Multivariate Volatility Modeling of Nigerian Bank Share Prices

This study aims at finding the optimal Multivariate Generalized Autoregressive Conditional Heteroscedasticity (MGARCH) model among Diag-BEKK, Scalar-BEKK and CCC that captures the dynamics of returns in Nigerian bank share prices, using the data of daily share prices of two highly capitalized banks in Nigeria listed on the platform of Nigerian Stock Exchange (NSE) which span from 2nd January, 2009 to 28th December, 2019. Multivariate Normal and Multivariate student-t log-likelihood functions were simplified using the BHHH and Marquardt algorithm and the optimal solution was obtained using the information criteria. The BHHH and Marquardt algorithm was implemented in the GARCH 7 software of Laurent. Finally, the findings of this study showed that the Diag-Bekk (1,1) with Multivariate student-t distribution was the overall best model out of the six model combinations. Multivariate volatility modeling is therefore, recommended for bank share prices in Nigeria, and even in other banks from emerging economies.