Oil and US GDP: A Real-Time Out-of-Sample Examination
We study the real-time Granger-causal relationship between crude oil prices and US GDP growththrough an out-of-sample (OOS) forecasting exercise; we do so after providing strong evidence ofin-sample (IS) predictability from oil prices to GDP. Comparing our benchmark model “withoutoil” against those “with oil” by way of both point and density forecasts, we find strong evidencein favor of OOS predictability from oil prices to GDP via our point forecast comparisons whenwe adjust our MSPEs to account for noise introduced under the null hypothesis that the parsimoniousbenchmark is the true data generating process. These results are consistent withwell-known IS results covering part of our OOS period, and also suggest that, in the 1990s and2000s, oil prices have had greater predictive content for GDP than in the mid to late 1980s. Byway of density forecast OOS comparisons, while we do not find statistically significant evidenceof such predictability from oil prices to GDP for the full 1970-2008 OOS period, our results qualitativelyalso suggest substantial time variation in this relationship; predictability from 1970 to1985, and increasing predictability near the onset of the Great Recession.
Link al Paper