The industry may be four years old and worth $7.5 billion worldwide, but for some firms weather risk and weather derivatives are still a new idea. Here, industry analyst Connie Paoletti presents a guide to the fundamentals of weather risk management

Weather derivatives have had the attention of energy and financial companies since the last visit of the El Niño weather phenomenon in 1997. But while some industry professionals are well versed in this area, others are only just being introduced to it. So an outline of the fundamentals of weather derivatives could prove very useful.

Weather variations influence the price of a commodity by influencing the demand for the commodity – for example, heating oil, orange juice, natural gas, cattle or electricity. But consider the correlation coefficient – the relationship between the commodity price and a weather parameter such as temperature – and you may find a correlation coefficient of around 0.5 (see figure 1).

A correlation coefficient of 0.5 means there is some indication that, for example, summer electricity prices increase as the temperature increases, but it is hardly near a perfect correlation of 1.0. We may conclude that circumstances apart from weather influence the price. Such circumstances may include generation outages, transmission constraints or government legislation.

However, when we compare the volume or demand for – rather than the price of – a commodity with temperature, the correlation coefficient may be greater than 0.95. Therefore, we may conclude that the pri-mary influence on demand for a particular energy commodity is the weather. Weather derivatives work as a volume risk management tool because consumption of a com-modity is closely tied to weather variations. Take, for example, a heating oil distributor that makes a certain profit margin on each gallon sold during a particular winter.

If the winter is warmer than normal, the distributor can expect less revenue due to the decrease in the number of gallons sold. In the case of a warmer-than-normal winter, the company’s base price for heating oil may go down, but that does not matter, because its sales volume is less.

During a cold winter, the demand for heating oil increases along with the cost per gallon. If the distributor’s price is not hedged, it is exposed to higher purchase costs. Therefore, weather derivatives should be used to complement price hedging.

The price and the volume hedge can work together to provide earnings stability for the company. Depending on the specific risk profile of a firm, weather derivative transactions may be constructed to provide the desired degree of protection. There is even an opportunity to embed or combine weather derivative costs and payments with the commodity price.

Strike level

The strike level is key to a weather derivative contract’s terms, because it indicates at which point the contract will necessitate a payout. It all depends on the amount of coverage a company requires.

The weather risk parameter, which refers to the type of weather event that affects earnings, must also be determined. Examples of such weather risk parameters include temperature, rainfall or precipitation, snowfall or snow depth, relative humidity, wind speed or wind chill, and actual data versus forecast.

The most popular weather measurement and reporting reference is the degree-day, which derives its value from the average temperature over a 24-hour period. The high and the low reported temperatures are averaged, and that average is either added or subtracted from a baseline temperature. The baseline is usually 65° Fahrenheit (18° Celsius).

An average temperature below 65°F implies that people will be turning on the heat and consuming more heating oil, natural gas or electricity to warm their homes. For every degree that the average temperature is below 65°F, one heating degree-day (HDD) is reported.

Most utilities have historical data on commodity usage, and can easily determine the impact of a warmer-than-usual winter in terms of dollars per degree-day. The normal winter time frame in the US is November through March. The lower the number of HDDs, the warmer the winter season.

Over a winter season, there may be a warmer-than-normal November, colder-than-normal December, normal January, warmer than normal February and colder than normal March to end up with what may be called a normal winter. As degree-days over seasons are cumulative, they tend to smooth out the extremes and are therefore less expensive to purchase.

Qualification and quantification of weather risk is possibly the most important and difficult thing to do with regard to structuring weather derivatives. Once weather risk is qualified in terms of temperature, precipitation or whichever parameter is relevant, it is important to quantify the risk. This is accomplished by performing a weather normalisation to determine a company’s earnings associated with a normal weather pattern.

The weather normalisation procedure involves the analysis of historical weather and revenue data to access the correlation between a specific weather parameter and earnings. Following normalisation, a company should determine the value of the quantifiable variable that is unacceptable based on its risk profile. The next step is to estimate the impact on company earnings of a weather pattern that is not within the "normal" range.

Then one must choose the weather tool that fits the company’s risk profile. It could be an exchange-traded contract, over-the-counter (OTC) swap, OTC option on degree-days or a defined event, such as temperatures dropping below freezing or snow depth exceeding a predetermined amount. Finally, a company must assess the amount of acceptable risk relative to the cost associated with mitigating that risk.

When the strike for a swap is selected, the swap strike is typically near the normal or mean. However, the swap provider is likely to put the swap strike slightly higher than the mean on the cooling degree-day (CDD) swap and slightly lower than the mean on the HDD swap. In doing so, they are essentially charging a premium for taking on the floating weather risk, although no money is paid upfront.

Practical example

Take the example of Burnitup, a fictitious natural gas distribution company. Burnitup is interested in reducing its earnings volatility to improve its share price performance and address controllable risks within the organisation.

Its revenues are affected by the margins they collect on natural gas throughput. The past several winters have been warmer than normal, so earnings have been down. While the forecast for this coming year is a return to normal winter weather patterns, the firm does not want to take any chances.

Hence, Burnitup wants to fix its weather-related earnings by fixing the number of HDDs for the coming winter season. Historical weather data shows the 30-year average at 4,943 HDDs and the 10-year average at 4,755 HDDs.

After studying the relationship between throughput and historical weather data, Burnitup determined that each HDD deviation from normal negatively affects earnings by $5,000 per HDD. It wants to fix the number of HDDs at the 30-year average of 4,943 for the winter season based on readings at the local airport.

The best swap strike proposed by the weather broker in the deal, due to the current warming trend, was based on the 10-year average of 4,755 HDDs. Burnitup agreed to a payout of $5,000 per HDD for the period November through March and a maximum payout of $2 million for a swap strike of 4,755 HDDs.

At the end of the contract term, the actual number of HDDs ended up being 4,307. Since the actual number of degree-days was below the swap strike, Burnitup will receive a payment. The payment will be equal to the difference between the swap strike and the actual data, multiplied by $5,000 per degree-day. This comes out as 448 HDDs multiplied by $5,000, which equals $2,240,000. But the payment will be limited to a maximum payout that is set at $2 million.

The $2 million will be added to earnings to make up for part of the deficit resulting from the warmer-than-normal winter. If the actual number exceeded 4,755 HDDs, Burnitup would have had to make a payment to the weather broker. The money to make this payment comes from the increase in Burnitup earnings due to the increase in throughput.

If Burnitup does not want to give up in-creased earnings during a colder-than-normal winter, but still wants protection against a warmer-than-normal winter, an analysis of the weather variability is required to determine the strike level.

The simplest way to measure overall variability is to calculate the standard deviation of the historical weather data. The standard deviations of the 30-year average and 10-year average are 363 and 339 HDDs, respectively. In selecting a strike for an OTC option, the strike is typically above the normal number of degree-days, or mean, and below normal for floors.

Out-of-the-money options with cumulative indexes are typically considered for most weather derivatives. And because they smooth out the volatility, they tend to be cheaper than other options. The strike level on a weather derivative can be anything when offered as an OTC transaction, but the standard strike level that is quoted is one half a standard deviation from the mean. Then the premium is determined based on the strike level, taking into account the reduced volatility.

Given the historical weather data for Burnitup, we calculate the targeted strike of an HDD floor by dividing the 10-year standard deviation by two and subtracting it from the 10-year average. So we have 339 divided by two, which is roughly 170, and subtract this from 4,755, which results in a strike of 4,585.

For the chosen location from November through March, an HDD floor with a strike level at 4,585 and a payout of $5,000 per degree-day was quoted to Burnitup at a premium of $300,000. As with all options, this premium is paid upfront and is non-refundable (see figure 2). Had Burnitup chosen the HDD floor as opposed to the swap, the formula would be to subtract the 4,307 from 4,585 to get 278 and multiply it by $5,000, making the payout $1,390,000.

A collar is a variation on the straightforward swap. A range – rather than a single strike level – is guaranteed. A collar is essentially constructed by buying one option and selling another. With a collar, companies lock in both a floor and a ceiling without being charged a premium.

On the Burnitup hedge, a costless collar could be constructed around the average by adding and subtracting one half a standard deviation from it. The range would be established between a floor of 4,585 and a ceiling of 4,925. If the actual number of HDDs is below the floor, Burnitup receives a payment. If the actual number of HDDs is above the ceiling, Burnitup makes a payment to the weather broker. If the actual number of HDDs is between the floor and ceiling, the result is no payout.

Connie Paoletti is an energy consultant at The Oxford Princeton Programme, a provider of training to the energy industry and beyond

Click here for further information from the Risk Waters Group.