Hello,
I'm running a BEKK GARCH model with the asymmetric option, and I'm wondering whether anyone can confirm the specification of the asymmetric part used in the instruction. I'm assuming that it is GJR (1993)
D' u(t-1) u'(t-1) D
where u(t) = Min[0, e(t)], e(t) being the residuals from the mean equations and u(t-1) u'(t-1) a Kronecker product.
The reason I ask is that I'm trying to replicate, as a check, the in-sample estimates of H(t) generated by RATS by working with the coefficient estimates but my results so far are very different. I can, however, replicate the estimated residuals from the mean equations exactly.
Any advice would be most appreciated.