U.S. GOVERNMENT BOND MARKET EFFICIENCY AND THE EXPECTATIONS HYPOTHESIS OF THE TERM STRUCTURE OF INTEREST RATES (UNITED STATES)

Under newly designed tests, evidence has been found against the joint hypothesis of the U.S. bond market efficiency, the expectations hypothesis of the term structure of interest rates and the stationarity of interest rates assumption in recent literature. Even so, there is no unanimous agreement on the empirical results and the causes of such unfavorable evidence. Shiller and Singleton, for example, have found that U.S. long-term interest rates are too volatile to be consistent with the joint hypothesis, while Flavin has shown a contrary result. Furthermore, even in the literature which agrees that the unfavorable evidence does exist, controversy remains regarding the cause of this finding. The existence of time-varying risk premia in long-term rates, the animal spirits of economic agents and the learning behavior by market participants have been suggested as the causes of the unfavorable evidence.; This dissertation seeks to (1) provide an explanation for the difference in test results between Shiller and Singleton on the one side and Flavin on the other; and (2) investigate the plausibility of various explanations for the evidence unfavorable to the joint hypothesis in the literature and search for some alternative explanations.; We find that the divergent conclusions drawn by Shiller and Singleton on the one hand and Flavin on the other is due to their different belief regarding the stationarity of U.S. interest rates and consequently their differing approach in conducting tests. For less controversial results which reject the joint hypothesis, we find that learning behavior by market participants is a very likely cause for such rejection. Additionally, we find that an alternative explanation, viz. the existence of measurement error in U.S. Treasury bond rates also provides some evidence as causing the rejection of the joint hypothesis. Finally, the time-varying risk proxies we examined cannot be supported as the cause of the evidence unfavorable to the joint hypothesis.