| Summary: | 碩士 === 國立雲林科技大學 === 財務金融系碩士班 === 91 === In the traditional econometric and time series models, it is generally assumed the variance of residuals are constant. Thus, the above assumption is used as the basis research and the making of inferences. But, in fact, many time series data of financial asset violate the hypothesis, and their residual are unstable and influenced by past impact. So, if the homoskedasticity of variance and autocorrelation phenomenon exist in the data, the use of the OLS method to estimate will cause biased results.
This thesis adopts Engle’s(1982) ARCH model and Bollerslev’s(1986) GARCH model. Our rearch targets are the spot EX rate of NTD to U.S. dollars and 10 days, 30 days, 60 days, 90 days, 180 days forward EX rate. Our research period is from 1993 Jan. to 2002 Dec. We separate three periods by different rest days in a week, and test whether the-day-of-the-week effect exists. Our empirical results are as follows.
1.In Taiwan’s U.S. EX rate market, the daily returns all present the
Distribution in an asymmetrically high peak, whether the spot EX rate or
forward rate.
2.The daily return of NTD to U.S. dollars are series correlated and the
variance is heterogenous, whether before or after implementation of the
policy which rest two days in a week.
3.We use the AIC standard to choose the appropriate ARCH or GARCH model to
test the day-of-the-week effect. In Taiwan’s foreign EX rate market, the
empirical results are different from the previous study that the return is
lower on Monday and higher on Friday.
4.After 2001 Jan. Taiwan changed its trading time. Other than the 60 days
forward EX rate, the day-of-the-week effect exists in all other days forward
EX rate. In total, the day-of-the-week effect existed in the U.S. EX rate
market whether before or after the policy implementation. And it means
Taiwan’s EX rate market violate the efficiency market hypothesis.
|
|---|
