Testing for Co‐explosive Behaviour in Financial Time Series

This article proposes a test to determine if two price series that each contain an explosive autoregressive regime consistent with the presence of bubble are related in sense linear combination them is integrated order zero. We refer such phenomenon as ‘co-explosive behaviour’, and propose based on stationarity testing framework. The allows episode one lead (or lag) other by number time periods. establish asymptotic properties statistic wild bootstrap procedure for obtaining critical values robust heteroskedasticity. Simulations show proposed has good finite sample size power performance. An empirical application detect whether co-explosive behaviour exists among set precious non-ferrous metals presented.