The author acknowledges with gratitude the helpful comments of Yoram Barzel, Steven N. S. Cheung, Stanley L. Engerman, Richard B. Freeman, Edward F. Meeker, William N. Parker, and many others on earlier versions of this paper. Nuffield College, Oxford, generously provided access to computer facilities for the statistical computations.

¹ In this paper the “South” includes the following 16 states: Delaware, Maryland, Virginia, West Virginia, North Carolina, South Carolina, Georgia, Florida, Kentucky, Tennessee, Alabama Mississippi, Arkansas, Louisiana, Oklahoma, and Texas. The “Upper South” includes Delaware, Maryland, Virginia, West Virginia, and Kentucky; the “Lower South” includes the other 11 (cotton growing) states.

² Because previous census data on tenure status were subject to substantial confusion, 1910 may be the earliest date for which our questions can be answered with much confidence. In the 1910 census of agriculture, “The plantation schedules obtained from landlords were carefully compared, with the general agricultural schedules obtained for the farms in the same plantation. In fact, one of the objects of the use of the plantation schedule was. to furnish a check upon the figures returned by the tenants and also to avoid duplication in the statistics. At prior censuses considerable dlfficulty was encountered from.the fact that owners of plantations, misunderstanding the inquiry, not infrequently gave information regarding trie entire plantation, while at the same time the information regarding the land operated by tenants was secured from the tenants, thus involving duplication of acreage, value of farm property, and even of the production of crops and other agricultural products.” See U.S. Bureau of the Census, Agriculture, 1910 (Washington: G.P.O., 1913), V, p. 879Google Scholar.

³ “Full tenants” are farmers who rent all the land they cultivate.

⁴ Banks, Enoch Marvin, “The Economics of Land Tenure in Georgia,” Columbia University, Studies in History, Economics and Public Law, XXIII (1905), p. 98Google Scholar.

⁵ Boeger, E. A. and Goldenweiser, E. A., “A Study of the Tenant Systems of Farming in the Yazoo-Mississippi Delta,” USDA Bulletin No. 337 (Jan. 13, 1916)Google Scholar; Spillman, W. J. and Goldenweiser, E. A., “Farm Tenancy in the United States,” USDA Yearbook, 1916 (Washington: G.P.O., 1917)Google Scholar; Dixon, H. M. and Hawthorne, H. W., “An Economic Study of Farming in Sumter County, Georgia,” USDA Bulletin No. 492 (Feb. 10, 1917)Google Scholar; Willard, Rex E., “A Farm Management Study of Cotton Farms of Ellis County, Tex.,” USDA Bulletin No. 659 (Jan. 17, 1918)Google Scholar; Haskell, E. S., “A Farm-Management Survey in Brooks County, Georgia,” USDA Bulletin No. 648 (May 1, 1918)Google Scholar; C. O. Brannen, “Relation of Land Tenure to Plantation Organization, USDA Bulletin No. 1269 (Oct. 18, 1924); Gray, L. C. and others, “Farm Ownership and Tenancy,” USDA Yearbook, 1923 (Washington: G.P.O., 1924)Google Scholar. See also Bizzell, William Bennett, “Farm Tenantry in the United States,” Texas Agricultural Experiment Station Bulletin No. 278 (April 1921), p. 103Google Scholar.

⁶ Cheung, Steven N. S., “Transaction Costs, Risk Aversion, and the Choice of Contractual Arrangements,” Journal of Law and Economics, XII (April 1969)Google Scholar.

⁷ Ibid., pp. 27, 29–30.

⁸ Like Cheung, I conjecture that a complete analysis of the choices among owner cultivation, share-rent tenancy, fixed-rent tenancy, and cultivation by wage laborers will require an extended consideration of the transaction cost differences inherent in these afternative contractual arrangements,. What I aim to show in the present discussion is that, when the land available for rental is fixed, relative risk aversion alone is sufficient to imply an unambiguous relation between risk and the mix of rental contracts.

⁹ Brannen, “Relation of Land Tenure to Plantation Organization,” pp. 30–31, 60–61.

¹⁰ An analysis of the regression residuals reveals that in North Carolina, share contracts were substantially more prevalent for both races than the equation predicts; in Florida, on the other hand, fixed-rent contracts were unaccountably prevalent, especially among the black tenants, and Alabama's black tenants deviated substantially from the prediction in the same way. The other residuals are too small to attract much interest.

¹¹ An analysis of the regression residuals in equation (9) reveals nothing, with the one exception that both black share-rent and black fixed-rent tenants in Georgia farmed unaccountably large acreages.

¹² Cheung, Steven N. S., “Private Property Rights and Sharecropping,” Journal of Political Economy, LXXVI (Nov./Dec. 1968), 1107–08, 1117–18Google Scholar.

¹³ Tenants were younger on the average than owner-operators, and therefore presumably less experienced. For age distributions in 1920, see Goldenweiser, E. A. and Truesdell, Leon E., Farm Tenancy in the United States (Washington: G.P.O.: 1924), p. 86Google Scholar. Tenants in general were probably less skillful farmers than were owneroperators. According to Dixon and Hawthorne, “only a small percentage of \tenants] have the ability to operate farms upon their own responsibility and make a real success at the business.” See Dixon and Hawthorne, “An Economic Study of Farming in Sumter County, Georgia,” p. 12.

¹⁴ Tenant lands, particularly those operated by share croppers, are said to have been generally more fertile than the lands of owner-operators. See Edwards, Thomas J., “The Tenant System and Some Changes Since Emancipation,” Annals of the American Academy of Political and Social Science, XLIX (Sept. 1913), 41Google Scholar; Goldenweiser and Truesdell, Farm Tenancy, p. 36.

¹⁵ The land values are from the census returns. The wage rates are supposed to represent the average wage rate of outdoor labor on farms, per month for men hiring for the year, without board, in 1909; see Holmes, George K., “Wages of Farm Labor,” USDA Bureau of Statistics, Bulletin 99 (Washington: G.P.O., 1912), p. 29Google Scholar. No distinction is made between the races, partly because appropriate wage data are not available for 1909, and partly because data for the late 1890's indicate that, after correction for racial differences of payment in kind and in money, the real wages of black and white agricultural wage earners were approximately the same. See Higgs, Robert, “Did Southern Farmers Discriminate?” Agricultural History, XLVI (April 1972), pp. 325–28Google Scholar, which is based on USDA Division of Statistics, “Wages of Farm Labor in the United States,” Miscellaneous Series, Bulletin 22 (Washington: G.P.O., 1901), pp. 22–34Google Scholar.

¹⁶ W. E. B. Du Bois’ opinion was: “That the average Negro laborer today is less efficient than the average European laborer is certain.” See his paper on “The Economic Future of the Negro,” American Economic Association Publications, 3d Ser. (Feb. 1906), p. 225; and also Alfred Holt Stone, “The Economic Future of the Negro: The Factor of White Competition,” ibid., pp. 243–94; and Dixon and Hawthorne, “An Economic Study of Farming in Sumter County, Georgia,” p. 35. For mortality data that suggest much poorer health among blacks than whites, see Hoffman, Frederick L., “Race Traits and Tendencies of the American Negro,” American Economic Association Publications, XI (Aug. 1896), pp. 39–60Google Scholar. The relatively high levels of morbidity and mortality among blacks were probably due in large part to a poor diet. See Atwood, W. O. and Woods, Chas. D., “Dietary Studies with Reference to the Food of the Negro in Alabama in 1895 and 1896,” USDA Office of Experiment Stations, Bulletin 38 (Washington: G.P.O., 1897)Google Scholar; Frissell, H. B. and Bevier, Isabel, “Dietary Studies of Negroes in Eastern Virginia in 1897 and 1898,” USDA Office of Experiment Stations, Bulletin 71 (Washington: G.P.O., 1899)Google Scholar. See also the budgets for black families presented in Bois, W. E. B. Du, ed., The Negro American Family (Atlanta: Atlanta University Press, 1909), pp. 114–22Google Scholar.

¹⁷ For a brief development of this theme, see Higgs, Robert, The Transformation of the American Economy, 1865–1914 (New York: John Wiley, 1971), pp. 107–14Google Scholar. See also Engerman, Stanley L., “Some Economic Factors in Southern Backwardness in the Nineteenth Century,” in Kain, John F. and Meyer, John R., eds., Essays in Regional Economics (Cambridge, Mass.: Harvard University Press, 1971)Google Scholar.

¹⁸ Indeed, the promise has been demonstrated on several occasions. For example, in their intensive study of two Mississippi counties in the 1930's, Davis and the Gardners concluded that “the system of relatively free competition in industry has resulted in a certain amount of variance between economic status and caste status, as a result chiefly of the organization of industry so as to obtain the highest possible rate of profit.” And “in so far as the system has prevented this full extension of the caste principle, it appears to have been operating upon two principles: (1) the principle of the sacredness of private property, and (2) the principle of free competition.” See Davis, Allison, Gardner, Burleigh B., and Gardner, Mary R., Deep South: A Social Anthropological Study of Caste and Class (Chicago: University of Chicago Press, 1941), pp. 461Google Scholar, 478. See also DeCanio, Stephen, “Agricultural Production, Supply, and Institutions in the Post-Civil War South,” Journal of Economic History, XXXII (March 1972), 396–98CrossRefGoogle Scholar.

¹⁹ See Becker, Gary S., The Economics of Discrimination (Chicago: University of Chicago Press, 1957)Google Scholar. See also Thurow, Lester C., Poverty and Discrimination (Washington: Brookings Institution, 1969), pp. 111–38Google Scholar. Thurow has challenged Becker's basic contention that discrimination is costly to the discriminator, arguing that “a white gain corresponds to a Negro loss and vice versa” (p. 112). However, this conclusion rests on the assumption (p. 118) that the entire white community acts as a discriminating monopolist. This makes the question revolve—as it ultimately must in any event—about whether the whites can enforce discriminatory practices that are costly to some, though perhaps not to all, white individuals. Aside from stressing the importance of governmental sanction for discrimination, Thurow's discussion of the enforcement problem does not really come to grips with this crucial question. The literature on the enforcement of cartel practices offers some suggestive analogies to this problem. See, for example, MacAvoy, Paul W., The Economic Effects of Regulation: The Trunk-Line Railroad Cartels and the Interstate Commerce Commission before 1900 (Cambridge, Mass.: M. I. T. Press, 1965)Google Scholar; Spann, Robert M. and Erickson, Edward W., “The Economics of Railroading: The Beginning of Cartelization and Regulation,” Bell Journal of Economics and Management Science, I (Autumn 1970), 227–44Google Scholar.

²⁰ It should be obvious—but experience shows that it is not obvious—that such analysis does not necessarily glorify the market mechanism or in any way sanction the exercise of racial discrimination. It is a device of scientific economics, devoid of any ethical content whatever.

²¹ For a useful discussion of relations between the racial caste system and the economic system, see Davis, Gardner, and Gardner, Deep South, pp. 15–58, 392–400, 454–82. Perhaps the most comprehensive discussion is still Myrdal, Gunnar, An American Dilemma (New York: Harper and Brothers, 1944), pp. 205–426Google Scholar. Despite its many virtues, however, Myrdal's classic work contains much that requires modification in the light of more recent developments in both economic theory and historical investigation.

²² Bois, W. E. B. Du, The Souls of Black Folk (New York: Fawcett, 1961 \originally published 1903]), p. 85Google Scholar.

²³ Baker, Ray Stannard, Following the Colour Line (New York: Harper Torchbooks, 1964 \originally published 1908]), p. 81Google Scholar.