How to measure hedging effectiveness and regulatory policy.
Stephen Maloney is with Towers Perrin’s Financial Risk team, where he is responsible for regulatory policy, compliance, and dispute resolution engagements involving commodity markets. He can be reached at firstname.lastname@example.org
Hedging programs promise protection against energy-market price spikes, and they can be important to the regulatory goal of sustainable, lowest long-term service cost. But how much price protection is enough in natural-gas markets? When is too much protection, well, too much? What is the most efficient utilization of risk capital when hedging energy supplies?
Questions of efficiency immediately beg questions of measurement and process control. Many CFOs will say utilities really need better guidance designing and adjusting their energy portfolios (both electric and natural gas). What metrics best characterize the dual goals of lowest service cost and optimal hedging performance? How much accuracy, sensitivity and metric reliability should be expected?
These questions are not rhetorical. The internal control structure and procedures established under Sarbanes-Oxley have improved financial report quality. These structures and procedures do not easily accommodate spreadsheet or ad hoc applications that often are used to structure hedges. CFOs need to have confidence that processes perform to a specified standard, are auditable, and produce reliable results.
Service at the Lowest Reasonable Long-Term Cost
Electrical distribution companies (EDCs) and local natural-gas distribution companies (LDCs) are tasked by regulation to provide reliable service at the lowest reasonable long-term cost and with minimal volatility. EDCs and LDCs have a legitimate concern that their operational and market decisions may be subjected to after-the-fact prudence reviews that could seem like a bureaucratic game of “hide the peanut.” Service providers need unambiguous performance standards. It’s tough running a process when the goalposts can move at any time.
Procurement auctions promise low cost. But as with any optimization problem, there’s a bit of “needle threading” involved. A local optimum is not necessarily global in nature. Small perturbations around the optimal can have surprisingly uneconomic effects. Optimization is not like horseshoes. A near-miss can be very non-optimal.
Regulators increasingly recognize that seeking optimal, integrated solutions can be counterproductive. Rather than relying solely on procurement auctions, a growing number of regulators have adopted the “portfolio approach” policy that may include procurement auctions for a portion of the secured supply along with a hedging program aimed at stabilizing prices and volatility. In many cases, the portfolio approach may operate within an incentive mechanism prescribing a general operating band, and some formula for allocating unanticipated costs or the benefits of exceptional performance between shareholders and customers.
A hedging strategy employs various physical and financial strategies with the goal of offsetting the risk of unfavorable price swings when purchasing energy. While comprising mostly options and swaps, financial hedging instruments can include futures, basis swaps, and fixed-price swaps involving natural gas and possibly other commodities whose price movements are known to be related to energy price movements.
In addition to financial instruments, weather derivatives and natural-gas storage provide important vehicles for mitigating price volatility through physical hedges against shortages and disruptions to pipeline operations.
Hedging is wholesome and good—just like apple pie. Unfortunately, while the concept of apple pie is difficult to fault, some apple pies can be pretty repulsive.
Consequently, while regulators generally are receptive to hedging, nobody is writing a blank check merely on a promise to do good things. Like CFOs, regulators must have confidence that a service provider’s business processes deploy the proper amount of “volatility insurance” at reasonable prices. Further, such processes must be sufficiently responsive to properly tune that price insurance as market conditions evolve.
So the regulatory standard is not always entirely clear, in part because hedging effectiveness is a business process issue as much as it’s a performance metric issue. Some jurisdictions have no policy. Others don’t object to hedging but reserve the option to review effectiveness under a general prudence standard.
Before considering regulator’s reviews, let’s take a quick look at natural-gas price market dynamics to see how current events may be fueling the policy debate.
Natural-Gas Price Dynamics
Hedging programs evolved in response to natural-gas price dynamics and seek to dampen price volatility. These dynamics reflect the combined effects of industry fundamentals and seasonal drivers.
To make matters worse, long-term industry fundamentals are not balanced. U.S. natural-gas production peaked in 2001 and is in decline. Yet demand for natural-gas-fired electric generation has grown more than 30 percent over that same period. Average wellhead prices roughly have tripled.
The current decade’s rise in natural-gas prices and volatility is particularly noticeable in the forward curves. We see rising prices and volatility year after year.
While demand has grown some 25 percent over the past decade, U.S. storage capacity barely has budged, increasing about a tenth as much. Thus the U.S. natural-gas supply chain is a “stiffer” system today and generally less capable of dampening perturbations.
A stiffer supply chain is more vulnerable to perturbations and more likely to respond with spikes and harmonics. A stiffer supply chain also tends to attract speculative trading practices. When taken to excessive levels, the supply chain can generate sudden, reasonable or unwarranted price changes with a reinforcing or “positive feedback” effect on price levels and volatility. The winter 2005-2006 price spikes most often associated with Hurricane Katrina’s disruptions are very evident in natural-gas price history.
Recalling its investigation of the winter 2005-2006 price spikes and Amaranth’s role in price volatility, the U.S. Senate Permanent Subcommittee on Investigations recently concluded “the increasing amount of financial speculation in energy markets has contributed to the steep rise in energy prices over the past few years.”1
The subcommittee further concluded that “Amaranth’s excessive speculation in natural-gas futures played a central role [in the unusually high prices and large differences between the winter and summer prices in 2006].”2 Companies that stayed too long in the futures market during this period generally paid more for winter gas. The subcommittee observed many of these costs were passed on to consumers.
Clearly, a hedging program mindful of market fundamentals would have been better equipped to cover its supply requirements through other markets. For example, as Amaranth bid up prices, it has been argued companies simply could have traded the spot market. In other words, hedging according to set formulas may not always result in efficient procurement. In light of these considerations, hedging programs are coming under closer scrutiny.
Regulator Views of Utility Hedging Programs
The California energy crisis put hedging programs on the regulatory radar screen. In a speech before the American Gas Association (AGA) in early 2001, then FERC Commissioner Linda Breathitt urged state regulators to investigate the benefits of permitting energy companies to hedge gas purchases to decrease volatility. The previous December, FERC recommended California utilities put 95 percent of their load in forward markets to minimize exposure to price volatility on the spot market.
An ever larger number of utilities are now hedging their supply requirements. AGA reports 88 percent of natural-gas utilities surveyed incorporate financial instruments as a hedge against market price volatility, compared to six years prior, when 55 percent of the survey sample reported using financial instruments as a hedge. AGA says 29 companies use options, 36 companies use fixed-price contracts, 26 companies use swaps, 19 companies use futures contracts, and 5 companies use weather derivatives. Companies that employed hedging programs generally saw positive impacts on their bottom lines and on their customers over the years. For example, FPL’s April 2004 8-K reported its hedging program contributed to the better-than-average performance in the northeast. Baltimore Gas & Electric reportedly hedges 10 to 20 percent of its winter gas requirements.
It’s difficult to dispute the benefits of hedging a sizable portion of indicated requirements when prices are trending up and volatility is high. It’s a simple matter of averting obvious hazards. It’s apple pie.
But risk aversion is not risk management. When the ramp-up slows and volatility settles, simple models and policies can lead to significant overhedging and unnecessary costs.
For example, Keyspan Energy Delivery New England bought two-thirds of the gas needed for the 2006-2007 winter in advance, at prices that turned out to be generally higher than they were as the winter season began. Massachusetts ratepayers paid markedly higher rates last winter to enjoy the benefits of more certainty in monthly bills.
This is not simply complaining about paying an insurance premium without filing a claim. Economic use of risk capital demands insurance premium purchases accurately reflect the risk. When the capital return from insurance falls below the cost of capital, self-insurance becomes attractive.
Similarly, the $2 monthly increase in residential bills would seem a bargain in a Katrina winter season. But should utilities routinely enter the winter season running 75 to 80 percent hedges with little regard to market fundamentals?
The California Public Utilities Commission (CPUC) thinks not. In rejecting the initial utility proposals for a long-term hedging program, the CPUC acknowledges the risks in natural-gas market instability. Nevertheless, the CPUC observes such risks do not implicitly justify spending enormous sums on hedging substantial portions of the utilities' gas portfolios.
The CPUC also cites the absence of evidence that the hedging programs were consistent with best practices, prevailing financial theory or statistics about the effects on natural-gas rates in other jurisdictions of purchasing such financial instruments. CPUC further notes the absence of an evaluation of what ratepayers would be willing to pay for various levels of protection from price spikes.
“The lack of supporting analysis is of particular concern because the utilities are proposing much higher levels of hedging purchases and spent large sums last year that provided almost no benefit to ratepayers.”3
California’s Regulatory Policy Initiatives on Hedging
In recent years, efficient natural-gas procurement in California was governed by various shareholder incentive mechanisms. These mechanisms include the gas procurement performance-based ratemaking (PBR) mechanism at San Diego Gas & Electric (SDG&E), the gas cost incentive mechanism (GCIM) at Southern California Gas (SoCalGas), and the gas core procurement incentive mechanism (CPIM) of Pacific Gas and Electric Co. (PG&E). Generally, these mechanisms are very similar.
For example, PG&E’s CPIM establishes procedures for evaluating and reporting gas procurement costs. Among other things, CPIM procedures provide methods for computing a tolerance band around the benchmark. By definition, costs within the band are judged reasonable. Gains or costs may be shared between shareholders and customers if actual costs fall outside the tolerance band.
Recall hedging seeks to manage costs within a tolerance band and would seem to fit naturally within this kind of incentive mechanism.
In 2006, settlements of certain class-action suits arising out of the California energy crisis and price spikes in the period March 2000 through May 2001 included proposals to modify the California incentive mechanisms. Currently before the CPUC for review and approval, these modifications would, among other things, exclude all financial transactions used to hedge natural-gas prices for any portion of the November through March period (winter hedges), and they would establish a long-term hedging program on behalf of core gas procurement customers beginning with winter 2007-2008.
If hedging generally would fit within a procurement incentive mechanism and is generally a good thing to do, why remove it? Well, one answer is that all policies have counterintuitive effects. In addition to greater efficiency, performance incentive mechanisms have been known to create incentives for performance that merely appear profitable. Hedging addresses volatility and not the expected net value of a portfolio. This consideration was a factor in CPUC’s previous decisions to unbundle hedging programs from performance incentive mechanisms.
The CPUC expects to render its decision late in 2007. If the CPUC establishes a long-term hedging program outside the current incentive mechanism, it will bring many practical performance issues into sharp focus and perhaps raise the bar on what constitutes efficient use of risk capital.
What Performance Standard for Hedging Effectiveness?
To begin with, we probably should distinguish between testing the efficacy of derivatives as hedging instruments that can be recognized in financial statements and testing the economic efficiency of hedging.
Financial Accounting Standard No. 133 (FAS 133) provides general guidelines for derivative mark to market and testing hedge effectiveness. FAS 133 tests comprise both historical performance (a retrospective test) and anticipated future performance (a prospective test) of the hedge.
Gains or losses on a derivative instrument not designated as an FAS 133 hedging instrument are recognized in current period earnings. Under hedge accounting, however, the combined profit and loss from the derivative and the hedged item are recognized in earnings in the same accounting period. In other words, hedging is not simply a matter of efficiently procuring energy supplies going into a winter season. It has implications on a utility’s financial statement (reported earnings) because not every item is marked to market in the current period.
The FAS 133 "80/125 Rule” judges a hedge effective if the ratio of the change in derivative value to hedged item values is between 80 and 125 percent. Many hedges fail this test under stable market conditions, and would not be recognized as effective for financial reporting purposes.
It’s best to consider FAS 133 hedge effectiveness tests as gatekeeper tests controlling access to financial statements. While FAS 133 hedge effectiveness may tell us something about the hedge instrument, so-called “effective hedges” may not dampen volatility. And FAS 133 tells us nothing about optimizing performance. Effectiveness and risk-capital efficiency are separable issues.
All of this leaves CFOs and regulators with a practical problem: When presented a collection of “effective hedges” within a variety of possible portfolios, which portfolio should be chosen?
There are many risk-based metrics for ordering portfolio preferences. Some companies use value at risk (VaR), which has its own problems. They examine how the previous day’s closing prices affect their portfolio and adjust their positions so as to satisfy supply requirements within a VaR limit established by risk management. Portfolios that reduce performance or increase VaR are less desirable compared to those that increase performance or reduce VaR.
In many cases, this strategy is crudely implemented through a pseudo “delta neutral” hedging strategy. A “delta neutral” hedging strategy seeks risk-neutral portfolios on the naïve assumption that energy markets are perfect, liquid markets unaffected by the effects of option pricing. Of course, naïve assumptions are not realistic, and unbiased, pseudo “delta normal” hedging practices are unlikely to be risk neutral. Worse, they can even be risk enhancing for many portfolios and scenarios.
Moreover, when presented with the dual problem of improving performance and reducing VaR, which goals dominate, by how much, and measured over what period? There needs to be more structure and rigor in methods used to measure risk and procedures employed to select risk-efficient portfolios.
Recall that regulatory policy seeks to maximize the cost savings in supply procurement while minimizing the risk. Proper implementation of this policy consequently means that utilities need a structured, documented process for selecting a mathematically optimized portfolio from among an extraordinary number of possible portfolios.
Just so we’re clear, optimization does not include running a program some number of times and performing a visual inspection for the “optimal” case. Unfortunately, life is not that easy.
Fortunately, whatever challenges exist in creating and managing business processes, implementing mathematical models with the requisite sophistication is much less an issue these days. A variety of standard stochastic optimization methods (e.g., branch-and-bound, Benders decomposition, Lagrangian relaxation, and genetic algorithms) have been shown to generate optimal portfolios under real-life conditions and require only a few minutes of computation using commercial libraries and compiled code.
While mathematics is fun, this is not a technical review paper. The key takeaway for policy and decision support purposes is recognizing that financial engineering algorithms are much more advanced than some people might think. They also are no longer just interesting academic artifacts, as current algorithms become far more accessible to real-life applications. For example, most stochastic optimization algorithms are implemented in standard mathematical libraries available to standard third-generation language compilers (e.g., C++, Java, and C#). Notably, these methods consider the range of uncertainties typically encountered in real life, including “noisy” forecasts and missing price data, and they remain surprisingly robust with real-life data.
In fact, given the current state of knowledge in software engineering and financial engineering, it’s ironic that many so-called “simplified” hedging approaches, thumb rules, and models in use today are anything but simple to explain. They are rarely subjected to rigorous validation, mostly unaudited, and often yield counterproductive effects. It’s even more remarkable how much money can be bet blindly on such heuristics based solely on the notion that hedging, like motherhood and apple pie, is good intrinsically.
In today’s market, a higher pedigree of risk-capital methods is long overdue. Maybe Sarbanes-Oxley financial reporting and audit requirements will drive the process. Maybe evolving regulatory policy, such as what could soon come out of California, will provide the requisite leadership. Aside from the positive effects of these external drivers, my own view is that electric distribution companies and LDCs themselves will raise the bar on hedging methods. After all, it is increasingly likely to be shareholder money on the line, if it isn’t already.
1. “Excessive Speculation in the Natural Gas Market,” Staff Report, United States Senate Permanent Subcommittee on Investigations, July 2007, p. 1.
2. Ibid., p. 49.
3. Administrative Law Judge (ALJ) Kim Malcolm, and President Michael R. Peevey, “Digest of Alternate Proposed Decision,” R.04-01-025 et al.: Re: Petitions to modify Decisions (D.) 05-10-043, D.05-10-015, D.04-01-047, D.02-06-023 and D.03-07-037 regarding gas hedging for the 2006-2007 winter season for Pacific Gas and Electric Company (PG&E), Southern California Gas Company (SoCalGas), and San Diego Gas & Electric Company (SDG&E), California Public Utilities Commission, July 18, 2006.