A Simple Panel Unit Root Test by Combining Dependent P-Values

This paper proposes a simple panel unit root test based on Zaykin et al.’s (2002) truncated product method. The test is powerful in cases where there are only a few large p-values, and is robust to a certain degree of cross-section dependence. Monte Carlo evidence shows good size and power properties relative to existing p-value combination tests. Unlike the previous tests, the new test allows to make stronger claims in the event of rejection of the null hypothesis. The proposed test is applied to a panel of 27 OECD real exchange rate series as well as to a group of inflation density forecasts in the SPF data.