Decision Making Retirement

For an understanding of why workers retire when they do, it is necessary to follow people’s work experiences as they approach and enter retirement. This is possible only with longitudinal surveys, in which the same people are interviewed over time. One commonly used source of this kind of information is the Health and Retirement Study (HRS), conducted by the University of Michigan, in which a sample of men and women between fifty-one and sixty-one years of age were interviewed in 1992 and then reinterviewed at two-year intervals. Much of the recent research on retirement relies on these data (information on the HRS is available on line at www.umich.edu/~hrswww/).

Longitudinal surveys have found that many workers do not go from full-time employment to not working at all, but rather to what some researchers have called partial retirement or bridge jobs which are part-time or short-term jobs. Part-time employment is particularly common among workers age sixty-five and older. Among workers age fifty-five to sixty-four, 83 percent work full-time; among workers age sixty-five and older, only 49 percent work full-time. An increase in workers aged sixty-five or older who have been at their current jobs less than one year is further evidence of the prevalence of bridge jobs. From 1987 to 1998, the proportion of workers age sixty-five and older who had been at their jobs less than one year increased from 10 percent to 16 percent. While some of these workers may be continuing their careers, albeit with different employers, it is likely that many are employed in bridge jobs.

There is evidence that women are more likely to go through this intermediate bridge stage of retirement. One study estimated that 40 percent of women (compared with 25 percent of men) hold a bridge job before retiring completely (see Quinn and Kosy, 1996). Defining partial retirement is again an issue. Because women are more likely to work part-time at any stage of life, they are more likely than men to be considered partially retired, even though they may not have reduced their work hours. However, when asked in surveys, more men than women described themselves as partially retired.

An added complication is that some people who are considered retired at one point in time may later return to work. Over 10 percent of women and 12 percent of men who were not working and considered themselves retired at the first HRS interview were employed two years later; an additional 1 to 2 percent were looking for work. An earlier study using a sample of workers who began receiving Social Security benefits in 1980 found that about 20 percent of both women and men were working when interviewed about two years later (see Iams 1986). Some of these workers may have been partially retired when they began receiving Social Security benefits, but others may have stopped working entirely and started working again when they found a good opportunity or felt they needed more income after they had stopped working.

A great deal of research on the factors that influence workers’ decisions to retire has focused on the incentive effects of Social Security and employer-provided pensions. Economic theory posits that workers will decide whether an additional year of work adds enough to their retirement income to balance the loss of a year of retirement leisure. When a worker becomes eligible for a pension or Social Security, the added earnings from another year of work may seem less rewarding. Evidence on retirement timing shows that the peak age of retirement for both men and women is sixty-two, when workers first become eligible for a retired worker’s benefit from Social Security. A second cluster forms at age sixty-five, when full benefits for many pensions, as well as Social Security, become available. Women are slightly less likely than men to retire at these peak ages. These results hold up whether using labor-force participation or self-definition as the measure of retirement.

Some workers can gain substantially higher Social Security or other pension benefits by continuing to work past these peak retirement ages. For women especially, determining whether continuing to work is advantageous can be complicated under Social Security rules. Women can only gain from additional work if Social Security benefits based on their own work records will exceed those from their status as wives or widows. Women who are married are entitled to either a benefit based on their own work record, or to a spouse benefit, which is equal to 50 percent of their husband’s benefit. If their own benefit is at or above the level of the spouse benefit, they usually gain from additional work, especially if they have been employed for less than the thirty-five years used to calculate benefits.

Based on the Health and Retirement Study, researchers estimate that about two-thirds of women will receive a benefit based on their own work record. However, a widow or widower may receive 100 percent of the deceased spouse’s benefit or a benefit based on their own work record, whichever is higher. Only 15 percent of these women had an earnings record equal to or higher than their husbands’ earning record, and thus 85 percent of women will receive a widow’s benefit based on their husband’s earning record if they outlive their spouse. The situation of divorced women is more complicated. Women must have been married for at least ten consecutive years to the same husband to be eligible for dependent benefits. If eligible, they are entitled to benefits under the same rules as wives when their former husbands are still living, and as widows after their husbands have died.

For employer-sponsored pensions, there is considerable variation in the incentives to retire. Many pension plans depend on years of service and earnings averaged over a stated period of years. These defined benefit plans are commonly offered by government, unions, and large private employers. These plans can encourage early retirement by failing to adjust benefits upward if they are taken at older ages, or by not counting years worked beyond some maximum number in calculating benefits. One study estimated that one-quarter of the decline in men’s labor-force participation between 1950 and 1985 could be explained by the extension of pension coverage (Samwick, 1998, cited in Quinn, 1999). Women are less likely than men to face such limits. In fact, because women often take time out of the labor force, many women need to work into their sixties and seventies to qualify for a pension.

However, employers are increasingly offering defined contribution pension plans, in which benefits are based on contributions made by employers and/or workers to an individual account. When the employee retires, benefits are calculated as the sum of all contributions plus interest, dividends, and capital gains (or losses). Between 1975 and 1997, the share of pension plan participants in defined contribution plans rose from 13 percent to 42 percent. Defined contribution plans do not appear to have the same disincentive effects as defined benefit plans. To the extent that defined contribution pensions continue to grow in popularity (relative to defined benefit plans), eligibility for private pensions will diminish as a factor affecting retirement decisions.

Health is another influence on the decision to retire for both women and men, but there is considerable dispute about its relative importance. Some health conditions clearly prevent work entirely, but many conditions, while making work more difficult, do not completely preclude it. For example, common aliments like arthritis may make older workers less employable long before they qualify for private or public disability benefits. In these cases, health status and the availability of retirement income may interact, decreasing the monetary gains from continuing employment while making work itself more burdensome and leisure more appealing. It has often been thought that health and disability problems at older ages primarily affect men in blue-collar jobs. However, some pink-collar jobs held by women, such as waitressing, practical nursing, and childcare, are also difficult for older people with even minor health problems.

The state of the economy has been shown to have an impact on men’s retirement decisions, and it probably influences women as well. Much of the research on retirement decision-making was undertaken in the 1970s and 1980s, which saw periods of relatively high unemployment. Older workers who lose their jobs tend to remain unemployed longer than younger workers, and they are also much more likely than others their age to retire. Employers who are downsizing their workforces often offer special retirement incentives to their older and (generally most costly) workers. In periods of rapid growth, of course, new jobs are much easier to find for those who lose their jobs. When employers find it difficult to fill vacancies, they may actively recruit older workers who would not previously have been considered. Probably part of the reason that labor-force participation of workers over age fifty-five was increasing in the 1990s was due to economic growth and an increasing demand for workers of all ages.

Marital status and the retirement decisions of a spouse can also influence the timing of retirement. Some of the earliest research on women’s retirement focused on unmarried women and tended to find few differences between these women and men. When researchers first began studying married women, they had to confront the possibility that couples might try to coordinate their retirement in order to enjoy leisure time together. Assuming that the husband was the primary earner, researchers expected the wife’s retirement to be influenced by that of her husband. Considerable research has, in fact, shown such an effect. However, women who stayed at home to raise children and then began to work later are less likely to retire when their husbands do, and married women who will become entitled to their own pensions if they continue to work are also less likely to retire when their husbands do. Some research has shown that a wife’s retirement affects when her husband retires.

Since women continue to be perceived as caregivers, the effect of the husband’s health on the wife’s retirement has also been investigated. Results on this issue have been mixed. Many researchers have found such an effect, but some have not. This kind of decision is probably more complex than it might first appear. There are two possibilities if a husband becomes ill or disabled: (1) the wife might quit work to care for her husband, or (2) she might continue to work and hire someone else to help with his care (as many men would do in case of a wife’s poor health). The ages of the spouses when the health problem occurs, as well as the earnings potential of the healthy spouse, would be important considerations. An additional consideration would be the source of the family’s health insurance—and whether the continued employment of the working spouse is necessary for continued insurance coverage. Workers who will not continue to be covered by health insurance if they should retire may be more likely to continue working.

Research on the effects of the poor health of parents and other family members is much less developed, partly due to data limitations. One recent study showed that women in their fifties who had a parent needing help with activities of daily living were more likely than others to leave the labor force. No such relationship was found for men who had a parent needing care.

Some parents have dependent children when they reach retirement age; with the trend toward later childbearing, this situation may become more common in the future. Few women have preschool children after they are in their fifties, but having school-age children is more common for these women. Parents must consider the cost of continuing to provide for these children into adulthood while still preparing for their own retirement. Having college-age dependent children has been shown to lead to delayed retirement for both women and men.

The most important influences on retirement vary considerably by race. Older African-American and Hispanic men are more likely to be unemployed or out of the labor force than white men. Older African-American women have rates of labor-force participation similar to white women, but Hispanic women are more likely to be unemployed or out of the labor market than are white or African-American women. Both African-American and Hispanic men and women are more likely to be influenced by push factors such as involuntary loss of a job or poor health. African-American and Hispanic workers are also more likely to be concentrated in low-wage occupations that are physically demanding and/or lack pension coverage.

There are a number of trends that suggest that the age of retirement may rise in the future. Americans are healthier and living longer. The shift in the proportion of the population working in manufacturing (rather than in service occupations) makes working with minor health problems feasible for more workers. The growth of defined contribution pension plans that reward employees for working longer is also an important trend. The extension of the age of eligibility for full Social Security benefits to sixty-seven, as well as the elimination of retirement incentives (including changes in the earnings test and the delayed retirement credit), may be expected to increase the average age of retirement, but the effects are expected to be small.

The state of the labor market is likely to have the most important impact on retirement. It is widely anticipated that labor shortages will develop as the large baby-boom generation reaches retirement age. In that case, some companies may offer incentives to older workers to delay retirement. Others may create attractive part-time positions for which they will recruit older adults. This scenario might not materialize, however, either due to a prolonged recession or to more rapid immigration due to increased recruitment of foreign workers by companies.

Polls indicate that well over 60 percent of workers intend to continue working past normal retirement age, with well over half citing ‘‘quality of life’’ rather than economic necessity as their reason. It is unlikely that labor force participation will increase as dramatically as these responses imply. Health, inability to find work, or other personal reasons will probably intervene in many cases. Workers citing quality-of-life reasons for delaying retirement will probably not continue to work unless they find a job they will enjoy.

An increase in Americans’ wealth is often cited as an important factor affecting the trend toward earlier retirement during the post— World War II period. Poverty among the elderly, though still high for some groups of women (and men), decreased markedly between 1959 and 2000. In 1959 the overall poverty rate for the elderly population was 35 percent, considerably higher than the rate for the whole population. In 1999 the poverty rate among the elderly was 9.7 percent, compared with an overall poverty rate of 13.3 percent. Most of the improvement in elderly poverty came in the first fifteen years of this period. By 1979, poverty among the older population had fallen to about 15 percent (17 percent for women). Since then, the decline has been much slower. Much of the improvement can be attributed to the increase in coverage and benefits made available through improvements in Social Security.