A0418
Title: Combining GARCH-MIDAS forecasts of US state-level volatility: The role of local and global EPU indices
Authors: Vincenzo Candila - University of Salerno (Italy) [presenting]
Oguzhan Cepni - Copenhagen Business School (Denmark)
Giampiero Gallo - NYU in Florence (Italy)
Rangan Gupta - University of Pretoria (South Africa)
Abstract: The role of both local (state-specific) and global economic policy uncertainty (EPU) is investigated in forecasting the volatility of U.S. state-level equity returns. A GARCH-MIDAS framework is adopted, which allows the inclusion of multiple EPU indices as low-frequency drivers of daily return volatility. To tackle the issue of identifying the most relevant predictors, an elastic net (EN) shrinkage technique is implemented that combines forecasts across different model specifications. Findings show that the combined model, which leverages information from both local and global EPU indices, consistently outperforms individual specifications, from both one-step- and multi-step-ahead perspectives. These results underscore the importance of accounting for both regional and global policy uncertainty when modeling volatility dynamics at the state level.
