I'm modelling the twin deficits via a Vector Error Correction Model, given that all the 4 variables included are I(1) and I find 1 cointegrating vector among them. I intend to ascertain the direction of causality between Budget and Current Account Balance. I have already tested for autocorrelation of the residuals through a LM test and normality.
Which other post-estimation tests, such as stability tests, must one report in an academic paper after a VECM estimation?
You can perform residual test like VEC heteroskedasticity test apart from autocorrelation LM test and normality test.
I agree with Mousumi but I will add RESET to avoid general non-linearity.
And concerning the VECM heteroskedasticy should one use ARCH or Breusch-Pagan Godfrey test?
You do both; after all, there is no harm in just presenting the results
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling