In terms of the model of producers' response, the data base, and the statistical technique, this paper represents an extension of the work begun by Franklin M. Fisher and Peter Temin in their article, “Regional Specialization and the Supply of Wheat in the United States, 1867–1914,” Review of Economics and Statistics, LII (May 1970), 134–149. The questions posed here concerning cotton and southern farmers are different, but without the findings of the Fisher and Temin study for purposes of comparison, the cotton results would be difficult to interpret. I wish to acknowledge Professors Fisher and Temin for their comments ana suggestions throughout the course of the research. Several deficiencies in an earlier version of this paper were pointed out by the participants at the April 1971 Cliometrics Conference in Madison, Wisconsin, particularly Richard Sutch, Roger Ransom, and Claudia Goldin, and by the anonymous referees. Responsibility for any remaining errors is entirely mine, however.

The results reported in this paper constitute a part of my doctoral dissertation, “Agricultural Production, Supply and Institutions in the Post-Civil War South,” which is forthcoming (with additions and revisions) from the M.I.T. Press, 1974.

¹ “Crop Notes, ” Southern Cultivator, XXXII (June 1874), 241Google Scholar.

² Farmers' and planters' magazines of the post-bellum years, such as Southern Cultivator and DeBow's Review, were filled with pleas for diversification mixed with praise for cotton's profitability. For a sample of these opinions, see “Diversity of Crops, ” Southern Cultivator, XXVII (June 1869), 174Google Scholar; “Grain and Stock vs. Cotton Culture,” ibid. (July 1869), 214; “The Philosopher's Stone—Pay as You Go,” ibid., XXXII (April 1874), 129; “Extent of Cotton Planting for Next Crop,” ibid. (reprinted from N.Y. Financial Chronicle, n.d.) (May 1874), 173; “American Cotton,” Southern Cultivator and Dixie Farmer [formerly Southern Cultivator], XLVII (February 1889), 100Google Scholar; “Our Opportunity,” ibid. (October 1889), 519; “Industrial Association of Mississippi,” DeBow's Review (January 1868), 83; “Agricultural Department,” ibid. (February 1870), 187–188; Grady, Henry, “Cotton and Its Kingdom,” Harper's New Monthly Magazine, LXIII (October 1881), 719–720Google Scholar; U.S. Industrial Commission, Reports of the Industrial Commission, Vol. X: Report of the Industrial Commission on Agriculture and Agricultural Labor (Washington: Government Printing Office, 1901), ccxlii–ccxliii;Google ScholarU.S. Congress, Senate, Report of the Committee on Agriculture and Forestry on Condition of Cotton Growers in the United States, the Present Prices of Cotton, and the Remedy; and on Cotton Consumption and Production, 53rd Cong., 3rd Sess., Report 986 (Washington: Government Printing Office, 1895), Vol. I, iii–xlivGoogle Scholar. A more extensive sample of similar lines of reasoning, including quotations, may be found in DeCanio, Stephen, “Agricultural Production, Supply and Institutions in the Post-Civil War South” (Unpublished Ph.D. Dissertation, Massachusetts Institute of Technology, September 1972), 127–143Google Scholar.

³ Hicks, John, The Populist Revolt (Lincoln: University of Nebraska Press, Bison Books, 1961), p. 48Google Scholar.

⁴ “Mr. Northen's Address,” Southern Cultivator and Dixie Farmer, XLVII (September 1889), 456Google Scholar; Enoch Banks, The Economics of Land Tenure in Georgia, Vol. XXIII, No. 1 of Studies in History, Economics and Public Law (New York: Columbia University Press, 1905), 55, 101Google Scholar; [pseud.], Nicholas Worth, “The Autobiography of a Southerner since the Civil War,” Atlantic Monthly, XCVIII (1906), 171Google Scholar.

⁵ U.S. Industrial Commission, Reports of the Industrial Commission, Vol. XV: Reports of the Industrial Commission on Immigration and on Education (Washington: Government Printing Office, 1901), 533Google Scholar, 553; Vol. X: Report … on Agriculture and Agricultural Labor, 62.

⁶ Ransom, Roger and Sutch, Richard, “Debt Peonage in the Cotton South after the Civil War,” JOURNAL OF ECONOMIC HISTORY, XXXII (September 1972), 641–669Google Scholar; Laird, William E. and Rinehart, James R., “Deflation, Agriculture, and Southern Development,” Agricultural History, XLII (April 1968), 122Google Scholar; Woodward, C. Vann, Origins of the New South, 1877–1913, Volume IX of A History of the South, eds. Stephenson, Wendell Holmes and Coulter, E. Merton (Baton Rouge: Louisiana State University Press, 1951), pp. 180–184Google Scholar. For an extensive nineteenth-century statement of the argument that the credit system was ultimately responsible for cotton overproduction, see Hammond, Matthew Brown, The Cotton Industry: An Essay in American Economic History (Ithaca, N.Y., 1897), pp. 141–226.Google Scholar

⁷ Ransom and Sutch, “Debt Peonage,” 643, 655, 656, 665.

⁸ Clark, Thomas D., “The Furnishing and Supply System in Southern Agriculture Since 1865,” Journal of Southern History, XII (February 1946), 37Google Scholar.

⁹ Ransom and Sutch, “Debt Peonage,” 659, 663–664. Of course, this is not the only evidence Ransom and Sutch advance to support their general contention that southern furnishing merchants exploited farmers in the credit market. Only one part of their broad hypothesis will be examined in what follows here, namely, whether southern crop-choice decisions were the consequence of a monopolistic credit system.

¹⁰ This possibility is discussed further in DeCanio, “Agricultural Production,” Chapter VI.

¹¹ Ibid., Chs. IV-V. Of course it is possible that cotton farms were simply located on the most fertile southern lands, and that the cotton farmers would have been able to outproduce the other farmers by an even greater margin had they diversified. But if this had been the case, there should have been a positive association between cotton culture and soil fertility net of any cotton-associated fertility difference, since non-cotton crops presumably would exhibit greater yields in these particularly fertile counties. No such association can be found in the aggregate. The most reasonable conclusion is that lands were “good” partly because they were capable of producing cotton, a valuable product in great demand throughout the world. See DeCanio, “Agricultural Production,” pp. 242–244 and Appendix 5.

² Ibid., Chs. IV-V.

¹³ U.S. Department of Agriculture, Prices of Farm Products Received by Producers, 3. South Atlantic and South Central States, Statistical Bulletin No. 16 (Washington: Government Printing Office, 1927Google Scholar). These prices are as of December 1 in the indicated years. The reported “prices paid to producers” are actually the prices at which the products first changed hands when sold by the producers (Statistical Bulletin 16, p. 2). Thus with elastic demand for agricultural products in the local markets, the net prices received by the fanners themselves diverged from the reported prices by the amount of the local transportation costs. But since the determinants of unit transportation costs for the various crops were largely identical and were also stable in the short run, the ratios of net prices received by farmers for different crops must have been nearly the same as the ratios of those prices in the local market.

¹⁴ Buck, Solon J., The Agrarian Crusade: A Chronicle of the Farmer in Politics (New Haven: Yale University Press, 1920), pp. 99–201CrossRefGoogle Scholar.

¹⁵ The graphs in Figure 1 are averages over nine states of the following indices defined for each state:

Graph #1: , where

= the USDA cotton price paid to producers, from the source of Note 9 above,

WP_t= the Warren-Pearson wholesale price index of all commodities, taken from Warren, George F. and Pearson, Frank, Prices (New York: John Wiley and Sons, Inc., 1933Google Scholar), Table 1, pp. 11–13.

Graph #2: = Undeflated USDA cotton price.

Graph #3: X_t/WP^t, where

X_t is a value-weighted (1899 base) index of the prices of five of the most important southern crops. Explicitly,

where i = cotton, corn, wheat, oats, and tame hay, and v¹ = the percentage contribution of each of these crops to the total value of output of all five in 1899. The are the USDA prices and WP^t = the Warren-Pearson wholesale price index.

Graph #4: , where Y_t is a value-weighted index of the prices of the four major cotton alternatives. Formally,

where i = corn, wheat, oats and tame hay, and w¹ = the percentage contribution of each of these crops in the total output from all four in 1899.

These four relative prices were first calculated for each state separately, then were averaged over North Carolina, South Carolina, Georgia, Tennessee, Alabama, Mississippi, Arkansas, Louisiana, and Texas. Florida was omitted from the average because its prices were often widely divergent from those of the other states in level. Florida was by far the least important agricultural state throughout the period, so its inclusion in the unweighted average price indices would have distorted them needlessly. Each relative price was multiplied by an appropriate scale factor so they could all be plotted on the same figure.

¹⁶ The course of the post-bellum deflation can be ascertained from the fall in the Warren-Pearson wholesale price index (see Note 15) or from examining Friedman, Milton and Schwartz, Anna Jacobson, A Monetary History of the United States, 1867–1960 (Princeton: Princeton University Press, 1963), Chart 62, p. 678Google Scholar.

¹⁷ Woodward, Origins, pp. 185–186

¹⁸ Nerlove, Marc, The Dynamics of Supply: Estimation of Farmers’ Response to Price (Baltimore: The Johns Hopkins Press, 1958Google Scholar). See also Fisher, Franklin M. and Temin, Peter, “Regional Specialization and the Supply of Wheat in the United States, 1867–1914,” Review of Economics and Statistics, LII (May 1970), 134–149Google Scholar. A survey of supply response studies in agriculture, with an emphasis on underdeveloped countries, is contained in Behrman, Jere Richard, Supply Response in Underdeveloped Agriculture: A Case Study of Four Major Annual Crops in Thailand, 1937–1963 (Amsterdam: North Holland Publishing Company, 1968), pp. 1–20Google Scholar.

¹⁹ S^t = the ratio of harvested cotton acres to the sum of harvested acres in ten of the major alternative southern crops: wheat, potatoes, barley, corn, oats, tobacco, buckwheat, tame hay, rye, and sweet potatoes. Omission of sugar from the possible alternatives is a potential source of bias in Louisiana, but no acreage series for sugar comparable to that of the other crops could be found. , where Y^t is as defined in Note 15. Since the prices are December 1 prices and the acreage values are for crop years, the labeling of the price and acreage values corresponds to the timing of the model. For example, crop year 1883 begins in 1883, so the December 1, 1882 price is the appropriate value of the lagged price variable. 1883 planting decisions would be based on the previous year's December 1 price according to equation (6). The sources for acreage data are as follows: (i) Cotton: U.S. Department of Agriculture, Agricultural Marketing Service, Cotton and Cottonseed: Acreage, yield, Production, Disposition, Price, Value, by States, 1866–1952, Statistical Bulletin No. 164 (Washington: Government Printing Office, 1955)Google Scholar; (ii) Wheat: U.S. Department of Agriculture, Agricultural Marketing Service, Wheat: Acreage, Yield, Production, by States, 1866–1943, Statistical Bulletin No. 158 (Washington: Government Printing Office, 1955)Google Scholar; (iii) Potatoes: U.S. Department of Agriculture, Bureau of Agricultural Economics, Potatoes: Acreage, Production, Value, Farm Disposition, January 1 Stocks, 1866–1950, Statistical Bulletin No. 122 (Washington: Government Printing Office, 1953Google Scholar).

Acreage and yield data for the other crops were not found in any of the regular Department of Agriculture series such as the Statistical Bulletins or the Miscellaneous Bulletins. These acreages were instead found in mimeographed circulars printed by the U.S. Department of Agriculture, Bureau of Agricultural Economics, titled and dated as follows: (iv) Revised Estimates of Barley Acreage, Yield and Production, 1866–1929 (February 1935); (v) Revised Estimates of Corn Acreage, Yield and Production, 1866–1929 (May 1934); (vi) Revised Estimates of Oats Acreage, Yield and Production, 1866–1929 (July 1934); (vii) Revised Estimates of Tobacco Acreage, Yield and Production, 1866–1929 (August 1935); (viii) Revised Estimates of Buckwheat Acreage, Yield and Production, 1866–1929 (August 1936); (ix) Revised Estimates of Tame Hay Acreage, Yield and Production, 1866–1929 (December 1936); (x) Revised Estimates of Rye Acreage, Yield and Production, 1866–1929 (October 1935); and (xi) Revised Estimates of Sweet Potatoes Acreage, Yield and Production, 1866–1929 (February 1937).

²⁰ Fisher and Temin, “Regional Specialization,” is the source of the comparable estimates for northern and western farmers.

²¹ Ibid., 143–144.

²² Ibid., 138.

²³ The estimation procedure follows Fisher and Temin. Their description of it, with notation and equation numbering changed to conform to the notation used here, is as follows:

[G]iven ρ, [(6) was lagged and multiplied by ρ, the product was subtracted from the original equation, and ordinary least squares was applied], choosing that value of ρ which gave the smallest resulting error sum of squares. This procedure uses the first observation only as a subtraction from the second. If one believes that the process generating the disturbances…was going on for a long time when the observation period started, then [this] procedure differs from maximum likelihood (assuming the v^t normally distributed in the usual way) in its treatment of the first observation, but is asymptotically equivalent to maximum likelihood. If, on the other hand, one believes that disturbances in the recent past before the first observation were differently generated, then [this] procedure yields the maximum likelihood estimator even in small samples. (Ibid., 140–141.)

²⁴ Cooper, J. P., “Asymptotic Covariance Matrix of Procedures for Linear Regression in the Presence or First Order Serially Correlated Disturbances,” Econometrica, XL (March 1972), 305–310CrossRefGoogle Scholar.

²⁵ It should be noted that the “quasi-t-statistics” reported in Fisher and Temin were calculated after the Cooper correction had been applied to the estimated standard errors. The sample sizes for the wheat supply functions (47 observations in most cases) were sufficiently large for the asymptotic properties of the Cooper correction formulas to apply. The difference is not important for the subsequent comparisons between the estimates of the wheat supply functions and cotton supply functions which will be undertaken below, since the (unbiased) estimated coefficients themselves will be compared, not their standard errors.

²⁶ Ignoring the disturbance for purposes of the proof,

where depending on whether . Then from the definition of k(n)

Since

this reduces to:

Observe that if δ = 0, k(n) reduces to:

which is equation (12) of the text. Also, if δ is small and n is not too large (so that α≈c and the last two terms in the numerator of k(n) are small relative to the middle terms), then (12) will hold approximately.

²⁷ Fisher and Temin, “Regional Specialization.” Their estimated wheat supply parameters which will be used in the subsequent comparisons are taken from their Table 1, pp. 142–143.

²⁸ See above, note 11.

²⁹ Cooley, Thomas F. and Prescott, Edward C., “Varying Parameter Regression: A Theory and Some Applications,” Annals of Economic and Social Measurement, forthcoming (October 1973)Google Scholar, contains a general discussion of this possibility and develops estimation methods for dealing with it.