Million Dollar Forecast Communication? Value of Temperature Uncertainty in Derivatives Pricing

Weather derivative contracts allow companies to hedge their weather risk and investors to make a profit through trading. The decision to buy or sell can be supported with forecast information, preferably including the possible errors. This study attempts to determine the value of different levels of uncertainty information. Idealized probability distribution forecasts were created for different amounts of unbiased error, strongly skewed (frontal) events, and two regime events. These forecasts were converted into ones with variable error bars, confidence levels, and a fixed range. The joint observed-forecast distributions were created assuming the original distribution forecasts were perfect. Bayesian inference was used to compute payoff probabilities for a number of typical Chicago Mercantile Exchange contracts. The value of the four forecast types versus climatology was determined using a range of possible decision and utility models focusing especially on high-payoff situations.