| Summary: | 博士 === 國立中正大學 === 國際經濟研究所 === 99 === This dissertation is comprised of two topics related to nominal exchange rate, one is about the predictability of nominal exchange rate and the other is for the correlation between equity and foreign exchange markets.
The first part of this dissertation adopts a simple average weight to combine forecasts from different fundamentals-based forecasts and then examines the statistical and economic significance of forecast combination under different forecast horizons. Empirical investigation based on monthly data over 1973-2009 results in several interesting findings. First, TR-based forecasts reject RW forecasts in six and five out of eleven countries at short and medium forecast horizons. M- and PPP-based forecasts reveal evidence of beating random walk at medium horizons. Fundamentals-based forecasts reveal little evidence to beat RW forecasts at long horizons. Second, combining forecasts from different fundamentals that have predictability over a specific horizon is generally promising to enhance the evidence of beating random walks over that horizon. Third, the average percentage of beating random walks at a specific forecast horizon by forecast combination increases with the number of fundamentals-based forecasts being combined. Fourth, increasing the number of forecasts being combined increases statistical significance of forecast combination. Fifth, forecast combination results in stronger economic significance than random walk regardless of horizons.
The second part of this dissertation provides a dependence-switching copula model to describe the dependence structure between stock and foreign exchange markets and obtains the following interesting results. First, exchange-rate exposure effects or portfolio rebalancing effects is observed for most countries at most times. Second, dependence and tail dependence between the two markets differ among the four combinations of market status, which is essential for safety-first agents investing globally, for correctly evaluating value-at-risk and for measuring systematic risk in financial crisis. Third, the smoothing correlation between the above two markets are similar for euro countries and the United Kingdom throughout the post-euro period. Finally, the dependence and tail dependence between stock and foreign exchange markets when both markets are booming generally differ with those when both markets are crashing.
|
|---|
