Modelling and forecasting the dynamics of the long-memory volatility relationship between FX spot and futures markets

This study investigated the long-memory volatility spillover effect between FX spot and futures markets, using a bivariate FIGARCH-DCC model. In particular, we considered the volatility dynamics of USD, Yen, and Euro spot and futures contracts on the Korea Exchange (KRX). Our empirical results indicate strong dynamic volatility correlations and long-memory volatility transmission between spot and futures returns. In addition, we analysed the forecasting results of univariate and multivariate GARCH-type models over multiple forecasting horizons (1, 5, and 20 trading days) to assess whether any model showed superior performance. We provide an analysis of the forecasting performance of two competing GARCH-class models: univariate and bivariate GARCH and FIGARCH models. We assessed the predictive accuracy of these models using the MSE and MAE as loss functions. The bivariate FIGARCH model with Student’s -distribution provided more accurate volatility predictions than the other GARCH models.