Three essays on empirical study of the term structure of interest rates

The modeling of the term structure dynamics is important for a variety of reasons. Forecasting is a first reason. The current yield curve contains information about future economic activity. Monetary policy constitutes a second reason for the term structure modeling. The transmission mechanism of monetary policy is related with the movements of the yield curves. The pricing and hedging of interest rate derivatives is a third reason. The price of many securities, such as coupon bonds, swaps, futures and options on interest rate is calculated based on some specification of term structure models. Bond portfolio diversification provides a fourth reason. Bonds and securities are traded in well-organized international market, to diversify bond portfolios in world market, we need the term structure models. This thesis is one more attempt to contribute to this discipline. Essay I (Chapter 2) reexamines the expectations hypothesis, which is a central view of the term structure of interest rates. The empirical failure of the expectations hypothesis has been well-documented. There are usually three interpretations for the failure of the hypothesis. These are, respectively, the failure of the rational expectations assumption and unlimited arbitrage assumption, the presence of time-varying risk premiums, and poor properties of the statistical tests in finite samples caused by peso problems. In this essay, I take into account peso problems and time-varying risk premiums by assuming that the data generating process is a Markov-switching vector autoregression model and investor's decision is based on a larger information set. In so doing, the testing of the expectations hypothesis is more robust. In particular, I find that the deviations from the expectations hypothesis are insignificant with my testing framework. The term structure of interest rates and macroeconomic activity are closely related. Hence, the term structure models should also be able to identify the economic forces behind these movements. On the other hand, the yield curve contains important information for forecasting the future path of the economy. Therefore, we should model the yield curve and macroeconomic variables jointly. Essay II (chapter 3) attempts to contributes to jointly modeling yield and macro factors. The model proposed in this essay extends the dynamic Nelson-Siegel model by incorporating regime shifts. To estimate the proposed model, A MCMC (Markov chain Monte Carlo) method is presented. The main finding is that there are significant bidirectional linkages between the yield curve and economic activity. Government securities are traded in well-organized international markets. Therefore, to what extent bond markets are integrated is a fundamental question in international finance. Furthermore, the degree of market integration is important for international public policy coordination. Essay III (Chapter 4) tries to answer this question. I propose a dynamic measure of bond-market integration based on the affine arbitrage-free dynamic Nelson-Siegel model. It contributes to the extant literature in several ways. First, my measure is consistent in both cross-section and time series. Second, this measure is theoretically consistent with no-arbitrage theory. Third, it is a dynamic measure. The empirical study demonstrates that global government bond markets are integrated to some extent.DOCTOR OF PHILOSOPHY (HSS