Essays on Money demand and the Predictability of Exchange rates and Stock prices
| Autor: | Yu-Hau Hu, 胡育豪 |
|---|---|
| Rok vydání: | 2007 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 95 This dissertation is composed of three independent essays. The subject of the first essay is the estimation of money demand in Taiwan. We modify the conventional money demand function by including a real exchange rate variable to reflect the effect of currency substitution. Empirical evidence indicates that the variable is crucial to the long-run stability of Taiwan’s money demand. After finding the failure of a linear error-correction model in describing the dynamics of Taiwan’s money demand, we apply a non-linear error-correction model to examine its dynamics and support the appropriateness of the non-linear model empirically. The subject of the second essay is to solve the Meese-Rogoff puzzle(1983), one of the well known puzzles in international economics. It concerns the weak relationship between nominal exchange rates and market fundamentals. We emphasize the importance of the Harrod-Balassa-Samuelson effect in modeling deviations from the purchasing power parity fundamental and re-evaluate market fundamentals in forecasting nominal exchange rates. Based on the period of post-Bretton Woods, we provide solid out-of-sample evidence to reject the random walk forecast model at medium-term and long forecast horizons and also find mild evidence for out of sample predictability of nominal exchange rate over short term. Our findings shed new light on understanding the Meese-Rogoff puzzle. Following the econometric technique in the second essay, we investigate the out-of-sample predictability of real US stock return based on the price-dividend (PD) ratio. Applying the Gordon(1962) growth model, a stable PD ratio provides an apparently plausible explanation for the predictability of stock return. However, if there are the structural changes in stock markets, investors will gradually adjust their expectation on real dividend growth and discount rates, which resulting in a trend movement in long-run equilibrium of the PD ratio. Therefore, it would not make sense to assume that the PD ratio always revert to its constant mean. Using the annual data of S&P500 real stock returns over the period of 1872-2004, we provide strong in-sample and out-of-sample evidence to support return predictability at all forecast horizons when the dynamics of the PD ratio are modeled appropriately. In addition, we show that our bootstrap tests have reasonable high power and correct size, and that the power of bootstrap tests does not increase with forecast horizons given the nonlinear framework. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|