This dissertation analyzes the time series properties of the term structure of interest rates. It presents a class of equilibrium models of asset returns, and investigates whether the empirical evidence in the term structure is consistent with implications drawn from such models. The dissertation consists of three essays.
The first two essays investigate the implications of allowing more general non-stationary structures in the stochastic processes of exogenous variables that underlie the uncertainty in the economy in a class of consumption-based, intertemporal asset pricing models. In contrast to prior studies, which routinely assume that the equilibrium characterizations of the model can be described with stationary fluctuations of economic activities as represented by simple growth rates or by Markhov Chains, I regard the fluctuations as deviations from potentially stochastic trends (Chapter 2) or as resulting from long-memory processes (Chapter 1). I use the theoretical economy to generate artificial time series of bond prices, which I then use to analyze the variations in risk and return implicit in bond returns and to conduct diagnostic tests of the model's implied restrictions with respect to the actual data. In light of findings offered in the empirical literature, I present the evidence that the distinction in the nature of stochastic process of exogenous variables is rather critical as it determines the uncertainty and persistence of shocks in the economy, and offer evidence that the implication for the behavior of term structure can be quite significant.
In the third essay (Chapter 3), I investigate the dynamic nature of the risk premia in the term structure. I seek to determine whether the risk premia in the term structure are consistent with characterizations of stationary processes or those of non-stationary processes. The distinction is important since an innovation in the first case has transitory effect while it has a permanent effect in the second case; and hence the distinction will have far-reaching implications for how the risk is priced in an asset pricing models. I show that modeling the risk premia with fractionally integrated time series models is very useful in delineating the nature of (non-)stationarity in the risk premia.