¹ Easterlin, Richard A. in his investigation of variations in income by states, “State Income Estimates,” in Kuznets, Simon and Thomas, Dorothy S., eds., Population Redistribution and Economic Growth: United States, 1870-1950 (Philadelphia, 1957), Vol. I, p. 748Google Scholar, has speculated that regional variations in prices would not have much of an impact upon his estimates of state incomes. Lebergott, Stanley, Manpower in Economic Growth: The American Record Since 1800, (New York, 1964)Google Scholar; Long, rence D., Wages and Earnings in the United States, 1860-1890, (Princeton, 1961)Google Scholar; Douglas, Paul H., Real Wages in the United States, 1890-1926, (Boston, 1930)Google Scholar, and Rees, Albert, Real Wages in ufacturing, 1890-1914 (Princeton, 1961)CrossRefGoogle Scholar in their debate over the course of real wages prior to World War I relied solely upon national indexes to deflate wages. An exception is Williamson, Jeffrey G., Late Nineteenth-Century American Developments General Equilibrium History (London, 1974)Google Scholar. In aggregative, general equilibrium, regional model of the U.S. economy in the late nineteenth century, Williamson includes cost-of-living indexes for the East and Midwest as variables. Our results support his conclusion that the lower costs of living in the Midwest resulted in higher real wages there. His differential in real wages between the East and Midwest in 1890 (estimated by simulation techniques) is about one third higher than ours in Table 3 (comparing simple averages of New England and the Middle Atlantic regions with the East and West North Central regions).

² The Bureau of Labor Statistics first published comparative cost-of-living indexes in 1948 in “The City Worker's Family Budget,” Monthly Labor Review, 66 (02 1948), 133–70Google Scholar. The initial series was discontinued in 1951 and another series was resumed in 1959 in “The Interim City Worker's Family Budget,” Monthly Labor Review, 83 (08 1960), 785–808Google Scholar. From their start through the present, these indexes show substantial variations in prices among the cities and regions of the United States. The variations are not mere curiosa; it has been argued that the substantial variations in nominal wages that exist in the present day United States are insubstantial when deflated by these indexes. See Coelho, Philip R. P. and Ghali, Moheb A., “The End of the North-South Wage Differential,” American Economic Review, 61, (12 1971), 932–37Google Scholar; idem, “The End of the North-South Wage Differential: Reply,” American Economic Review, 63 (Sept. 1973), 757-62.

These indexes should not be confused with the intercity indexes the Bureau of Labor Statistics also publishes. The intercity indexes take as a base a given year for each city. The worker's family budget takes a given city (or an average) as the base, thus facilitating a comparison of the cost of living among cities in absolute terms that are not possible with the intercity indexes.

³ Coelho, Philip R. P. and Shepherd, James F., “Differences in Regional Prices: The United States, 1851-1880,” this Journal, 34, (09 1974), 551–91Google Scholar; and idem, “Regional Differences in Real Wages: the United States, 1851-1880,” Explorations in Economic History, 13, (Apr. 1976), 203-30.

⁴ There were actually two Aldrich Reports: United States Congress, Senate, Retail Prices and Wages, Senate Report No. 986, 52nd Cong., 1st Sess., 3 parts (Washington, D.C., 1892)Google Scholar; and United States Congress, Senate, Wholesale Prices, Wages, and Transportation, Senate Report No. 1394, 52d Cong., 2d Sess., 4 parts (Washington, D.C., 1893)Google Scholar. All references in this paper are to the first Aldrich Report on retail prices and wages.

⁵ Wages of farm laborers were included, however, but rural prices were not (see footnote 17).

⁶ Aldrich Report, p. I.

⁷ Prices were collected in Jackson, Mississippi, but wages were not.

⁸ Thirty of the 32 localities from which the wages from the special industries were collected were in the group of 69 towns from which the occupational wage data came. Two, Prattville, Alabama, and Lynn, Massachusetts, were not. In addition to the data already mentioned, the Aldrich Report contained a detailed budget study of 232 families from 11 cities and towns in 8 states; prices of farm products collected from many primary markets; additional retail prices of most of the 215 commodities for May 1, 1892, from three localities-Chicago, Illinois, Dubuque, Iowa, and Fall River, Massachusetts; and prices of 160 commodities in England for June 1889 and September 1891.

⁹ United States Department of Agriculture, Wages of Farm Labor in the United States: Results of Nine Statistical Investigationsfrom 1866 to 1892, Misc. Series, Report No. 4 (Washington, D.C., 1892)Google Scholar; idem, Wages of Farm Labor in the United States: Results of Eleven Statistical Investigations, 1866-1892, Misc. Series, Bull. 22 (Washington, D.C., 1901); and idem, Wages of Farm Labor in the United. States: Results of Twelve Statistical Investigations, 1866-1902, Misc. Series, Bull. 26 (Washington, D.C., 1903).

¹⁰ , Coelho and , Shepherd, “Differences in Regional Prices,” pp. 575-76 and 583Google Scholar.

¹¹ See , Coelho and , Shepherd, “Differences in Regional Prices,” pp. 563–65Google Scholar.

¹² See ibid., pp. 588-91, for the price indexes of the major commodity groups in the period 1851-1880. For contemporary evidence concerning the regional variability of housing costs, see United States Bureau of the Census, Statistical Abstract of the United States, 1973 (Washington, D.C., p. 357Google Scholar.

¹³ , Coelho and , Shepherd, “Differences in Regional Prices,” pp. 568–69Google Scholar.

¹⁴ The only southern towns from which consumer expenditure survey data are taken are Richmond, Virginia, and Savannah, Georgia. No survey data came from western towns. Table D-1 of our appendix lists these cities and towns.

¹⁵ These and subsequent data presented in this section are price relatives computed from the Aldrich Report data.

¹⁶ See , Coelho and , Shepherd, “Differences in Regional Prices,” 572 n. and 578–82Google Scholar for evidence on the convergence of regional prices and the emergence of the South as a low price region. For more recent evidence see Sherwood, Mark K., “Family Budgets and Geographic Differences in Price Levels,” Monthly Labor Review, 98 (04 1975), 8–15Google Scholar.

¹⁷ The only rural southern cost-of-living index for this period is found in Ransom, Roger L. and Sutch, Richard, One Kind of Freedom: The Economic Consequences of Emancipation (London, pp. 217–19Google Scholar. The only rural southern prices used in this index are two for food-prices of shelled corn and bacon. Retail shelled corn prices are not given in the Aldrich Report, so the only price that may be compared is that of bacon. For 1890 Ransom and Sutch suggest an average cash bacon price for rural Georgia of Vh cents per pound (p. 239). Prices in Savannah in 1890 for bacon ranged from 9 to 15 cents per pound, with a median of 10 cents, and a common range of 10-12% cents (Aldrich Report, pp. 280-81). Comparison of this one price thus suggests a substantial urban-rural difference in 1890.

However, it should be noted that southern cities were growing. In the South there was a migration from rural to urban areas. Furthermore, the evidence on the wages of farm laborers indicates that rural wages were substantially below urban. These facts lead us to expect that our conclusion about real wages in the South (especially the South Atlantic Region) would not be reversed by including data from rural areas.

¹⁸ Coelho and Shepherd, “Differences in Regional Prices.”

¹⁹ Ibid., p. 591.

²⁰ See our Appendix C, available on request.

²¹ , Coelho and , Shepherd, “Differences in Regional Prices,” p. 59Google Scholar.

²² Ibid., pp. 569-70, 572, and 579. In our earlier paper we had been skeptical about this result. Nevertheless, all the evidence supports the conclusion that urban areas in the South Atlantic region were more costly places in which to live. As stated above, however, this conclusion is subject to the qualification concerning the possible upward bias in the South Atlantic price indexes due to the use of fixed weights and the absence of rental data. Housing costs may have been lower in the South (see Table C-1 and the discussion in our Appendix C).

²³ The number of observations for data from the Aldrich Report in each occupation and the standard deviation of nominal wages in each occupation are reported in Appendix Tables E-l and E-2 The Aldrich Report did not list the number of workers in each occupation. Consequently, each wage observation is given equal weight in our tabulations. This means that our unweighted averages will not be the same as the actual average wage paid in each occupation in each region, which ideally would be computed by weighting money wages paid by the number of workers to which each wage was paid.

²⁴ Once more, the data on farm labor should be used cautiously. Aldrich Report farm wages were included in the calculations for mean wages in the United States, but USDA data were not included (see footnote 3 of Appendix A for further clarification).

²⁵ Fogel, Robert W. and Engerman, Stanley L., Time on the Cross (Boston, 1974)Google Scholar, argue that slavery did not handicap blacks in acquiring skills as much as was suggested in the traditional literature. A wave of criticism has been leveled at Fogel and Engerman, however. See, for example, David, Paul A., et. al., Reckoning With Slavery: A Critical Study of the Quantitative History of American Negro Slavery, (New York, 1976)Google Scholar.

²⁶ Coelho and Shepherd, “Regional Differences in Real Wages.”

²⁷ , Rees, Real Wages in Manufacturing, p. 33Google Scholar.

²⁸ , Douglas, Real Wages in the United States, pp. 208–11Google Scholar. This estimated daily wage was obtained by dividing the weekly wage calculated by Douglas ($10.88) by the number of hours worked (Douglas, p. 208), and multiplying the result by the average number of hours worked in a day estimated by , Rees, Real Wages in Manufacturing, p. 33Google Scholar.

²⁹ In another related research project, the classification of the 1890 census into the two-digit Standard Industrial Classification (SIC) is underway by the authors. In gathering the data, some anomalies were found. For example, in Minnesota the “Fruit and Vegetable, Canning and Preserving” industry employed 111 workers and total wages were $2,700. See United States Congress, House of Representatives, Eleventh Census of the United States: 1890, Manufacturing Industries; Part I: Totals for States and Industries, House Miscellaneous Documents, 1891-92, 52d Cong; 1st Sess., Vol. 50, Pt. 12 (Washington, D.C., 1896), p. 198. If this is taken as the average wage for a full-time yearly worker, the result is patently ridiculous. Fairly obviously, this was a seasonal industry. Taking these figures to be wages for full-time workers imparts a substantial downward bias to mean wages.

³⁰ Coelho, and Shepherd, , “Differences in Regional Prices,” p. 569Google Scholar. The 1880 and 1890 indexes compared are:

The 1880 index for the South Atlantic region represented only Jacksonville, Florida, and hence was not included in the comparison.

³¹ Ibid., pp. 576-77.

³² Ibid., p. 583.

³³ See Wilson, Lonny L., “Intercity Wage and Cost-of-Living Differentials in the United States, 1889-1939,” unpublished Ph.D. dissertation, Department of Economics, University of Iowa, 1973Google Scholar. Wilson examined the relationship of city size to money wages and the cost of living. He found that money wages tended to increase with city size, but that at no time during 1889-1939 did the cost of living bear any relationship to city size. Subject to qualifications about the limited number of consumption goods contained in his cost-of-living estimates, Wilson cautiously concluded that real wages increased with city size during this period.

³⁴ The city size classes, SC, were based upon the following geometric progression suggested by Wilson (ibid., p. 15), using 25,000 as an arbitrary base:

SC = 25,000 2ⁿ, where n = - 2, - 1, 0, 1, 2, 3, 4, 5.

After examining the results, which showed no relationship between the cost of living and city size, and in view of the small number of observations in each class, we consolidated the eight size classes into the four shown in Table 4.

³⁵ United States Congress, House of Representatives, Eleventh Census of the United States: 1890, Manufacturing Industries; Part I: Totals for States and Industries, House Miscellaneous Documents, 1891-92, 52d Cong; 1st Sess., Vol. 50, Pt. 12 (Washington, D.C., 1896)Google Scholar.

³⁶ The United States was divided into the nine aforementioned regions. There were only eight regional coefficients because putting in the ninth would make the system perfectly collinear. The omitted region was the Middle Atlantic; the intercept includes the impact of this region along with other autonomous forces.

³⁷ The corrected total sum of squares (total variation of the equation) in nominal terms (in all the wage equations) was approximately 770,000, whereas in real terms it was approximately 530,000.

³⁸ The South Atlantic and New England regions were the only regions that were significant in both real and nominal terms in the second set of equations (columns 3 and 4). Both regional coefficients increased when expressed in real terms. Equations were also tested which had “value added” per employee; this variable was constructed by subtracting from the value of output the cost of materials. The R2 did improve and the variable was highly significant, but this variable was also highly collinear with wages per employee. This is to be expected; the value added of a firm consists of its wage payments, capital costs, and profits. The largest part of these three items would be wages.

³⁹ Note that this result was not obtained in our 1851-1880 study, “Regional Differences in Real Wages.” Here the differences in the cost of living between the Northeast and the Midwest more than offset higher money wages in the former region.

⁴⁰ Pomeroy, Earl, The Pacific Slope (New York, 1966), p. 166Google Scholar.