Energy organization sentiment and oil return forecast

Abstract

This study investigates the role of energy organization sentiments for oil return forecasts. First, we construct organization sentiment indexes using ChatGPT, a large language model, which enables us to extract sentimental information from the oil market reports issued by the International Energy Agency (IEA) and the Organization of the Petroleum Exporting Countries (OPEC). We found that organization sentiment indexes have a significantly negative impact on future oil price changes, and the information in OPEC's sentiment dominates that in the IEA's sentiment. The significance survives in models controlled by well-known oil pricing factors, e.g., oil market fundamentals, financial factors, and consumer and investor sentiments. The organization sentiment indexes Granger cause changes in oil production decisions, where oil production is identified as the channel through which the organization sentiment indexes influence future crude oil returns. We also found that the impact of organization sentiments is time-varying depending on investor sentiments and the market returns but mostly remains significant for both the in-sample fit and out-of-sample forecasts. Oil market participants, e.g., oil consumers, producers, and investors, can refer to the proposed organization sentiment indexes while trading crude oil to improve their utility. The inclusion of OPEC sentiment yields 2.40 % of certainty equivalent return gain, which is increased to 2.56 % with the addition of IEA sentiment. The findings of this study imply that the IEA should review its role and influence to maintain energy security effectively, and OPEC should track the profitability of its production adjustments.