Between 1900 and 1950 U. S. life expectation at birth increased from 47 to 68 years. Then for the next two decades further progress in longevity came to a virtual halt. Although there were some minor changes in the age-specific death rates of both men and women, they had little impact on the expectation of life at birth. During this interregnum a number of thoughtful analysts reviewed the progress in mortality over the preceding century, pointing out why the century-long decline in mortality rates was unique and could not be repeated: Nearly all the gains that could be made from the elimination of death from infectious diseases below age 60 had been made. Short of a dramatic breakthrough biologically, it was doubtful that declines in mortality rates at old ages could be as large as those that had already transpired. Indeed, by the early 1960s there was some evidence of a relative deterioration in the mortality rates of persons aged 40 to 70 in a study of Western countries, which appeared to be due to the effects of smoking (Stolnitz 1956/57; Keyfitz 1977; Preston 1970).

It was not until the end of the 1970s that demographers became aware that a new decline in mortality was underway, concentrated this time at older ages. Evidence of a downturn in the death rates of the elderly was contained in medicare data which showed that beginning in 1968 and continuing through the end of the 1970s, mortality rates at age 65 and over were declining by two percent per year, and the most rapid advances were concentrated among those aged 85 and older. This development was so unexpected by demographers and epidemiologists that it set off intense discussions, akin to those stimulated two decades earlier as population specialists became aware of the baby boom (Wilkin 1981). The new round of research focused not only on the explanation for the improvement in mortality rates but on how long the decline might continue and whether an increase in the burden of chronic disease was a necessary consequence of the increase in life expectation at older ages (Verbrugge 1984 and 1989; Wilson and Drury 1984; Svanborg 1988; Guralnik et al. 1989; Riley 1989 and 1990; Rothenberg and Koplan 1990).

The most far-reaching aspect of the discussion was the new debate over whether or not the life span (the ultimate length of the life of a species) is fixed, and if so how long the human life span was. In a celebrated paper Fries (1980), reasserted the prevailing gerontological view that although life expectation has increased from 47 to 73 during the twentieth century, the life span was fixed. On the assumption that the Gompertz curve (which relates the log of nqx to age) was linear at adult ages, he estimated that the life span was fixed at 85-7 years. Consequently age 85 was the upper limit of life expectation and it would be achieved by a rectangularization of the survivorship (1x) curve: Virtually all deaths in a cohort would be compressed into a few years in the neighborhood of age 85. He also argued that movement of life expectation toward the ideal life span (85) would not increase the proportion of the elderly population that was disabled because the onset of chronic diseases could be postponed (the morbidity curve would also be rectangularized) through changes in lifestyle and biomedical interventions (cf. Olshansky et al. 1991).

back to top

That paper touched off a highly productive debate which is still in progress. One important aspect of his argument was that chronic diseases had not only replaced acute infectious diseases as the principal medical problem, but that these chronic conditions were independent of the acute infectious diseases. They were instead "problems of accelerated loss of organ reserve" (p.132), part of the natural process of senescence that preceded mortality. Since this upper limit of life was fixed at age 85-7 years, the most that could be accomplished by lifestyle changes and medical intervention was the compression of morbidity against the rigid ceiling of 85-7 (cf. Fries 1983 and 1989).

Although the issues raised by Fries have not yet been resolved, much of the evidence accumulated by investigators during the last decade militates against the notion of a genetically fixed life span or, if it is fixed, suggests that the upper limit is well above 85. Vaupel's study of Danish twins indicates that genetic factors account for only about 30 percent of the variance in age of death (Vaupel 1991a). His study of Swedish males who lived to age 90 indicates that the death rate at that age has declined at a rate of about 1 percent per annum since 1950, a finding that is contradictory to the rectangularization of the survivorship curve (Vaupel 1991b). Two recent studies of insect populations (Carey et al. 1992; Curtsinger et al. 1992) indicated that variation in environmental conditions had a much larger effect on the life span than genetic factors and revealed no pattern suggestive of a fixed upper limit. Collectively, these studies do not rule out genetic factors but suggest something much less rigid than the genetic programming of absolute life span: An emerging theory combines genetic susceptibility of various organs to cumulative insults as a result of exposure to risk.

Recent studies indicate that age-specific rates of chronic conditions above age 65 are generally falling, but they do not support the proposition that the span of years during which individuals will be afflicted by chronic diseases is being compressed. According to Manton, Corder, and Stallard (1993) the rate of disability among the elderly in the U. S. declined by 4.7 percent between 1982 and 1989. Put on a decade basis, this rate of decline is quite similar to the long-term rates of decline between 1910 and 1985-88 in chronic conditions among elderly veterans (Fogel, Costa, and Kim 1994). The finding is consistent with the growing body of evidence indicating that chronic diseases at later ages are, to a considerable degree, the result of exposure to infectious diseases and other types of biomedical and socioeconomic stress early in life. It is also consistent with the predicted decline of about 6 percent per decade in chronic diseases based on the Waaler surface in ill health reported below (pp. 314 - 316) (cf. Blair et al. 1989; Manton, Stallard, and Singer 1992; Manton and Soldo 1992).

Much of the current research is now focused on explaining the decline in chronic conditions. Part of the emerging explanation is a change in life styles, particularly the reduction in smoking, the improvement in nutrition, and the increase in exercise, which appear to be involved in reducing the prevalence of coronary heart disease and respiratory diseases. Another part of the explanation is the increasing effectiveness of medical intervention. This point is strikingly demonstrated by data comparing cure rates from chronic disease of veterans of the Union Army age 65 and over in 1910 with World War II veterans of the same age in the mid 1980s (Fogel, Costa, and Kim 1994).

back to top

Recent work in demographic and agricultural history indicates that the modern decline in mortality rates began about 1700 and coincided with a remarkable increase in the food supply. Another parallel development was an increase in body size by over 50 percent, with both stature and BMI contributing to this change. The new evidence also indicates interregnums and even reversals in these processes, some lasting as long as half a century. These findings have given rise to a theory called "technophysio evolution." The theory focuses on a synergism between technological and physiological improvements that has produced a form of human evolution that is biological but not genetic, rapid, culturally transmitted, and not necessarily stable. The process, it is argued, is still ongoing in both rich and developing countries.

Unlike the genetic theory of evolution through natural selection, which applies to the whole history of life on earth, technophysio evolution applies only to the last three hundred years of human history, and particularly to the last century. Although technophysio changes have been very rapid from an evolutionary perspective, little of this pattern of change is visible over a couple of decades, which is about half the length of a typical biomedical career. Over such relatively brief spans human physiology may seem to be more fixed than it has in fact been intergenerationally. What is called for then is evidence which makes it possible to chart secular trends in chronic disease over the past 3 or 4 generations, the period of most dramatic changes in both technological and physiological improvement.

Experiments on rats and other animal models indicate that controlling the environment can greatly alter physiological functioning and expand the average length of life by as much as threefold. Of course, rats cannot control their environment, so their life span under natural circumstances has not changed. Homo sapiens, however, do control their own environment to a large degree and they have greatly altered their own physiology and average life span in just a relatively few generations. The current pace of technological change within and beyond the biomedical sciences, as well as the continuing growth in stature and the reduction in mortality rates at older ages, suggests that technophysio evolution has not yet run its course.

back to top

There is also mounting evidence that improvements in human physiology over the past several generations have made a major contribution to economic growth by increasing both the thermodynamic and physiological efficiency of workers. The physiological factor pertains not only to past economic growth in the U.S. and other rich countries, but to currently developing countries as well. Height and BMI are significant determinants of the wages of men in urban Brazil (Thomas and Strauss 1992). Using data for rural south India, Deolaliker (1988) finds that the elasticity of wage rates with respect to heights is in the range of 0.28 to 0.66 and the elasticity of farm output with respect to BMI is as large as 2. Market productivity in India is heavily influenced by caloric intake during the season of peak labor demand and by BMI during the off peak season (Behrman and Deolaliker 1989). However, Haddad and Bouis (1991) report that wages in the rural Philippines are strongly influenced by height not BMI. In an extension, Foster and Rosenzweig (1991) find that height and calories have particularly large effects on piece rate wages.

Thus far the evidence developed from the Union Army sample indicates that average health among men 65 years of age or older has been improving since 1910 (Fogel, Costa, and Kim 1994). This implies that secularly rising retirement rates cannot be attributed to worsening health. Nor can secularly rising retirement rates be attributed mainly to the increased availability of private pensions and of Social Security, because the increase in male retirement rates dates from 1900 and 41 percent of the increase occurred prior to 1950 (see the series in Moen 1987). Among the explanations currently being investigated are an increased demand for leisure arising from higher income, and from the growth of the entertainment and tourism industries.

The work so far has demonstrated that the Union Army records provide a unique opportunity to examine the impact of the first large government transfer on labor force participation rates. Union Army pensions were available regardless of labor force participation status. The Civil War pension program stipulated no work disincentive, and pensions did not depend upon past wages. Therefore, Union Army pensions can be used to estimate a pure income effect on labor supply. The generosity of Union Army pensions varied with the degree of incapacity, and for a given degree of incapacity, the pension amount was substantially higher if the veteran could trace his disability to the War. Therefore, the effect of pensions on labor supply can be disentangled from that of health. The findings, therefore, reveal the effect of income growth on retirement, and so bear on income effects arising from Social Security retirement and disability programs. Comparisons with more recent cohorts are being used to investigate whether differences in elasticities of labor force non-participation with respect to income reflect a change over time in the demand for retirement or whether studies of more recent cohorts have underestimated the effect of pension size because of inadequate controls for health and for the tax penalty on work after retirement under current Social Security provisions.