Analyst forecasts and price discovery in futures markets: The case of natural gas storage

Abstract

We investigate analyst forecasts in a unique setting, the natural gas storage market, and study the contribution of analysts in facilitating price discovery in futures markets. Using a high-frequency database of analyst storage forecasts, we show that the market appears to condition expectations regarding a weekly storage 452 Gay, Simkins, and TuracJournal of Futures Markets release on the analyst forecasts and beyond that of various statistical-based models. Further, we find that the market looks through the reported consensus analyst forecast and places differential emphasis on the individual forecasts of analysts according to their prior accuracy. Also, the market appears to place greater emphasis on analysts' long-term accuracy than on their recent accuracy. forecast accuracy. We consider a few alternative weighting schemes and find supporting evidence that the market looks through the consensus forecast and places differential emphasis on the forecasts of analysts according to their prior accuracy. Further, it appears that the market focuses more on analysts' longterm forecast accuracy than their recent accuracy.Our investigation also entails two sidebars that create interesting information environments for analyzing market behavior. First, gas in storage is highly cyclical and involves two calendar periods during which supply and demand fundamentals alternate in importance and thus present differing challenges to analysts. During the "injection season" (April to October), storage typically accumulates because demand is lower during the warmer months and new inventories are being added in advance of the next winter heating season. During the "withdrawal season" (November to March), inventories are drawn down due to high consumption demand in the winter heating season. Demand shocks are mainly weather driven and are larger in magnitude and less predictable than supply shocks, which are typically technology driven. Hence, weekly storage changes are typically greater and more difficult to predict during the withdrawal season. 3 We find that analyst forecasts are indeed less accurate and more dispersed during the demand-driven withdrawal season than during the supply-driven injection season.Second, we consider a major change in responsibility for conducting the weekly survey. The survey was originated in January 1994 by the American Gas Association (AGA)-the industry trade group. In May 2002, the U.S. Department of Energy's "Energy Information Administration" (EIA) assumed responsibility from the AGA. The EIA implemented several changes to the survey procedures with the intention of reducing noise and producing more accurate estimates. We find that analyst accuracy improved and forecast dispersion decreased following the EIA takeover.Though we do not claim that our weighting schemes are optimal, our study adds to the literature on combining forecasts to produce a superior forecast. A seminal paper by Bates and Granger (1969) first suggested and proposed a method to optimally combine forecasts. This study spawne...

Noisy Inventory Announcements and Energy Prices

“…The OLS estimate of the response coefficient for natural gas futures is about −2.4, implying that an unexpected 1% increase in natural gas in storage results in a 2.4% drop in natural gas futures prices. This estimate is somewhat larger in absolute value than the corresponding estimate for an earlier sample period in Gay et al (). The ITC estimate of the response coefficient for natural gas is almost four times as large as the OLS estimate.…”

“…To measure the unexpected change in inventory, previous studies use the Bloomberg consensus forecasts. Gay et al () provide evidence that this measure of market expectations is an improvement over forecasts obtained from historical inventory data using statistical approaches. However, the resulting measure of the unexpected change in inventory contains measurement error.…”

“…Linn and Zhu () find a large increase in volatility of natural gas futures returns after the release of the Natural Gas Storage Report. Gay, Simkins, and Turac () show that a 1% unexpected increase in natural gas inventory results in an approximately 1% drop in the natural gas futures price. Chang, Daouk, and Wang () find a significant response of crude oil futures returns to unexpected inventory changes immediately after the announcement.…”

“…Chang, Daouk, and Wang () find a significant response of crude oil futures returns to unexpected inventory changes immediately after the announcement. Using cross‐sectional data on analyst forecasts of inventory, both Gay et al () and Chang et al () provide evidence that prices react more strongly to forecasts of analysts with better prior forecast accuracy. Halova () finds that unexpected changes in natural gas inventory have a statistically significant effect on prices of crude oil futures.…”

The impact of crude oil inventory announcements on prices: Evidence from derivatives markets

“…These effects are also examined in the context of monetary conditions, OPEC meetings, inventory storage reports, and demand and supply shocks. Gay, Simkins, and Turac (2009) also study the natural gas storage market and show that futures prices are responsive to the surprise component of the EIA announcement. Horan, Peterson, and Mahar (2004) examine implied volatility of options on crude oil futures surrounding OPEC meetings and find that implied volatility embedded in option prices drift upward prior to scheduled information releases and drop afterwards.…”