Do oil price changes contain useful predictive information about the U.S. bear stock market?

Abstract

This paper investigates the predictive power of $\kappa$-month oil price changes (characterizing oil market trends) for the U.S. bear stock market. We find from both in-sample and out-of-sample evaluations that $\kappa$-month oil price changes with $\kappa >12$ can strongly predict the bear stock market, but conventional oil predictors fail to have predictive power. In particular, a higher current oil price relative to its level $\kappa$ months ago will induce a higher probability of bear stock market in the future and the 48-month oil price change performs best at most of the forecast horizons considered. These results are robust to different subsamples, oil price series, estimation schemes for out-of-sample analysis, and phases of the business cycle. Moreover, the information provided by the 48-month oil price change covers that of inflation rate and does not completely overlap with that provided (individually or jointly) by non-oil predictors so that forecast improvement can be achieved by taking it into account.