Predicting Realised Volatility Around Covid-19 in Nordic Markets

Abstract

This study evaluates the performance of different models in predicting volatility in the Finnish and Swedish stock markets, particularly focusing on their adaptation to the Covid-19 pandemic. The study covers the period between 2018 and 2023. The model set contains a variety of methods, ranging from historical and implied volatility to the GARCH(1,1) model. The overall model performance is assessed using robust measures such as RMSE and MAE, along with regression analysis. Our findings reveal that the GARCH(1,1) model was the superior forecasting method for the OMXH25, with implied volatility proving to be the least effective forecasting method. Interestingly, the implied volatility forecast was the most reliable method for the OMXS30, with the historical volatility forecast being the weakest forecasting method. Additional research shows that both indices’ volatility was significantly affected by Covid-19, with increased cases and deaths related to Covid-19 leading to increased volatility and increased vaccinations against Covid-19 leading to decreased volatility.