Abstract

Although partial retirement usually refers to workers voluntarily transitioning from full-time work to complete retirement, health or financial problems may be causing these changes. Using data from the Health and Retirement Study, I find that partial retirees are less healthy than similar full-time workers and report that their health constrains their ability to work. In addition, over 26 percent of partial retirees return to work after retirement. The decline in hourly wages for partial retirees who were previously retired is substantially worse than it is for partial retirees who transitioned directly from full-time work. Regression analysis indicates that divorced women and older workers with defined contribution pension plans have an increased probability of returning to work or postponing retirement.

Partial retirement, referring to the reduction in hours as workers age, has become a more widely recognized phenomenon by economists, but less attention has been paid to what causes people to partially retire. It is possible that partial retirement is truly a way of gradually reducing hours in transition to retirement. It is also possible that other workers are responding to financial or health problems. In addition, less interest has been given to examining the experience of partial retirees and other older workers who have returned to the labor force after being fully retired. Because partial retirement affects many older workers, it is important to examine possibly significant variations in their experiences.

Previous research has found that a considerable number of older workers partially retire. Gustman and Steinmeier (1984) estimate that approximately one third of older white men will be partially retired at some point in their lives. Quinn (2003) estimates that between one third and one half of older workers will be employed in a bridge job (defined as a job held after a career job and that is held less than ten years). Using data from the Health and Retirement Study (HRS), I find that 20 percent of older Americans who were working full-time in 1992 became partially retired at some point by 2002.

Though much of the current literature on partial retirement emphasizes the advantages it may offer, research has also shown that such jobs often have some drawbacks. Quinn (1996) finds that bridge jobs generally have lower wages and benefits. Additionally, bridge jobs usually represent a movement down the career ladder to less-skilled work. Aaronson and French (2001) find evidence that partial retirees earn significantly lower wages; they estimate that an older male worker who reduces his hours from forty to twenty per week will face an hourly wage penalty of 25 percent. Other research addresses the labor demand for older workers. Using an employer survey along with data from the HRS, Hutchens and Chen (2003) conclude that the availability of partial retirement may not be as high as it seems because employers are much less willing to offer partial retirement if it involves flexible scheduling, such as job sharing, and includes benefits.

A recent report by the Government Accountability Office (GAO) also finds that many older workers do not believe they have opportunities to partially retire with their current employer (GAO 2005). In focus groups of older workers and retirees, very few indicated their employer would offer them an opportunity to gradually or partially retire. Many of the retired focus group participants felt their employment opportunities were limited to low-wage, low-skilled jobs. Although policy makers often advocate partial retirement as one possible solution to the fiscal pressures of an aging population (Burtless and Quinn 2002), this research suggests there are key questions left unanswered as to the quality and quantity of jobs that older workers may find and obtain.

How and Why Do Older Workers Partially Retire?

The literature on partial retirement characterizes the phenomenon as voluntary transitions made by older workers because they want to work less but are not ready for full retirement. For many partial retirees, this depiction appears to be accurate—data from the first six waves of the HRS reveal that 33.57 percent were working full-time in the period immediately prior to par-tial retirement. However, a significant number did not enter partial retirement through the traditional route. Over 26 percent of partial retirees actually returned to work from full retirement. Another almost 14 percent were either unemployed, disabled, or not in the labor force before they were partially retired. For these groups of older workers, it is possible that partial retirement represents adverse employment or underemployment.

Aside from those who were once fully retired, other workers may also be partially retired for adverse reasons. The loss of a job or health problems could cause older workers to reduce their hours when they would otherwise have preferred to work full-time. Among partial retirees who changed employers, 15 percent report having left their job involuntarily—their route to partial retirement began with a layoff or a business closure. Almost 55 percent of partial retirees who changed jobs report leaving their last job because of retirement. They may have changed employers in order to access a defined benefit pension or because of an inability to reduce hours at their career job. Alternatively, partial retirees may change employers to reduce stress or increase job satisfaction by pursuing a prior passion (Raskin and Gettas 2003). Thus, most partial retirees who have changed employers report doing so for voluntary reasons; however, a significant number do so involuntarily, and this group has not been adequately addressed in the literature.

The economic situation among partial retirees is starkly different based on the nature of their transition. Sharply lower hourly wages of partial retirees who have gone back to work after retirement would suggest that these workers may have been pushed into this transition. Upon retirement individuals should have higher nonlabor income than before and, thus, a higher reservation wage. To examine their change in wages, I compare the individual's last full-time hourly wage to the hourly wage earned in partial retirement. I find that partial retirees who were fully retired face a much larger wage penalty than those who transitioned directly from full-time work. Comparing median hourly wages, older men who returned to work from retirement experienced a decline of $6.78 in their hourly wage—a drop of 39.51 percent from their last reported full-time wage, when evaluated at the median. Older men who went directly from full-time work to partial retirement, however, earned $0.46 less per hour, which represents a 2.81 percent decline in hourly wages, evaluated at the median. The situation was similar for older women. Women who returned to work from retirement experienced a 17.44 percent drop in hourly wages, compared to an 8.68 percent drop for women who transitioned directly from full-time work to partial retirement.

Negative financial shocks may also cause older Americans to return to work or postpone retirement. In particular, the substantial decline in the stock market in 2000 may have had profound effects on retirement decisions. As seen in Table 1, the average change in the reported value of stocks held by those aged 6065 was sharply negative in 2002. It is important to note, though, that 68 percent of people in this age group do not report owning stocks in the HRS. However, the vast majority of individuals aged 6065 report having financial assets (not including housing wealth), and the average change for this measure was negative in both 2000 and 2002.

Alternatively, older workers may be forced into partial retirement because they have health problems that interfere with holding a full-time job. A drop in productivity due to poor health may explain some of the decline in hourly wages. In order to test if partial retirees have more health problems than full-time workers, I use probit regressions to control for factors such as age, education, and race. I use detailed measures provided by the HRS covering a variety of health problems. The regression results indicate male partial retirees are significantly more likely to have health problems in almost every category (see Table 2). Compared to men who are working full-time, men who are partially retired are more likely to have a problem with mobility and large muscle activity. Partially retired men are also more likely to have problems with gross and fine motors skills and Activities of Daily Living (ADLs). Women who are partially retired are more likely to have problems with mobility/large muscle activity as well as gross and fine motor skills than their full-time counterparts. Both male and female partial retirees are significantly more likely to report that health limits their ability to work. The likelihood that health problems constrain the ability to work increases by 14 and 12 percentage points for partially retired men and women, respectively. Honig and Hanoch (1985) found similar evidence of health constraints on full-time work among partial retirees in their study using data from the Retirement History Survey (RHS). The evidence of significantly poorer health among partial retirees may imply they have reduced

 

productivity due to health limitations, which could be one factor contributing to lower wages.

Factors that Affect Working Later in Life

In order to identify factors that cause people to return to work from retirement, I use the following probit model:

value of savings, and out-of-pocket medical expenditures. The Income_it variable is a composite measure of a change in income received from a variety of sources including pension, annuity, alimony, inheritance, rental, dividend, interest, Social Security Disability Insurance (SSDI), Supplemental Security Income (SSI), Social Security retirement, unemployment insurance, worker's compensation, and other government transfers.

Negative financial or health shocks may also occur while an individual is still working and cause him/her to postpone retirement. Therefore, I perform a similar regression analysis to explain factors that influence whether or not an individual works past his/her planned retirement age. The explanatory variables are the same, except that the indicator for whether or not his/her last job ended involuntarily is not included.

Data

I use data from the first seven waves of the HRS, spanning from 1992 to 2004, and the RAND contribution file to estimate the regression in equation (1). The sample consists of individuals born between 1924 and 1947. The dependent variable indicates if there is a change in labor force status from one wave to the next (that is, from time period t to t+1). To ensure that the explanatory variables are capturing the individual's status prior to this change, I use demographic data from time period t and change in health, financial, and income variables from time period t-1 to t.

Regression Results

So, what factors cause older men and women to stay retired or reenter the labor force? Regression analysis indicates that women are significantly more likely to go back to work if they are divorced, which may represent an unmeasured negative financial shock. African American women also have a higher probability of returning to work, as do women who have a spouse still in the workforce. Male retirees who experience negative heath changes or who did not finish high school are less likely to return to work, however, possibly because their employment opportunities are limited to low-wage or physically demanding jobs.

There is some evidence that a decline in savings increases the likelihood of returning to work. However, the results for the change in IRA and stock wealth variables have the predicted negative sign but were not statistically significant. These results are consistent with the findings of Maestas (2004), which suggests that returning to work is not primarily a response to financial shocks. Additional evidence from Kezdi and Sevak (2004) reveals that retirement is an absorbing state—retirees are more likely to reduce consumption rather than return to the workforce—even in response to financial losses sustained during the recent stock market downturn. Kezdi and Sevak suggest that once a person retires it is very difficult to reenter the workforce. This view concurs with statements of retirees participating in focus groups about work and retirement (GAO 2005). Many of these retirees reported their employment opportunities were limited to low-wage, low-skilled jobs.

If retirement is largely an absorbing state, it may be more pertinent to examine factors that cause older workers to continue working past their expected retirement date. According to HRS data, almost 38 percent of older Americans postpone retirement—40 percent of women and 35 percent of men work longer than they planned.

Regression analysis indicates pension type and retiree health insurance have significant effects on the probability of postponing retirement. The presence of a defined contribution pension plan increases the likelihood individuals will continue to work after their expected retirement age by almost 8 percentage points for women and 10 percentage points for men. Because defined contribution pensions, such as 401(k)s, have much lower average and median values than defined benefit pensions, older workers may need to stay in the labor force because they cannot afford to retire. Defined contribution plans also do not have age-specific retirement incentives, so older workers can continue working without diminishing the value of their lifetime pension benefits.

Retiree health insurance, however, reduces the probability of postponing retirement by about 11 percentage points for older male workers. Particularly in the face of rapidly rising health care costs, having retiree health insurance protects an individual from having to work longer to finance health care consumption.

For women, being divorced increases the likelihood of staying in the workforce. Divorced women are almost 11 percentage points more likely to postpone retirement. Similarly, Munnell and Jivan (2005) find that divorce significantly increases the probability of women aged 5161 being employed. As they explain, divorce is a negative financial shock due to the loss of the husband's earnings. (See the Appendix Tables 1 and 2 for additional regression results.)

Conclusions and Policy Implications

Based on this research, policy makers should consider often significant drawbacks present in partial retirement jobs. In light of these drawbacks, policy makers could work on initiatives to improve the ability of older workers to attain higher-quality jobs. Such policies include job training for older workers, more effective age discrimination laws, and improving the jobs themselves by giving employers incentives to provide better wages and benefits in the form of pension and health care coverage. Alternatively, policy initiatives could focus on improving the financial security of older Americans, particularly for vulnerable populations such as divorced women and older workers with only defined contribution pension plans, so they are able to have a sustainable retirement.

Note

Sharon Hermes is an economist with the U.S. Government Accountability Office (GAO). The views of the author do not necessarily reflct those of the GAO. The research contained in this paper is an extension of dissertation research by the author.