I'm testing out the Pesaran (2000) method for including exogenous variables in cointegration. It says it assumes weak exogeneity of the exogenous variable. Do I have to test for that - even if inclusion of exogenous variables just serve the purpose of robustness tests?