It is also possible to examine the effect of the introduction of the income tax in a cross-sectional/time-series environment. Such an analysis is possible because states introduced income taxes at different times, as noted in Table 1.
Accordingly, the relation of state-government spending both to the presence of income taxation and to several variables important in the literature on the growth of government is examined. The variables are for the years 1961, 1971, 1981, and 1991. There are assumed to be fixed effects throughout, so that ordinary least squares provides a BLUE estimator. The left-hand panel of Table 5 contains the results for the estimation of the following equation:
LOGTE = a + b INCOME + c POP + d TAX + e URBAN LOGTE is the natural logarithm of total expenditure by the state government, net of revenue received from the federal government. INCOMEis personal income, which has universally been found to be correlated with higher government spending as a share of gross national products in cross-national studies (e.g., Erlich and Lui, 1999; Peltzman, 1980), and URBAN is the percentage of the state’s population that lives in urban areas, as defined by the U.S. Bureau of the Census. This variable has at least once been found to be a significant explanatory variable for national-government spending (Kau and Rubin, 1981). These variables all come from the Statistical Abstract of the U.S. (various years). Finally, TAX is a dummy variable that takes a value of one if the state has any income tax, corporate or individual, in effect at the time of the observation.4
Per capita income is positively related at a highly significant level to the amount of government spending, a result that is in agreement with the growth-of-government literature. Population is of course positive at a highly significant level, a result that would be in agreement with almost any theory of endogenous government. The urban population is also significant. However, the presence of income taxation also yields, with a high degree of confidence, faster growth in government spending independently of these factors.
The results for total revenue collected by state governments, depicted in the right-hand column of Table 5, are similar. The dependent variable is the natural logarithm of revenue collected by state governments. The results are all the same in terms of signs and statistical significance for coefficients, but the overall performance of the model, as measured by adjusted R2, is slightly better. Thus, the results clearly suggest that the presence of income taxation at the state level does promote a greater degree of government spending and revenue. The next section explores some reasons why this might be true.