Page 1 of 1

seasonality in VEC

Posted: Sun Dec 06, 2009 8:09 am
by soungl
Hi

I have two monthly varialbes - employment and wages at the county level. Wage data is released with a lag of 6 months. I'd like to forecast the current values of wages (for the past six months) before it is reported. Since these two varilabes are tested to be cointegrated of order 1, I ran VECM with using them as endogenous variables (and a month dummy as an exogenous variable).

Residual tests suggest strong autocorrelation in the residuals. In particular, the correlogram shows huge spiks in the 12th lag (see attachment). How may I address this serial correlation in the residuals. The model already includes the month dummy to capture the seasonality. I think I should try seasonal differencing. Is there an option for this in VECM?

Thank you.