Canadian stock returns and the term structure of interest rates

This study examines the relationship between stock returns and the term structure of interest rates in a Canadian setting. Following Zhou's study of the U.S. market (Federal Reserve Board, 1996), the hypothesis tested is the excess return hypothesis, which states that expected returns move one-for-one with interest rates. This relationship is explored using nominal and real return data for Canadian stocks and bonds.  In addition, the study examines the ability of U.S. interest rates to predict Canadian stock returns. 

A 2SLS model with instrumental variables is used to estimate these relationships in order to avoid possible endogeneity and, thus, obtain consistent and unbiased coefficients, inasmuch as ex post interest rates are used in the regressions. In addition, because of limitations in the availability of data for Canadian bond yields, the study tests for possible structural changes over the period examined (1956-2008). As well, Newey-West standard errors are calculated in order to adjust for serial correlation in the stock return data.

The S&P/TSX monthly price index is used to compute log stock returns. The zero coupon bond yields for Canadian treasuries with matching maturities are used to proxy for Canadian interest rates. McCulloch and Kwon's (1993) data set of zero coupon bond yields implied by the yield curve for U.S. treasuries is used as a proxy for U.S. interest rates for the period 1956-1991. 1991-2008 U.S. bond yields were obtained from the the CRSP database. Canadian and U.S. CPI data were used to compute real stock returns and real interest rates. 

The study finds that nominal Canadian interest rates do not have any predictive power vis-a-vis nominal Canadian stock returns. In contrast, real interest rates do appear to be related (negatively) to real stock returns, at least for long time horizons. As for the relationship between U.S. interest rates and Canadian stock returns, the relationship is negative, albeit statistically insignificant, in the nominal case, but positive and significant for real stock returns and real interest rates.