Weather derivatives are financial instruments that can be used by organizations or individuals as part of a risk management strategy to reduce risk associated with adverse or unexpected weather conditions. The difference from other derivatives is that the underlying asset (rain/temperature/snow) has no direct value to price the weather derivative.
Contents |
Farmers can use weather derivatives to hedge against poor harvests caused by drought or frost; theme parks may want to insure against rainy weekends during peak summer seasons; and gas and power companies may use heating degree days (HDD) or cooling degree days (CDD) contracts to smooth earnings. A sports event managing company may wish to hedge the loss by entering into a weather derivative contract because if it rains the day of the sporting event, fewer tickets will be sold.
Heating degree days are one of the most common types of weather derivative. Typical terms for an HDD contract could be: for the November to March period, for each day where the temperature rises above 18 degrees Celsius keep a cumulative count of the difference between 18 degrees and the average daily temperature. Depending upon whether the option is a put option or a call option, pay out a set amount per heating degree day that the actual count differs from the strike.
The first weather derivative deal was in July 1996 when Aquila Energy structured a dual-commodity hedge for Consolidated Edison Co.[1] The transaction involved ConEd's purchase of electric power from Aquila for the month of August. The price of the power was agreed to, but a weather clause was embedded into the contract. This clause stipulated that Aquila would pay ConEd a rebate if August turned out to be cooler than expected. The measurement of this was referenced to Cooling Degree Days measured at New York City's Central Park weather station. If total CDDs were from 0 to 10% below the expected 320, the company received no discount to the power price, but if total CDDs were 11 to 20% below normal, Con Ed would receive a $16,000 discount. Other discounted levels were worked in for even greater departures from normal.
After that humble beginning, weather derivatives slowly began trading over-the-counter in 1997. As the market for these products grew, the Chicago Mercantile Exchange introduced the first exchange-traded weather futures contracts (and corresponding options), in 1999. The CME currently trades weather derivative contracts for 18 cities in the United States, nine in Europe, six in Canada and two in Japan. Most of these contracts track cooling degree days or heating degree days, but recent additions track frost days in the Netherlands and monthly/seasonal snowfall in Boston and New York. A major early pioneer in weather derivatives was Enron Corporation, through its EnronOnline unit.
In an Opalesque video interview, Nephila Capital's Barney Schauble discusses how some hedge funds have now begun focusing on weather derivatives as an investment class. Counterparties such as utilities, farming conglomerates, individual companies and insurance companies are essentially looking to hedge their exposure through weather derivatives, and funds have become a sophisticated partner in providing this protection. There has also been a shift over the last few years from primarily fund of funds investment in weather risk, to more direct investment for investors looking for non-correlated items for their portfolio. Weather derivatives provide a pure non-correlated alternative to traditional financial markets.
There is no standard model for valuing weather derivatives similar to the Black-Scholes formula for pricing European style equity option and similar derivatives. That is due to the fact that underlying asset of the weather derivative is non-tradeable which violates a number of key assumptions of the BS Model. Typically weather derivatives are priced in a number of ways:
Business pricing requires the company utilizing weather derivative instruments to understand how its financial performance is affected by adverse weather conditions across variety of outcomes (i.e. obtain a utility curve with respect to particular weather variables). Then the user can determine how much he is willing to pay in order to protect his/her business from those conditions in case they occurred based on his/her cost-benefit analysis and appetite for risk. In this way a business can obtain a 'guaranteed weather' for the period in question, largely reducing the expenses/revenue variations due to weather. Alternatively, an investor seeking certain level or return for certain level of risk can determine what price he is willing to pay for bearing particular outcome risk related to a particular weather instrument.
The historical payout of the derivative is computed to find the expectation. The method is very quick and simple, but does not produce reliable estimates and could be used only as a rough guideline. It does not incorporate variety of statistical and physical features characteristic of the weather system.
This approach requires building a model of the underlying index, i.e. the one upon which the derivative value is determined (for example monthly/seasonal cummulative heating degree days). The simplest way to model the index is just to model the distribution of historical index outcomes. We can adopt parametric or non-parametric distributions. For monthly cooling and heating degree days assuming a normal distribution is usually warranted. The predictive power of such model is rather limited. A better result can be obtained by modelling the index generating process on a finer scale. In the case of temperature contracts a model of the daily average (or min and max) temperature time series can be built. The daily temperature (or rain, snow, wind, etc.) model can be built using common statistical time series models (i.e. ARMA or Fourier transform in the frequency domain) purely based only on the features displayed in the historical time series of the index. A more sophisticated approach is to incorporate some physical intuition/relationships into our statistical models based on spatial and temporal correlation between weather occurring in various parts of the ocean-atmosphere system around the world (for example we can incorporate the effects of El Niño on temperatures and rainfall).
We can utilize the output of numerical weather prediction models based on physical equation describing relationships in the weather system. Their predictive power tends to be less or similar to purely statistical models beyond time horizons of 10–15 days. Ensemble forecasts are especially appropriate for weather derivative pricing within the contract period of a monthly temperature derivative. However, individuals members of the ensemble need to be 'dressed' (for example with gaussian kernels estimated from historical performance) before a reasonable probabilistic forecast can be obtained.
A superior approach for modelling daily or monthly weather variable time series is to combine statistical and physical weather models using time-horizon varying weight which are obtained after optimization of those based on historical out-of-sample evaluation of the combined model scheme performance.