Sources of Time Variation in the Covariance Matrix of Interest Rates

The main objective of this paper is to study the sources of time variation in the covariance matrix of interest rates. We depart from the traditional standard deviation–correlation decomposition of covariances and investigate whether time variation in the covariance matrix of bond yield changes is caused by time-varying eigenvalues and/or eigenvectors. On the basis of a formal testing procedure, we find that common factors display a clear time-varying volatility over the past three decades. Most notably, we observe that the switches in monetary policy that take place with the appointment of a new Federal Reserve chairman play an important role in characterizing the time variation in the loadings on the common factors that drive interest rates.