Testing the efficiency of oil price forecast revisions in times of COVID-19 and the Russia–Ukraine conflict

Abstract

We investigate weak- and strong-form efficiency in fixed-event forecast revisions for Brent and WTI prices using proprietary microdata from Energy & Metals Consensus Forecasts™ by Consensus Economics®. Our findings indicate forecasters mostly revise independently of past revisions, suggesting weak efficiency. Contributing to the strong-form efficiency literature, we compile data on 75 publicly available variables, which capture COVID-19, the Russia–Ukraine conflict, macroeconomic, financial, and oil market indicators. To ensure the information available to forecasters matched what was realistic at the time of their predictions, we lagged the variables to account for publication delays. Additionally, we added another lag to each variable, doubling the information set from 75 to 150 variables. This constitutes a significant effort in comprehending the information accessible to crude oil forecasters. Employing innovative multiple testing and penalised regression methods to address variable selection in a data-rich environment, we find that, conditional on passing weak efficiency, support for strong-form efficiency is limited. Notably, analysts incorporate past variable values, including COVID-19 and Russia–Ukraine conflict metrics, in their revisions. Our econometric modelling sheds light on how analysts’ decision-making adapt to changing market conditions, sociopolitical developments, and critical information.