Elsevier

Economics Letters

Volume 132, July 2015, Pages 125-128
Economics Letters

Oil price forecastability and economic uncertainty

https://doi.org/10.1016/j.econlet.2015.04.023Get rights and content

Highlights

  • Information on economic policy uncertainty matters in predicting oil price changes.

  • Nonlinearities are revealed in their relationship via a TVP-VAR approach.

  • We compare the forecastability of VAR models vs. benchmark AR & RW models.

  • The results indicate that TVP-VAR models outperform other alternatives.

Abstract

Information on economic policy uncertainty does matter in predicting the change in oil prices. We compare the forecastability of standard, Bayesian and time-varying VAR against univariate models. The time-varying VAR model outranks all alternative models over the period 2007:1–2014:2.

Introduction

Hamilton (2008) indicates that nine out of ten recessions in the US since World War II have been preceded by an increase in oil prices. Interestingly, Hamilton (2009) goes as far as arguing that a large proportion of the recent downturn in the US GDP during the ‘Great Recession’ can be attributed to the oil price shock of 2007–2008. Stock and Watson (2003) also show the ability of oil price in predicting growth and inflation. A recent strand of literature emphasizes the role of economic policy uncertainty (EPU) on real activity (Bloom, 2009), which in turn affects oil-price movements, as depicted in Kang and Ratti (2013). To the best of our knowledge, this is the first attempt to forecast the change in oil prices using a news-based measure of EPU. This measure, developed by Baker et al. (2013), relies on an automated text-search process of large US newspapers and identifies articles that use words related to economic policy, regulation and uncertainty. In our approach we compare the ability of VAR, standard Bayesian VARs and time-varying parameter VARs, against random-walk and univariate AR models of real changes in oil prices over the monthly out-of-sample period 2007:1–2014:2, using an extended in-sample period of 1900:1–2006:12. The paper is organized as follows: Section  2 briefly presents the various econometric models and Section  3 discusses the data and results; finally Section  4concludes.

Section snippets

Econometric models

The econometric models used include the classical and Bayesian VAR, a time-varying VAR with stochastic volatility (TVP-VAR) and a new TVP-VAR with Markov-switching heteroscedasticity as in Bekiros and Paccagnini (2014). Our benchmark models are the random-walk (RW) and an AR(p) model.

Data and results

The two variables of concern comprise real oil prices obtained by dividing the Western Texas Intermediate (WTI) by the Consumer Price Index (CPI), and EPU. We analyze the ability of the EPU to forecast real oil price changes over the period 2007:1–2014:2, i.e., during the recent global crisis, using an in-sample period spanning 1900:1–2006:12. Data is obtained from the Global Financial Database. The EPU variable is based on two overlapping sets of newspapers (e.g., Wall Street Journal, NY

Conclusions

The importance of oil prices in determining movements of US growth and inflation is well-established, hence accurate forecasting is of paramount importance. Moreover, recent works in the literature advocate in favor of economic policy uncertainty driving oil-price fluctuations. Against this backdrop, we compare the forecastability of various uni- and multivariate models of real oil returns and EPU. Our results indicate that TVP-VARs outperform the others in all horizons till two-years-ahead

References (14)

There are more references available in the full text version of this article.

Cited by (98)

  • Outlook of oil prices and volatility from 1970 to 2040 through global energy mix-security from production to reserves: A nonparametric causality-in-quantiles approach

    2022, Resources Policy
    Citation Excerpt :

    Aloui et al. (2016) apply a Copular methodology to document that an increase in EPU can cause either positive or negative effects on oil returns, depending on the economic situation. Bekiros et al. (2015) apply VAR models to identify that EPU enhances the out-of-sample predictability of oil price changes. Van Robays (2016) use a threshold VAR model to find that higher uncertainty increases the sensitivity of oil prices to oil demand and supply shocks.

View all citing articles on Scopus
1

Tel.: +33 01 53 63 36 00, +39 055 4685 916; fax: +33 01 45 44 40 46, +39 055 4685 902.

View full text