 2025 2035 2045 2055 2065 2075 2085 Sources: Bureau of Labor Statistics and authors’ compilations of data from Social Security actuary. 5. Correcting Labor Supply Projections for Older Workers Could Help Social Security and Economic Reform reports rely on historical Current Population Survey (CPS) data provided by the Bureau of Labor Statistics (SSA Chief Actuary 2011). These data are disaggregated by age and sex to capture effects of specific demographic trends on various sections of the population. Variables that influence SSA’s projections include the effect of early and normal retirement ages, marriage rates, divorce rates, spousal labor force participation, disability prevalence, and unemployment rates. Published projections of labor force participation by the Social Security actuary have painted sometimes grim and at best modest outlooks for participation rates among older workers, but how well have these matched reality? Figure 2 plots labor force participation projections of previous Trustees Reports against BLS-reported rates among 65- to 69year-old men since 1985. In this period, actuals have almost continuously surpassed projections, with projections almost always implying a flattening of the overall upward trend that does not occur. Calculations for older groups generally provide a similar tale. The one exception is the group just before Social Security eligibility age (55–60). This group’s prevalence rate of disability insurance receipt is at least partially related, an issue we do not pursue further here. All this is not to question the utility of using historic trends in forecasting, but to point out that the future does not always resemble the present. As seen in figure 2, Social Security allows for some change from current levels, but these projections almost always imply a relatively flat trend line outward from the year of projection. Projections used to head slightly downward, but more recently they head slightly upward. The 2012 Trustees Report projects slightly higher overall participation rates among men and women over age 16, citing in part higher life expectancies.3 The 2011 Social Security Technical Panel recommended that the actuaries take more account of the factors likely to affect labor supply—for instance, the shift from defined benefit to defined contribution pensions in the private sector. Workers with defined contribution plans stay in the workforce longer since they tend to have less security that their pensions will last (Johnson and Steuerle 2003). However, the panel’s recommendations only went halfway: they recognized modern influences on labor supply but still failed to note the flaw that derives from ignoring labor demand. The failure of Policy Reform to Take Advantage of the Skills and Talents of Older Workers Failure to fully recognize the shifts occurring in labor force participation among older workers creates a blind spot for policymakers exploring options for much-needed reform in Social Security and labor policy. Policy has come a long way in removing some institutional barriers that discourage work, but much more can be done. For example, Social Security reform could make it easier for workers to take a partial retirement benefit while continuing to work part time, or it could allow workers who defer retirement to purchase a higher Social Security annuity more easily than the today’s clumsy arrangement of earnings tests and delayed retirement credits (Steuerle 2010). Greater labor force participation at older ages aids not just in the solvency of our entitlement programs, but increases personal income and general revenues as well. At the same time, if low-earning workers continue to be encouraged to retire earlier and earlier relative to life expectancy, while higher-earning workers remain in the workforce longer, current policies could lead to increased inequality in the incomes of the old. Reform should therefore consider other policies, even independently of any shortfall in trust fund balances. These include raising the earliest retirement age, which now signals that people are old when they are 62 and on average have very long remaining life expectancies, and increasing minimum benefits for workers with low lifetime earnings. Older workers are to the first half of the 21st century what women were to the last half of the 20th: the largest underused source of labor and human capital in the economy. Few policymakers are taking this extraordinary possibility into account in their reform packages. In an increasingly informationand service-based economy, older workers represent a valuable source of knowledge and experience that employers will tap, especially if Social Security and related economic reforms attempt to channel rather than obstruct these forces. • Acknowledgments The authors are indebted to the Social Security Office of the Actuary for their very generous assistance in providing both past and current projection data. They are also grateful to Richard Johnson for valuable insight and comments. 6. Correcting Labor Supply Projections for Older Workers Could Help Social Security and Economic Reform notes References 1. When it comes to the budget, for instance, the Social Security actuaries and the Congressional Budget Office more traditionally have only reported on changes on Social Security trust funds, not on what is often a larger and more important effect on income taxes. We had been encouraging both offices for some time to start showing these other economic effects. Butrica, Barbara A., Karen E. Smith, and C. Eugene Steuerle. 2006. “Working for a Good Retirement.” Retirement Project Discussion Paper 3. Washington, DC: The Urban Institute. 2. Technically, models typically calculate trends for specific age/sex groups and then derive a population estimate by taking those trends and reweighting by the relatives size of different age/sex cohorts over time. As shown in the anecdotal examples above, this methodology mistakenly assumes that the changing weights themselves do not affect the trend lines for the various age/sex groups. It ignores, for instance, how labor demand could more easily allow early retirement when the birth rate was higher or when women increasingly entered the workforce. 3. In other reports, we have suggested that projections over time should also be made on the basis of remaining life expectancy and relative life expectancy, which are often better measures of people at equivalent levels of health over the decades than is chronological age. Note that the trend toward higher labor force participation would be lower if measured by equivalent life expectancy rather than equivalent chronological age, an issue we do not pursue further here. See Cushing-Daniels and Steuerle (2007) and Steuerle and Spiro (2010). Congressional Budget Office. 2012. “Raising the Retirement Ages for Medicare and Social Security.” Issue brief. Washington, DC: Congressional Budget Office. Costa, Dora L. 1999. “Has the Trend Toward Early Retirement Reversed?” Paper presented at the First Annual Joint Conference for the Retirement Research Consortium, Washington, D.C., May 20–21. Cushing-Daniels, Brendan, and C. Eugene Steuerle. 2007. “Retirement and Social Security: A Time Series Approach Based on Remaining Life Expectancy.” Washington, DC: The Urban Institute. Johnson, Richard W. 2012. “Social Security Claims Edged Down in 2011.” Retirement Security Data Brief 5. Washington, DC: The Urban Institute. Steuerle, C. Eugene, and Adam Carasso. 2001. “A Prediction: Older Individuals Will Work More in the Future.” Straight Talk on Social Security and Retirement Policy 32. Washington, DC: The Urban Institute. Steuerle, C. Eugene, and Christopher Spiro. 1999. “Adjusting for Life Expectancy in Measures of Labor Force Participation.” Straight Talk on Social Security and Retirement Policy 10. Washington, DC: The Urban Institute. Steuerle, C. Eugene, Christopher Spiro and Richard W. Johnson. 1999. “Can Americans Work Longer?” Straight Talk on Social Security and Retirement Policy 5. Washington, DC: The Urban Institute. Technical Panel on Assumptions and Methods. 2003. Report to the Social Security Advisory Board, Washington, D.C., October. ———. 2011. Report to the Social Security Advisory Board, Washington, D.C., September. Johnson, Richard W., and C. Eugene Steuerle. 2003. “Promoting Work at Older Ages: The Role of Hybrid Pension Plans in an Aging Population.” Working Paper 2003-26. Philadelphia, PA: Pension Research Council. U.S. Social Security Administration, Board of Trustees. 2012. The 2012 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Federal Disability Insurance Trust Funds. Washington, DC: U.S. Social Security Administration. Munnell, Alicia H. 2006. “Policies to Promote Labor Force Participation of Older People.” Working Paper 2006-2. Chestnut Hill, MA: Center for Retirement Research at Boston College. U.S. Social Security Administration, Office of the Chief Actuary. 2011. Long-Range OASDI Projection Methodology. Washington, DC: U.S. Social Security Administration. Steuerle, C. Eugene. 2007. “The Seven Deadly Sins in Aging Policy and Research: A Cautionary List for Policymakers and Prognosticators.” In Labor Supply in the New Century, June 2007, edited by Katharine Bradbury, Christopher L. Foote, and Robert K. Triest. Boston, MA: Federal Reserve Bank of Boston. Zedlewski, Sheila R. 2007. “Will Retiring Boomers Form a New Army of Volunteers?” Perspectives on Productive Aging Brief 7. Washington, DC: The Urban Institute. ———. 2010. “How Social Security Can Costlessly Offset Declines in Private Pension Protection.” The Government We Deserve opinion column. Washington, DC: The Urban Institute. 7. About the Authors Program on Retirement Policy C. Eugene Steuerle is an Institute fellow and Richard B. Fisher Chair at the Urban Institute. http://www.retirementpolicy.org Caleb Quakenbush is a research assistant at the Urban Institute. The Program on Retirement Policy addresses how current and proposed retirement policies, demographic trends, and private-sector practices affect the well-being of older individuals, the economy, and government budgets. Copyright © July 2012 The views expressed are those of the authors and do not necessarily reflect those of the Urban Institute, its trustees, or its funders. Permission is granted for reproduction of this document, with attribution to the Urban Institute. uRbAn InSTITuTE 2100 M Street, nW ● Washington, dC 20037-1231 (202) 833-7200 ● paffairs@urban.org ● www.urban.org 8.