Identifying time variability in stock and interest rate dependence

The correlation between stock markets and interest rates has been discussed in numerous studies in the past, with differing results in terms of strength and direction of the relationship. This paper uses models of the multivariate GARCH type which allow for time-variability and regime changes in correlation. All estimated models allowing for timevarying correlation complement each other in identifying time-varying patterns found in the (co-)movement between the variables. Furthermore, we provide evidence for both large changes in correlation, as well as for the existence of regimes between which correlation may move. Our result of a dominant time factor indicates a transition in market structures over time, which is in line with observations in the markets and which may be seen as an explanation for previously differing results. -- Time-varying correlation ; regime transition ; multivariate GARCH ; smooth transition ; cross-asset correlation ; non-linear estimation