Taming Turbulence: Forecasting Through the Structure of Volatility

Abstract

This paper develops a structural framework for forecasting GDP that explicitly incorporates macroeconomic volatility as an informative signal. The approach links aggregate output volatility to the evolving patterns of sectoral interdependence, decomposing volatility into idiosyncratic and synchronization components that capture both sector-specific shocks and their systemic transmission. These volatility measures are then integrated into mixed-frequency (MIDAS) nowcasting models to dynamically adjust the weight assigned to high-frequency indicators. Embedding volatility dynamics in this way improves the stability and reliability of GDP forecasts, reducing overshooting and enhancing performance during periods of economic stress. The results highlight how monitoring volatility in real time can strengthen both economic forecasting and policy assessment, particularly when uncertainty and sectoral dispersion are high.

Writting up