I have a trivariate Panel VAR system with the following variables: LnGDP (Natural log of GDP) Fiscal Expenditure (as % of GDP) Interest rate ( in %). LnGDP and interest rate are stationary at levels but fiscal expenditure is not. The first difference of the Fiscal Expenditure is stationary. In addition to this, when I add drift in stata for checking stationarity of fiscal expenditure, it leads to Stationary series. My first question is if I run a Panel VAR model based on the above information, what can be the future consequences? My second question is provided the above regression is non-spurious, how do we interpret the coefficients in the model