¹ Smith, James, “Race and Human Capital,” American Economic Review, 74 (09. 1984)Google Scholar; Smith, James and Welch, Finis, Closing the Gap: Forty Years of Economic Progress for Blacks (Santa Monica, 1986)Google Scholar; Smith, James and Welch, Finis, “Black Economic Progress After Myrdal,” Journal of Economic Literature, 27 (06 1989), pp. 519–64Google Scholar; Welch, Finis, “Education and Racial Discrimination,” in Ashenfelter, Orley, ed., Discrimination in Labor Markets (Princeton, 1973)Google Scholar; Margo, Robert, “Race and Human Capital: Comment,” American Economic Review, 76 (12. 1986), pp. 1221–24Google Scholar; Margo, Robert, “Race, Educational Attainment and the 1940 Census,” this JOURNAL, 46 (03. 1986), pp. 189–98Google Scholar; and Margo, Robert, “Educational Achievement in Segregated School Systems: The Effects of ‘Separate-but-Equal,’” American Economic Review, 76 (09. 1986), pp. 794–801.Google Scholar

² Myrdal, Gunnar, An American Dilemma (New York, 1944), p. 392.Google Scholar

³ See, for example, Rowan, Richard, The Negro in the Textile Industry (Philadelphia, 1970), pp. 65–68Google Scholar; Anderson, Bernard, The Negro in the Public Utilities Industries (Philadelphia, 1970), pp. 162–71Google Scholar; Dickerson, Dennis C., Out of the Crucible: Black Steel Workers in Western Pennsylvania, 1875–1980 (Albany, 1986), pp. 240–62Google Scholar; Fogel, Walter, The Negro in the Meat Packing Industry (Philadelphia, 1970), pp. 101–10Google Scholar; Northrup, Herbert, The Negro in the Tobacco Industry (Philadelphia, 1970), pp. 2–5, 60–84CrossRefGoogle Scholar; Rubin, Lester, The Negro in the Shipbuilding Industry (Philadelphia, 1970), pp. 112–19Google Scholar; King, Carol and Risher, Howard W. Jr, The Negro in the Petroleum Industry (Philadelphia, 1969), pp. 26–29Google Scholar; and Northrup, Herbert, The Negro in the Rubber Tire Industry (Philadelphia, 1969), pp. 30–39.Google Scholar

⁴ See Jacoby, Sanford M., Employing Bureaucracy: Managers, Unions, and the Transformation of Work in American Industry, 1900–1945 (New York, 1985)Google Scholar; and Sundstrom, William, “Internal Labor Markets Before World War I: On the Job Traming and Employment Promotion,” Explorations in Economic History, 25 (10. 1988), pp. 425–45.CrossRefGoogle Scholar

⁵ See Arrow, Kenneth, “The Theory of Discrimination,” in Ashenfelter, Orley and Rees, Albert, eds., Discrimination in Labor Markets (Princeton, 1973)Google Scholar; Arrow, Kenneth, “Some Mathematical Models of Racial Discrimination in the Labor Market,” in Pascal, Anthony, ed., Racial Discrimination in Economic Life (Lexington, 1972)Google Scholar; and Lang, Kevin, “A Language Theory of Discrimination,” Quarterly Journal of Economics, 101 (05 1986), pp. 363–82.CrossRefGoogle Scholar

⁶ See Aigner, Dennis J. and Cain, Glen, “Statistical Theories of Discrimination in the Labor Market,” Industrial and Labor Relations Review, 30 (04. 1977), pp. 175–87.CrossRefGoogle Scholar

⁷ Milgrom, Paul and Oster, Sharon, “Job Discrimination, Market Forces, and the Invisibility Hypothesis,” Quarterly Journal of Economics, 102 (08. 1987), pp. 453–76.CrossRefGoogle Scholar

⁸ Lundberg, Shelly J. and Startz, Richard, “Private Discrimination and Social Intervention in Competitive Labor Markets,” American Economic Review, 73 (06 1983), pp. 340–47.Google Scholar

⁹ On these modes of thought, see Gould, Stephen J., The Mismeasure of Man (New York, 1981)Google Scholar; Fredrickson, George M., The Black Image in the White Mind (Middletown, CT, 1971)Google Scholar; and Fredrickson, George M., The Arrogance of Race (Middletown, CT, 1988).Google Scholar

¹⁰ Johnson, Charles S., The Negro in American Civilization (New York, 1930), p. 81.Google Scholar

¹¹ Smith, , Race and Human Capital; Dale Hienstand, Economic Growth and Employment Opportunities for Minorities (New York, 1964)Google Scholar; Dutcher, Dean, The Negro in Modern Industrial Society (New York, 1931)Google Scholar; Northrup, Herbert et al. , Negro Employment in Basic Industry (Philadelphia, 1970)CrossRefGoogle Scholar; and Northrup, Herbert and Rowan, Richard, Negro Employment in Southern Industry (Philadelphia, 1970).CrossRefGoogle Scholar

¹² See Heckman, James, “Heterogeneity and State-Dependence,” in Rosen, Sherwin, ed., Studies in Labor Markets (Chicago, 1981), pp. 91–139Google Scholar; and Eliwood, David T., “Teenage Unemployment: Permanent Scars or Temporary Blemishes,” in Freeman, Richard and Wise, David, eds., The Youth Labor Market Problem (Chicago, 1982).Google Scholar

¹³ See David, Paul, “Clio and the Economics of QWERTY,” American Economic Review, 75 (05 1985), pp. 332–37Google Scholar; David, Paul, “Why and When Does History Matter?” this JOURNAL, 50 (06 1990), forthcomingGoogle Scholar; Arthur, W. Brian, Ermoliev, Yuri M., and Kaniovski, Yuri M., “On Generalized Urn Schemes of the Polya Kind,” Cybernetics, 19 (03 1983)Google Scholar; and Arthur, W. Brian, Ermoliev, Yuri M. and Kaniovski, Yuri M., “Path-Dependent Processes and the Emergence of Macro-Structure,” European Journal of Operational Research, 30 (03 1987).Google Scholar

¹⁴ The sample of firms is drawn from the U.S. Housing Corporation survey of firms in late 1918. The manuscripts are housed in the National Archives. These data are unique because the survey collected firm-level information on the racial composition of work forces for the last weeks of 1913, 1916, 1917, and “last payroll” (late 1918—usually Oct. 1918). The dates are actually the first weeks of 1914, 1917, 1918, and “last payroll,” but it is best to think of the observation as the results of the previous year's employment experiences. This is the interpretation used throughout this article and is the reason why the dating convention mentioned in the text is adopted.Google Scholar

¹⁵ Firms were then placed into two-digit Standard Industrial Classification categories. All are two-digit level except Fertilizer and Cottonseed Oil. These two are singled out because they are such large employers of blacks in the South.Google Scholar

¹⁶ Firms' addresses were used to locate them on ward maps for 1920, and ward-level data on the racial composition of population were gathered from the Bureau, U.S. of the Census, 14th Census of the United States, 1920, vol. 3: Population-Composition and Characteristics of the Population By States (Washington, DC, 1920), pp. 226, 447.Google Scholar

¹⁷ Even within the rural South the degree of geographic concentration by race was high. In 1910, 45 percent of southern blacks lived in counties where they were the majority of the population. Only 4.6 percent lived in counties with less than 12.5 percent black, while 12.1 percent lived in counties that were more than 75 percent black. See Higgs, Robert, Competition and Coercion: Blacks in the American Economy, 1865–1914 (Chicago, 1980), pp. 24–32.Google Scholar Typically, blacks were concentrated in rural southern counties growing cotton with a mule and plow on small plots of land rented annually from landlords. See Whatley, Warren, “Southern Agrarian Labor Contracts as Impediments to Cotton Mechanization,” this JOURNAL, 42 (03. 1987), pp. 45–70.Google Scholar Some of the voluminous literature on the isolation of southern blacks from northern industrial labor markets includes Wright, Gavin, Old South, New South. Revolutions in the Southern Economy since the Civil War (New York, 1986)Google Scholar; Higgs, Robert, Competition and Coercion, pp.24–32Google Scholar; Mandle, Jay, The Roots of Black Poverty (Durham, 1978)Google Scholar; Myrdal, An American Dilemma, pp. 182–201Google Scholar; and Thomas, Brinley, “Negro Migration and The American Urban Dilemma,” Migration and Urban Development (London, 1972), pp. 140–69.Google Scholar

¹⁸ U.S. Department of Labor, The Negro at Work during the World War and Reconstruction (Washington, DC, 1921), p. 68.Google Scholar Also see U.S. Department of Labor, Negro Migration in 1916–1917 (Washington, DC, 1919)Google Scholar; Scott, Emmett J., Negro Migration during the War (New York, 1969)Google Scholar; Kennedy, Louise, The Negro Peasant Turns Cityward (New York, 1930)Google Scholar; and Henri, Florette, Black Migration: Movement North 1900–1920 (New York, 1976).Google Scholar

¹⁹ U.S. Dept. of Labor, Negro Migration in 1916–1917, pp. 68, 88, 105. In 1910 total black employment in Detroit was only 3,310. The labor department estimates that Detroit received more than 25,000 new black inhabitants during 1916. New Jersey also received over 25,000, and Ohio received more than 37,000. During 1917 Detroit received another 30,000 and Chicago added approximately 50,000. The Housing Corporation sample shows that between 1914 and 1916 the black share of employment in Cincinnati increased by 20 percent and most of this probably occurred during 1916. During 1917 alone the increase was 34.6 percent.Google Scholar

²⁰ See Higgs, Robert, “The Boll Weevil, the Cotton Economy, and Black Migration, 1910–1930,” Agricultural History, 50 (04 1976), p. 491.Google Scholar

²¹ Between 1900 and 1910 the outmigration rates for blacks in Virginia, Kentucky, and North Carolina were 14.8, 13.8, and 8.3 percent. Between 1910 and 1920 they were 16.1, 18.5, and 9.8 percent. In Alabama and Mississippi the rates increased from 6.3 and 5.7 percent in the 1900s to 11.9 and 12.7 percent in the 1910s.Google Scholar See Vickery, William, “The Economics of the Negro Migration, 1900–1960” (Ph.D. diss., University of Chicago, 1969), p. 26.Google Scholar The fact that unemployed southerners displaced by the boll weevil tended to migrate first may explain why the migration left the north-south gap in unskilled wages virtually unaffected. See Wright, Gavin, Old South, New South, pp. 64–69.Google Scholar

²² The following account is taken from U. S. Dept. of Labor, Negro Migration in 1916–1917.Google Scholar

²³ Flora Gill found that migration was a response to economic incentives, but the financial constraint caused by poverty and imperfect capital markets bore heavily on the decision to move. See Gill, Flora, “The Economics of the Black Exodus” (Ph.D. diss., Stanford University, 1974).Google Scholar

²⁴ Between Nov. 1, 1916, and May 1, 1917, one railroad is reported to have carried 12,731 black workers to the North. A labor recruiting firm founded in Bessemer, AL, in Feb. 1917 purchased a recruiter's license (which had risen to $1,050 to deter such a practice) and by June 1917 had shipped 4,456 blacks to northern and eastern states. See U.S. Dept. of Labor, Negro Migration in 1916–1917.Google Scholar

²⁵ McClelland, Edward S., personnel director, “The Negro in the Westinghouse Electric and Manufacturing Corporation,” Opportunity, 1 (01. 1923), pp. 22–23.Google Scholar

²⁶ Knapp, J. W., personnel director, Bethlehem Steel Corporation, “An Experiment with Negro Labor,” Opportunity, 1 (02 1923), pp. 19–20.Google Scholar For reports on other similar episodes, see The Chicago Race Commission, The Negro in Chicago (Chicago, 1922)Google Scholar; The Detroit Bureau of Government Research, The Negro in Detroit (Detroit, 1926)Google Scholar; Farnham, Dwight, “Negroes as a Source of Industrial Labor,” Industrial Management, 56 (08. 1918), pp. 123–29Google Scholar; and Johnson, The Negro in American Civilization.Google Scholar

²⁷ Detroit Bureau of Government Research, The Negro in Detroit; Chicago Race Commission, The Negro in Chicago; and U.S. Dept. of Labor, The Negro at Work during the World War, p. 10.Google Scholar

²⁸ A lower cost of living in the South may explain the lower entry-level wage in Atlanta, but it would only exacerbate any wage premium skilled white workers in Atlanta might have enjoyed over skilled whites in Cincinnati. This finding becomes more puzzling when we note that whites in Atlanta were not more educated than whites in Cincinnati. In 1920 the illiteracy rate in Atlanta was 1.2 percent for whites over the age of 10. In Cincinnati it was 0.4 percent. In 1910, 79.3 percent of whites in Atlanta between the ages of 6 and 14 were in school. In Cincinnati the figure was 90.9.Google Scholar See U.S. Bureau of the Census Thirteenth Census of the United States, Population (Washington, DC, 1913), vol. 3, part 2, pp. 418–19, and vol. 2, p. 400Google Scholar; and U.S. Bureau of the Census, Fourteenth Census of the United States, Population (Washington, DC, 1914), vol. 3, pp. 222, 784.Google Scholar

²⁹ In Atlanta a similar pattern prevailed: black workers found jobs in industries that employed large percentages of common, unskilled labor.Google Scholar

³⁰ Baruch, Bernard M., American Industry During the War (New York, 1941), chaps. 4, 6, and appendices 5–10Google Scholar; and Rockoff, Hugh, Drastic Measures: A History of Wage and Price Controls in the United States (Cambridge, 1984), chap. 3.CrossRefGoogle Scholar

³¹ This regional wage gap is one way to interpret the split labor market hypothesis. See Bonicich, Edna, “A Theory of Ethnic Antagonism: The Split Labor Market,” American Sociological Review, 37 (10 1973)Google Scholar; and Bonicich, Edna, “Advanced Capitalism and Black/White Race Relations in the United States: A Split Labor Market Interpretation,” American Sociological Review, 41 (02 1976).Google Scholar

³² The comparison, by skill category, shows the following: The figures are from U.S. Department of Labor, The Negro at Work during the World War, pp. 40–49.Google Scholar

³³ See Perlman, Jacob and Frazier, Edward K., “Entrance Rates of Common Laborers in 20 Industries, July 1937,” Monthly Labor Review, 45 (12 1937), pp. 1494, 1497. In the North the average white wage for entry-level workers in 20 industries was 55.8 cents, that for blacks was 56.6 cents. In the South the figures were 43.4 and 34.5 cents.Google Scholar

³⁴ See Bailer, Lloyd C., “The Negro Automobile Worker,” Journal of Political Economy, 51 (10. 1943), pp. 415–29CrossRefGoogle Scholar; and Northrup, The Negro in the Rubber Tire Industry, pp. 30–39.Google Scholar

³⁵ See Foner, Phillip S., Organized Labor and the Black Worker: 1619–1973 (New York, 1974)Google Scholar; Northrup, Herbert, Organized Labor and the Negro (New York, 1944)Google Scholar; and Bois, W. E. B. Du, The Negro Artisan (Atlanta, 1902).Google Scholar

³⁶ Fishback, Price, “Segregation in Job Hierarchies: West Virginia Coal Mining, 1906–1932,” this JOURNAL, 44 (09. 1984), pp. 755–74.Google Scholar

³⁷ See Dewey, Donald, “Negro Employment in Southern Industry,” Journal of Political Economy, 60 (08. 1952).CrossRefGoogle Scholar

³⁸ Chicago Race Commission, The Negro in Chicago, pp. 365–66, 385–91.Google Scholar

³⁹ The probability is equal to Wⁿ. Robert Higgs uses this formula to explain the variation in integration by scale in 1909 Virginia. See Higgs, Robert, “Firm Specific Evidence on Racial Wage Differentials and Workforce Segregation,” American Economic Review, 67 (03. 1977).Google Scholar

⁴⁰ Both cities had racially segregated residential patterns. In Cincinnati blacks were concentrated along the waterfront, and in Atlanta they were concentrated in the western-middle section of the city. As one study of workers in Cincinnati found, “less than one-third were in walking distance of their jobs,” and “when going to their jobs and returning home…the street cars are crowded, jammed, loaded-down, and decorated like bequilled porcupines with tools.” Some paid up to 10 percent of their wages on transportation.Google Scholar See Miller, Zane, Boss Cox's Cincinnati (New York, 1968), p. 32.Google Scholar For Atlanta, see Preston, Howard, Automobile Age Atlanta (Athens, 1981).Google ScholarThe Housing Corporation data show statistically significant cross-sectional variation in racial wage ratios for firms in both cities. Of the 12 wards in Atlanta with firms that employed blacks, four had average firm-level racial wage ratios that were significantly different from the city average and represented 46 percent of the firms in the sample that employed blacks. In Cincinnati, 9 of the 20 wards had average racial wage ratios significantly different from the city average and covered 35 percent of the firms that employed blacks.Google Scholar

⁴¹ Myrdal, An American Dilemma, pp. 392–93.Google Scholar Others had similarly optimistic views. In 1906 W. L. Bulkley described the situation in New York City: “The northern employer is too prone to turn off the colored applicant with the bland assurance that he himself would have no objection, but his white workmen would disrupt the business if a black competitor were forced upon them…. This intolerant attitude against the Negro workman is largely a matter of fad and fancy. Upon the show of firmness on the part of the employer it would soon vanish away.” Bulkley, W. L., “The Industrial Conditions of the Negro in New York City,” Annals of the American Academy, 27 (1906), p. 87.CrossRefGoogle Scholar

⁴² Farnham, “Negroes as a Source of Industrial Labor,” p. 128.Google Scholar

⁴³ Chicago Race Commission, The Negro in Chicago, pp. 393–98.Google Scholar

⁴⁴ Milgrom and Oster generate an equilibrium where workers who are undervalued initially are discriminated against in promotion because promotion reveals their true abilities to potential “raider” firms. The resulting allocation of labor in the firm is inefficient but profitable because firms are able to reap rents from their invisible workers. See Milgrom and Oster, “Job Discrimination, Market Forces, and the Invisibility Hypothesis.” One way to reap those rents without revealing this to the workers responsible is to segregate high-quality invisible workers and low-quality visible workers into different low-level jobs so that comparisons between them are never made.Google Scholar

⁴⁵ Economists working in both the Marxian and Coasian traditions have long emphasized how worker management influences labor performance, making a priori and independent measure of worker ability problematic. See Gintis, Herbert, “The Nature of Labor Exchange and the Theory of Capitalist Production,” Review of Radical Political Economics, 8 (Summer 1976), pp. 36–54CrossRefGoogle Scholar; Bowles, Samuel, “The Production Process in a Competitive Economy: Walrasian, Neo-Hobbesian, and Marxian Models,” American Economic Review, 75 (03 1985), pp. 16–36Google Scholar; Coase, Ronald, “The Nature of the Firm,” Economica, 4 (11 1937), pp. 387–405CrossRefGoogle Scholar; and Williamson, Oliver, “The Organization of Work,” Journal of Economic Behavior and Organization, 1 (03 1980).Google Scholar

⁴⁶ Houze, J. O., personnel manager, The National Malleable Casting Company, “Negro Labor in the Industries,” Opportunity, 1 (01 1923). pp. 20–22.Google Scholar

⁴⁷ Knapp, “An Experiment with Negro Labor,” p. 19. For similar accountsGoogle Scholar, see McClelland, “Negro Labor in the Westinghouse Electric and Manufacturing Corporation”Google Scholar; Meier, August and Rudwick, Elliott, Black Detroit and the Rise of the UAW (New York, 1979), pp. 10–12Google Scholar; and Dickerson, Out of the Crucible, chap. 5.Google Scholar

⁴⁸ Farnham, “Negroes as a Source of Industrial Labor,” pp. 124–25.Google Scholar

⁴⁹ To demonstrate the probability structure of state dependence in outcomes, think of all firms having identical urns of balls at time t, each ball having a value y_t(say, the number of black workers hired at time t). A firm randomly chooses a ball from the urn which happens to have the value y_t*. It then adds to the urn more balls with values greater than y_t*. In the next period the firm randomly chooses a ball which happens to have the value y _{t + 1}**. The outcome in period t + 1 is state dependent because the outcome in period t increased the probability that y _{t + 1}** is greater than y_t*.Google Scholar

⁵⁰ Myrdal, An American Dilemma, p. 392, fn. 6.Google Scholar

⁵¹ It is best to think of BLACK_it, as the result of the previous period's employment activity. BLACK ₁₈ is the result of decisions made between Jan. 1, 1916, and Oct. 1918. BLACK ₁₆ is the result for the period Jan. 1, 1914, to Dec. 31, 1916, and so forth. Ideally we would want continuous-state, continuous-time data so we could estimate instantaneous state dependence. Given our data limitations, the best we can do is measure the average state dependence emanating from the average experience of the previous period. The dates are chosen to make the periods approximately equal in length.Google Scholar

⁵² Cross-sectional data cannot estimate this equation because each firm has its own intercept, producing more coefficients than observations. On the other hand, estimation of the equation without the inclusion of δ_it will create an upward bias in the estimated state-dependence coefficient unless that part of δ_it that is correlated with BLACK_{it - 1} is fully captured by a linear combination of the other independent variables. One way to handle this is to fix the individual component and coefficients over time and assume Cov(n_it, n_it - 1) = 0. Now simple differencing eliminates δ_i. Time-invariant variables drop out of the equation, but here we are interested in θ as a measure of state dependence and θ is still identified.Google Scholar

⁵³ Moreover, the LF variable now measures the black percentage of population in the city, not the firm's ward, so as to pick up the effect wartime changes in the supply of black workers may have had on changes in black employment.Google Scholar

⁵⁴ LF(16 – 18) captures this. Since it is the same for all firms in the city it enters the regression as a constant, and the constant is never significant statistically.

⁵⁵ See Scott, Negro Migration during the War, chap. 8.Google Scholar

⁵⁶ Fogel, The Negro in the Meat Packing Industry, p. 105.Google Scholar

⁵⁷ Northrup, The Negro in the Tobacco Industry, pp. 73–74.Google Scholar

⁵⁸ Northrup, The Negro in the Rubber Tire Industry, p. 30.Google Scholar