The evolution of retirement.

Costa, Dora L.

Not only are more men living past age 65 in America today than ever before, but American men also have been abandoning the labor force at ever younger ages. The retirement rate of American men over the age of 64 has risen rapidly from a mere 25 percent at the end of the last century to over 80 percent today. At the same time the very nature of retirement has changed. For most individuals retirement is no longer a time of withdrawal from all activities and of dependence on family and friends; rather it is a time of discovery, personal fulfillment, and relative independence. In the past, such a retirement experience was limited to the wealthy few who could afford it. Now it is an option available to the majority of workers.

That most men now can look forward to a period of personal fulfillment at the end of their working lives is one of the achievements of our century, but such a retirement is expensive, and financing it poses budgetary dilemmas. Approximately 80 percent of elderly households receive over half of their income from Social Security, and Social Security is facing a fiscal crisis. If men continue to abandon the labor force at ever younger ages, the crisis is likely to be even more acute. To understand whether retirement rates will continue to rise, we must examine how retirement has evolved from 1880 to the present. Retirement rates were rising throughout this period. In fact, 41 percent of the long-run rise in retirement rates occurred before the postwar growth of Social Security and private pension plans. In my forthcoming book, The Evolution of Retirement: An American Economic History, 1880-1990 (University of Chicago Press for NBER, 1998), I therefore investigate the factors that have fostered rising retirement rates.(1)

Income and Retirement

Retirement requires income, whether in the form of state-provided retirement or disability benefits, private pensions, income from other family members, or assets. Researchers have investigated the role that each of these income sources plays in the retirement decision, largely using cross-sectional data for the years after the 1960s. But, because 70 percent of the rise in retirement among men older than 64 occurred before 1960, only large increases in benefits could have enticed those remaining in the labor force to have withdrawn.

Previous researchers have not been able to examine the impact of income on the retirement decision prior to the 1960s because the necessary data has been unavailable. Fortunately, a longitudinal dataset that follows Union Army recruits of the American Civil War from their youth to their death can be generated from census records and from records of the Union Army pension program. At the beginning of the century, Union Army pensions were the most widespread form of assistance to the elderly, serving about a quarter of the population over age 64 in 1900. I estimate the income effect of Union Army pensions on retirement rates.

I find that pensions had a substantial impact on retirement rates both in 1900 and in 1910. My findings suggest that the high labor force participation rates of older men prevailing at the turn of the century arose because retirement incomes were too low to fully support them and, as retirement incomes have risen, so have retirement rates. I attribute much of the long-term increase in retirement rates to the rising incomes of the elderly. Their wages, and hence their savings and pensions, have increased, as have government transfers.

However, increased income is not the sole explanation for the rise in retirement. In fact, I show that the income elasticity of retirement has fallen over time. Whereas rising retirement incomes could explain up to 90 percent of the increase in retirement rates between 1900 and 1930, they could account for only half of the increase between 1930 and 1950, and for almost none of the more recent increase. Workers now may be less responsive to changes in transfer income because they are no longer close to subsistence levels; instead, they reach retirement age with enough to satisfy their consumption needs. Alternatively, by establishing age 65, and later age 62, as an "official" retirement age, Social Security may have led individuals to want to retire at that age and therefore reduced the effect of income on the work decision. Finally, retirement also has become more attractive because men are less circumscribed in their choice of leisure time activities. Mass tourism and mass entertainment have increased the variety of recreational activities and lowered their price.

Other Explanations

I consider several other explanations for increased retirement rates, including worsening average health of the population. I show that retirement rates rose despite declines in the burden of chronic disease. Between 1910 and 1983, the prevalence of heart disease among men above the age of 64 fell from 75 to 40 percent; that of musculoskeletal disease from 68 to 48 percent; and that of respiratory disease from 42 to 30 percent. Between 1935 and 1992, rates of blindness fell by about one third. The elderly have benefitted from advances in medical technology, fewer occupational hazards, and better conditions in early life. At the same time, health has become less important to the retirement decision. Because we now can better control chronic conditions, and because physical job requirements have been reduced, those in poor health are more likely to participate in the labor force relative to those in good health than was the case in 1900 and in 1910. Age 65 therefore may no longer be as appropriate a demarcation of old age as it was in the first half of the century, when the typical health of a 65-year-old was very poor.

Declines in part-time work, nonfarm self-employment, and farming do explain the rise of retirement since 1880 either. The proportion of 65-to-74-year-old employees who work part-time has risen from 15 percent in 1940 to 47 percent in 1990. The fraction of the labor force that is self-employed has fallen, but only since the 1960s have older self-employed workers been more likely to remain in the labor force than wage and salary workers. The lower retirement rates of farmers relative to non-farmers are also a recent phenomenon. Using longitudinal data on Union Army veterans, I show that in 1900 and 1910 farmers were no less likely to retire than non-farmers and that, upon retirement, farmers moved to a nearby town.

One factor that accounts for up to one-fifth of the increase in retirement rates of men over age 64 since 1900 is the increased duration of unemployment spells. Unemployment within state of residence had a substantial effect on the retirement of men over age 64 in 1900, and on men aged 50 to 64 in 1980. But, the unemployed would not have been able to retire unless they had income sources other than wages. In fact, high unemployment within state of residence was much more likely to induce Union Army veterans versus non-veterans to leave the labor force.

The Retirement Lifestyle

A man who retired in 1880 could expect a very different life from that of a man retiring today. Close to half of retired men in 1880 were living in the households of their children or other relatives, whereas today that figure is only 5 percent. By examining data on the living arrangements of Union Army veterans, I show that, at the beginning of the century, men would have preferred to remain independent of their families. The majority simply could not afford to do so, though. I argue that rising retirement incomes explain the decline in the percentage of men older than 64 living in the households of their children and the narrowing of differences in living arrangements by retirement status since the beginning of the century. But, I also show that changes in income now have a relatively smaller effect on the living arrangements decision than they did in 1910, perhaps because independent living is now relatively inexpensive. The growth of retirement communities in low cost living areas, the declining price of transport and of communication with family members, and the rise in private and state social support services, among other factors, have lowered the price of living alone.

Independent living may be not only cheaper than it was in the past but also more attractive. A leisurely retirement lifestyle is now often made possible by resettlement to a community with a better climate or other environmental amenities, or to one with a low cost of living. As independent living has become more attractive, this in turn may have increased the attractiveness of retirement.

The typical worker now looks forward to retirement (or at least the first few years of it), because retirement has become a time for travel and recreation. Leisure time activities are now pursued more widely across all income classes because of rising incomes and because technological change has not only lowered the price of existing products, but also has created new products that lower the "quality-adjusted price" of entertainment. Technological advances also have lowered travel time and thereby decreased the time-cost of entertainment. Using consumer expenditure surveys, I show that because recreational goods and leisure are complements, the lower price of recreation (in both time and money) may have increased the demand for retirement.

The elderly in part have financed their retirement through public monies. First it was through Union Army pensions, then in the late 1920s and early 1930s many states provided pensions to the needy aged. These pensions later were replaced by Social Security Old Age Assistance and Old Age Insurance. The growth of all of these programs was made possible by the availability of revenue resources and was spurred in part by increasingly well-organized elderly pressure groups. As the population ages, the elderly may become an even more powerful political force. But, as their numbers rise, it will become increasingly difficult for the young to finance a lengthy retirement for the old. The continued provision of the retirement lifestyle to which we have grown accustomed is increasingly likely to lie with individuals.

1 See also "Pensions and Retirement: Evidence from Union Army Veterans," Quarterly Journal of Economics (May 1995); "Agricultural Decline and the Secular Trend in Retirement Rates," Explorations in Economic History (October 1995); "Health and Labor Force Participation of Older Men, 1900-1991," Journal of Economic History (March 1996); "A Theory of Technophysio Evolution, With Some Implications for Forecasting Population, Health Care Costs, and Pension Costs," Demography (February 1997); and "Displacing the Family: Union Army Pensions and Elderly Living Arrangements," Journal of Political Economy, forthcoming December 1997.

Costa is an NBER faculty research fellow and an associate professor of economics at MIT. Her profile appears in this issue.