Specification sensitivities in the Markov-switching unit root test for bubbles

The aim of this article is to provide some empirical guidelines for the practical implementation of the Markov-switching augmented Dickey-Fuller (MSADF) test proposed by Hall et al. (J Appl Econom 14:143-154, 1999) for detecting explosive bubble behavior. We conduct simulation studies to compare the performance of the MSADF test under different error variance specifications, namely the constant variance and regime-dependent variance assumptions. An empirical application to the money base, consumer price and exchange rate in Argentina reveals the practical importance of the error variance specification on the MSADF test outcomes.