Crude oil price changes and the United Kingdom real gross domestic product growth rate: An out-of-sample investigation

Abstract

We evaluate the impact of changes in the price of crude oil on the United Kingdom (U.K.) real gross domestic product (GDP) growth rate by way of an out-of-sample forecasting analysis. We compare the performance of several nonlinear models and determine, which aspects of nonlinearities are most useful for obtaining forecast improvements. Likewise, our approach takes into account the possibility that relative predictive performance can vary over the out-of-sample period. Results based on quarterly data from 1974q1 through 2018q4 illustrate that our conclusions depend on the definition of forecast improvement and whether we rely on pairwise or multiple forecast comparison. For instance, it is very difficult to find evidence that point forecasts exploiting crude oil price variables are statistically significant more accurate than point forecasts produced under the benchmark. On the other hand, the null hypothesis of no population-level predictability is borderline rejected for certain nonlinear crude oil price variables. We also observe notable differences between using real-time and ex-post revised GDP data with regards to local out-of-sample performance. The predictive power associated with the more successful crude oil price measures appears to concentrate in the early 1990s and around the onset of the Great Recession.