A Monte Carlo Comparative Simulation Study for Identification of the Best Performing Panel Cointegration Tests

Abstract

In this paper the performance of nine panel cointegration tests, having the null hypothesis of no cointegration, with respect to weighted average rank scores under the whole space of alternative using Monte Carlo simulations have been carried out. Our results indicate that PdPtp, PAWS and PdP_V tests are the only three best performing tests among all panel cointegration tests whether time and cross sectional dimensions are small, medium or large. However, PDFTstar, PDFTrhostar and PdGtp panel cointegration tests have also identified as best performer at large cross sectional dimensions.