Multiple Comparisons and Joint Significance in Panel Unit Root Testing with Evidence on International Interest Rate Linkage

This paper adds to the issue of inference regarding potentially nonstationary panels where units are correlated. Recently, it has been proposed to tackle this problem by computing individual p-values and combining them to an overall joint significance. We adopt and illustrate this fairly general approach allowing for competing means to account for cross-correlation when analyzing samples of N = 10 international interest rate differentials of different maturities. Alternatively, we investigate the approach of multiple testing or multiple comparison that has rarely been employed in econometrics. The advantages are that the computation of p-values is not necessarily required, and that one may identify for which specific unit a null hypothesis of interest may be considered as violated while controlling the overall significance level of the multiple testing problem. This comes at the price of an increased computational burden.