A Response Surface Analysis of Critical Values for IPS Panel Unit Root Tests

Abstract

The paper develops an algorithm to generate the critical values through Monte Carlo simulations which will be computationally efficient as opposed to the traditional simulation techniques used in the earlier panel unit root studies. The results from the simulation experiments are used to construct the response surface regressions in which the critical values depend on both cross-sectional and time units. The predictability of the response surface regressions is compared with reported IPS critical values. The usefulness of these results is illustrated through an empirical example of purchasing power parity test.