The Relationship between Carry Trade Returns and Foreign Exchange Risk

• Main Authors: FAN,YU-WEI, 范友韋
• Other Authors: CHIEN,MEI-SE
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/94907385053391859098

碩士 === 國立高雄應用科技大學 === 金融系金融資訊碩士班 === 103 === This paper employs the panel smooth transition regression (PSTR) model of Gonz´alez (2005) to investigate the explaining ability of foreign exchange risk measured on the return to the carry trade strategy. The empirical evidence is based on the framework of Cenedese et al. (2014), whereby the measure of foreign exchange risk is the market variance that can be decomposed into two components: the cross-sectional aggregate variance and the cross-sectional average correlation of all bilateral exchange rates. The sample data include the spot exchange rates and forward exchange rates of 48 economies from 1997 to 2014. Our main findings are as follows. First, strong evidence shows that there are significant intertemporal non-linear relationships between the excess returns and variance measures of the carry trade. Second, there are similar results of PSTR for the cases of the global portfolio and the developing economies portfolio, confirming that the relationship between risk and market return is positive for the first and second regimes while it is negative for the third regime which different from aggregate variance level. Third, the results of PSTR also support that the relationship between risk and market return is positive for the advanced economies portfolio, but the coefficients are insignificant. In fact, the coefficients of the advanced economies portfolio are far smaller than the coefficients of the other two portfolios, implying that investors will seek smaller risk-taking returns for the advanced economies portfolio. Fourth and finally, for the global portfolio and the developing economies portfolio, the coefficients of average correlation show that the expected excess return will increase for the first and third regimes in different aggregate variance level, while it will decrease if the excess returns are more correlated between other economies in portfolio. As to the advanced economies portfolio, the average correlation results are similar with the other two portfolios, but are insignificant.