Essays on the Causal Relationship Between Short-Term and Long-Term Interest Rates

Abstract

This thesis is about the causal relationship between interest rates. In chapter 1, with the help of time-series econometrics and by applying linear Granger causality tests based on the Toda-Yamamoto approach, the linear causality directions between the federal funds rate and five different interest rates during the last seven business cycles in the U.S. are investigated. We also examine the linear Granger causality directions between the overnight rate and five other interest rates during the last three business cycles in Canada. In chapter 2, the Diks and Panchenko Granger causality test is applied to explore the nonlinear causality effects between the short-term and long-term interest rates. By combining nonlinear causality effects with the linear ones which are found in the first chapter, it is seen that during the related periods in the U.S. and Canada, the most common Granger causality direction between short-term and long-term interest rates is a bidirectional one. Moreover, our findings show that during recent periods, the federal funds rate and overnight rate Granger cause other interest rates significantly. In chapter 3, the rolling window strategy is employed to detect the linear and nonlinear Granger causality relationship between the federal funds rate and the 10-year government bond rate, during different time horizons, investigating whether these causalities change with the passing of time. Our findings show that during different time horizons, there is a significant two-way Granger causality relationship between these interest rates. Although we have a different interpretation of the existence of bidirectional causation between short-term and long-term interest rates, this conclusion provides some support to some post-Keynesian structuralists viewpoints like Pollin (2008). However, Pollin's claim indicating that with the passing of time the significant causality effects of the federal funds rate to the market rates becomes insignificant is not supported by the current thesis findings because our results demonstrate that these causality effects have not been diminishing over the most recent business cycles.