Regime switching in the term structure of interest rates

There is an enormous literature committed to developing models to explain and forecast interest rates and bond prices. Affine term-structure models comprise a large component of this literature. In these models bond prices are linear functions of one or more state variables. Notable for their tractability, affine models have largely been unsuccessful in accounting for the average slope of the yield curve. The difficulty is partly due to the fact that the parameters of the model are held constant over the modeling period. However, there is growing evidence that these parameters are time-varying, and many studies indicate that regime-switching processes are a promising way to model time variation. This dissertation incorporates regime-switching in interest rates into a fully-specified term structure model.; In Chapter 2 we present a term structure model where expectations of market participants vary with changes in economic regimes. The benefit of employing regime-switching is that it allows an economic interpretation of the factors driving changes in interest rates. Even without explicitly incorporating macroeconomic variables, we can relate regime changes in interest rates to changes in monetary policy or the economy. We find that affine-yield models are more successful in terms of in-sample pricing errors and out-of-sample forecasts once regime changes are taken into account. The model in Chapter 2 makes significant progress in replicating the shape of the average term structure.; In Chapter 3 we extend the model to allow macroeconomic variables to convey information about the underlying economic regime. We accomplish this by allowing the transition probabilities to depend on a vector of observable macroeconomic variables. This is natural extension of the model in Chapter 2. We can track the effect of the macroeconomic variables on the yield curve by monitoring the progression of the transition probabilities over time, much in the same way the business cycle literature employs the transition probabilities to convey information about turning points in business cycles. We find that the inclusion of economic variables improves the performance of the regime-switching term structure model in terms of in-sample fit and out-of-sample forecasts.