^{page 514 note 1} “It would suggest that when we have done the necessary research, the over-all contribution of government investment in the nineteenth century will be a modest one”; North, Douglass, Growth and Welfare in the American Past (Englewood Cliffs, N. J.: Prentice-Hall, 1966), p. 104.Google Scholar

“State and local activity and finance were limited during the colonial period and the succeeding half century…State governments during the second half century ceased to be an active developmental factor limiting their activities to supervising municipal operations and regulating private enterprise… For the first few years of national government, Federal expenditures, tax revenues and borrowing, exceeded state and local expenditures, tax revenue and borrowers. The rapid movement of the population, the growth of cities, the extension of the suffrage, and the ascent of laissez-faire philosophy terminated for a time the predominance of Federal activity and finance. Beginning in the 1820's government activity,… except during wartime, grew more rapidly at the periphery than at the center”; Studenski, P. and Krooss, H. E., Financial History of the United States (New York: McGraw-Hill, 1952), pp. 2 and 6-7.Google Scholar

^{page 515 note 2} Take, for example, the case of the expenditure on education. If perfect mobility of labor existed, labor trained in high school-expenditure areas would flow to the areas of low expenditures until marginal products in the two regions were equated, and the region financing the added education would reap no greater benefits than those garnered by the regions spending nothing at all.

^{page 515 note 3} In the future we expect to examine variations in the functional breakdown of receipts and expenditures as well as to attempt to estimate the differential growth that resulted from alternative policies.

^{page 515 note 4} Intercensal population estimates are based on linear interpolations.

^{page 515 note 5} For a discussion of the data and the problems of estimation, see Appendix A. Because of the difficulties in allocating military expenditures, no attempt has been made to determine the spatial distribution of lie federal budget for the year 1865.

^{page 516 note 6} The regional breakdown employed in this paper is the standard Census Bureau classification: Region I (New England) includes the states of Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont; Region II (the Middle Atlantic) includes New Jersey, New York, and Pennsylvania; Region III (the East-North-Central) Illinois, Indiana, Michigan, Ohio, and Wisconsin; Region IV (the West-North-Central) Iowa, Kansas, Missouri, Minnesota, Nebraska, and North and South Dakota; Region V (the South Atlantic) Delaware, Florida, Georgia, Maryland, North and South Carolina, Virginia, and West Virginia; Region VI (the East-South-Central) Alabama, Kentucky, Mississippi, and Tennessee; Region VII (the West-South-Central) Arkansas, Louisiana, Oklahoma, and Texas; Region VIII (the Mountains) Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah, and Wyoming; and Region IX (the Pacific) California, Oregon, and Washington.

^{page 517 note 7} Between 1860 and 1880 while national population increased 67 per cent, New England's population rose by only 27 per cent.

^{page 517 note 8} The three states of Maine, Vermont, and New Hampshire, for example—with a combined population about equal to that of South Carolina—were receiving twenty times as much in federal pensions.

^{page 518 note 9} A regression model that makes per capita expenditures or receipts a function of time and urbanization shows a significant improvement in its multiple correlation coefficient when a set of regional dummy variables is added.

^{page 519 note 10} In some sense this conclusion is at variance with that reached by Engerman, Stanley L.. See “The Economic Impact of the Civil War,” Explorations in Entrepreneurial History, sec. ser.; III, No. 3 (1966), 191Google Scholar . We recognize, of course, that Engerman is concerned with redistribution over income groups and our interest is spatial; however, given regional income differences, these are not always different.

^{page 519 note 11} Throughout most of the following discussion no attempt will be made to distinguish between receipts and expenditures, because the models of receipts yield almost die same results as do those of expenditures. Instead we will talk of government activity and specify those few times when the receipt models yielded significantly different results than the expenditure models.

In general, the models take per capita state expenditures or receipts as the dependent variable and examine the impact of time, urbanization, location, and income on it. The cross-sectional models exclude time (needless to say), and income is included only in the cross-sectional model for 1840, 1880, and 1900.

^{page 520 note 12} The same phenomena appear in the case of 20 of the 25 states whose fiscal histories span three fourths of a century or more when they are examined independently.

^{page 520 note 13} Income and urbanization are not as closely correlated as one might suppose. For 1880 the simple coefficient of determination between the two is only .24, and for 1900 it is .54.

^{page 520 note 14} There are only 16 observations in 1840, and no model that was tried appears to predict very well for 1900.

^{page 520 note 15} “State Expenditure and Revenue, 1902-1950,” a paper delivered to the Third International Economic History Conference in Munich, Germany, 1965.

^{page 522 note 16} The actual levels are well below those of Regions I and II, and 15 to 20 per cent below those of Regions III and IV. A model using nothing but time and regional variables and excluding Maryland and the reconstruction years from the South shows expenditure in Region V to be $1.72 below New England and that in Region VI to be $1.85 below, while the average for Regions II, III, and IV was only $1.17 below.

^{page 522 note 17} Post-reconstruction thinking may well have had some long-run effects. In North Carolina, for example, where peacetime expenditures averaged less than 50 cents per capita in both the preceding and succeeding decades, they reached $8.30 in 1869 and averaged over $4.50 for the three years 1868-70.

^{page 523 note 18} The regressions presented in this section are based on tax receipts rather than on total receipts, because the former are available for all six censuses while the latter are not. However, for the years 1890 to 1902, when both were available, an analysis of variance suggests that there are no regional differences in the ratio of tax to total receipts. Moreover, in both these cases the regression models produced similar results for tax and total receipts.

The models themselves were similar to those employed in the state analysis, but the weight was on the cross-sectional rather than on the aggregate models. In general the dependent variable was per capita tax receipts and the independent was location, urbanization, and income.

^{page 523 note 19} No matter whether treated as a linear or a dummy variable, it contributed significantly to the “explanation” of the level of per capita tax receipts.

^{page 524 note 20} In the two decades from 1890 to 1910 the number of cities over 500,000 increased by 100 per cent, and the number over 100,000 rose by 150 per cent.

^{page 524 note 21} Easterlin's figures were used for per capita state incomes. For 1902 the 1900 estimates were employed. See Easterlin, R., “Interregional Differences in Per Capita Income, 1840-1950,” in Trends in the American Economy in the Nineteenth Century (Princeton, N. J.: Princeton University Press for NBER, 1960).Google Scholar

^{page 525 note 22} Because of the break in federal data, the years 1861 to 1869 are excluded.

^{page 525 note 23} In the case of Region I the jump is from about $10 to $30; for II from $10 to $20; for III from $7.50 to $15; for IV from $10 to $15; and for V from $5 to $15. Only in the cases of Regions VI and VII is there a break in the pattern. In the case of the former the upward trend spans the entire period, and in the latter there is no substantial increase after 1837.

^{page 526 note 24} Hughes, J. R. T. and Rosenberg, Nathan, “The United States Business Cycle before 1860: Some Problems of Interpretation,” Economic History Review, XVI (03 1963), pp. 487–88Google Scholar . It should be noted, however, that Hughes and Rosenberg consider only ordinary expenditures and receipts.

^{page 527 note 1} Because of changing definition of a fiscal year, the 1843 figure is the stated 1843 total plus one half the 1844 figure.

^{page 528 note 2} Bogart, Ernest L., Financial History of Ohio (New York: Longmans Green, 1924)Google Scholar ; Boyle, James E., The Financial History of Kansas (Madison: University of Wisconsin, 1908)Google Scholar ; Ely, R. T., Taxation in American States and Cities (New York, 1888)Google Scholar ; Eastman, F. M., Taxation for State Purposes in Pennsylvania (Philadelphia, 1898)Google Scholar ; Fankhauser, Wm. C., A Financial History of California (Berkeley: University of California Press, 1913)Google Scholar ; Hannah, Hugh S., A Financial History of Maryland (Baltimore: Johns Hopkins University, 1907)Google Scholar ; Hollander, Jacob H., (ed.), Studies in State Taxation (Baltimore: Johns Hopkins University, 1900)Google Scholar ; Jewett, Fred E., A Financial History of Maine (New York: Columbia University Press, 1937)Google Scholar ; Phelan, R. V., The Financial History of Wisconsin (Madison: University of Wisconsin, 1908)Google Scholar ; Rhodes, M. C., History of Taxation in Mississippi (Nashville, Tenn.: Geo. Peabody, 1930)Google Scholar ; Snyder, W. P., Compendium and Brief History of Taxation in Pennsylvania (Harrisburg, Penn.: 1906)Google Scholar ; Sydenstricker, Edgar, A Brief History of Taxation in Virginia (Legislative Reference Bureau of Virginia, 1915)Google Scholar ; and Taff, N. O., History of State Revenue and Taxation in Kentucky (Nashville, Tenn.: Geo. Peabody, 1931)Google Scholar.

^{page 552 note 3} Inclusion of time improves the R2 significantly, to .73, but produces insurmountable problems of interpretation.

^{page 552 note 4} The 1.44 figure is the average ratio of total local receipts to local tax receipts from the 1902 census. An analysis of variance on the interregional differences in these ratios suggests that there were no significant regional differences, and the K-S normalcy tests suggest that the 48 ratios were normally distributed about their 1.44 average. The 1920 average was used in preference to that of 1890 because (1) industrial receipts were reported net in the 1890 census and (2) estimated unreported receipts are a much higher proportion of the total in the 1890 census.