Are you thinking of event studies here or....? Best wishes, David Booth

David is correct, you need to clarify what you are trying to do. When one is talking about unit roots, one is usually talking time series that are solely explained by the observed history of the time series. These models do not usually include external associated variables such as dummy variables representing external events. I can image a model with a non-stationary constant that changes during each policy period or perhaps different ARMA parameters during each policy period. But usually dummy variables are used in a classical regression perspective which clashes with the ARIMA perspective. Are you thinking space-state models?