Abstract:
We investigated the presence of regimes in volatility of returns of the Ghana Stock Exchange index using
single- and two-regime Markov-switching threshold GARCH with skewed- and student-t innovations
separately for model fit. We found that the 2-regime threshold GARCH(1,1) with skewed student-t
innovations provide a better fit to the data by using the deviance information criterion (DIC) to discriminate
among the candidate models. There are two clear regimes with different statistics describing the volatility of
returns for the low and high regimes. Incorporating regime switching thus avoids the practice of the single
regime choice which pulverises the unconditional volatility through complex averaging leading to the
overestimation and underestimation of risk during the low and high regimes respectively.
