Challenges in pricing weather derivatives: A study

Abstract

Weather conditions have significant impact on various industries such as energy, power, agriculture, leisure, retail, etc. According to recent studies, up to twenty percent of US GDP is influenced by weather conditions. Thus, a huge sector of industries is exposed to weather risks. Weather derivatives are special types of derivative contracts whose payoff depends on a weather variable such as temperature, humidity, etc. These have emerged as an effective instrument to hedge risks related to weather conditions. These are different from insurance contracts in that the buyer does not need to have a weather sensitive production to hold a weather derivative unlike the insurance contracts. Also, in the case of weather derivatives, the payoff does not depend on the actual loss suffered by the holder of the instrument; it only depends on actual weather conditions. Due to such flexibility, weather derivatives are fast gaining popularity and the market is continuously increasing. Pricing of these derivatives is a crucial issue since the underlying of these derivatives is not tradable. Thus the market for weather derivatives is incomplete and traditional pricing approaches such as Black Scholes model are not applicable. Various other pricing models such as valuation based on temperature modeling, historical Burn analysis and actuarial pricing model have been applied by various people to price the weather derivatives. In the report, we have briefly talked about weather derivatives, challenges in pricing them and the existing pricing approaches. Then we have developed an auto regressive model of temperature by using temperature data of Cincinnati Northern Kentucky Airport for the period January 1997 to December 2006. This model has been used to forecast the values of HDD and CDD indices using Monte Carlo Simulation techniques to arrive at the price of weather futures. The obtained prices have been compared with the actual prices of these futures on the Chicago Mercantile Exchange (CME). Further, the temperature data has been deseasonalized to develop a new model and it has been found out that deseasonalizing the temperature data yields better results as compared to the seasonal data.