American Political Science Review Vol.96,No.3 FIGURE 1.Interindustry Variation in Wages 40 35 20 15 8 10 1800 1820 1840 1880188019001920194019601980 2000 Year -Annual eamings in 15 industries (Census)-Annual eamings in 20 industries(Census) Hourty wage rates for unskilled workers (NICB)Hourty eamings of prod.workers (BLS) -Annual eamings of prod.workers (Census)-Daity wages of common labor(Weeks) There are good reasons for exercising caution when on annual wages of production workers in two-digit SIC examining wage and profit differentials,since they may industries is readily available from the Department of partly reflect other features of factor markets besides Commerce,12 hourly earnings for production workers mobility (these issues are discussed further below).It are calculated after 1947 by the Bureau of Labor Statis- is the size of industry rents that is key for the political tics(BLS),13 and separate data on hourly wages for un- story here,however,and wage and profit differentials skilled workers between 1920 and 1937 were compiled are the clearest measure we have of whether such rents by the National Industrial Conference Board(Glasser actually exist.9 1940,36). Using each of the data series to calculate coefficients Interindustry Variation in Wages and Profits of variation across industries yields an interesting set of results.The data,shown in Figure 1,indicate two broad Following Long(1960)I use data on wage payments trends:a general decline in interindustry variation in reported in the decennial census to calculate annual wages over the course of the nineteenth century,con- wages for workers in major manufacturing industries sistent with a marked rise in interindustry labor mobil- (approximates of the modern two-digit Standard In- ity,and a general increase in wage variation beginning dustrial Classification [SIC]categories)for each census sometime between the 1910s and the 1930s,indicating year beginning in 1820.10 I also calculated average daily a steep decline in interindustry mobility more recently. wages of"common laborers"in each of these industries These different trends have been noted separately from the payroll records of firms compiled in the Weeks by analysts focusing on particular eras (e.g.,Atack report of 1886.11 After the turn of the century,evidence Bateman,and Margo 2000;Bell and Freeman 1991), and the evidence of sizable differences in wages across industries in recent years is also consistent with much Hiscox(2002)providesa detailed discussion and treatment of these recent work by labor economists using more detailed measurement issues and a more detailed analysis of all the available evidence on historical trends in U.S.factor mobility. survey data on individual workers (e.g.,Dickens and 10 I began with the 17 industries examined by Long(1960,72-73) for the period 1860 to 1890,amending the list to extend the series and Pennsylvania only(since all data were entered manually for each for 15 of these industries for which data are available over the period hrm). 1820 to 1910.I then created a separate series for 20 industries,adding 12 See the U.S.Department of Commerce's,Census of Manufactures five categories that were excluded from Long's study but for which and Annual Survey of Manufactures (various years).Beginning in data exist over the full span of years.All lists,and original data,are 1900,earnings data are reported for 15 two-digit SICindustries;from available from the author. 1 The Weeks report was published as U.S.Congress,House(1886).I Employmen an amin calculated simple averages across firms in Massachusetts,New York, (various years). 595American Political Science Review Vol. 96, No. 3 Year +Annual earnings in 15industries (Census) +Annual earnings in 20 industries (Census) 4Houriy wage rates for unskilled workers (NICB) tHouriy earnings of prod. workers (BLS) I+Annual earnings of prod. workers (Census) tDaily wages of cwnmon labor (Weeks) There are good reasons for exercising caution when examining wage and profit differentials, since they may partly reflect other features of factor markets besides mobility (these issues are discussed further below). It is the size of industry rents that is key for the political story here, however, and wage and profit differentials are the clearest measure we have of whether such rents actually exist.9 Interindustry Variation in Wages and Profits Following Long (1960) I use data on wage payments reported in the decennial census to calculate annual wages for workers in major manufacturing industries (approximates of the modern two-digit Standard In￾dustrial Classification [SIC]categories) for each census year beginning in 1820." I also calculated average daily wages of "common laborers" in each of these industries from the payroll records of firms compiled in the Weeks report of 1886." After the turn of the century, evidence Hiscox (2002) provides a detailed discussion and treatment of these measurement issues and a more detailed analysis of all the available evidence on historical trends in U.S. factor mobility. lo I began with the 17 industries examined by Long (1960, 72-73) for the period 1860 to 1890, amending the list to extend the series for 15 of these industries for which data are available over the period 1820 to 1910. I then created a separate series for 20 industries, adding five categories that were excluded from Long's study but for which data exist over the full span of years. All lists, and original data, are available from the author. l1 The Weeks report was published as U.S. Congress, House (1886). I calculated simple averages across firms in Massachusetts, New York, on annual wages of production workers in two-digit SIC industries is readily available from the Department of commerce,12 hourly earnings for production workers are calculated after 1947 by the Bureau of Labor Statis￾tics (BLS),'~ and separate data on hourly wages for un￾skilled workers between 1920 and 1937 were compiled by the National Industrial Conference Board (Glasser 1940,36). Using each of the data series to calculate coefficients of variation across industries yields an interesting set of results. The data, shown in Figure 1, indicate two broad trends: a general decline in interindustry variation in wages over the course of the nineteenth century, con￾sistent with a marked rise in interindustry labor mobil￾ity, and a general increase in wage variation beginning sometime between the 1910s and the 1930s, indicating a steep decline in interindustry mobility more recently. These different trends have been noted separately by analysts focusing on particular eras (e.g., Atack, Bateman, and Margo 2000; Bell and Freeman 1991), and the evidence of sizable differences in wages across industries in recent years is also consistent with much recent work by labor economists using more detailed survey data on individual workers (e.g., Dickens and and Pennsylvania only (since all data were entered manually for each firm). l2 See the U.S. Department of Commerce's, Census of Manufactures and Annual Survey of Manufactures (various years). Beginning in 1900, earnings data are reported for 15 two-digit SIC industries; from 1947, they are reported for 19 industries. l3 See the U.S. Bureau of Labor Statistics' Employment and Earnings (various years)