Test for Tax Effects in the Pricing of Financial Assets

This paper tests four competing hypotheses regarding taxation and the pricing of financial assets: no differential tax effects in the pricing of stocks or bonds; the equilibrium condition described by Miller (1977); the two-parameter after-tax pricing equations of Brennan (1970) and Litzenberger and Ramaswamy (1979); and dividend-related clientele effects described by Elton and Gruber (1970) and Litzenberger and Ramaswamy (1980). These hypotheses make substantially different predictions about the impact of taxes on financial decisions; however, when stated in terms of returns the hypotheses imply parameter restrictions with the same functional form. The commonality of the parameter restrictions allows simultaneous testing of all four models by testing the form of the parameter restrictions. This approach differs from that of previous researchers who have by and large implicitly assumed that the parameter restrictions implied by the respective models hold (see Elton and Gruber 1970; Black and Scholes 1973, 1974; Litzenberger and Ramaswamy 1979, 1980; Miller and Scholes 1982). Conditional on this assumption, parameter estimates have been made and This paper tests four models describing the impact of taxes on financial asset returns: no-tax effects; Miller's equilibrium; the twoparameter-pricing models of Brennan and Litzenberger and Ramaswamy; and taxinduced-dividend clientele effects. These tests are carried out by casting the hypotheses in terms of systems of equations and identifying the parameter restrictions implied by each of the hypotheses. The systems approach is particularly useful for testing for clientele effects because we are able to identify withinand between-equation restrictions without specifying the marginal tax rate of the clientele which holds a particular security. My results suggest that none of tax-motivated hypotheses is descriptive of the data. * This paper is a distillation of several research projects I have undertaken in the last two years. Special thanks are in order to G. Constantinides, E. Fama, J. Ingersoll, A. Madansky, and M. Scholes. I owe a particularly important debt to Merton Miller for help and encouragement. Finally, I would like to thank Steve Buser and E. Han Kim.