I frequently used Eviews for econometric analysis, and would like to move on to Python for some automatic work. Now I wonder how can I achieve Tramo/Seats seasonaly analysis with regressors in statsmodels?
To be specific, I have a monthly series of economic data (Y), but it suffer some known seasonal effect like spring-festival-effect, which I have quantified them and summarized as another series (X). In Eviews, I clicked Y serie, then proc -> Seasonal Adjustment -> Tramo/Seats..., in the pop-up windows, other than Tramo/Seats Tab, you have Regressors Tab, which you can add X serie as seasonal component or separate additional component, and then OK, you can got the perfect seasonal adjusted values.
In statsmodels, I used sm.tsa.x13_arima_analysis Function:
sm.tsa.x13_arima_analysis(Y, exog = X, forecast_years = 0, prefer_x13 = True)
But the result is far from satisfying. It seems the exog = X options has no effect at all. The results both with or without exog = X option resemble that from the Eviews Tramo/Seats analysis without indicating X as seasonal component as regressor.
Can anyone tell me why?
P.S I checked gretl tramo/seats analysis. It does not have adding regressors functions neither.
