Adaptive local VAR for dynamic economic policy uncertainty spillover

Abstract

Economic uncertainty has far-reaching global effects, especially during major crises. This paper introduces an adaptive local vector autoregressive (VAR) model to better understand how uncertainty spreads across countries. The proposed model identifies periods when economic spillovers remain stable, allowing for more precise estimation of their dynamics. Unlike traditional approaches that rely on fixed rolling windows or use all past data, our method adjusts to current conditions, reducing bias and improving accuracy. Using monthly data on Economic Policy Uncertainty (EPU), we show that this approach captures dynamic spillovers more effectively, particularly during unprecedented events like the COVID-19 pandemic and the Global Financial Crisis (GFC). These findings highlight the need for flexible tools in policy-making to address rapidly changing global risks. These findings underscore the importance of flexible tools for analyzing spillovers and allow for further exploration of adaptive methods in economic uncertainty analysis.