No Chills or Burns from Temperature Surprises: An Empirical Analysis of the Weather Derivatives Market

Abstract

This article examines the efficiency of a relatively new market, the weather derivatives market. The weather market is an especially interesting market to study, since it is relatively new and less liquid and whose underlying has no fundamental value. We examine the weather derivatives futures market traded on the CME (Chicago Mercantile Exchange) in both HDD (heating degree days) futures contracts and CDD (cooling degree days) futures contracts in 18 cities across the United States. We examine the efficiency of the weather market in three ways. The first way is by comparing the market's implied forecasts for the weather against other forecasts, including variations of a historical forecasts and forecasts based upon the National Weather Service 7-day MOS prediction model. The second way is by looking at surprise temperatures across cities and across different days of the month and subsequent returns from buying or selling those weather futures contracts. We examine the properties of these returns to determine whether we observe overreaction in the weather derivatives market. The third way is by looking at weather patterns across different cities, i.e. cross auto-correlation of actual weather patterns compared to derivative contracts. In this article, we find that the weather derivatives market is quite efficient despite its lack of depth. It also might indicate that markets with shorter horizons and an impossibility of inside information will generally be more efficient.

Keywords: Weather derivatives, weather forecasting, market efficiency, HDD, CDD

JEL Classification: G0, G14