III. Income taxation and federal government spending in the United States Several time series illustrate the remarkable change in U.S. federal-government revenue after the implementation of the income tax. Table 4 shows the percentage of total federal government revenues derived from customs, excise, employment and individual and corporate income taxes since 1792. Once introduced, income taxes soared immediately, before being modestly dislodged in share terms by the growth of employment-based taxes in the 1950s and 1960s, mostly to fund Social Security and later Medicare. Individual income taxes in particular now account for almost half of federal-government revenue, and individual and corporate income taxes are almost sixty percent.
Figure 1 depicts the growth in the logarithm of real per capita federal government spending from 1850-2000. Also included is the trend line derived from conducting a naïve regression, i.e. one measuring only the relation between spending and elapsed time, of this value on the years 1850-1916. The estimated equation for this period is
LOGG = 3.36627 + 0.01444 t, (5)
where t is the years elapsed since 1850 (¯2 = 0.2652) . This implies that, even ignoring the other factors that might account for government growth, per capita spending has a natural momentum of growth of roughly 1.44 percent prior to the introduction of the income tax. As can be readily seen from the figure, growth in government spending after 1916 is greater than before. In fact, running the same regression for the period 1917-2000 yields
LOGG = 5.38516 + 0.04199 t, (6)
with ¯2 = 0.8210. Thus, particularly in the period in which income taxation exists, the naïve model performs surprisingly well in explaining the growth of government spending. (6) implies an annual growth rate of government spending of roughly 4.20 percent, roughly three times as high as the rate without an income tax. A Chow test performed with sub-samples of 1850-1916 and 1917-2000 yields a test statistic for a structural break at this time of F = 43.20292 (p < 0.01).
Defense and non-defense spending However, the fact that the income tax was introduced in 1916 means there is one powerful contrary hypothesis for the differential growth rate that must be considered. 1917 was the year the U.S. entered the First World War, and a case can certainly be made that this conflict was a watershed event with respect to the U.S.’s role in the world. While the American colonial experience, and hence engagement with the larger world, went back at least to the taking of Spanish colonies during the Spanish-American war, it was the Great War that first drew American soldiers overseas to fight a war that had begun without them, and which launched a long ongoing era in which U.S. national interests were defined around events all over the world. If Versailles set the stage for World War II, which in turn set the stage for the Cold War, with its concomitant massive projection of U.S. military might around the world, then perhaps the rise in government spending after 1916 was primarily a function of the wider role the U.S. began to take in the world after that time. Perhaps the income tax was even seen as necessary to support the imminent entry into the war.3
To assess this possibility, Figures 2 and 3 show the same growth rate for the logarithms of both real per capita defense and non-defense spending for the same period. In fact, the structural break in 1917 is far more pronounced for non-defense than for defense spending. In Figure 2, defense spending, the naïve model estimated for 1850-1916 and applied to 1917-2000 is again depicted, along with actual logarithmic per capita defense spending. Figure 3 depicts the same data and result for non-defense spending, exclusive of interest on the national debt. For defense spending the estimations for 1850-1916 and 1917-2000, respectively, are
LOGGNON = 4.50194 + 0.04698t. (1917-2000, ¯2 = 0.9165) (8b)
Only in the case of non-defense spending before 1917, when the permanent defense establishment was much smaller, is the adjusted R2 low. Otherwise, the naïve model does strikingly well both in explaining spending growth and in demonstrating a difference between the period before and after 1916. The rise in the rate of growth in spending is substantial for both defense and non-defense spending. In both cases, the introduction of the income tax is associated with a rise in annual spending to in excess of four percent. The simple test does not allow refutation of the idea that the increase in U.S. defense spending is caused by considerations external to the introduction of the income tax (e.g., exogenous historical events such as the rise of fascism and communism), but the corresponding rise in non-defense spending suggests that even if this hypothesis is true income taxation nonetheless precedes a higher rise in domestic spending for reasons, presumably, unrelated to such external geopolitical considerations.