We have been supported in this project by our respective institutions, Washington State University and Whitman College, and we would like to express our special thanks to W. Robert Sloan for yeoman work in developing the computer program. Roger Ransom and Richard Sutch assisted us in obtaining data and sources. Helpful comments upon an earlier draft of this paper were received from Marvin Fishbaum, William Moss, Samuel Williamson, an unidentified referee, and members of the Economic History and Development Workshop at the University of British Columbia. Janet Elfring and Hamideh Zonnoor assisted us with graphs, tables, and proofreading. All remaining errors are solely our responsibility.

¹ Our data came primarily from the following sources: United States, Department of Interior, Census Office, Tenth Census of the United States: 1880, Vol. XX, Report on the Statistics of Wages in Manufacturing Industries, with Supplementary Reports (also in 47th Cong., 2d Sess., House Misc. Doc. No. 42, Vol. 13, Part 20 [Washington, D.C.: U.S.G.P.O., 1886]).Google Scholar This volume, including the supplementary report, “Report on the Average Retail Prices of Necessaries of Life in the United States,” was compiled under the direction of Joseph D. Weeks (hereafter referred to as the Weeks Report). It is the Weeks Report which was relied heavily upon by Mitchell, Wesley C., Gold, Prices, and Wages Under the Greenback Standard (Berkeley, California: University of California Press, 1908Google Scholar; reprinted by Augustus M. Kelley, 1966), and by Hoover, Ethel D., “Retail Prices After 1850,” Trends in the American Economy in the Nineteenth Century, N.B.E.R., Studies in Income and Wealth, Vol. 24 (Princeton, N.J.: Princeton University Press, 1960), pp. 141–90.Google Scholar The data from the Weeks Report came from 85 towns and cities located principally in the New England, the Middle Atlantic, and the East North Central Regions (see Appendix Table II for a listing of these towns).

² In a recent article, Gourvish, T. R. (“The Cost of Living in Glasgow in the Early Nineteenth Century,” Economic History Review, Second Series, XXV [February 1972], 65–80)Google Scholar found substantial variations in prices between London and Glasgow during the early nineteenth century. His conclusions confirm our suspicions about using data from one region (or national data) to deflate wages in another region. See also, Gourvish, , “A Note on Bread Prices in London and Glasgow, 1788–1815,” Journal of Economic History, XXX (December, 1970), 854–860.Google Scholar

³ Easterlin, Richard A., “Interregional Differences in Per Capita Income, Population, and Total Income, 1840–1950,” Trends in the American Economy in the Nineteenth Century, N.B.E.R., Studies in Income and Wealth, Vol. 24 (Princeton, N.J.: Princeton University Press, 1960), pp. 73–140.Google Scholar Also see Easterlin, Richard A., “Regional Income Trends, 1840–1950,” Harris, Seymour, editor, American Economic History (New York: McGraw-Hill, 1961), pp. 525–547Google Scholar, reprinted in Fogel, Robert W. and Engerman, Stanley L., eds., The Reinterpretation of American Economic History (New York: Harper and Row, 1971), pp. 38–49.Google Scholar

⁴ Long, Clarence D., Wages and Earnings in the United States, 1860–1890, N.B.E.R., General Series, No. 67 (Princeton, N.J.: Princeton University Press, 1960).Google Scholar

⁵ This, of course, presumes that migrants accurately perceived and reacted to differences in money income and living costs. Our results indicate that differences in living costs amounted to a part of the differentials in money incomes. For example, our estimates show an average difference in price levels of about 16 percent between New England and East North Central towns (New England being higher) for 1851–1880 (see Table 4 below), whereas Easterlin's estimated differentials in per capita money income for these two regions are considerably larger than our differentials. (Easterlin's differentials expressed as a percentage of U.S. average per capita income are 65, 74, and 39 percentage points for 1840, 1860, and 1880, respectively, in favor of New England; Easterlin, “Regional Income Trends, 1840–1950,” in Harris, American Economic History, p. 528.) However, Easterlin's figures for per capita income in the South have been questioned in a recent article by Gunderson, Gerald, “Southern Ante-Bellum Income Reconsidered,” Explorations in Economic History, X (Winter 1973), 151–76.CrossRefGoogle Scholar

We do not suggest that current differentials in real income were the only motives to which potential migrants reacted. Certainly, expectations of increased future income after settlement had proceeded, or a desire to own land, may have played important motivating roles in internal migration. However, we believe that it is important to estimate how much living costs differed before one speculates on how much they might have mattered to migrants, or what role they may have played.

⁶ In conjunction with our emphasis on cost-of-living differences between geographic locations, one should note that our concern is not with the relation between the cost of living and city size. The latter is another major variable which is related to the urban cost of living, and which is a valid subject of investigation. The question of the cost of living and city size does arise indirectly at the end of this article, however, because we must deal with questions about the representativeness of the particular sample of cities and towns we have in each region.

⁷ For an excellent discussion of existing nineteenth-century price data, see Hoover, Ethel D., “Wholesale and Retail Prices in the Nineteenth Century,” Journal of Economic History, XVIII (September 1958), 298–316.CrossRefGoogle Scholar

⁸ This index is based upon a budget calculated for a four-person family at an intermediate living standard. The budget is a so-called standard budget, which varies from city to city, but supposedly measures equivalent standards of living in each. Differences in the cost of food, for example, reflect not only differences in price levels, but differences in regional preference patterns in the choice of foods which meet required nutritional standards. Such indexes are open to criticism because the differences among them reflect not only differences in prices, but differences in weights which are based upon estimates of the quantities consumed. United States, Department of Labor, Bureau of Labor Statistics, Handbook of Labor Statistics, 1970, Bulletin No. 1666 (Washington, D.C.: U.S.G.P.O., 1970), p. 326.Google Scholar Anchorage, Alaska, was the highest at 141 compared with nonmetropolitan southern cities (cities between 2,500 and 50,000 inhabitants) which had a rating of 86. These were, of course, widely separated cities, and Alaska is a notoriously high cost-of-living state. Nevertheless, significant differences existed between cities much closer geographically. For example, six percentage points separated Baltimore and Washington (96 and 102, respectively), two cities less than 40 miles apart. Thirteen points separated New York (110) and Pittsburgh (97), which are 378 miles from one another. Baltimore and New York were 14 points apart and 188 miles distant. It should be noted that Perloff, Harvey S., et al., Regions, Resources, and Economic Growth (Lincoln, Nebraska: University of Nebraska Press, 1965), pp. 538–41Google Scholar, do not agree with our contention. They state: “Such evidence as is available suggests that the prices of consumer goods in cities do not differ appreciably among regions” (p. 540). They do agree, however, that ruralurban cost-of-living differences are large.

⁹ Orshansky, Mollie, “Counting the Poor: Another Look at the Poverty Profile,” Social Security Bulletin, XXVIII (January 1965), 9–10Google Scholar; and Orshansky, , “Who's Who Among the Poor: A Demographic View of Poverty,” Social Security Bulletin, XXVIII (July 1965), 9–10.Google Scholar

¹⁰ See Koffsky, Nathan, “Farm and Urban Purchasing Power,” Studies in Income and Wealth, Vol. XI, National Bureau of Economic Research (Princeton, New Jersey: Princeton University Press, 1949), pp. 153–78.Google Scholar Koffsky estimated that a farm budget in terms of urban prices would cost 30 percent more, and that an urban budget cost 14 percent more than it would in terms of farm prices. D. Gale Johnson assumes a difference of 25 percent between the cost of a farm budget in farm prices and in urban prices in “Functioning of the Labor Market,” Journal of Farm Economics, XXXIII (February 1951), 76.Google Scholar

Various studies exist which pertain to the cost of living in particular rural areas. See for example, University of Minnesota, Agricultural Experiment Station, The Cost of Living on Minnesota Farms, 1905–1914, by Peck, F. W., Bulletin No. 162 (August 1916)Google Scholar; and Iowa State College, Agricultural Experiment Station, Cost of Living on Iowa Farms, by Von Tungeln, George H., Thaden, J. E., and Kirkpatrick, E. L., Bulletin No. 237 (June 1926).Google Scholar

¹¹ These questions are dealt with at length in a paper by one of the authors. See Coelho, Philip R. P., “Illusory Regional Income Differentials,” Working Paper Series No. 67 (March 1972)Google Scholar, School of Business Administration, The University of Western Ontario, London, Ontario; The Annals of Regional Science, forthcoming.

¹² For a discussion of the definitional problems see Meyer, John R., “Regional Economics: A Survey,” American Economic Review, LIII (March 1963), 20–26.Google Scholar

¹³ If the initial expansion were started by an increase in the supply of nonexportable, or partially exportable, consumer goods (such as food), then die price level, in the expanding region initially would fall, reflecting the lower consumer good prices. Immediately thereafter, a sequence of events similar to that which we depict below would take place. Mobile factors would begin to migrate, lowering price levels in the region from which emigration was occurring, and increasing price levels in the region to which immigration was taking place. Equilibrium would occur when prices in each region reached levels at which no further net migration was induced.

¹⁴ A larger proportion of the population will be participating in the labor force in A than in B because the opportunity cost of not working has risen in A as wages have risen. This presumes an upward-sloping supply curve of labor.

¹⁵ Other things being the same (including the costs of migration) more skilled workers and those families which have working wives find the returns greater simply because they earn more. Younger people find the returns greater than older people because they can expect to be working over a longer period of years. Families with fewer children, and single persons, find the costs of migration to be lower.

¹⁶ In this paper our main concern is the development of regional price indexes. We also examine, however, regional differences in prices of certain groups of related commodities, such as food, clothing, rents, and so forth (see the discussion in section II).

¹⁷ Easterlin's definitions are the same as those defined by the Bureau of the Census except that Maryland and Delaware (we have no observation for Delaware) are included in the Middle Atlantic Region rather than the South Atlantic Region. See Easterlin, “Regional Income Trends,” in Fogel and Engerman, eds., The Reinterpretation of American Economic History, p. 39.

¹⁸ Mitchell, Gold, Prices, and Wages, p. 65.

¹⁹ Hoover, “Retail Prices After 1850,” p. 155.

²⁰ Lebergott, Stanley, Manpower in Economic Growth (New York: McGraw-Hill, 1964), pp. 338–39.Google Scholar

²¹ The only exception to this rule of excluding prices that were not annual averages for commodities which may have experienced seasonal variations was for Boston. Boston prices were unspecified as to date, but they were given in ranges. From the data themselves, it appeared that these were high and low prices for the year; consequently we used the midpoint of the range to represent an annual average price.

²² Lebergott, Manpower, p. 341. Lebergott uses Pilot Knob, Missouri, as an example. It was a small mining town where company-owned housing was rented to employees. His criticism may well be relevant for Pilot Knob and other company towns, but he does not make a case that applies to the majority of the towns.

²³ In computing the index of rental prices, we used all the Weeks Report data for rental prices as the basis for this index. For each year (1860–1865) we took the average price of each type rental unit and combined them (equally weighted) to get an index taking the prices prevailing in 1860 as the base (United States 1860 = 100). The index of rental prices for the years 1860–1865 (respectively) is as follows (the results are expressed in percentile points): 100; 100; 114; 121; 132; 142.

²⁴ Rees, Albert, Real Wages in Manufacturing, 1890–1914 (Princeton, N.J.: Princeton University Press, 1961), p. 101.CrossRefGoogle Scholar

²⁵ For a discussion and listing of nineteenth-century budget studies pertaining to the United States, see Williamson, Jeffrey G., “Consumer Behavior in the Nineteenth Century: Carroll D. Wright's Massachusetts Workers in 1875,” Explorations in Entrepreneurial History, IV (Winter 1967), 98–135.Google Scholar For an exhaustive bibliography of family expenditure studies for the United States and foreign countries published prior to 1935, see United States, Department of Agriculture, Studies of Family Living in the United States and Other Countries: An Analysis of Material and Method, by Williams, Faith M. and Zimmerman, Carle C., Miscellaneous Publications, No. 223 (Washington, D.C.: U.S.G.P.O., 1935).Google Scholar

²⁶ Massachusetts, , Bureau of Statistics of Labor, Sixth Annual Report of the Bureau of Labor Statistics, Massachusetts Public Document No. 31 (March 1875)Google Scholar; and United States, Department of Labor, Seventh Annual Report [1891] (Washington, D.C.: U.S.G.P.O., 1891).Google Scholar See Williamson, “Consumer Behavior,” for an excellent discussion of Wright's 1875 Massachusetts survey as well as later ones conducted under his direction as Commissioner of Labor. The percentage distributions by major category of expenditure for these and other nineteenth-century budget studies are reported in the two Aldrich Reports: United States Congress, Senate, Retail Trices and Wages, Senate Report No. 986, 52nd Cong., 1st Sess., part 1 (Washington, D.C.: U.S.G.P.O., 1892), p. LIIIGoogle Scholar; and United States, Congress, Senate, Wholesale Prices, Wages, and Transportation, Senate Report No. 1394, 52nd Cong., 2nd Sess., part 1 (Washington, D.C.: U.S.G.P.O., 1893), pp. 86–87.Google Scholar Also, see Hoover, “Retail Prices After 1850,” for a discussion of these sources. Stanley Lebergott, Manpower, pp. 344–45, uses the 1875 Massachusetts budget study in compiling a price index for 1860–1880. A “normal” family was defined as “one which has no boarders or dependents, does not own its dwelling place, has an expenditure given for rent, fuel, lighting, clothing, and food, has both a husband and wife, and has not more than five children, no one of which is over 14 years of age.” ([Aldrich Report] Wholesale Trices, Wages, and Transportation, p. 61.)

Alternative sources which have been used by others seemed less appropriate for our purposes. The Bureau of Labor Statistics index used by Douglas, Paul H. derived its initial weights from budget studies carried out in 1901 (Real Wages in the United States, 1890–1926, [New York: Houghton-Mifflin and Company, 1930]).Google Scholar The weights used by Silbering in his study of wages in nineteenth-century England gave no weight to housing (“British Prices and Business Cycles, 1779–1850,” Review of Economic Statistics, V, Supplement 2 [October 1923], p. 234).Google Scholar Alvin Hansen's study of real wages in America also gave no weight to housing for earlier years in the nineteenth century, and for the period 1840–90 he used weights derived from family budgets in the Aldrich Reports (“Factors Affecting the Trend of Real Wages,” American Economic Review, XV [March 1925], p. 29Google Scholar; and “Correction,” American Economic Review, XV [June 1925], p. 294).Google Scholar

²⁷ Hoover, “Retail Prices After 1850,” p. 151.

²⁸ The commodities and services which Hoover includes in her index and which we exclude were not in the Weeks Report. Hoover obtained these data from a monograph by Adams, Thurston M. (Vermont, Agricultural Experiment Station, Prices Paid by Vermont Farmers for Goods and Services and Received by Them for Farm Products, 1790–1940; Wages of Vermont Farm Labor, 1780–1940, Statistical Supplement, Bulletin 507 [February 1944]).Google Scholar The data were solely from Vermont; consequently, we were unable to utilize them in the construction of regional price indexes. The commodities and services excluded were overalls, shoe repairs, fruit, medical care, and newspapers. These comprised 6.3 percent of all items in Hoover's index. In adapting Hoover's weights to our use, we merely dropped these commodities and services and assigned their weights to all other commodities proportionately. Because all but one of the excluded commodities and services were in the clothing and “other” categories (overalls, shoe repairs, medical services, and newspapers), these categories declined in relative importance while the other categories increased.

²⁹ Hoover, “Retail Prices After 1850,” pp. 149–151; and Lebergott, Manpower, p. 344.

³⁰ The expenditures included in the “other” category in the Aldrich Report, and the percentage each comprised of total expenditure were:

Aldrich Report, Wholesale Prices, Wages, and Transportation, p. 63.

³¹ The indexes resulting from varying the weights are available from the authors upon request. Also in order to test the sensitivity of our indexes to changes in regional definitions, the regional definitions were varied somewhat. In doing so, Maryland was removed from the Middle Atlantic Region, and West Virginia from the East South Central Region. The resulting indexes for these two regions were not materially affected by these deletions. The indexes are not sensitive to the particular regional definitions used.

³² Simply by changing the base year a switch may occur in the rank ordering. In the following example taken from Keynes, John Maynard (“Rents, Prices, and Wages,” Economic Journal, XVIII [September 1908], 472)CrossRefGoogle Scholar, instead of using geographical locations we can (mentally) substitute years to illustrate the problem.

Suppose rents have a weight of 1 and prices have a weight of 4 (total weights equal 5), and we have the following data using London as the base (London = 100):

Change the base from London to Ireland and the following results appear:

Ireland has a higher real wage than London if Ireland is the base. In our investigation, this is one of the main reasons for using two widely separated base years. It is the use of different base years that causes the different results in the indexes for the South Atlantic Region for the years 1869 and 1871 (see Tables 2 and 3).

³³ If a commodity (or commodities) was (were) missing from a group, the weight of that particular commodity was reassigned within the group as follows: let x_i be an observation in the j-th group, let be a missing commodity in the j-th group, then is the additional weight given to x_i for all i's in the j-th group.

If an entire group(s) was (were) missing, the weight of the group(s) was (were) reapportioned as follows: let be an empty group, let x_j be a non-empty group, then is the additional weight assigned to each group (x_j). Within the group these additional weights were assigned in proportion to the original weights.

³⁴ The index may be expressed in the following notation:

³⁵ The only town in the West North Central region for which we consistently have data in the 1850's is St. Louis, and St. Louis prices were relatively high. Prices in the rest of the region may have been lower, especially prices in rural areas. As more data for more towns became available after 1860, this may be one of the reasons that the estimated cost of living in the West North Central Region declined relative to other regions. Lebergott, Manpower, p. 346, notes other problems with the data from Cedar Rapids, Iowa.

³⁶ As noted above, one weakness of this study of regional price behavior is the small number of southern towns included in the Weeks Report. Very little can be done, at this point, to remedy this weakness until more southern price data (together with price data from regions west of the Mississippi) are assembled. Another source (Young, Edward, Labor in Europe and America [Philadelphia: S. A. George and Company, 1875])Google Scholar does exist which contains prices from all regions (including the South and West) for three years, 1867, 1869, and 1874. Young was chief of the United States Bureau of Statistics in the Treasury Department, and had worked under David A. Wells (see Lebergott, Manpower, p. 263). One difficulty in using Young's data is that he does not state how, or from what cities, the data were collected; he simply lists them by state (he does, however, specify the towns from which his wage data came, and price data may have come from the same towns). Lebergott does defend the soundness of data in another work by Young. We have computed the following regional indexes from Young (using the Weeks data for the entire United States in 1880 as the base):

The Young data do not confirm our expectations of prices in the South Atlantic Region. Other evidence, cited below, from later studies suggests this region, and the South in general, does become a relatively low cost-of-living area, but not until the 1930's. A firm conclusion, however, awaits further research.

³⁷ See the discussion of these differences above in section II, especially footnote 28.

³⁸ In examining these charts it should be noted that the sub-indexes are not only subject to all the problems and limitations of the overall price indexes, but these problems and limitations are magnified in the sub-indexes. For example, observations of prices in the 1850's are less frequent, especially for the West North Central and East South Central Regions. As in computing the overall index, when an observation of a particular commodity in a sub-index was missing, the weight of the missing commodity was assigned proportionately to other commodities for which prices were observed in that year. This is the reason for the erratic, behavior of the clothing and textile index (Figure 3) for the West North Central region in years prior to 1857.

³⁹ See the discussion of rents in the text below, and especially the source cited in footnote 58.

⁴⁰ Davis, Lance E., “The Investment Market, 1870–1914: The Evolution of a National Market,” Journal of Economic history, XXV (September 1965), 355–99CrossRefGoogle Scholar, found that interregional differences in interest rates did exist in the nineteenth century, but, unlike our cost-of-living differentials, that these differences narrowed in the latter part of the nineteenth and early twentieth centuries.

⁴¹ The variance was calculated from Table 2 (1880 = 100). The South Atlantic Region was not included in the calculation for the years 1876–1880 in order that the regions compared be the same.

⁴² The equation estimated was log V_t = α + βt, where V_t is the variance of the regional price indexes in year t, α a constant, and t the years 1851 to 1880. In our estimates of the parameters, the sign of β, the coefficient of time, was positive but not significantly different from zero at the 10 percent level of confidence.

⁴³ An F-test (using the data from Table 2) indicated that the variance for the years 1864 and 1865 was statistically different from the variance for the entire period. There may be valid objections to the use of this test because it picks up both the secular trend and the regional dispersion (see the above discussion and qualifications of the F-tests used to test for convergence of regional price levels over time). However, an analysis of year-to-year variance reveals that the years 1864 and 1865 had the smallest absolute variance with the exception of 1879 and 1880; it will be recalled that in 1864 and 1865 the price levels in each region reached their peaks for the entire thirty-year period.

The narrowing of regional price differentials during the Civil War is an interesting phenomenon. Apparently, the Civil War inflation had such a great impact upon the economy that the regional differences were overwhelmed by the nation-wide inflationary pressures. One might wonder whether a similar narrowing of regional price differences occurs whenever a rapid change in the price level takes place, such as during and after World War I, in the depression of the 1930's, and during the early post-World War II years.

⁴⁴ United States, Department of Commerce and Labor, Bureau of Labor, Eighteenth Annual Report of the Commissioner of Labor, 1903; Cost of Living and Retail Prices of Food, 58th Cong., 2nd Sess., House Doc. No. 747, Vol. 113, Part II (Washington, D.C.: U.S.G.P.O., 1904).Google Scholar

⁴⁵ Ibid., p. 655.

⁴⁶ Great Britain, Board of Trade, Cost of Living in American Towns, reprinted by United States, Congress, Senate, Senate Document No. 22, Vol. 4, 62nd Cong., 1st Sess. (Washington, D.C.: U.S.G.P.O., 1911).Google Scholar

⁴⁷ Ibid., p. lxxviii. The indexes are:

Food was given three times the weight of rent in computing the combined indexes for each region, and New York City was used as the base ( = 100). The numbers in parentheses indicate the number of cities included in each region. Most of the 28 cities were larger commercial or industrial centers; all were east of, or bordered, the Mississippi River. The combined indexes infer that only the East North Central Region exhibited significantly lower costs of living, living costs in other regions being fairly close to one another. The source of this lack of regional variation in living costs is due to the similarity of food prices among regions (as measured by this study); only ten percentage points separate the region exhibiting the lowest food costs from the highest-cost region. It is a combination of low rents together with the lowest food costs of any region which makes the East North Central Region the lowest combined cost-of-living region. If one looks at the individual cities, it is clear that rents varied much more than food prices. Rent indexes for the individual cities ranged from a low of 44 for Muncie, Indiana, to 101 for St. Louis, Missouri, whereas the food price indexes ranged only from 91 for Detroit, Michigan, to 109 for Atlanta, Georgia. Because there was a good deal of variation among cities within each region, especially with regard to rents, the interregional variations were substantially less than those among individual cities. With a few exceptions, rents seemed to be positively correlated with city size while there seemed to be little relation between food prices and city size. This probably is the major reason why New England exhibits low rents compared with the other regions; most of the New England towns included in the study, with the exception of Boston, were smaller.

⁴⁸ The National Industrial Conference Board (N.I.C.B.) published numerous studies on the cost of living in the United States from 1914 to 1940. See, for example, The Cost of Living in.the United States (New York: N.I.C.B., 1925)Google Scholar; The Cost of Living in Twelve Industrial Cities (New York: N.I.C.B., 1928)Google Scholar; and Beney, M. Ada, Cost of Living in the United States, 1914–1936, N.I.C.B. Studies, No. 228 (New York: N.I.C.B., 1936).Google Scholar

⁴⁹ The Cost of Living in the United States (1925), p. 29.

⁵⁰ Ibid. The United States = 100. When these index numbers are averaged by region, the variations among regions are less:

(The number in parentheses indicates the number of towns included in each regional index.) The magnitudes of the regional differences are similar to the 1909 study (footnote 47), but there is a shift in relative positions. The East North Central and the East South Central Regions appear to have become more expensive while the West North Central Region appears lower.

⁵¹ The Cost of Living in Twelve Industrial Cities (1928).

⁵² Ibid., p. 50.

⁵³ Ibid., pp. 52–53.

⁵⁴ The Cost of Living in the United States, 1914–1936 (1936), pp. 62–67.Google Scholar Indexes of other major groups of commodities are not given by city, so it is not possible to compute overall regional cost-of-living indexes.

⁵⁵ Stecker, Margaret L., Intercity Differences in Costs of Living in March 1935, 59 Cities, W.P.A., Division of Social Research, Monograph No. 12 (Washington, D.C.: U.S.G.P.O., 1937), pp. 162–63.Google Scholar This study was based upon estimated Sving costs for a 4-person manual worker's family at a maintenance level in 59 U.S. cities. Fixedweight budgets were constructed and used for two levels of living. The maintenance level was estimated to represent “normal or average minimum requirements for industrial, service, and other manual workers; the emergency level takes into account certain economies which may be made under depression conditions” (pp. xii–xiii). The emergency-level budget gave an almost identical difference between the highest and lowest cost of-living cities of 22.7 percentage points. Under both budgets, Washington, D.C., had the highest estimated cost of living of the 59 (the cost-of-living index was 112.2 for the maintenance level, and 112.3 for the emergency level, compared with 100.0 for the average of the 59 cities). Mobile, Alabama, had the lowest estimated cost of living using the maintenance-level budget (index of 89.6), and Wichita, Kansas, was the lowest using the emergency-level budget (also 89.6). However, when these cities were grouped according to the regional definitions we use and the unweighted mean of the index numbers taken, the regional differences are not as large as those among individual cities.

⁵⁶ The towns are Boston, Bridgeport (Conn.), Philadelphia, Buffalo, Baltimore, Newark (Jersey City was the town in the Weeks Report), Cleveland, Cincinnati, Indianapolis, Columbus (Ohio), Peoria, St. Louis, Cedar Rapids, Jacksonville, Louisville, and Memphis. Based upon these 16 cities, the following regional indexes are obtained (for a maintenance-level budget):

(The number of towns included in each regional index is indicated by the number in parentheses.) If one groups all 59 cities by region, the results are not much different. The W.P.A. study groups all 59 cities by the Bureau of Census definition of regions; the resulting regional index numbers are (for the maintenance-level budget):

The relative differences between the South Central region and the northern regions are about the same as for the 16 cities in both the W.P.A. study and the Weeks Report, but the differences among the northern regions are smaller. The indexes are not strictly comparable because of differences in regional definitions. We have included Delaware and Maryland in the Middle Atlantic region, and West Virginia in the East South Central region. The census definition includes all of these states as well as the District of Columbia in the South Atlantic region. It would appear that regional variations among these 59 cities in 1935 are less than shown by our indexes based upon the Weeks Report data, but possibly little changed from the evidence examined above for earlier years in the twentieth century. The relative positions of the different regions change once again, however. Clearly, only the southern regions stand out as lower cost-of-living areas with New England being higher.

⁵⁷ United States, Department of Labor, Bureau of Labor Statistics, Handbook of Labor Statistics, 1972, Bulletin No. 1735 (Washington, D.C.: U.S.G.P.O., 1972), p. 318.Google Scholar Individual cost-of-living indexes are given for 40 metropolitan areas (urban areas with a population in excess of 50,000). If a simple average is calculated for each region from the individual city indexes, the index for New England is 110 compared with the index for the three southern regions combined of 91 percentage points (the United States urban average equals 100 percent). The indexes for the other regions fall within a range of 98 to 102 (these indexes are unweighted means; Anchorage, Alaska, and Honolulu, Hawaii, are not included in the average for the Pacific Region). Usually, the cost-of-living indexes for the nonmetropolitan areas (urban areas of 2,500 to 50,000 persons) are lower than the indexes of the metropolitan areas. Housing costs vary more among regions than any other component of living costs. One should remember the basis upon which these indexes have been constructed and the criticism to which they are subject (see footnote 8 above).

⁵⁸ See, for example, the differences in average monthly rents in manufacturing towns in 1860 and 1861 in New England and Middle Atlantic states estimated in Martin, Edgar W., The Standard of Living in 1860 (Chicago: University of Chicago Press, 1972), p. 422.Google Scholar

⁵⁹ One recent study which has examined the trend in real wages during the latter half of the nineteenth century is Miller, M. Amata, “Regional Trends in Real Wages, United States, 1860–1880,” (unpublished Ph.D. dissertation, University of California, Berkeley, 1971).Google Scholar Miller estimates regional rates of change in retail prices and real wages, but does not think it feasible to compare regional differences in levels of consumerprices and real wages.

⁶⁰ The Weeks data have been coded, placed upon computer cards, verified, and taped. Arrangements can be made to transfer the data to any researcher who desires to work with it by writing to Coelho at Washington State University.