Illuminating the uncovered interest rate parity

This study investigates the relationship between ex post exchange rate changes and interest differentials for a wide set of currencies for overnight, 1-month, 3-month, and 6-month interest rates. The first contribution of this paper is to estimate the uncovered interest parity coefficients with different techniques other than OLS. Quantile Regression and Seemingly Unrelated Regression are employed to smooth the outliers, and restrict the error term structure. Rolling regression with moving windows is used to examine the stability of the coefficients over time. The dynamicity from the moving windows allows us to study the consistency of the UIP performance through the full period. Half-life tests are applied to investigate the persistence of the UIP deviation. The second contribution of this paper is its exploration of the causes of the deviation. The relationship between ex post uncovered interest differential and macroeconomic and policy variables is investigated.;Our evidence about the validity or rejection of the uncovered interest parity is highly diverse and inconclusive. The goodness of fit for each country's regression is found to be very low for most countries. The sub-period coefficient estimates fluctuate in a high degree. However, the UIP hypothesis can not be rejected for more than 50% of the time for the majority of the countries. Rolling regression reveals that UIP evidence is not as bad as we thought.;One striking finding is the impact of the stock market variables on the UIP deviation. A rising stock market index reduces the deviation significantly, whereas, a more volatile stock market increases the deviation. This finding confirms the presence of risk premiums and forecasting errors that lead to the failure of UIP. Other key findings include the negative effect on the deviation by inflation, and positive effect by currency crisis.