¹ Williamson, J. G., “The Relative Costs of American Men, Skills, and Machines: A Long View,” Discussion Paper No. 289–75, Institute for Research on Poverty, University of Wisconsin-Madison (July 1975)Google Scholar. In the pages that follow, I shall refer at length to the data contained in that paper. It is available on request. So too are two longer working papers upon which the present version is based.

² Peter Lindert and I are collaborating on a paper that will survey our knowledge of American wealth and income distribution over the past three centuries. Among other things, the paper will establish a very high correlation between the structure of wages, class pay differentials, income distribution statistics by size as well as wealth distribution trends. The “urban inequality index” in Figures 1 and 2 is based on a linked series measuring the ratio of skilled to common labor. See Williamson, “The Relative Costs,” Tables 6 and 14, pp. 22 and 25.

³ The full employment forces associated with inflation are very important in any accounting of the income leveling after 1939, but only the cost of living incidence of the1 inflation is at issue here.

⁴ See, for example, Seers, D., Changes in the Cost-of-Living and the Distribution of Income Since 1938 (Oxford, 1949)Google Scholar and Brittain, J. A., “Some Neglected Features of Britain's Income Leveling,” American Economic Review, 50 (May 1960), pp. 593–603Google Scholar.

⁵ Cost of living indices abound for isolated regions and occupational groups. For example, the 1890–1929 period supplies at least five such indices for “average urban workers.” Albert Rees' index was preceded by the contributions of Willford King and Paul Douglas, and Douglas himself cites seventy-three publications dealing with cost of living estimates limited to the 1890–1926 period alone. See Rees, A., Red Wages in Manufacturing, 1890–1914 (New York, 1961)CrossRefGoogle Scholar and Douglas, P., Real Wages in the United States (Boston, 1930), pp. 656–667Google Scholar. With the exception of King, however, none of these studies attempts to compare cost of living changes by socioeconomic class. King supplies indices for five socioeconomic classes, 1909–1928, in his The National Income and Its Purchasing Power (New York, 1930), ch. 3, pp. 65–72Google Scholar. To my knowledge, his is the only serious attempt to confront the issue prior to the 1930's. King's pioneering contribution has been all but ignored by subsequent researchers.

Aside from the recent paper by Coelho, Philip and Shepherd, James, “Differences in Regional Prices: The United States, 1851–1880,” Journal of Economic History, 34 (September 1974), 551–591CrossRefGoogle Scholar, there are no relevant studies whatsover for the nineteenth century. I use the Coelho-Shepherd price data extensively below.

⁶ Taking all consumption expenditures of the average workingman in 1918–1919 as a base of 100, the unskilled labor intensity of food ranked 117 while apparel, consumer durables, and miscellaneous consumer goods ranked 80, 68, and 79 respectively. Williamson, J. G., “Who Pays for the Services of the Working Poor?” Discussion Paper No. 334–76, Institute for Research on Poverty, University of Wisconsin-Madison, February 1976Google Scholar.

⁷ I am in essence appealing to the very old chestnut that characterizes industrialization as “labor-saving” (and thus skills- and machine-using). Nominal inequality results as the relative demand for low-wage labor declines. These historical episodes may also be associated with heavy immigration. If so, the nominal inequality trends will be even more striking. The impact on the relative prices of labor-intensive necessities may be somewhat moderated in the very short run by “population sensitive” demand effects.

⁸ Habakkuk, H. J., American and British Technology in the Nineteenth Century (Cambridge, 1962)Google Scholar; Kindleberger, C. P., Europe's Postwar Growth: The Role of Labor Supply (Cambridge, 1967)CrossRefGoogle Scholar; and Lindert, P., Fertility and Scarcity in America (Princeton, 1976, forthcoming)Google Scholar.

⁹ This characterization of the driving force behind structural change certainly has a legion of adherents. See, for example, Kuznets, Simon, Modern Economic Growth (New Haven, 1966), pp. 98–101Google Scholar. I am making a somewhat different appeal, however, since my interest is in distribution and not structural change. A key assumption to my use of the “technological dualism” thesis in the text is the following: total factor productivity growth is most rapid in sectors which are most intensive in machines and skills. It seems innocuous and it plays a dominant role in, for example, Habakkuk, American and British Technology, pp. 50, 61, and 160–161. One example of unbalanced technological progress in American history may serve to motivate this discussion. As a case in point, compare Robert Gallman's estimates of total factor productivity growth in antebellum agriculture with Robert Zevin's and Paul David's estimates for cotton textiles: Gallman, R. E., “Changes in Total U.S. Agricultural Factor Productivity,” Agricultural History, 46 (January 1972), 208Google Scholar; Zevin, R. B., “The Growth of Cotton Textile Production After 1815,” in Fogel, R. W. and Engerman, S. L., eds., The Reinterpretation of American Economic History (New York, 1971), p. 146Google Scholar; David, P., “Learning By Doing and Tariff Protection: A Reconsideration of the Case of the Ante-Bellum United States Cotton Textile Industry,” Journal of Economic History, 30 (September 1970), 521–601CrossRefGoogle Scholar. The annual rate in manufacturing exceeds that of agriculture by 3.05 percent from 1815 to 1859. Judging by Figure 1, the first inequality surge in America coincided with very unbalanced rates to total factor productivity growth. The same is true of the second inequality surge, 1899–1929, when Kendrick's figures document a differential of 1.40 percent. From 1869 to 1899, on the other hand, the differential is only 0.30 percent and this, of course, in a period of quiescence in inequality trends. As a final contrast, the “revolutionary income leveling” after 1929 coincided with balanced rates of total factor productivity performance since the differential was −0.16 percent.

Long run trends in inequality and rates of technological dualism seem to correlate very well.

¹⁰ Ashton, T. S., “The Treatment of Capitalism by Historians,” in Hayek, F. A., ed., Capitalism and the Historian (London, 1954), p. 51Google Scholar.

¹¹ On the relative costs of antebellum machines, see my “Watersheds and Turning Points: Conjectures on the Long Term Impact of Civil War Financing,” Journal of Economic History, 34 (September 1974), 636–661CrossRefGoogle Scholar and “The Relative Costs.” For the 1900's, see my “The Sources of American Inequality, 1896–1948,” Review of Economics and Statistics, 58 (forthcoming 1976)Google Scholar.

¹² This paragraph makes many assertions about the factor intensity of nineteenth century industries. They strike me as highly plausible, but we have only early twentieth century evidence to confirm these assertions. The direct and indirect unskilled labor cost share in producer durables sectors was 0.13 in 1919, the lowest “raw” labor intensity anywhere in America. In contrast, the figure for the consumption of farm products was 0.24 in the same year. Williamson, “Who Pays for the Services of the Working Poor?” The direct unskilled labor cost shares for the same year were the following: agriculture (0.36), mining and quarrying (0.29), and manufacturing (0.26). By 1929, the range was even wider: agriculture (0.40), and manufacturing (0.19). These last mentioned data can be found in my “Demand and the Distribution of Income: America, 1913–1929,” paper presented to the Sixth International Congress on Economic History, Copenhagen, Denmark (August 19–23, 1974).

¹³ Some readers may take objection to this argument by pointing out that transportation was one sector which was technologically dynamic, was very capital intensive, but which favored the cost-of-living of the urban poor. After all, did not costreducing productivity changes in internal transport drive down the relative price of urban foodstuffs? Not necessarily. Most of the gain accrued to the farmer and he does not play an explicit role in our accounting for American urban inequality experience.

¹⁴ All commodity price indices are Warren-Pearson wholesale quotations originally gathered mainly in New York City. The “price” of servants was approximated by the nominal daily wage of common labor. The interested reader can find far greater detail reported in my two earlier working papers: “Prices and Urban Inequality: American Cost of Living by Socioeconomic Class, 1820–1948,” Paper No. EH 74–26, Graduate Program in Economic History, University of Wisconsin, Madison (August 1974); “Strategic Wage Goods, Prices, and Inequality,” Discussion Paper No. 294–75, Institute for Research on Poverty, University of Wisconsin, Madison (September 1975). For Dorothy Brady's budget share estimates, see her “Consumption and the Style of Life,” in Davis, L. E., et al., American Economic Growth: An Economists' History of the United States (New York, 1972)Google Scholar.

¹⁵ See Appendix Table. The growth rate is much lower when the 1820–1856 years are examined. Real unskilled wages grew at 0.82 percent per annum over the 1820–1856 period. The 1.21 figure, therefore, is to be viewed as an upper bound.

¹⁶ Stanley Lebergott reaches a different conclusion. He found a much more modest rise in real wages, even taking into account that our estimate is an upper bound. He estimated the rise to be only 17 percent up to 1850. Up to 1857, he found no rise at all. Lebergott, S., Manpower in Economic Growth (New York, 1964), pp. 149–150Google Scholar. Paul David estimates a real wage improvement 1818–1860 of about 1.9 percent per annum in his “The Growth of Real Product in the United States Before 1840,” Journal of Economic History, 27 (June 1967), Table 7, p. 178Google Scholar. The discrepancy between us lies primarily in the cost of living index used. David's purpose was restricted to an estimate of the real cost of farm labor. Our purpose is to measure the real income improvement of the urban poor.

¹⁷ Calculated from David, “The Growth of Real Product,” Table 1, p. 155.

¹⁸ Gallman, R. E., “The Pace and Pattern of American Economic Growth,” in L. E. Davis, et al., American Economic Growth: An Economists' History of the United States (New York, 1972)Google Scholar.

¹⁹ Up to this point we have been content simply to establish whether high income families suffered smaller cost of living increases than poor families during phases of trending nominal inequality. The result is forthcoming regardless of our definition of “high income.” How do we translate this finding into a quantitative measure of the importance of price changes on real income differentials? Unfortunately, the secular income distribution evidence for the nineteenth century is almost exclusively limited to wage structure measures and these supply only the most limited earnings range. Obviously, the impact of differential cost of living patterns will depend on the range of income classes examined; the wider the range, the greater the variance in budget weights, and thus the larger the difference in cost of living performance. Since we wish to make broad inferences about trends in urban inequality, it seems sensible to apply the “rich” cost of living index to the “top” of our nominal pay structure in assessing the likely quantitative importance of urban prices. We do so for much of the remainder of this paper.

²⁰ The rate of quality improvement is assumed to apply to all nonfood and nonservice expenditures. Robert Gordon tells us that consumer durables improved in quality at a rate roughly equal to 2.5 percent per annum, 1935–1953. The size of the downward bias on the ante-bellum relative cost-of-living of the poor, is simply (.50 – .35) × (2.5 percent) = +.38 percent per annum. The two figures in parentheses refer to budget weights of rich less that of poor and they are the weights reported in Table 3. For the twentieth century evidence, see R. J. Gordon, “Measurement Bias in Price Indexes for Capital Goods,” Review of Income and Wealth, 17 (June 1971), Table 4, p. 144.

²¹ Williamson, “The Sources of American Inequality.”

²² The rise is even more dramatic on the farm. The farm cost of living index rose by 41 percent, reinforcing the urban bias. Only after 1890 are the data sufficient to augment our analysis of urban socioeconomic classes.

²³ This paragraph may appear to argue that real unskilled wages rose during the Civil War. On the contrary, taking 1860 = 100, the real wage for unskilled workers was 77.5 in 1864; for skilled workers it was 74.5. Thus, these new cost of living indices socioeconomic class do not erase “Mitchell's Paradox.” (While real wages declined during the Civil War, the Appendix Table shows that they rose during World Wars and II. Family annual earnings is another matter entirely.) Nor should the decline in real wages imply the “wages share” diminished during the war. In sharp contrast to twentieth-century wars, Civil War financing relied far more heavily on indirect taxes both internal and external). As a result, real disposable income fell for all social classes as public military expenditures rose markedly as a share in GNP. Furthermore, family incomes among poor urban households may have behaved differently than average earnings of the employed unskilled—depending on who entered the military. The point remains that unskilled workers fared considerably better in real terms than did skilled workers, professionals or property income recipients.

²⁴ Garrett, P. W., Government Control Over Prices (Washington, D.C., 1920), p. 35Google Scholar.

²⁵ A more extensive argument can be found in my “Demand and the Distribution of Income” where these demand effects for the 1914–1929 period are discussed in detail.

²⁶ Some might well argue that the rich could easily absorb the penalty, but this paper avoids such judgments.

²⁷ Not quite, since between 1914 and 1919 the relative price of food declined. It did rise up to 1918.

²⁸ Nor does it lie with price controls. My analysis of cost of living looks only at the years 1936 and 1948. With few exceptions, all wartime price controls were lifted by late 1947.