The power of standard panel cointegration statistics may be affected by misspecification errors if proper account is not taken of the presence of structural breaks in the data. We propose modifications to allow for one structural break when testing the null hypothesis of no cointegration that...