Non-Segmented Equilibria Under Differential Taxation: Evidence from the Canadian Government Bond Market

Abstract

Tax-clientele effects induce segmented equilibria and these affect the estimation of the term structure of interest rates. The identification of these effects depends on the deviations of bond prices from the value of their discounted cash flows. However, bond prices contain noise that makes it difficult to determine whether the deviations are the result of statistical error or of clientele effects. Evidence from a deterministic test capable of detecting the feasibility of a non-segmented equilibrium is used to form hypotheses concerning the frequency and timing of tax effects in the Canadian bond market during the period 1964 through 1986. Tax reforms introducing capital gains taxation in Canada in 1972 generate additional hypotheses regarding the timing of tax effects. The empirical results from regression tests confirm that non-segmented equilibria occur in those months that the feasibility test indicated they were feasible. Further, the sample period is divided into an early period with virtually no evidence of tax effects and a later period characterized by tax effects. This division is roughly linked to the introduction of capital gains tax in 1972 and the increased interest rate volatility in the later period.