Oil price forecasts and security analysts: evidence from the oil industry

ABSTRACT

This study examines the relation between the U.S. Energy Information Administration (EIA) oil forecasts and analyst earnings forecasts for oil firms. Analyst forecast biases (revisions) are positively related to EIA oil forecast biases (revisions) but only for firms with greater analyst coverage. For these firms, we also document a positive association between analyst revisions and future oil prices. This association becomes insignificant after the introduction of concurrent EIA oil forecast revisions, suggesting that security analysts might not possess incremental information on future oil prices.

KEYWORDS:

• Oil forecasts
• security analysts
• earnings forecasts
• forecast bias

JEL CLASSIFICATION:

• Q40
• G24

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Initially, EIA releases the STEO report every quarter, but starting in 1997, the release frequency changes to monthly on the first Tuesday following the first Thursday of each month.

² While WTI is overwhelmingly traded in nearby futures contracts, we find that the spot rates and the near-term futures rates are highly correlated. Over our sample period, the correlation coefficient is.9905 for daily rates and.99997 for monthly rates.

³ To further investigate the issue, we run a fixed-effect regression with the firm-effect fixed, using firm quarterly earnings as the dependent variable, and the oil prices of the current quarter and the two quarters after as the independent variables. We find the coefficient of the current quarter oil prices to be significantly positive, suggesting higher actual EPS in quarters with higher oil prices. As to the two future oil prices, we do not find them to be significant. Therefore, the current-quarter oil prices should have an overwhelming impact on the firm earning of the contemporaneous quarter. As a result, analysts should overwhelmingly consider the current quarter oil prices in their forecast of the current quarter earnings.

⁴ This partitioning method requires a firm-quarter to have 12 or more analysts to be included in the well-covered sample. It is possible for firms to drop from the sample intermittently. As we cover a long sample period of 17 years, some firms might grow or dwindle, causing significant changes in analyst coverage. For example, Continental Resources (NYSE: CLR) first appears in the sample in 2007 with only three analysts, but the number grows to more than 11 at the end of 2009. An example of losing analyst coverage is W&T Offshore (NYSE: WTI), who consistently appears in the well-covered sample only in the 2008–2009 period, but has then consistently lost coverage and only records three analysts at the end of the sample period according to I/B/E/S. These firms appear in the well-covered sample when they have 12 or more analysts for a given quarter, and drop when not. Admittedly, our partition standard based on the median coverage of firm-quarters might inadvertently exclude some firm-quarters from the well-covered sample even when the firm’s fundamentals are not dramatically different. In unreported tests, we perform robustness checks with alternative definitions of well-covered firms. Specifically, we first apply the median number of average analyst coverage at the firm level, instead of at the firm-quarter level, and define firms with the average number of analysts above the median (8.78) as well-covered. Secondly, we start from our original sample and then add the firm-quarters from firms already in the sample but with 10 or 11 analysts. For both samples, the main results are largely unchanged.