¹ See Bateman, Fred, Foust, James D., and Weiss, Thomas J., “Large Scale Manufacturing in the South and West, 1850–1860,” Business History Review, XLV (Spring, 1971), 1–17CrossRefGoogle Scholar; “The Participation of Planters in Manufacturing in the Antebellum South,” Agricultural History, XLVII (April, 1974), 277–297; and “Profitability in Southern Manufacturing: Estimates for 1860,” Explorations in Economic History, XII (July, 1975), 211–231. See also Bateman, and Weiss, , “Comparative Regional Development in Antebellum Manufacturing,” Journal of Economic History, XXXV (March, 1975), 182–208CrossRefGoogle Scholar; “Market Structure Before the Age of Big Business: Concentration and Profit in Early Southern Manufacturing,” Business History Review, XLIX (Autumn, 1975), 312–336; and “Manufacturing in the Antebellum South,” in Uselding, Paul, ed., Research in Economic History, 1 (Greenwich, Conn., 1976), 1–44.Google Scholar A monograph, Bateman and Weiss, Industrialization in the Slave Economy, is forthcoming from the University of North Carolina Press.

² While the subject of profitability is discussed in several of the papers, it is most comprehensively examined in Bateman and Weiss, “Manufacturing in the Antebellum South,” in Uselding, ed., Research in Economic History.

³ Part way through their collaboration, Professor Foust left academic life, so some of the profitability studies included him as co-author while others, notably the piece in Research in Economic History, did not. Referring to the “BFW” studies is sometimes inaccurate since Prof. Foust is included when in fact he was not involved, but we regarded this as less objectionable than excluding Prof. Foust's initial altogether.

⁴ On southern agriculture, see, Conrad, Alfred and Meyer, John, “The Economics of Slavery in the Antebellum South,” Journal of Political Economy, LXVI (April, 1958), 312–336Google Scholar; Fogel, Robert W. and Engerman, Stanley, Time on the Cross (Boston, 1974)Google Scholar, or Vedder, Richard K. and Stockdale, David, “The Profitability of Slavery Revisited: A Different Approach,” Agricultural History, XLIX (April, 1975), 392–404.Google Scholar These studies show rates of return on slaves and/or agricultural capital of from 5.7 to 10.6 per cent. Interest rates in 1860 were about 5 per cent on long-term government or high grade municipal bonds, and about 6 per cent on higher grade railroad bonds. See Homer, Sidney, A History of Interest Rates (New Brunswick, N.J., 1963), 309–310.Google Scholar

⁵ Phillips, Ulrich B. “The Economic Cost of Slaveholding in the Cotton Belt,” Political Science Quarterly, XX (June, 1905), 257–275.CrossRefGoogle Scholar

⁶ Income (as opposed to wealth) statistics for the antebellum period are scanty. For three perspectives on the subject, see Soltow, Lee, “Economic Inequality in the United States in the Period from 1790 to 1860,” Journal of Economic History, XXXI (December, 1971), 122–139Google Scholar, or Soltow, , Men and Wealth in the United States (New Haven, 1975)Google Scholar; Robert Gallman, “The Pace and Pattern of American Economic Growth,” in Davis, Lance, Easterlin, Richard, and Parker, William, eds., American Economic Growth: An Economist's History of the United States (New York, 1972)Google Scholar; and Lindert, Peter H. and Williamson, Jeffrey G., “Three Centuries of American Inequality,” in Uselding, Paul, ed., Research in Economic History, Vol. I (1976), 69–123.Google Scholar

⁷ The estimate of the growth in manufacturing capital was calculated from decennial census statistics of manufacturing capital for 1850 and 1860. See United States Department of the Interior, Census Office, Report on the Manufacturing Industries in the United States, Part I (Washington, D.C., 1895), 4.Google Scholar If BFW are correct in their estimates of profits and the new estimates of manufacturing capital formation are also approximately accurate, some two-thirds of profits were paid out as dividends, assuming firms did not alter the proportion of capital financed by debt. We could not find evidence indicating that dividend payouts typically were that high or that firms moved significantly away from debt financing over time. On the latter point, Paul McGouldrick suggests that the debtequity ratio rose in cotton textiles in the late antebellum period in New England Textiles in The Nineteenth Century, (Cambridge, Mass., 1968), 14.

⁸ Bateman and Weiss, “Manufacturing in the Antebellum South,” in Research in Economic History, 8. In part, certain states may have not been examined because census manuscripts have been destroyed. This is the case of Rhode Island, for example. Other moderately important states excluded from the BFW sample included Illinois (for 1850) and Georgia.

⁹ By any input or output criteria, manufacturing in the New England and Middle Atlantic states was more than twice as large as in the southern and western states, yet there were many more firms from the South and West (3,545) in the 1860 sample than from New England and the Middle Atlantic states (2,060). See Bateman and Weiss, “Manufacturing,” in Research, 4 and 8. The results on profitability by region are summarized on p. 25. An unknown bias in the BFW study results from the fact that 30 per cent of firms were in just two industries, lumbering and flour milling.

¹⁰ Bateman and Weiss, “Manufacturing,” in Research, 21.

¹¹ Or so Bateman and Weiss argue. Ibid., 25.

¹² Ibid.

¹³ Ibid., 6.

¹⁴ This might have some bearing on trying to explain why large firms were apparently less profitable despite the presence (in 1860 at least) of constant returns to scale.

¹⁵ By “inevitable human errors” we are not suggesting that BFW were unusually careless. We are just expressing the view that it is not obvious that the census enumerators were necessarily grossly more inaccurate in summing data than, say, BFW. Mistakes in aggregation of numbers did not end with the introduction of computer-based scholarship in the mid-twentieth century. Moreover, a 100-per cent representation of firms, ceteris paribus, seems preferable to a sample of fewer than 5 per cent of firms that is not strictly random in character.

¹⁶ Manufactures in the United States in 1860 (Washington, D.C., 1865), 742.

¹⁷ Report on the Manufacturing Industries in the United States, 86–87.

¹⁸ Ibid., 4 and 8. A strong case be made for relating working capital and miscellaneous costs to output, not fixed capital. We chose at this point to replicate the procedures used by BFW.

¹⁹ The depreciation formula is that used by BFW.

²⁰ Wright, C. D. and Hunt, W. C., History and Growth of the United States Census Office (Washington, D.C. 1900), 313.Google Scholar

²¹ See Report on the Manufacturing Industries of the United States, Part I, 4.

²² Ibid.

²³ In several later calculations, we increased our estimate of miscellaneous costs to 12 per cent of gross output ($226 million). While taxes may have been lower in 1860 than 1890, other items were probably more important, relative to output, in 1860 than 1890, most notably interest expense and probably rental payments. Interest rates were much higher in 1860 than in 1890. For example, railroad bonds yielded 6 per cent in 1860 but only about 3.3 per cent in 1890. See Homer, A History of Interest Rates, 310, 316. Both the presence of higher interest rates in 1860 and probably deflationary expectations in 1890 would lead us to predict that nominal rents in 1860 were higher than in 1890. Power costs also may have declined over time owing to improvements in technology.

In reading the McLane Report as well as industry studies for the antebellum period, one gets the impression that “Miscellaneous costs” were indeed important. Payments of funds to other businesses for contract work were common. For example, the White Stone textile factory of Providence, R.I., paid $2,500 for “bleaching and bailing goods” in 1831, some 5.4 per cent of output. Another 10 per cent of output went for transportation expenses. See U.S. House Documents No. 308, Vol. 7, Documents Relative to the Manufacturing of the United States, Part I, (Washington, D.C., 1833), 947–948. Frequent references are made in the literature to royalty payments of the use of patented technology. See, for example, Cole, Arthur H. and Williamson, Harold F., The American Carpet Manufacture (Cambridge, Mass., 1941), 61.Google Scholar Also, the “putting out” system was still common in some manufacturing in 1860. Blanche Hazard, for example, dates the beginnings of the “factory stage” in boot and shoe manufacture to 1855 in The Organization of the Boot and Shoe Industry in Massachusetts Before 1875 (Cambridge, Mass. 1921), 97–98. Its presence may have had the effect of increasing miscellaneous costs of firms, depending on how they classified the work performed under such a system. The lack of detail in the Census with regard to the various miscellaneous cost items makes it a rather crude source for calculating profits.

²⁴ See “Manufacturing in the Antebellum South,” in Research in Economic History, 18–19 for some discussion of the BFW procedures.

²⁵ Wright and Hunt, History and Growth of the U.S. Census Office, 313. The instructions listed are actually for the 1850 Census but the authors assert elsewhere that no major changes occurred between 1850 and 1860.

²⁶ See, for example, the studies by Fogel and Engerman, Time on the Cross, and Vedder and Stockdale, “The Profitability of Slavery Revisited,” Agricultural History, on agriculture. For a good study of antebellum manufacturing profitability, see MeGouldrick, New England Textiles. All of these studies show rates of return of around 10 per cent for 1860.

²⁷ We assume various combinations of the following: manufacturing capital equal to the BFW assumption; manufacturing capital equal to our earlier cited assumption ($1,624 million); manufacturing capital equal to $1,330 million, which assumes one half of working capital was in fact reported; miscellaneous costs equal to 10 per cent of total output; miscellaneous costs equal to 12 per cent of total output; labor costs equal to the amount reported in the Census and used by BFW, labor costs increased by 21.89 per cent ($81 million). It should be repeated that these calculations are on the basis of published Census data for all firms; the rates of return are weighted by the size of firm, unlike with the BFW results.

²⁸ The 23 per cent return assumes that all the calculations of costs made by BFW were correct, but that they grossly overstated manufacturing capital (by 52 per cent, using $2,035 million instead of $1,330 million). If the BFW cost calculations are correct but the true amount of manufacturing capital was $1,624 million, one obtains a rate of return of about 17 per cent. On the other hand, if one assumes that BFW's capital estimate is correct but that they understated labor and miscellaneous costs, one obtains a rate of return of a bit below 6 per cent; if one assumes that they understand only labor costs, the estimated return is 7.5 per cent.

²⁹ This conclusion, of course, does not necessarily conflict with the BFW finding that the return on investment was higher in southern manufacturing than in northern manufacturing. We must confess some skepticism on that point, however. Labor costs equalled 31 per cent of manufacturing capital in 10 southern states and the District of Columbia, compared with 46 per cent for the rest of the country, if the Census numbers are to be believed, in 1860. This startling difference may be explained by lower wages in the South, but we are concerned that slave labor costs were not fully stated. Either slaves should have been counted as manufacturing capital, with their maintenance expenses recorded as a labor cost, or, if not counted as capital, their full cost as a labor input should have been enumerated, including the opportunity cost of the value of slave output (e.g., the hire rate.) Slaves were important in Southern manufacturing. Half of Virginian factory workers were slaves, according to Robert, Joseph C., The Tobacco Kingdom (Gloucester, Mass., 1965), 198.Google Scholar Claudia Goldin makes similar remarks in Urban Slavery in the American South 1820–1860. (Chicago, 1976), 21. Some arithmetic exercises suggest that if understatement of slave labor costs were substantial, much if not all of the North-South rate of return differential can be explained away.

³⁰ Bateman and Weiss report a national profit rate three percentage points lower in 1850 than in 1860, which is consistent with our calculations of the difference in those rates. While we believe their estimates are probably biased, we know of no reason why the magnitude of the bias should have changed substantially over time.

³¹ Bateman and Weiss, “Manufacturing in the Antebellum South,” 8. Slight decreasing returns are observed for 1850 and constant returns in 1860.

³² In a Cobb-Douglas production function of the form V = A L^α K^β where V = value added, L = labor, K = capital, α the output elasticity of labor input and β the output elasticity of capital input, the marginal productivity of capital is Regression analysis is used to estimate α and β the output elasticities of the two inputs.

³³ The data came from BFW, “Manufacturing,” 8, and Manufactures in the U.S. in 1860, 742.

³⁴ For example, Paul David estimates the output elasticity of his capital variable to have been between .11 and .27, depending on the variation of the model used. David was looking at six antebellum textile firms. His estimates are highly consistent with ours. See David, Paul, “Learning by Doing and Tariff Protection: A Reconsideration of the Case of the Ante-Bellum U.S. Textile Industry,” Journal of Economic History, X, (September, 1970), 565.Google Scholar Paul Uselding, estimating Cobb-Douglas functions from time series data, obtains values for the output elasticity of the capital input of from .18 to .29, depending on the model used. Again, these values are similar to those observed in our statistical estimation and markedly below those observed by BFW. See Uselding, Paul, “Technical Progress at the Springfield Armory,” Explorations in Economic History, IX (Spring, 1972), 291–307.Google Scholar In a comment on the Uselding paper, David Klingaman, Richard Vedder, and Lowell Gallaway estimated Cobb-Douglas production functions for iron goods industries, obtaining capital elasticities of between .19 and .31; Klingaman, , Vedder, , and Gallaway, , “The Ames-Rosenberg Hypothesis Revisited,” Explorations in Economic History, XI, (Spring, 1974) 311.CrossRefGoogle Scholar

³⁵ The data are for several eastern states, including Massachusetts, Connecticut, New Hampshire, and Pennsylvania. The data are located in state capitals or at the National Archives and were primarily gathered by our colleague David Klingaman, whose assistance we gratefully acknowledge.

³⁶ McGouldrick, New England Textiles.

³⁷ Ibid., 94. Profits as a percent of net worth are estimated by McGouldrick to have been 12.7 for 1860, slightly higher than the 11.5 per cent estimate for all industry obtained using BFW procedures or 10 per cent obtained using “refined” BFW procedures.

³⁸ Ames, Edward and Rosenberg, Nathan, “The Enfield Arsenal in Theory and History,” Economic Journal (December, 1968).CrossRefGoogle Scholar For other studies that emphasize the importance of raw materials see Habakkuk, H. J., American and British Technology in the Nineteenth Century (Cambridge, England, 1967)Google Scholar and David, Paul, Technical Choice, Innovation, and Economic Growth (Cambridge, England, 1975).Google Scholar

³⁹ A more detailed exposition of the various models can be found in Vedder, Richard, Gallaway, Lowell, and Klingaman, David, “Discrimination and Exploitation in Antebellum American Cotton Textile Manufacturing,” Uselding, Paul, ed., Research in Economic History, Vol. III (Greenwich, CT., 1978) 217–262.Google Scholar

⁴⁰ The McLane Report data are exclusively for Massachusetts and New Hampshire, while the 1850 data are for Connecticut.

⁴¹ Homer, A History of Interest Rates, 304–305, observes interest rates on highgrade bonds of between 4 and 6 per cent throughout the period 1830–1860.

⁴² Cyclical variation in profits was relatively pronounced in antebellum manufacturing. In the 1836–1860 period, the annual rate of return for Baker sample cotton textile firms varied from -5.6 to +19.1 per cent. McGouldrick, New England Textiles, 94. This is consistent with the casual empiricism of Bezaleel J. Taft, writing in 1832: “The average rate of profit for the last three years … may be set down at 7 percent per annum. In 1829, say nothing; in 1830, 15 percent, and in 1831, 6 percent.” See the McLane Report, 81. Taft finds the variation largely attributable to tariff legislation.

⁴³ Calculated from McGouldrick, New England Textiles, 94.

⁴⁴ Bateman and Weiss, “Manufacturing in the Antebellum South”.