Commerce,Coalitions,and Factor Mobility September 2002 FIGURE 2.Interindustry Variation in Profits 140 120 100 60 40 20 0 18001820 184018601880 19001920194019601980 2000 Year --Annual profits,of capital in 15 industries (Census)--Annual profits,of capital in 20 industries (Census) Annual profits per man-hour (Census) --Corporation after-tax profits,of net worth (SEC) Katz 1987;Krueger and Summers 1988).14 As Figure 1 according to their main activities into two-digit SIC indicates,the size of these much-discussed "indus- industries.16 For earlier years,following Bateman and try rents"trended downward markedly during ear- Weiss (1981),I used census manuscripts to calculate lier stages of industrialization and upward only more profits(value-added minus wage costs)as a percentage recently.15 of the capital invested for firms in each of the major There is very little direct evidence on firm profits in manufacturing industries in each census year.After different industries prior to 1909,when federal taxes 1919,the Department of Commerce ceased reporting were first imposed on corporate incomes(Epstein and data on capital invested,but from 1947 reports total Gordon 1939,122).Beginning in 1933,data from an- man-hours consumed per year for each industry,and nual reports on corporation profits (as percentages of these can be used as a proxy for total investments.17 net worth and equity)are available from the U.S.Secu- Figure 2 charts coefficients of variation in profits rities and Exchange Commission (SEC),categorized across manufacturing industries using these different data series.The results generally match the pattern ex- 14 Only very basic controls can be applied in the aggregate data to hibited in the wages data.There was a general decline account for heterogeneity in skill levels across industries.There is in interindustry variation in profits over most of the strong evidence,however,that interindustry differences in skill mixes nineteenth century,indicating a sharp rise in capital are quite stable over time and controlling for a greater range of indi- mobility,but then a long-term increase in profit differ- vidual skill variables is not important for estimating the relative size entials beginning some time between the 1880s and the of differentialsover time.See Hiscox 2002 and Krueger and Summers 1987. 1910s,indicating a significant decline in interindustry is Note too that the latter trend fits with evidence of a long-term de- capital mobility since then.18 The evidence suggesting cline in quit rates among manufacturing workers since 1919(Hiscox high levels of capital specificity in recent years matches 2002:Ragan 1984)and with survey data on job tenure that show that the number of years spent on the same job by the average worker rose substantially between 1950 and 1990.Workers aged 55 to 64 1 The data are reported by the U.S.Securities and Exchange Com- were at their jobs an average of 16.0 years in 1991,compared with mission,in Survey of American Listed Corporations:Corporation 9.5 years in 1951;those aged 45 to 54 had been at their jobs an av- Profits (various years),and the U.S.Department of Commerce,in erage of 12.2 years in 1991,up from 7.9 years in 1951;and for those Statistical Abstract of the United States (various years). in the 35 to 44 age bracket the average tenure rose to 7.9 years in This follows Alt et al.1999.Note that the industry lists used for cal- 1991 from 4.3 years in 1951.Data are from the Employee Benefits culations of profit variation are identical to those used in the analysis Research Institute:see The Economist,January 28,1995.Economists of wages. have noted that these data clash violently with the widely held per- 18 There are no controls here for cross-industry differences in risk ception that the U.S.workforce has become increasingly mobile in or demand shocks,but Hiscox (2002)reports matching results us- response to globalization and technological change;see reports in ing measures of profits disaggregated to the four-digit SIC level to The Economist,January 28.1995,and in The New York Times,April estimate equations and control for industry-specific risk and demand- 1993. side variables. 596Commerce, Coalitions, and Factor Mobility September 2002 FIGURE 2. Interindustry Variation in Profits 0 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 Year Ib~nnual profits, % of capital in 15 industries (Census) +Annual profls, % of capital in 20 industries (Census) +Annual profits per man-hour (Census) Katz 1987: Krueger and Summers 1988).14 As Figure 1 indicates, the size of these much-discussed "indus￾try rents" trended downward markedly during ear￾lier stages of industrialization and upward only more recently.'' There is very little direct evidence on firm profits in different industries prior to 1909, when federal taxes were first imposed on corporate incomes (Epstein and Gordon 1939, 122). Beginning in 1933, data from an￾nual reports on corporation profits (as percentages of net worth and equity) are available from the U.S. Secu￾rities and Exchange Commission (SEC), categorized Only very basic controls can be applied in the aggregate data to account for heterogeneity in skill levels across industries. There is strong evidence, however, that interindustry differences in skill mixes are quite stable over time and controlling for a greater range of indi￾vidual skill variables is not important for estimating the relative size of differentials over time. See Hiscox 2002 and Krueger and Summers 1987. l5 Note too that the latter trend fits with evidence of a long-term de￾cline in quit rates among manufacturing workers since 1919 (Hiscox 2002; ~agan 1984) and with survey data-on job tenure that show that the number of years spent on the same job by the average worker rose substantially between 1950 and 1990. Workers aged 55 to 64 were at their jobs an average of 16.0 years in 1991, compared with 9.5 years in 1951; those aged 45 to 54 had been at their jobs an av￾erage of 12.2 years in 1991. up from 7.9 years in 1951; and for those in the 35 to 44 age bracket the average tenure rose to 7.9 years in 1991 from 4.3 years in 1951. Data are from the Employee Benefits Research Institute; see The Economist, January 28.1995. Economists have noted that these data clash violently with the widely held per￾ception that the U.S. workforce has become increasingly mobile in response to globalization and technological change: see reports in The Economist, January 28. 1995, and in The New York Times. April 1993. +Corporation after-tax profits, % of net worth (SEC) according to their main activities into two-digit SIC industries.16 For earlier years, following Bateman and Weiss (1981), I used census manuscripts to calculate profits (value-added minus wage costs) as a percentage of the capital invested for firms in each of the major manufacturing industries in each census year. After 1919, the Department of Commerce ceased reporting data on capital invested, but from 1947 reports total man-hours consumed per year for each industry, and these can be used as a proxy for total investments." Figure 2 charts coefficients of variation in profits across manufacturing industries using these different data series. The results generally match the pattern ex￾hibited in the wages data. There was a general decline in interindustry variation in profits over most of the nineteenth century, indicating a sharp rise in capital mobility, but then a long-term increase in profit differ￾entials beginning some time between the 1880s and the 1910s, indicating a significant decline in interindustry capital mobility since then.18 The evidence suggesting high levels of capital specificity in recent years matches '"e data are reported by the U.S. Securities and Exchange Com￾mission, in Survey of American Listed Corporations: Corporation Profits (various years), and the U.S. Department of Commerce. in Statistical Abstract of the United States (various years). l7 This follows Alt et al. 1999. Note that the industry lists used for cal￾culations of profit variation are identical to those used in the analysis of wages. l8 There are no controls here for cross-industry differences in risk or demand shocks, but Hiscox (2002) reports matching results us￾ing measures of profits disaggregated to the four-digit SIC level to estimate equations and control for industry-specific risk and demand￾side variables