¹ The best contemporary accounts are Clark, Jane Perry, The Rise of a New Federalism (Columbia, 1938)Google Scholar and Key, V. O., The Administration of Federal Grants to States (Chicago, 1937)Google Scholar. See also Benson, George C. S., The New Centralization (New York, 1941)Google Scholar and Bitterman, Henry J., State and Federal Grants-in-Aid (Chicago, 1938).Google Scholar For a current view of New Deal federalism see Patterson, James T., The New Deal and the States (Princeton, 1969). All of these studies focus on the national and state responses to the New Deal separately, rather than on the interaction of national, state, and local governments, the focus of this paper.CrossRefGoogle Scholar

² These figures, and most of the data in the paper, were taken from U.S. Department of Commerce, Historical Statistics on Government Finances and Employment (Washington, D.C., 1969), hereafter referred to as Historical Statistics on Government. Unless otherwise noted, all data on national, state, and local government expenditures are from this source. The figure is essentially the same if revenues are used, or if, prior to 1942, military expenditures are included.Google Scholar

³ The relation between national, state, and local expenditures is slightly different when intergovernmental expenditures are attributed to the receiving government rather than the originating government or when revenues are considered, but the basic shift in the 1930s still dominates either series.Google Scholar

⁴ The figures in the text are for net state and local expenditures: total expenditures minus revenues from other levels of government. The gross figures on expenditure growth are: state growth of 84 percent ($2.8 billion to $4.5 billion) and local government growth of 20 percent ($6.4 billion to $7.5 billion). Non-militaly federal expenditures grew by 138 percent ($3.6 billion to $8.4 billion). These gross state and local figures, as well as the non-military federal expenditures, are the basis for calculating the percentage of total growth attributable to individual government functions in Table 1.Google Scholar

⁵ Cooperatively administered programs varied widely in their administrative structure. It was not necessarily the case that the direct grant programs allowed the state and local governments more flexibility than the indirect grants. The administration of the major programs is discussed in some detail in the following section.Google Scholar

⁶ The Treasury data were collected by the Office of Government Reports for the years 1933 to 1939, and the Treasury Department of 1932 and 1940. For additional analysis of these data, see Arrington, Leonard, “Western Agriculture and the New Deal,” Agricultural History (10. 1970)Google Scholar, Reading, Don, “New Deal Activity and the States 1933 to 1939,” this JOURNAL (12. 1973)Google Scholar, and Wright, Gavin, “The Political Economy of New Deal Spending: An Econometric Analysis,” Review of Economics and Statistics (02. 1974).CrossRefGoogle Scholar

⁷ WPA expenditures account for all of the increase in national direct highway expenditures and a significant portion of the increase in direct natural resource expenditures.Google Scholar

⁸ See Nourse, Edwin G., Davis, Joseph S., and Black, John D., Three Years of the Agricultural Adjustment Administration (Washington, D.C., 1937)Google Scholar, Nourse, Edwin G., Government in Relation to Agriculture (Washington, D.C., 1940)Google Scholar, and Saloutos, Theodore, The American Farmer and the New Deal (Ames, 1982).Google Scholar

⁹ Clark, New Federalism, p. 205.Google Scholar

¹⁰ For the WPA see Howard, Donald S., The WPA and Federal Relief Policy (New York, 1943)Google Scholar and MacMahon, Arthur, Millet, John, and Ogden, Gladys, The Administration of Federal Work Relief (Chicago, 1941).Google Scholar For the CCC see Salmond, John A., The Civilian Conservation Corp (Durham, 1969).Google Scholar A great deal of information on the WPA can be found in Works Progress Administration, Final Report on the WPA Program, 1935–1943 (Washington, D.C., 1946)Google Scholar and the annual Report on the Progress of the WPA, published between 1936 and 1943. The Social Security programs and their early development are described in detail in Natural Resources Planning Board, Security, Relief, and Work Policies (Washington, D. C., 1942).Google Scholar See also Altmeyer, Arthur J., The Formative Years of Social Security (Madison, 1966)Google Scholar and Douglas, Paul M., Social Security in the United States (New York, 1936).Google Scholar Annual information on the Social Security programs can be found in the United States Social Security Board, Annual Reports (1935–1942) (Washington, D.C.).Google Scholar

¹¹ There was a type of WPA project that did not require a state or local sponsor, these were the “federal” projects. The WPA federal projects were primarily white-collar projects and arts projects (theater, history, literature, and so on). After 1939 Congress almost completely eliminated WPA federal projects.Google Scholar

¹² The debate over the “political” uses of WPA funds became highly charged several times. In 1938 the race for a Kentucky Senate seat was fought over the alleged use of WPA to support the Roosevelt candidate. Later investigations did find abuses of WPA funds, not by Hopkins (the WPA administrator) but by the candidate. It also came out that his opponent had abused highway expenditures for the same political purposes. These kinds of charges were a permanent part of relief expenditures under FERA and the WPA. See Charles, Searle F., Minister of Relief, Harry Hopkins and the Depression (Syracuse, 1963)Google Scholar, MacMahon, Millet, and Ogden, Federal Work Relief, especially Ch. 12, and Howard, WPA, pp. 746–52.Google Scholar

¹³ The actual allocation policies followed by FERA and the WPA were never firmly set down. Occasionally this caused friction with Congress, which pressured Hopkins unsuccessfully to reveal exactly how he was distributing the funds. See Senate Report No. 1, 76th Congress, 1st. Session, as well as the discussion in Howard, WPA, pp. 586–604, and MacMahon, et. al., Federal Work Relief, pp. 222–25. Their general conclusion is that both FERA and the WPA did match implicitly. I test this proposition directly in the following section.Google Scholar

¹⁴ As Howard wrote, “A further cause for regret on the part of those who would like to see federal agencies create jobs wherever they appear to be needed is the limitation imposed upon the WPA by Congress which prevents the employment of workers in any given area unless local or state agencies can be induced to initiate projects and contribute a substantial proportion of their costs.” Howard, WPA, p. 546.Google Scholar

¹⁵ The TVA and the regular rivers and harbors projects were not cooperatively administered and are not discussed here.Google Scholar

¹⁶ See Bitterman, State and Federal Grants-in-Aid, for a discussion of highway programs.Google Scholar

¹⁷ Williams, Kerwin, Grants-in-Aid Under the Public Works Administration (New York, 1939).Google Scholar

¹⁸ United States Housing Authority, Housing and Urban Redevelopment (Washington, D.C., 1940)Google Scholar and Federal Housing Administration, A Handbook on Urban Redevelopment for Cities in the United States (Washington, D.C., 1941).Google Scholar For the REA see, Brown, D. Clayton, Electricity for Rural America: The Fight for the REA (Westport, 1980).Google Scholar

¹⁹ See Gramlich's, article “Intergovernmental Grants: A Review of the Empirical Literature,” in The Political Economy of Fiscal Federalism, edited by Oates, Wallace E. (Lexington, 1977)Google Scholar for an extensive review of the grants literature. Recent work has utilized more complicated theoretical and empirical techniques than those used here. See Winer, Stanley L., “Some Evidence on the Effect of the Separation of Spending and Taxing Decisions,” Journal of Political Economy (02. 1983), 126–40,CrossRefGoogle Scholar and McGuire, Martin C., “A Method for Estimating the Effect of a Subsidy on the Receiver's Resource Constraint: With an Application to U.S. Local Governments,” Journal of Public Economics (08. 1978), 25–44.CrossRefGoogle Scholar

²⁰ See the theoretical exposition in Wilde, James A., “Grants-in-Aid: The Analytics of Design and Response,” National Tax Journal (06 1971), 143–55.Google Scholar

²¹ See Gramlich, “Intergovernmental Grants.”Google Scholar

²² Gavin Wright, “The Political Economy or New Deal Spending: An Econometric Analysis.”Google Scholar

²³ Using Wright's notation where E_i is the expected gain in the probability of Democratic victory with a 1 percent shift in the distribution towards the Democrats, D_i is the Democratic share of the vote, V_i is electoral votes in state i, Pr is the probability of Democratic victory given the historical distribution of the Presidential vote in a state, and Pr₁ is the probability of Democratic victory when that distribution is shifted 1 percent towards the Democrats. The political productivity measure is then calculated by dividing E_i by 1 percent of the total vote in each state: where T_i is the total vote in state i.

²⁴ These types of equations are usually estimated by 2SLS or 3SLS. In the estimates that follow for state and federal expenditures an error components method is used. This allows the error terms in equations (4a) and (4b) to have both state and time-specific components. This estimator is described in detail in Fuller, W. A. and Battese, G. E., “Estimation of Linear Models with Crossed- Error Structure,” Journal of Econometrics (05 1974), 67–78. These estimates were calculated with the Time Series Cross Section (TSCS) package in SAS.CrossRefGoogle Scholar

²⁵ A limited annual series on local government finances was collected by the census, Financial Statistics of Cities, but this series only includes large metropolitan areas. Since large cities have considerable different relations with state and national governments than do small cities, such a sample is not representative of all local governments. In the analysis presented here expenditure data for the states for the years 1937 to 1940 were taken from U.S. Bureau of the Census, Financial Statistics of States, 1937, 1938, 1939, and 1940 (Washington, D.C., 1940, 1941, 1942. and 1943).Google Scholar State and local data for 1942 were taken from U.S. Bureau of the Census, Governmental Finances in the United States 1942 (Washington, D.C., 1945).Google Scholar State and local data for 1932 were taken from U.S. Bureau of the Census, Financial Statistics of State and Local Governments, 1932 (Washington, D.C., 1935).Google Scholar The national grant data were taken from U.S. Office of Government Reports, Vol. 10, for the years 1933 to 1939, and from U.S. Department of Treasury, Annual Report of the Secretary of the Treasury (Washington, D.C.) for 1932, 1940, and 1942. The Wright variables, PP_i, V_i, and the W_i are the same as described in the Wright article. These data were graciously supplied by Wright.Google Scholar

²⁶ Grants and expenditures for agriculture and for unemployment compensation are not considered as separate categories. Although both of these programs are clear examples of how financial incentives of the national programs affect state expenditures, the relation between the two cannot be tested via a regression since there are zeroes on one side of the ledger in both programs.Google Scholar

²⁷ The question of equalization was at the heart of several debates over policies in Social Security, relief, and agricultural programs. The problem was that the “neediest” individuals usually lived in states with the lowest levels of state and local government expenditure. To equalize the provision of services meant to give states that spent less larger national grants, which of course, created incentive problems. During the New Deal, equalization was not explicitly made a part of the Social Security program. See Altmeyer, Social Security, and the discussion in Howard, WPA, pp. 533–604.Google Scholar

²⁸ In the regressions the coefficients of primary interest are those on the endogenous variables: state expenditures and national grants. A number of other techniques (including 2SLS and 3SLS) and specifications (including first differences and log linear) were used to estimate these equations, and the only relationship between endogenous variables significantly affected by different specifications was the sign of the state expenditure variable in the national equation. It was positive, negative, and zero under alternative specifications. We cannot say much about the effect of total state or local expenditures on total national grants. On the other hand, the coefficient estimates on the endogenous variables in the relief, highway, and state expenditure equation were quite stable under alternative specifications, although the significance of the results was reduced somewhat in the first difference specification (these results are available on request). We have very little understanding of the national government allocation process, and while the Wright variables, on the whole, do fairly well, they clearly do not capture all of what is going on.Google Scholar

²⁹ Categorical grants made by the Social Security Board are perfectly matched by state expenditures for categorical relief. The estimates for relief grants and expenditures are qualitatively the same if these Social Security grants and state expenditures are excluded, although including all national relief grants and all state relief expenditures seems to be the appropriate specification, since the WPA reportedly took all relief programs into consideration when making grants. Howard, WPA, pp. 597–98.Google Scholar

³⁰ The national grants do not include grants for highway projects made by either the WPA or PWA. Interestingly, when those grants are included with the formula grants there is a positive relation between state expenditures and national grants. The WPA and PWA did match, and this shows up. The data on project breakdowns is somewhat crude, however, making it difficult to separate expenditures on highway projects, and I do not have a great deal of faith in those results.Google Scholar

³¹ All of the variables are first differences, and are defined as in Table 3. Urban, Black, Density. and Depend are differences in 1930 and 1940 values. Aid, Income, Local Debt, and Property Tax are differences in 1932 and 1942 values. Local Debt is per capita local government debt and Property Tax is the percentage of all local taxes derived from Property Tax. Other specifications included a three-equation system with a national, state, and local equation. Both first differences and levels were used. In all of the specifications there was a negative effect on grants to local governments on local government expenditures.Google Scholar

³² In this light it would appear that Congress would prefer, other things equal, not to give up any of its power to allocate funds. It is interesting to note that the early, albeit unsuccessful, relief bills proposed to allocate relief funds by a formula controlled by Congress. See Wallis, John Joseph, Relief and Unemployment in the Great Depression (Unpublished Ph.D Dissertation, University of Washington, 1981).Google Scholar

³³ See Patterson, James T., Congressional Conservatism and the New Deal (Lexington, 1967)Google Scholar and Freidel, Frank, FDR: Launching the New Deal (Boston, 1973), pp. 436–53.Google Scholar