Title: Forecasting volatility with the asymmetric GARCH-MIDAS model
Authors: Vincenzo Candila - Sapienza University of Rome (Italy) [presenting]
Alessandra Amendola - University of Salerno (Italy)
Abstract: Recently proposed, the GARCH-MIDAS model has gained much attention in volatility modelling literature. It allows us to forecast the volatility of a financial instrument by considering two multiplicative components, a short- and long-run one. The long-run component filters the information provided by some additional exogenous variables, observed at lower frequencies than those of the financial instrument of interest, into the short-run component. In this framework, the exogenous variables are usually some macroeconomic determinants of the variable of interest or some measures of the state of the economy. So far, the GARCH-MIDAS model only considers an average effect of these variables on the volatility of the dependent variable. However, positive and negative variations of the exogenous variables may have different effects on the volatility. And, in terms of risk management, the knowledge that only a given direction in the exogenous variable variation would produce a larger increasing of volatility could be very useful. Thus, the proposal is to forecast the volatility by considering an asymmetric GARCH-MIDAS specification, in order to investigate the impacts of exogenous variables increasing or decreasing have on the volatility of the variable of interest.