High-frequency volatility forecasting in emerging markets: a comparative approach

Abstract:

This paper tackles volatility forecasting using high-frequency intraday data from four financial markets from Eastern Europe, namely the Bucharest, Warsaw, Budapest and Prague stock exchanges employing three different methodologies: a HEAVY (High-fRequency-bAsed- VolatilitY) framework, a GARCH-type one and a stochastic volatility (henceforth SV) framework. By calculating realized volatility and comparing it with HEAVY, GARCH and SV forecasts we find that the HEAVY model does not outperform the GARCH forecasts for three of our series, the BET index being the only one for which the HEAVY forecasts perform better. At the same time, forecasting volatility in an SV framework seems to outperform the HEAVY model for the Warsaw Stock Exchange. The current study contributes to empirical research on volatility forecasting through the means of a comparative analysis for emerging Eastern European markets. To our knowledge, the paper is the first one to tackle volatility forecasting using high frequency HEAVY model within a cross-regional comparison of this sort.