Modelling Volatility Cycles: the (MF)^2 GARCH Model

Abstract

RCEA aims to further independent, advanced research in Economics, Econometrics and related fields and to promote contact between economists, econometricians and scientists from other fields all over the world. Research at RCEA is conducted to enlighten scientific and public debate on economic issues, and not to advance any economic, political or social agenda. In this respect, RCEA mission is particularly concerned with raising awareness and stimulating discussion on the changes required to make capitalism sustainable along its economic, environmental, human, social and political dimensions

Long-Term Volatility Shapes the Stock Market's Sensitivity to News

“…In this model, returns are stationary, and multi-step ahead volatility forecasts can be easily computed. In addition, in line with the volatility feedback effect, Conrad and Engle (2022) show that major news events, whether good or bad, lead to upward revisions in expected long-term volatility.…”

“…The drawback of the GARCH-MIDAS, however, is that multi-step volatility forecasts are difficult to compute (because forecasts are needed for the explanatory variables in the long-term component) and that relevant variables may change over time. The MF2-GARCH of Conrad and Engle (2022) overcomes those shortcomings by modeling the long-term component as a function of the shortterm component's volatility forecast errors. In this model, returns are stationary, and multi-step ahead volatility forecasts can be easily computed.…”

“…Hence, the appropriate modeling of the conditional variance of cash flow news is crucial. We draw on recent developments in the literature on volatility models and assume that the conditional variance of cash flow news follows a multiplicative factor multi-frequency GARCH (MF2-GARCH) process (Conrad and Engle, 2022). In this model, the conditional volatility is decomposed into a short-and a long-term component.…”

“…In this model, the conditional volatility is decomposed into a short-and a long-term component. While the shortterm component captures day-to-day movements in volatility, the persistent long-term component is closely related to macroeconomic and financial conditions and behaves counter-cyclical (see Conrad and Engle, 2022). In contrast to other multiplicative GARCH models, the MF2-GARCH ensures that cash flow news is covariance stationary and multi-step ahead volatility forecasts can be easily computed.…”

Realised Volatility Forecasting: Machine Learning via Financial Word Embedding

“…Many studies have attributed news as a major contributor to volatility (Engle and Ng, 1993;Engle and Martins, 2020;Conrad and Engle, 2021). In recent years, researchers have shown an increased interest in using natural language processing (NLP) and machine learning (ML) methods to extract relevant information from textual data such as news.…”