Integration and causality in US mortgage and T-bond markets

Abstract

This paper re-examines the issues of integration and causality in US mortgage and T-bond markets by using the well-known cointegration and error correction methodology. It employs monthly data from January 1980 through June 1993. The unit root test reveals nonstationarity in 30-year nominal mortgage rates and 30-year nominal T-bond yields. The DF tests affirm cointegration between these two variables. The estimates of the associated error correction model depict unidirectional long-run as well as short-run Granger causality that runs from the 30-year T-bond market to the 30-year mortgage market. Reversible short-run feedbacks are also observed between the two markets.