TESTING INCIDENCE OF A DIFFERENTIAL TAX ON INCOME FROM CAPITAL IN THE UNITED STATES
BASTANI, SHAROKH BOMAN
Doctor of Philosophy thesis
This thesis was an attempt at resolving the controversy regarding the incidence of the corporation income tax (CIT) in the United States. Three alternative hypotheses reflecting opposing points of view were constructed. The first hypothesis was the neoclassical hypothesis, which states that when the CIT is imposed, or its rate steeply increased, the net (after tax) share of capital declines in the short run, whereas in the long run it remains close to this lower level. Thus, the corporations cannot shift the tax either in the short run or in the long run. The second hypothesis was the Krzyzaniak-Musgrave (K-M) hypothesis, which states that the net share of capital remains constant (it may increase slightly) after the imposition of the CIT. Furthermore, through time it continues to remain constant until long run equilibrium is attained. Hence, the corporations shift the tax in full. The third hypothesis was the Krzyzaniak-Musgrave-Stein (K-M-S) hypothesis, which states that the short run outcome corresponds to the K-M case, while in the long run the neoclassical outcome is obtained. Econometric models were constructed in order to test these three hypotheses. In the long run it is not possible to identify systems of equations (due to the scarcity of near exogenous variables). Therefore, it was necessary to use reduced form models. Two reduced form models were constructed: (1) the 'crude' model which was used to make a choice between the three alternative hypotheses; (2) the distributed lag model which was used to determine the length of time required for long run adjustments to take place, and to throw some light on the role of expectations. Using data for the case of the United States economy for the years 1929 through 1978, in the models, the following results were obtained: (1) among the three rival hypotheses, the data supported the K-M-S hypothesis; (2) long run adjustments occur rapidly (in three time periods); and (3) expectations (based on past tax rate changes) do not play a major role. The results of the crude model are highly reliable due to the powerful nature of our tests. On the other hand, the distributed lag models turned out to be unsatisfactory because their solution is weak and they do not take into consideration the changes in the slopes and the intercepts. Thus, from the results of this thesis and those of Krzyzaniak's papers the theory of the incidence of the CIT needs to be reconstructed while keeping in mind the impossibility theorems.