Hostname: page-component-78c5997874-mlc7c Total loading time: 0 Render date: 2024-11-03T00:42:48.029Z Has data issue: false hasContentIssue false
  • English
  • Français

The Labor of Older Americans: Retirement of Men On and Off the Job, 1870–1937

Published online by Cambridge University Press:  03 March 2009

Roger L. Ransom  and
Richard Sutch
Roger L. Ransom
Affiliation:
Professor of History, University of California, Riverside, CA 92521
Richard Sutch
Affiliation:
Professor of Economics, University of California, Berkeley, CA 94720.
Get access Rights & Permissions [Opens in a new window]

Abstract

Labor force participation rates for American men sixty and over are estimated for the period 1870 through 1937. They suggest a higher frequency of retirement and quite different trends in the incidence of retirement than have usually been supposed. Evidence is also presented to establish that many older industrial workers changed to less renumerative and less demanding occupations late in their working life. This “on-the-job retirement” may have made the transition from employment to full retirement less sudden than today.

Type
Articles
Information
The Journal of Economic History , Volume 46 , Issue 1 , March 1986 , pp. 1 - 30
Copyright
Copyright © The Economic History Association 1986

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 We are referring largely to the discussion of this topic by social historians. A sampling can be found in: Achenbaum, W. Andrew, “The Obsolescence of Old Age in America, 1865–1914,” Journal of Social History, 8 (Fall 1974), pp. 48–62;CrossRefGoogle Scholarand Old Age in the New Land: The American Experience Since 1790 (Baltimore, 1978);Google ScholarFischer, David Hackett, Growing Old in America: The Bland-Lee Lectures Delivered at Clark University (expanded ed., New York, 1978); original ed., 1977);Google ScholarHaber, Carole, “Mandatory Retirement in Nineteenth Century America: The Conceptual Basis of a New Work Cycle,” Journal of Social History. 12 (Fall 1978). pp. 77–97:CrossRefGoogle Scholarand Beyond Sixty-Five: The Dilemma of Old Age in America's Past (New York, 1983):Google Scholarand Graebner, William, A History of Retirement: The Meaning and Function of an American Institution: 1885–1978 (New Haven, 1980).Google ScholarEconomic historians have done comparatively little work on the history of retirement in the United States, but those who have looked at this topic seem to accept the conventional wisdom on the issue. For example, see Weaver, Carolyn L., “On the Lack of a Political Market for Compulsory Old-Age Insurance Prior to the Great Depression.” Explorations in Economic History, 20 (July 1983). pp. 294–328.CrossRefGoogle Scholar

2 Graebner, History of Retirement, pp. 215–17.Google Scholar

3 Fischer, Growing Old in America, pp. 135–53.Google Scholar

4 The life-cycle hypothesis is associated with Franco Modigliani's research on saving. Modigliani's extensive collection of articles spanning the past four decades has been reprinted in Abel, Andrew, ed., The Collected Papers of Franco Modigliani, vol. 2, “The Life Cycle Hypothesis of Saving” (Cambridge, Mass., 1980).Google ScholarFor a discussion of the aggregate rate of saving in the nineteenth century see Davis, Lance E. and Gallman, Robert E., “The Share of Savings and Investment in Gross National Product During the 19th Century in the U.S.A.,” Fourth International Conference of Economic History, Bloomington, 1968 (Paris, 1973).Google Scholar

5 Mushkin, S. J. and Berman, Alan, “Factors Influencing Trends in Employment of the Aged,” Social Security Bulletin, 10 (08. 1947), table 1, p. 19.Google ScholarMushkin and Berman's figure for 1910 is an interpolated estimate taken from Edwards, Alba M., Comparative Occupation Statistics for the United States, 1870 to 1940, U.S. Bureau of the Census, Sixteenth Census, 1940, “Special Report” (Washington, D.C., 1943), p. 93.Google Scholar

6 The Mushkin and Berman numbers have been relied upon by many observers. David Hackett Fischer cites them to support his argument that a downward trend in labor force participation began at least as early as 1870, Growing Old in America, p. 142.Google ScholarFischer's source for these numbers is a textbook by Gagliardo, Domenico, American Social Insurance (revised ed., New York, 1955), p. 32. Gagliardo's source was the article by Mushkin and Berman. Andrew Achenbaum, “Obsolescence of Old Age,” has also cited the version presented in Gagliardo.Google Scholar

7 Separate tabulations were reported for men and women. Our focus here is only on the labor force participation of men; a complete examination of household labor supply would, of course, require an examination of the labor force participation of women as well. The nineteenth-century history of women's labor force participation over the life cycle is an important subject in its own right which has recently received a good deal of attention. The interested reader should consult Oppenheimer, Valerie, The Female Labor Force in the United States (Berkeley, 1970);Google ScholarFraundorf, Martha Norby, “The Labor Force Participation of Turn-of-the-Century Married Women,” this Journal, 39 (06 1979), pp. 401–17;Google ScholarDegler, Carl, At Odds: Women and the Family in America: 1776 to the Present (New York, 1980);Google ScholarRotella, Elyce, From Home to Office: U.S. Women at Work, 1870–1930 (Ann Arbor, 1981);Google ScholarWandersee, Winifred, Women's Work and Family Values, 1920–1940 (Cambridge, Mass., 1981);CrossRefGoogle ScholarCiancanelli, Penelope, “Women's Transition to Wage Labor: A Critique of Labor Force Statistics and Reestimates of the Labor Force Participation of American Women” (Ph.d. diss., New School for Social Research, 1983);Google ScholarGoldin, Claudia, “The Changing Economic Role of Women: A Quantitative Approach,” Journal of Interdisciplinary History, 13 (Spring 1983), pp. 707–34;CrossRefGoogle Scholarand Cookingham, Mary, “Working After Childbearing in Modern America,” Journal of Interdisciplinary History, 14 (Spring 1984), pp. 773–920.CrossRefGoogle Scholar

8 Instructions to enumerators have been reprinted in Bureau, U.S. of the Census, Twenty Censuses: Population and Housing Questions, 1790–1980 (Washington, D.C., 1979). The quotations are taken from pp. 19–20.Google Scholar

9 U.S. Census Office, Ninth Census, 1870, The Statistics of Wealth and Industry of the United States III (Washington, D.C., 1872), p. 805.Google Scholar

10 The 1870 census took no explicit notice of unemployment. In 1880 an attempt was made to enumerate the months of unemployment, but results of this inquiry were neither published nor tabulated. U.S. Bureau of the Census, Twelfth Census, 1900, Special Reports: Occupations at the Twelfth Census (Washington, D.C., 1904), p. ccxxv. It is probable that the unemployed who were seeking employment for the first time in 1870 and in 1880 were not counted as productively occupied. This is a difference from the modern treatment which would influence the comparability of the figures for teenagers (and perhaps adult women), but it had negligible impact on the statistics for adult males.Google Scholar

11 U.S. Census Office, 1870, Statistics of Wealth and Industry, p. 832;Google Scholarand U.S. Gensus Office, Tenth Census, 1880, Statistics of the Population of the United States at the Tenth Census (06 1, 1880) (Washington, D.C., 1882), p. 714.Google Scholar

12 The census study was undertaken in the 1940s by Durand, John D. and Goldfield, Edwin, U.S. Bureau of the Census, Sixteenth Census, 1940, Population: Estimates of the Labor Force, Employment, and Unemployment in the United States, 1940 and 1930 (Washington, D.C., 1944)Google Scholarand the results were reported in greater detail in Durand, John D., The Labor Force in the United States, 1890–1960 (New York, 1948), tables A-2 and A-6, pp. 199 and 208.Google ScholarWe have used the labor force figures reported by Durand and the age tabulations of the population reported in U.S. Bureau of the Census, Fifteenth Census, 1930, Population V, “General Report on Occupations” (Washington, D.C., 1933), table 2, p. 115. We have distributed those reported with “age unknown” to specific age categories in proportion to those with reported ages.Google Scholar

13 Andrew Achenbaum apparently feels that this problem was severe since he has rejected the 1870 and 1880 tabulated results as “so flawed” as to be “useless,” Old Age in the New Land, p. 179. Unfortunately. Achenbaum provides no details of how he reached this conclusion.Google Scholar

14 U.S. Census Office, 1870, Statistics of Wealth and Industry, p. 832;U.S. Census Office. 1880, Population, p. 704;Google ScholarU.S. Census Office, Eleventh Census, 1890, Report on the Population of the United States at the Eleventh Census: 1890, Part 2(Washington, D.C., 1897), p. 115;Google ScholarDurand, John, Labor Force in the United States, p. 199, and U.S. Bureau of the Census, 1930, Population V, p. 115.Google Scholar

15 We have also conducted a review of the scholarly work by Edwards, Alba, Comparative Occupation Statistics for the United States;Google ScholarDurand, John, Labor Force in the United States;Google ScholarLong, Clarence, The Labor Force Under Changing Income and Employment (Princeton, 1958);Google ScholarMiller, Ann Rattner and Brainerd, Carol, “Labor Force Estimates,” in Kuznets, Simon and Thomas, Dorothy Swain, eds., Population Redistribution and Economic Growth: United States, 1870–1950 I, “Methodological Considerations and Reference Tables” (Philadelphia, 1957);Google Scholarand Lebergott, Stanley, Manpower and Economic Growth: The American Record Since 1800 (New York. 1964). We found no comments to suggest that these authors suspect that an undercount of adult men had occurred. This statement, however, would not be true for the case of women and children. There is evidence that female occupations were systematically underenumerated throughout the nineteenth and early twentieth centuries.Google Scholar

16 U.S. Census Office, 1870, Statistics of Wealth and Industry, p. 798.Google Scholar

17 U.S. Census Office, 1880, Population, p. 704.Google Scholar

18 U.S. Census Office, 1890, Population, 2, p. cxxii. As far as we are aware, little attention has been focused upon the 1870 and 1880 tabulations after the changes in definitions were introduced in 1890. Alba Edwards made an extensive retrospective analysis of the occupation statistics for the Census Bureau in his Comparative Occupation Statistics for the United States. Edwards had no complaints about the reliability of the occupational coverage for older men in 1870 and 1880. but when he came to make his own retabulation of the occupation-by-age figures Edwards began with 1890 because of the lack of comparability of the age classifications used at the two earlier censuses with those that followed (pp. 92–93). Most subsequent research has followed Edwards's lead and begun detailed analysis with the returns from 1890.Google Scholar

19 U.S. Bureau of the Census, Twenty Censuses, p. 29.Google Scholar

20 ibid., p. 30, italics in original.

21 ibid., p. 35.

22 Enumerators in 1910, 1920, and 1930 were warned not to return occupations for the permanently retired, U.S. Bureau of the Census, Twenty Censuses, p. 44. However, this apparently did not prevent significant numbers of those who would be considered retired by today's standard from reporting a source of income which the census at the time interpreted as a gainful pursuit. In addition to the categories listed above, it is likely that some retired individuals were listed as “boarding house keepers” even though they merely rented out a room in their house to supplement their retirement income. In what follows we have, however, made no correction for this potential overcount.Google Scholar

23 U.S. Bureau of the Census, 1900, Occupations, pp. ccxxv, ccxxxiii. See also our discussion of the census of 1900 below. The census bureau study undertaken by Durand and Goldfleld demonstrated that the 1930 labor force was less than the number enumerated as gainful workers by the census of 1930. U.S. Bureau of the Census, 1940, Estimates of the Labor Force.Google ScholarA portion of the difference was explained by “recently retired or disabled persons who were reported as having gainful occupations although they were no longer working or seeking work,” Durand, The Labor Force in the United States, pp. 197–198. Although this group had a small impact on the overall statistics, it had a considerable impact on the figures for older men.Google Scholar

24 The Center for Studies in Demography and Ecology of the University of Washington has drawn from the manuscript schedules of population in 1900 a l-in-760 sample of the national population which is available for public use. We have obtained this sample of 99.034 individuals from the Inter-university Consortium for Political and Social Research. The original sample was collected by Samuel Preston under the auspices of a grant from the National Science Foundation. For a detailed description of the 1900 sample see Graham, Stephen N., “The 1900 Public Use Sample Users' Handbook Prepared at the Center for Studies in Demography and Ecology, University of Washington,” Inter-university Consortium for Political and Social Research 7825 (draft version, 1981). We are grateful for access to this data, and note that the Center for Studies in Demography, the National Science Foundation, and the Inter-university Consortium bear no responsibility for the analysis or interpretations presented here.Google Scholar

25 In 1900 the U.S. census departed from the practice of every preceding and subsequent census and recorded the age of individuals not—as is conventional—at their last birthday, but instead at their nearest birthday. For this reason, we have calculated age for the purposes of Table I from the information reported to the census on the year and month of birth of each individual. Since we know the official census date was June I, 1900, it is an easy matter to calculate the conventional age of each individual. The figures in Table I for the men 60 and over use this calculated age so that the results will be comparable to figures from other censuses, however, use of the reported age in an alternative calculation produced no substantial difference in the result.Google Scholar

26 The census of 1900 did not report the gainfully occupied rate for men 60 and over. The figure in the text may be compared to 80.4 percent reported in 1890, U.S. Census Office, 1890, Population. p. cxii. The 1900 census did publish a tabulation for the group of men 65 and over. Their gainfully occupied rate was reported as 68.4 percent, U.S. Bureau of the Census, 1900, Occupations, p. cxviii. Our census sample inplies a gainful occupation rate of 75.2 percent for men 65 and over. This suggests either that the sample over estimates the number of older employed men or that the census officials edited some occupations from the tabulations in a way not discernable to those who collected the sample.Google Scholar

27 The census of 1900 recorded the number of months of unemployment experienced during the preceding 12 months. According to the instructions issued to the enumerators, the census law “does not contemplate that this question shall apply solely to the principal occupation in which the person may have been engaged during the year, but it is the intent to find out the number of months (or parts of months) during which a person ordinarily engaged in gainful labor was not employed at all,”Google ScholarU.S. Bureau of the Census, 1900, Occupations, p. ccxxvii. Nevertheless, individuals appeared in the manuscripts reporting a gainful occupation but with 12 months of reported unemployment. For the purposes of Table 1, we counted as “permanently unemployed” anyone reporting six months or more of unemployment during the preceding year.Google ScholarWe note that census officials expressed confidence in the reliability of the 1900 unemployment statistics and that the period from June 1899 to May 1900 was considered prosperous,ibid., pp. ccxxxiii–ccxxxiv.

28 Temporary unemployment was calculated by dividing the total number of months of unemployment during the previous year reported by men in the labor force by 12.Google Scholar

29 The data used to plot Figure 2 are based on reported age and are presented in Ransom, Roger L. and Sutch, Richard, “Retirement and Changes in Occupation with Advancing Age: American Men in the Late Nineteenth Century,” Working Papers on the History of Saving, 4 (University of California, Berkeley, 09 1985), table B, p. 51.Google Scholar

30 U.S. Bureau of the Census, Sixteenth Census, 1940, Population IV, “Characteristics by Age: Marital Status, Relationship, Education, and Citizenship,” Part I (Washington, D.C., 1943), table 24, p. 90.Google Scholar

31 Ibid, p. 3.Google Scholar

32 Ibid.Our own analysis of the age distributions reported by the 1950 census has convinced us that the problem lay entirely with the 1940 census.Google Scholar

33 For a discussion of the 1937 census see Durand, Labor Force in the United States, pp. 96–99.Google ScholarThe original census returns are reported in U.S. Office of the Administrator of the Census of Partial Employment, Unemployment, and Occupations, Census of Partial Employment, Unemployment, and Occupations, 1937 IV, “The Enumerative Check Census” (Washington, D.C., 1938). A description of the methods we have used to estimate labor force participation from the check census results is available from the authors upon request.Google Scholar

34 Durand's 1890 estimate of labor force participation may be contrasted to the gainful employment rate of 73.8 percent reported by the census that year. U.S. Census Office, 1890, Population, pp. cxxi–cxxii.Google Scholar

35 Stanley Lebergott has suggested that there should be little difference between the gainful occupation and labor force measures of the labor supply in the 1930s. However, his conclusion referred to the total labor force and should not be applied to a particular segment of it. Essentially. Lebergott's observation was that substantial numbers of student and female seasonal workers who were counted in the labor force after 1945 but largely missed by the enumeration of gainful workers almost exactly offset the conceptual differences between the two concepts emphasized by Durand and Goldfield. See Lebergott, Manpower in Economic Growth, pp. 398–403. Even if we grant Lebergott's point, the Durand-Goldfield estimates for the male labor force over 60 would be left largely unaltered by any adjustment.Google Scholar

36 Durand himself warned readers that these estimates “would be much less reliable than those for 1930,” Labor Force in the United States, pp. 197–198.Google ScholarDurand's methods were criticized at the time by Long, Clarence D. although Long's own estimates were only slightly different, “The Labor Force in Wartime America,” National Bureau of Economic Research, Occasional Papers, 14 (03. 1944), table 1, p. 9.Google Scholar

37 Bancroft, Gertrude, The American Labor Force: Its Growth and Changing Composition (New York, 1958), pp. 38–39.Google Scholar

38 See, for example, Mushkin and Berman, “Trends in Employment of the Aged”;Google Scholarand Dorfman, Robert, “The Labor Force Status of Persons Aged Sixty-Five and Over,” American Economic Review, 44 (May 1954), pp. 634–44.Google Scholar

39 The figures in Table 4 are based on the published census returns and the 1900 census sample. We have employed the definitions of the agricultural sector suggested by Miller, Ann and Brainerd, Carol, “Labor Force Estimates,” p. 382. Our estimates of the percentage of the labor force engaged in agriculture differ somewhat from the estimates made by Miller and Brainerd (p. 609),Google Scholarand the alternative estimates calculated by Lebergott, Stanley, “Labor Force and Employment, 1800–1960,” National Bureau of Economic Research, Output, Employment, and Productivity in the United States After 1800: Studies in Income and Wealth, 30 (Princeton, 1966), table 2, p. 119. We prefer to employ the census estimates in this context to preserve comparability with the results presented in Table 2. Use of either alternative, however, would not materially affect the results.Google Scholar

40 The 1930 census reported gainful workers, but not labor force. Durand constructed estimates of the total labor force, but provided no occupational breakdown. Therefore, we distributed Durand's labor force estimate between agriculture and non-agriculture using the proportions reported by the census that year for the gainfully occupied (25. I percent in agriculture for all males and 36.3 percent for men 60 and over).Google Scholar

41 Wolfgang Kleber has suggested this mechanism played an important role in the case of Germany. In that country many middle-aged workers shifted out of industrial jobs into the service sector. Kleber concluded that over the entire period from 1882 to 1978 “the increase in the working male population in the service sector occurred not so much through the recruitment of younger occupied persons as much as through mobility during occupational life,” “Labor Force Change in Germany Since 1882: A Life-Cycle Perspective,” Explorations in Economic History, 22 (Jan. 1985), pp. 109–10.Google Scholar

42 In principle, we would prefer data which follow the career paths of a sample of individuals as they age, but such information, if available, has eluded our search. Nevertheless, the crossectional patterns are striking and provide strong quantitative support of the view that downward occupational movement was quite common the late nineteenth century. There is also evidence of a non-quantitative nature to bolster our interpretation. Tamara Hareven has reported nineteenthcentury career patterns which support our intepretation of the cross-section data, “The Last Stage: Historical Adulthood and Old Age,” in Erikson, Erik, ed., Adulthood (New York, 1976), p. 208.Google ScholarThe phenomenon of downward mobility has also been mentioned by Hershberg, Theodore et al. ,, “Occupation and Ethnicity in Five Nineteenth-Century Cities: A Collaborative Inquiry,” Historical Methods Newsletter, 7 (06 1974), pp. 204–11;CrossRefGoogle ScholarHaber, Carole, “Mandatory Retirement,” p. 78;Google ScholarDahlin, Michel, “From Poorhouse to Pension: The Changing View of Old Age in America, 1890–1929” (Ph.D. diss., Stanford University, 1982), fn. 3, p. 86;Google Scholarand Weiler, N. Sue, “The Aged, the Family, and the Problems of a Maturing Industrial Society: New York, 1900–1930” (Ph.D. diss., University of Illinois, Chicago, 1983), pp. 107–14.Google Scholar

43 Massachusetts Bureau of Statistics of Labor, Ninth Annual Report of the Bureau of Statistics of Labor, February, 1878, Public Document 31, Part VI (Boston, 1878), pp. 155–262.Google Scholar

44 The proportions of elderly are measured here by the percentage of all workers over 20 who were over 60. Workers under 21 years of age have been excluded because different practices with regard to apprenticeships in skilled occupations were reflected in unsystematic reporting of the occupations of minors.Google Scholar

45 The original source presents a very detailed age and occupational breakdown. When the complete distributions are compared, the age distribution for industrial workers can be considered to have been drawn from a different population than the distributions for unskilled workers, artisans, and workers in the building trades. In each case the differences are significant at the 99.9 percent confidence level when tested with the Kolmogorov-Smirnov procedure. The unskilled and artisan distributions cannot be distinguished by this test even when set at the 90 percent confidence level. For more detail see Ransom and Sutch, “Retirement and Changes in Occupation,” appendix D, p. 54.Google Scholar

46 Massachusetts Bureau of Statistics of Labor. Annual Report, 1878, Part III, pp. 40–49.Google Scholar

47 The growth rates of the occupations in Table 6 are measured by the total number of men within each occupational title.Google Scholar

48 Bureau, Michigan of Labor and Industrial Statistics, Sixth Annual Report of the Bureau of Labor and Industrial Statistics (Lansing, 1890).Google Scholar

49 For a detailed description of the data set on Michigan furniture workers see Ransom, Roger L. and Sutch, Richard, “Labor Supply, Income, and Saving over the Life Cycle: A Survey of Male Workers from the Furniture Industry in Michigan, 1890,” History of Saving Project Research Memorandum, 3 (University of California, Berkeley, 11. 1985).Google Scholar

50 Uselding reported these findings as a rejection of the claims made by muckrakers that industrial jobs were unsafe and unhealthy, “In Dispraise of the Muckrakers: United States Occupational Mortality, 1890–1910,” Research in Economic History, 1 (1976), pp. 348–49.Google Scholar

51 The procedure for calculating the age-corrected death rates is described by Uselding, “Occupational Mortality,” pp. 342–44.Google ScholarThe data on 1890 occupational mortality is from U.S. Census Office, Census, Eleventh, 1890, Report on Vital and Social Statistics at the Eleventh Census: 1890 (Washington, D.C., 1896), Part I, table 15, pp. 972–73.Google Scholar

52 This finding is corroborated by the shape of the age-income profiles reported by other researchers for late nineteenth-century American workers in a variety of different occupations. See, for example, Hannon, Joan Underhill, “The Immigrant Worker in the Promised Land: HumanCapital and Ethnic Discrimination in the Michigan Labor Market, 1889–1890” (Ph.D diss., University of Wisconsin, Madison, 1978) and “Ethnic Discrimination in a 19th-Century Mining District: Michigan Copper Mines, 1888,”Google ScholarExplorations in Economic History, 19 (Jan. 1982), pp. 28–50;Google ScholarHaines, Michael R., “Industrial Work and the Family Life Cycle, 1889–1890,” Research in Economic History, 4 (1979) and “The Life Cycle, Savings, and Demographic Adaptation: Some Historical Evidence for the United States and Europe”Google Scholarin Rossi, Alice S., ed., Gender and the Life Course (Hawthorne, N Y., 1985).Google Scholar

53 We have attempted to investigate one possible explanation of the trend. It is possible that the health of older men was improving during this period even though life expectancy at age 60 was not (see Table 3). If so, then labor force participation rates of older men may actually have risen, industry by industry, as fewer workers would have been forced by illness or infirmity to retire. Such a rise would be obscured in the aggregate statistics because of the shift away from agriculture. We were unable to find support for this conjecture, however, and the low rates of dependency among older men reported by Daniel Scott Smith for 1900 argue against this interpretation. Smith's pioneering study concluded that “the family in the direct sense of household co-residence of adult generations was the welfare institution for old people in 1900.” Yet only one-sixth of married people 55 and over in his study lived with their married children.Google ScholarSmith, , “Life Course, Norms, and the Family System of Older Americans in 1900,” Journal of Family History, 4 (Fall 1979), p. 296.CrossRefGoogle Scholar

54 It should be possible to discriminate between the two alternative explanations suggested here with an industry-by-industry study of changes in the relative wages and the age distributions of older workers in the early decades of the twentieth century, since some industries were more heavily influenced by the seniority movement than others. Such research would take us beyond the scope of this article. However, we hope that the findings reported here will stimulate further work along these lines.Google Scholar