Natural gas storage forecasts: Is the crowd wiser?


Inventory information is critical for natural gas market participants because it directly measures the supply and demand fundamentals. The more accurate this information is, the more investors can either hedge the price movements triggered by news releases or speculate on it. Professional analyst forecasts from Reuters, for example, play a crucial role in helping market participants to predict the weekly U.S. Energy Information Administration’s (EIA) natural gas storage changes.

Recently there has been an emergence of crowdsourced forecasts that aim to compete with professional analysts. Online platforms, like Estimize, forecast economic indicators and earnings estimates across developed and major emerging markets. Estimize has provided crowdsourced forecasts for the EIA Natural Gas Storage report since the first quarter of 2014. However, it is unclear whether natural gas market stakeholders can benefit from this, considering such information alongside forecasts provided by professional analysts.

We examined the effectiveness of crowdsourced forecasts against professional forecasts for natural gas storage changes. We employed several measures of forecast accuracy and found that, on average, crowdsourced predictions were less accurate than those of the professionals. Investigating market reactions (price, volume, and volatility) to the accuracy of consensus forecasts, we document an absence of market reaction to the Estimize consensus forecast when controlling for Reuters. Our findings, therefore, indicate that crowdsourced forecasts do not add information to natural gas stakeholders beyond what is already contained in professional analyst forecasts.

We offer three factors for the above findings:

  1. more significant divergence of opinions among crowd analysts,
  2. lower incorporation of the publicly available information in crowdsourced relative to professional forecasts, and
  3. a greater need for social-recognition that leads crowd analysts to issue more extreme forecasts.

These factors point toward anti-herding behaviour which is more apparent among crowd analysts than professionals and helps to explain the accuracy gap between the crowdsourced and professional forecasts.

Our research suggests that it is unlikely that crowdsourced forecasts would better predict the EIA’s natural gas storage announcements. The gap in accuracy between Reuters and Estimize forecasts narrows over time.  We do not rule out the possibility that, as crowd platforms mature, crowdsourced forecasts may offer a useful incremental source of information and even outperform professional forecasts. However, this is not what we observe over our sample period.

This overview is based on the paper entitled "Natural gas storage forecasts: Is the crowd wiser?" authored by Adrian Fernandez Perez, Alexandre Garel, & Ivan Indriawan

Alternative link: doi:10.5547/01956574.41.5.afer