The empirical studies : the day-of-the-week effect on foreign exchange market - application of autoregressive conditional heterocedasticity model

• Main Authors: Li, Chian-Jang, 李乾彰
• Other Authors: Chen,Se-Kung
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/16022894713199509475

碩士 === 國立臺灣大學 === 國際企業學系 === 86 === THESIS ABSTRACT In this paper, we examine the daily behavior of exchange rate with ARCH model which was developed by Engle in 1982 and GARCH model which was developed by Bollerslev in 1986. We use daily exchange returns of US dollars , Japanese yen , Hong Kong dollars , Deutsche marks , UK pound sterling , South Korean won , Swiss francs spanning from 1988 to 1996 to discuss : (1) the stationarity of exchange rate data , (2) the statistical properties of daily exchange returns of seven foreign currencies and (3) to test t Day-of-the-Week effect with ARCH and GARCH model . As a result of the findings from , the empirical analysis : First of all , it is a test of stationarity of daily exchange rate data by using unit root test with DF and PP method , we can not reject the hypothesis which have unit root situation initially. Additionally , further analysis reveal that the data have stationary properties as the first difference . Secondly , in order to investigate the statistical properties of daily exchange returns data , the empirical findings suggest that the data are neither independent nor identically distributed. Furthermore, there are autocorrelation and heteroskedasticity with overall data. Thus , the conclusion that an ARCH for all currencies'' daily exchange returns can be reached as the LM test statistics are significant which even at the order of six . Finally , we choose the best model of each currency by using Likelihood ratio test and Schwarz information criterion .The conclusion is US dollars-ARCH(4)，Japanese yen-ARCH(4),Hong Kong dollars-ARCH(4)，Deutche marks-ARCH(4)，UK pound sterling- GARCH(1,1)，South Korean won-ARCH(5)，Swiss francs-ARCH(4)。 In order to investigate whether Day-of-the-Week effect exists , autoregressive conditional heteroskedasticity model is adopted .The empirical findings indicate that daily exchange return of Hong Kong dollars tends to decline on Monday and increase on Friday as weekend effect of US stock market which French found in 1980 . In the meantime , Japanese yen doesn''t have any Day-of-the-Week effect at all while the rest of the currencies have some degree of day-of-the-week effect .