The RATS Software Forum

I am running a bivariate VAR-BEKK-GARCH and the off-diagonal elements of matrix B, the GARCH coefficients indicating persistence of volatility spillovers, are negative. I thought BEKK produced only positive coefficients. Is there any mistake that I am making?

Thanks,
Humyra

Your interpretation. BEKK produces positive definite covariance matrices no matter what the coefficients and no matter what the data. There is no reason the coefficients themselves need to be positive and, in fact, the signs of the coefficients aren't identified. See

https://estima.com/ratshelp/garchmvrpf. ... utput_BEKK