GARCH(1,1) model with the Yeo-Johnson transformed returns

Abstract

One of the most popular class in modeling financial volatility is GARCH-type model. Several extensions to the basic GARCH model have been developed to be more flexible in capturing various characteristics of financial time series data. In this study, a new class of GARCH models is proposed by applying Yeo-Johnson transformation to the return series. The proposed model is estimated by employing adaptive random walk Metropolis (ARWM) method in Markov chain Monte Carlo (MCMC) scheme. Our empirical results on the GARCH(1,1) models showed that the proposed model outperformed the initial model in fitting ten different international stock indices.