Journal of Statistics Applications & Probability


Effective modelling of extreme financial losses is a key investment strategy required by investors for successful assessment of risk in any financial market. This study compares the modelling capabilities of two extreme value theory (EVT) models via the conditional extreme value’s (CEV’s) GPD (generalized Pareto distribution) and point process for risk management and risk forecasting in the BRICS (Brazil, Russia, India, China and South Africa) equity markets. Prior to the application of the two EVT models, heteroscedasticity in the BRICS returns was filtered out using the generalized autoregressive conditional heteroscedasticity (GARCH) model. The findings reveal that under the GPD model, the risks in the five BRICS equity markets can all be modelled by the Gumbel class of distributions. Under the point process approach however, the risk in the Russian equity market can be modelled by the Fre ́chet-Pareto class of distributions, while the risks in the Brazilian, Indian, Chinese and South African equity markets can be modelled by the Weibul class of distributions. Furthermore, in terms of risk levels, the findings show that the Russian IMOEX market is the most risk-prone, while the least risky is the Indian NIFTY market, with the remaining three markets in between them. That is, the Russian IMOEX market has the highest level of risk, followed by the South African JALSH market, then the Chinese SHCOMP, Brazilian IBOV and Indian NIFTY markets, respectively.

Digital Object Identifier (DOI)