| This paper investigates the theoretical issue regarding pricing FX risk with a model based on stochastic discount factors. The model implies that when exchange rate is treated as the price for an investment asset, FX risk premium depends on the correlation between expected future exchange rate and the domestic SDF. When treated as the relative price between domestic and foreign consumption asset, exchange aqrate provides information about the ratio of different money measures. The FX premium is determined by the relative volatility of two economies and the correlation between. FX premium is nearly zero when both countries are in steady growth and the international risk sharing mechanism works well. FX premium exhibits great changes as long as either one of the countries is suffered from unexpected fluctuations. Variance adjustment are needed when one calculates FX premiums of the same currency pair under different money measures.The paper verifies the theoretical conclusions with the spot and forward exchange rate of GBP/USD, AUD/USD and USD/JPY during the sample period from 1993 to 2007. Unlike the widely documented auto-correlated FX risk premium with conditional heteroskedasticity, this paper points out that in the sample period, FX risk premiums for both GBP/USD and AUD/USD are no different from white noise series. Heteroskedasticity caused by short term economic fluctuations appears in JPY/USD but is finally proven to be non-lasting.Correlation tests and Granger Cause-and-Effect tests informs little about the risk factors of FX risk, while a combined method of SDF and GMM does contribute. Though macro factors are proven to be insignificant in short-term exchange rate pricing, micro factors, including stock price changes and interest rate changes do affect exchange rate. With USD as the quoting currency, the exchange rate depends on the interest rate difference between US and the currency issuing country, and also depends on systematic risk changes of the currency issuing country. |
PDF Full Text Request