Page 1 of 1

bivariate TGARCH -- misspecified (?)

Posted: Mon Feb 10, 2014 5:24 am
by espritz
Hey there,

I am currently studying the causal relationship between nominal and real uncertainty of inflation and output growth, with extra exogenous variables. To do so, I estimated a VAR-TGARCH model (I previously tested the asymmetry with Engle-NG procedure).

I estimated the appropriate VAR which leaves the residuals uncorrelated and heteroskedastic.

- Honestly I think I did something wrong, since the equation mean shows not-significant coefficients and the sum of variance equation coefficients is > 1.

- Secondly, I am not sure about which kind of arch restriction coefficient I should consider, since the help guide does not explain that: I just know it's useful to test both BEKK And VECH models.

How can I recognize when something "goes wrong"? I would appreciate your help.