Statistical results firmly refute the notion that the 2000 drop in retirement was linked to the stock market.
Conventional wisdom - backed by some statistical studies - holds that a plunging stock market causes workers to postpone retirement, and conversely that a booming market inspires early retirement. But in Bulls, Bears, and Retirement Behavior (NBER Working Paper No. 10779), authors Courtney Coile and Phillip Levine find no evidence that changes in the stock market drive aggregate trends in retirement behavior.
Coile and Levine analyze the relationship between stock market performance and retirement over the past two decades, giving special attention to the Wall Street boom years of 1995-9 and the bust period of 2000-2. In their study, they use data from three sources: the Health and Retirement Study (HRS), a longitudinal survey of near-retirement age households that began in 1992; the Current Population Survey (CPS), a monthly survey that forms the basis of most published U.S. labor statistics; and the Survey of Consumer Finances (SCF), a triennial survey focused on household wealth holdings.
The authors take advantage of what amounts to a double experiment in which groups with greater exposure to the stock market are predicted to be more likely to retire during the boom and less likely to retire during the bust relative to other groups. Their overall goal is to estimate the relationship between market fluctuations and aggregate changes in retirement, as opposed to estimating individual wealth effects.
Coile and Levine begin by documenting trends in retirement patterns and show that there was a large drop in retirement rates in 2000, the year the stock market fell precipitously. However, they point out that retirement rates were also lower during the boom years of the late 1990s than they had been in the early 1990s or 1980s. Then they present descriptive statistics detailing the stock market holdings of older households in the late 1990s. They show that relatively few older households had large stock holdings prior to the market drop in 2000 and that these households would have to have been extremely sensitive to stock losses for the stock market decline to have caused the observed drop in retirement rates
Next, the researchers compare the impact of stock market fluctuations on the retirement behavior of individuals who are likely to have been differentially affected by changes in the market, such as persons with and without defined contribution pension plans. Evidence supporting an impact of the stock market on retirement decisions would require that those who are more likely to own stock are more likely to retire in boom periods and less likely to retire in bust periods. The researchers further apply the same strategy to an analysis of labor force re-entry decisions. In both cases, they fail to find the expected pattern.
Coile and Levine conclude that their results do not necessarily contradict previous studies that have found wealth effects associated with changes in the stock market or other wealth shocks. Indeed, they say it is almost certain that some individuals experienced large drops in wealth because of the market bust and as a result postponed retirement. But Coile and Levine suspect that this is a fairly narrow segment of the population; as their analysis shows, most workers of near-retirement age have few if any stock assets. Therefore, they say, it seems unlikely that even a substantial labor supply response by that group could be driving changes in aggregate labor market trends.
Coile and Levine acknowledge that their results do not explain the drop in labor market activity among older workers in the bust year of 2000. They theorize that the drop may have been merely a realization of a longer-term decline in retirement among older workers. However, they maintain that their statistical results firmly refute the notion that the 2000 drop in retirement was linked to the stock market.
The researchers also assert that, despite their findings, stock market fluctuations are likely to have broader implications for individual's behavior and well-being. Recent retirees, workers nearing retirement, and workers further from retirement all may respond in any number of ways, such as changing their level of consumption, altering savings and investment activities, updating expectations about leaving a bequest, adjusting longer-term retirement plans, or reconsidering if and how long spouses should work.
Indeed, Coile and Levine believe that the drop in wealth resulting from the market bust in 2000 was likely reflected in changes in consumption and possibly in other behaviors to a much greater degree than changes in labor supply of near-retirement age workers or recent retirees. "The results of our analysis," write Coile and Levine, "provide little support for an impact of the boom and bust on retirement or labor force re-entry.
-- Matt Nesvisky