Costless Collars: A Poor Hedge for Utility Customers

Costless collars are an attractive low-cost strategy for speculators who have a directional view on the underlying market. They are not a prudent strategy for a utility looking to reduce commodity price risk to end-use distribution customers. Utilities that employ a costless collar hedging strategy are likely exposing their customers to excess price risk and higher commodity costs versus employing a hedging strategy that uses simple forward contracts. Utility company and utility commission documents available through the Internet suggest the many electric and natural gas utilities employed costless collar hedge strategies on behalf of their customers between 1998 and 2008. This article examines the performance of a costless collar hedging strategy as compared to a forward contract hedging strategy for a hypothetical natural gas utility within the 1998 through 2008 time frame. Key assumptions to this comparison are that the utility has no view on future market prices and that the utility employs a dollar-cost averaging strategy to place hedges.

Collar Mechanics

A collar is a combination of a call purchase and a put sale or vice versa, where the options have the same underlying volume and expiration. A costless collar is achieved when the proceeds from the option sale offset the cost of the option purchase. A costless collar hedging strategy may appear attractive to utilities because there is no premium required and utility customers enjoy some participation should wholesale natural gas prices move lower. For a utility hedging against rising commodity costs, the potential hedge strategy would be the purchase of an out-of-the-money call and the sale of an out-of-the-money put. Mechanically, the call purchase protects from rising commodity costs when underlying market price becomes higher than the call strike. Similarly, prices can move lower until they become lower than the put strike which becomes the commodity price floor for the utility customer.

In the diagram above, the utility would have no financial impact from a (costless) collar strategy if underlying market prices remain between lines A and B. When underlying market prices increase beyond line A (call strike), the utility will have a positive return from the call option purchase. This positive return will be offset by increase commodity costs in the physical market. If market prices decline below line B (put strike), the utility will have a negative return from the put option sale. This negative return will be offset by lower commodity costs in the physical market. The risk profile for the collar above is largely similar to the profile for a long (purchased) forward contract. In fact, the purchase of an at-the-money call and the sale of an at-the-money put, with the same volume and expiration, is the synthetic equivalent of a long forward contract.

A costless collar is attractive to a speculative investor who has a directional view of the market. First, the transaction has no net premium costs. Second, if the investor is right in the market view, the long option will increase faster in value than the corresponding directional position in the underlying market. Similarly, if the investor is wrong on the market direction, the losses from short (sold) option can accumulate faster than the corresponding directional position in the underlying market. The costless collar provides the investor with more leverage than a vanilla long or short forward position. Increased leverage equates to increased risk, which is part of the reason costless collars are not appropriate for utility hedging strategies.

Dollar Cost Averaging

Unless a utility has a prevailing price view, the best long-term hedging strategy on behalf of customers is a linear-decline (dollar cost averaging) strategy. Under a linear decline hedging strategy, an equal percentage of monthly exposures are purchased at regular intervals for specified number of months into the future, with limited regard to market conditions. For example, assume a utility forecasts that customers will use 240,000 MMBtu of natural gas in each month for the next two years. The company will establish an initial hedge where next month’s requirements (January 2008 under INITIAL HEDGE in the graphic below) will be fully hedged and each successive month will be hedged 1/24th less than the prior hedge. Once the initial hedge is established, each subsequent hedge will have the same volumetric amount for all months as shown in the 2ND HEDGE and 3RD HEDGE boxes below. Note that this strategy performs the best when all months receive the same weighted average price each for each transaction. If this process is continued in perpetuity, each month will have 23/24th identical price hedges to the prior month and 23/24th identical hedges to the next month.

Under this strategy, months which are 12 months apart (i.e. January 2008 and January 2009) will have 50 percent of the same hedge components. If this program were to be changed from a 24-month program to a 48-month program, hedges for expected monthly usage would be 47/48th identical and months which are 12 months apart would have hedges that are 77 percent identical. Present day exchange-based and over-the-counter (OTC) natural gas markets could accommodate a 48-month hedging program for most major U.S. natural gas hubs.

The graphs below are based on daily price and settlement data from the New York Mercantile Exchange (Nymex) as if a linear decline hedging strategy had been executed on the business day closest to the 15th of the month, every other month for the period October 1998 through March 2008. The key point from the graphs is that as the hedge horizon increases from 6 months, to 12 months, and finally to 24 months, the month-to-month and year-to-year natural gas expense volatility decreases significantly if hedged is this manner. This is illustrated on the graphs below. The pink line on each graph shows the hedged price of gas as compared to the market price of natural gas on the blue line.

It is clear from the graphs above that a utility customer would have more commodity price stability under a strategy hedging forward 24 months simultaneously versus strategies that hedge less than 24 months.

Collars versus Forwards

The same data and the same logic can be used to estimate the performance of a hedging strategy using collars as compared to a strategy using forwards. For this example, I’ll assume put-call parity, which means that value of an out-of-the-money call has the same value as the out-of-the-money put which is equidistant from the underlying market for a given expiration date. In actuality, put-call parity does not hold in most markets because underlying prices have non-normal distributions. Underlying price distributions and the impact on collar pricing will be discussed a little later in this paper. On the graph below, it is evident that a $1.00 costless collar hedging strategy exposes utility customers to greater commodity price volatility than a simple hedging strategy using forwards.

Mathematically, the standard deviation of monthly price changes for the un-hedged position is $1.24 for the 10-year period in the graph above. This compares to a standard deviation of monthly price changes of $0.37 for the costless collar hedge and $0.16 for the hedge using forwards. It is important to note that the forward hedging strategy resulted in lower average costs than the collar hedging strategy because the general trend in prices in the example period was rising. If the prices trended lower in the example time period, the collar strategy would have resulted in lower prices. In either case, the strategy using forwards will result in less cost volatility.

In the hedge comparison, I assume that daily price changes are normally distributed (put-call parity exists) and no transaction costs. In reality, underlying natural gas price changes have a slightly right-skewed distribution, meaning that an out-of-the-money call will be slightly more expensive than the equivalent out-of-the-money put. This differential in actual costs may be increased by market liquidity and/or collar width. The chart below shows the actual daily percentage price changes for prompt month Nymex natural gas futures for October 1998 through September 2008. Notice the circle which contains one day price spikes which have no comparable events on the left side of the distribution.

Firms that regularly buy and sell options will have multiple option pricing models which rely on different assumptions about the distribution of underlying price changes (normal, lognormal, etc.). Without a sophisticated option pricing model, it should be evident from the graph above that the risk associated with declining prices is less than the risk associated with increasing prices. A utility that is purchasing a call and selling a put has limited downside risk and unlimited upside potential. The counterparty will likely want to be compensated for taking the other side of this transaction. The counterparty compensation (implied cost) will come from the utility’s sold put being closer to at-the-money than the utility’s purchased call in order to achieve a costless collar. Generally speaking, the more out-of-the-money the call and put strikes, the more compensation that will be sought by the counterparty. This is because price probability distributions are less reliable as the distance from the mean (at-the-money price level) increases and because the counterparty has less ability to offset the risk.

Experience and Back Testing

As the Risk Manager for the Kinder Morgan Inc.’s Retail Division, I witnessed the market pricing of costless collars first hand from 2001 through 2007. Following the winter of 2000/2001, the Colorado Public Utility Commission (CPUC) called for more aggressive hedging strategies by natural gas utilities that served Colorado customers. The CPUC staff was particularly intrigued by the no premium feature and by the potential downside price participation offered by costless collars. They recommended that Kinder Morgan Retail institute a hedging program using these structures. In the fall of 2001, Kinder Morgan Retail obliged the CPUC staff and hedged a percentage of the winter natural gas purchase requirements using costless collars. At inception, it was clear that Kinder Morgan Retail was paying a significant implied premium to the market, as represented by the put strike, compared to what was represented by the underlying price distribution. In the years following 2001, Kinder Morgan Retail continued to pay a significant implied premium to the market as compared to what would be expected from the underlying price distribution. The results of the RFQ, including the non-winning offers, were provided to the CPUC every year. I should note that Kinder Morgan Retail customers were far better off when the utility hedged with costless collars than if the utility had not hedged all. However, it is unlikely that Kinder Morgan Retail’s costless collar hedging strategy reduce cost volatility as well as a strategy that used simple forwards. Confidentiality limits the ability to provide a more thorough analysis in this specific example.

The expected premium versus premium paid is largely a function of market liquidity. At Kinder Morgan Retail, the costless collars were tied to natural gas for delivery at price points in the Rocky Mountains. In more liquid markets, such as Nymex natural gas futures, much less implied premium will have to be paid to the market to execute a costless collar hedge. Even if there is no implied premium (perfect put-call parity), customers protected with hedges using costless collars are still subject to more cost volatility than customers protected with hedges using forwards. This is clearly demonstrated in the Nymex futures charts above. It is easy for utilities or for the public utility commission to back-test the performance of an options’ based hedging strategy. Simply adjust the historic supply costs by removing the financial impact of the option trades and replace them with hypothetical forward trades where the volume and transaction date (time) are identical to the options. For the utility or utility commission seeking to reduce cost volatility for end-use distribution customers, analysis will show that costless collars are not a prudent long-term hedging strategy when compared to a hedging strategy using simple forwards.