Why Do Oil Prices and Output Move Together? State-Dependent Evidence

Vollständiger Abstract

Worum geht es in dieser Arbeit?

This paper reexamines the relationship between oil prices and U.S. industrial production using monthly data from January 1974 to August 2025. We document three empirical patterns. First, the full-sample correlation between oil price changes and industrial production growth is positive, contrary to the conventional view. Second, this masks substantial time variation: the correlation is negative before the mid-1980s but mostly positive thereafter. Third, positive co-movement becomes substantially stronger during recessions. Using a structural vector autoregression (SVAR) that decomposes oil price movements into supply, global demand, and oil-specific demand shocks, we examine the forecast error variance decomposition (FEVD) of the real price of oil. The full-sample baseline shows oil prices are dominated by oil-specific demand shocks, with global demand shocks playing a modest role. In contrast, state-dependent results reveal that during recessions, global demand shocks account for a substantially larger share of oil price forecast error variance. Because global demand shocks move oil prices and output together, this shift explains why positive co-movement is stronger during recessions. Rolling estimation further shows that the long-run shift from negative to positive co-movement reflects changes in the transmission of oil shocks rather than a secular increase in the importance of global demand shocks.