I agree with the general point of Levine and Mitchell's paper. Predictions of future labor force participation rates that are based on extrapolations of past trends are unlikely to be very useful if any interesting changes are underway. Roent research about the determinants of retirement behavior and casual eoipiricism about likely trends in these determinants suggest what current data already seem to indicate - that the early retirement trend may have already abated or reversed. And accurate predictions about the future require a comprehensive model of retirement behavior that is currently beyond our grasp.
Within this general framework of agreement, I have several comments about the, material in the paper, and a criticism about a major focus that is ffiissing - the nature of the jobs that older workers are likely to hold in the future. The latter is the topic of the last section.
I think that the authors shortchange the importance of post-career employmont. The extent of it obviously depends on the definition of a career job. Using the Retirement History Study, Rich Burkhauser, Dan Myers and I defined a career job as one held for at least 10 years on which an individual was working full time (35 or more hours per week). We found that during the 1970s, a quarter of the wage and salary workers and half of the self-employed did something other than leave the labor force when they left full time career employment ( Quinn, et. al., 1990, pp. 173-178). Wage and salary workers who stayed employed usually found a new job; the self-employed were most likely to move to part-time status on their career jobs. Chris Ruhm ( 1990b) defined the career job as the longest job held over the life-time, and concluded that fewer than half of the of the RHS workers retired directly from their career jobs. In either case, this seems to be an important phenomenon, and one that may well grow in the future.
Levine and Mitchell quote our research to conclude that "most of these individuals spend only a short time in this partial retirement status." I had used exactly the same table they referenced to conclude that "these transitional jobs were of sufficient duration to represent an important part of the retirement process" ( Quinn, et. al., 1990, pp. 189). I am reminded of the scene in Annie Hall in which Woody Allen and Diane Keaton (his wife in the movie) were asked separately about their frequency of sex. "Hardly ever," responds Woody, "about three times per week.""All the time," says she, "about three times per week." In fact, I think partial retirement may well grow in importance in the future, and that it should be a major part of the research agenda.
Concerning the impact of poor health, I fear that the authors' words may be misinterpreted. Although I agree that "increasingly poor health cannot explain the long-term trend toward earlier retirement in the United States," I do not agree that it "there is little support for the view that health is a major factor explaining early retirement." Health is especially important in explaining the behavior of very retirees, and is always a significant explanatory variable in behavioral equations ( Kingson, 1982; Sammartino, 1987). A large minority of retirees always name health when asked in questionnaires - 25 percent in the most recent