Modeling the volatility of stock indexes and predicting VaR during COVID-19 pandemic

As COVID-19 spreads globally and becomes a pandemic disease, it has severely impacted the capital markets of major economies around the world, posing severe challenges to financial risk management. In this thesis, we study the ARIMA-GARCH model and the Markov switching GARCH model and its application in financial risk management. This thesis verifies and compares the predictive ability of ARIMA-GARCH models and Markov switching GARCH models on value at risk through empirical research. Especially incorporating the drastic fluctuations caused by the COVID-19 into the forecasting scope, we found that the ARIMA-GARCH model has basically no predictive ability, and the Markov switching GARCH model still has a good forecasting ability. Then we try to further improve the predictive power of the model by adding one additional regime. However, the performance of the model did not improve significantly. Key words: COVID-19, ARIMA-GARCH model, Markov switching GARCH model, value at risk, stock indexes