In this project we construct and analyze a new measure of industrial metal price uncertainty, the metal uncertainty index (MUI), following and extending the methodology of Jurado et al. (2015), which defines uncertainty as the unpredictable component of future outcomes. The index is constructed from spot and futures prices of five industrial metals -- copper, aluminium, nickel, zinc, and lead -- together with a broad set of macroeconomic, financial, and commodity-market variables that characterize developments in the global metal sector. We conduct a range of empirical analyses for industrial production, equity markets, metal prices, and the real effective exchange rate of the US dollar, including impulse response functions and forecasting exercises for the global economy, the United States, and the euro area. To account for potential nonlinearities, we employ a threshold vector autoregressive (TVAR) model that distinguishes between low- and high-metal-uncertainty regimes. In addition, we incorporate alternative uncertainty measures, including financial uncertainty and geopolitical risk, to complement the role of metal-specific uncertainty. The impulse response analysis reveals that the effects of metal uncertainty shocks are strongly regime dependent. Specifically, the magnitude, persistence, and, in the case of stock markets, even the direction of responses differ across low- and high-metal-uncertainty regimes, with the effects generally being more pronounced during periods of elevated metal uncertainty. The only notable exception is the real effective exchange rate of the US dollar, whose response is stronger in the low-uncertainty regime. More broadly, the responses to shocks in MUI differ from those associated with financial uncertainty suggesting that the two indices capture distinct underlying sources of uncertainty. We further evaluate the forecasting performance of models incorporating the MUI alongside other uncertainty measures under both linear and nonlinear specifications. The results show that models including the MUI frequently outperform competing alternatives, particularly for the global economy, across a range of forecast horizons and evaluation criteria. Moreover, TVAR models often rank among the best-performing specifications, underscoring the importance of accounting for regime-dependent dynamics in forecasting macro-financial variables. Overall, the findings indicate that metal uncertainty contains valuable information, enhancing both our understanding of macro-financial transmission mechanisms and the accuracy of forecasts, particularly when nonlinearities are explicitly incorporated.
Presentations
- 18th International Conference on Computational and Financial Econometrics (CFE 2024), London, December 2024
- 19th International Conference on Computational and Financial Econometrics (CFE 2025), London, December 2025
- Final presentation of the project funded by OeNB Anniversary Fund, Institute for Advanced Studies (IHS), Vienna, June 2026