Uncovered Interest Parity (UIP) in the mean and the variance of the NIS/US$ Exchange Rate
This paper examines the simultaneous short-term relationships between the NIS/$ exchange rate and local interest rates, the dollar interest rates and the interest-rate differentials along the interest rates term structure. The test is performed using the ARCH (autoregressive conditional heteroskedasticity) procedure of two variables with a bivariate threshold ARCH (BV-TARCH) that incorporates a set of equations describing the relationship between the exchange rate and interest rates. This procedure enables us to analyze the effect of each of the variables on the other regarding the expectation and the conditional variance. It also enables us to identify asymmetry in the effects of shocks on the series. The second part of the paper examines the Uncovered Interest Parity (UIP) hypothesis along the term structure of the interest-rate-differential.
The analysis of the results of the procedure used on weekly data for the years 1999-2005 shows that in the mean equation the NIS/$ exchange rate has a positive effect on the long-term domestic interest rates and on the interest-rate differential, but the reverse relationship does not hold. Moreover, the impact of positive shocks in the local interest rates on the conditional variance was greater than negative ones. This is probably the result of the asymmetrical monetary policy during most of the sample period: sharp increases of interest rates versus moderate declines.
Examination of the UIP during the sample period shows that changes in the interest-rate differential, particularly for the long term, predicted the changes in the exchange rate better than did the differential itself. These findings can be explained by the significant contraction of the interest-rate differential that occurred during the sample period compared with the relatively small depreciation in the NIS/$ exchange rate in that period.