An empirical inquiry into the sources of risk premia in foreign exchange markets

This research investigates the comparative explanatory power and structural dimensions of the variables of two competing models to explain the risk premium in foreign exchange markets. The two models, the economic model which focuses on purchasing power parity (PPP) and the asset pricing model which emphasises the equity risk premium, will be analysed by using the transfer function, principal component analysis, and cointegration analysis, applied to the same sample data. The empirical evidence reveals that the strong correlation between deviations from PPP and the foreign exchange risk premium observed in prior work is a spurious relationship that emerges due to a shared variance component related to changes in the exchange rate. This result refutes the previous finding that deviations from PPP play a much greater role than real interest rate differentials in explaining the foreign exchange risk premium. Nevertheless, the net influence of deviations from PPP, of which a shared variance component has been deducted, has statistically significant explanatory power. Principal component analysis tends to support the independence of the two models and implies the possibility of coupling the two models for a better empirical estimation. Augmented Dicky Fuller tests reveal the stationarity of most variables examined, except for the real interest rate differential, which is integrated of order one. This invokes a serious problem in the economic model for real interest differentials, because non-stationary variables cannot be robust explanatory variables over time for stationary dependent variables. Therefore, the findings of this research invalidate the results of previous empirical work that relies on the stationarity of the real interest rate. However, inflation differentials are identified to be a valid economic risk factor over the analysis period and the evidence for the equity market risk premium is mixed