Even the volatility is volatile. And that can play havoc with hedging.

Jeff Skilling resigned from Enron over a year ago-after power prices in markets serving California had fallen 90 percent in three months.

But in July, Bank of America won approval from the Treasury Department to offer cash-settled electricity derivatives-with a former Enron regional director at the head of the desk.

So what has changed, and what hasn't?

Well, power prices, for one. Some may believe they are quieter now, than when California imploded. But more predictable? Hardly.

If anything, power prices are becoming even less predictable, as if they ever were. To explain, let's consider an overly simplified fictitious example that shows where we've been and where we're going.

• "Wild" Prices. Suppose for a few years that power quadruples in price every summer, and then each winter falls back to 25 percent of that peak. Is that volatile? No, it is not. Yes, the price seems "wild"-moving at times by a factor of four-but the pattern is there for all to see. That's not volatility.
• "Mild" Prices. Now suppose instead that the pattern changes after a few years so that power prices only double in summer. But some summers the price doesn't change at all, or even drops a little. And some winters see a steady or even slightly rising price, but you never know when. Power prices seem more "quiet" in this era, but there's less of a pattern. Now that's volatility.
• Reality. Now consider a third reality-what the numbers are telling us today. Today it seems that even the relative degree of unpredictability is changing. Prices seem to be moving back and forth in random fashion between periods of greater relative predictability, and then less relative predictability. It's not so much the size of the price change on the wildest trading day, but more the fact that patterns are extremely difficult to determine. And just when you think you've found one, it vanishes, and in comes a new regime, equally unfathomable in different ways.

In short, we always knew that electricity was the most volatile of traded commodities, but what we've got now is something more-

What does that mean for risk management?

In this article we take stock of price behavior in major wholesale power markets over the last several years, and examine how the markets behaved and whether or how they might be changing. The major markets in terms of volume of trade are "Into Cinergy," which is the volume market leader by a wide margin, and PJM, in second place. Following those two markets, in no particular order, are "Into Entergy," Mid Columbia, Palo Verde, and "Into ComEd."

What we find are two distinct phenomena that are nevertheless difficult to reconcile. First, electricity markets in different geographic regional areas appear to be converging in terms of price, indicating that they are becoming increasingly well-connected and well-integrated. This fact suggests that markets are continuing to evolve and grow in ways consistent with the FERC goal of developing large, regional, increasingly interconnected regional transmission organizations (RTOs). (Note: There does appear to be a pronounced and lingering premium in PJM, judging from the relatively large basis that persists between PJM and Cinergy, ComEd, and Entergy.)

Second, volatility appears to have become more variable over time, as measured across discrete time periods, according to a study of price behavior for Into Cinergy and Into ComEd.

All of this makes hedging more difficult. It may not be enough simply to invest in some forward and long-term contracts. Purchase the contract on the wrong day, and the advantage could be lost-just as a private investor can lose much of the potential of a future bull market if he should buy his entire stake on the wrong day.

One way that companies have avoided this danger in power markets is to use some sort of dollar-cost averaging approach for completing forward deals. The company randomly chooses days over an extended period of time for closing deals for forward months.

In fact, it is clear from an examination of past price history that large increases in price risk may last for only a few days. Therefore, a company may want a forward contract with a term of only a week or weekend during a month, and not for an entire month. The shorter-term standard contracts could, of course, be combined into monthly or several-month contracts.

Nevertheless, at the end of the day, it is not always possible to figure out exactly what has happened.

Many thought that Enron and other major marketing companies regularly were able to set price levels and elevate price volatility, judging from what happened in California at the worst of the crisis. And that's why regulators continue to focus on major weaknesses in the wholesale part of the business.

Yet, wholesale power price levels and volatility in the first six months of 2002 were very similar to what they were in the first six months of 1999 when Enron et al. were active and growing, and the sky was blue. It was only when conditions were extreme-when regulators lacked the means or the resolve to intervene-that these companies seemed able to control price by restraining supplies and creating real or artificial constraints on systems. By contrast, during much of the time (when conditions appeared normal), these traders may have served a very useful role.

But enough of such subjective assumptions. Let's turn to the numbers, to read what they say.

#1-Power Price Levels Vary Greatly

As shown in Figure 1, the extraordinarily high wholesale prices seen from June 2000 to June 2001 in western power markets (such as Palo Verde in Arizona, near the southern California border, and Mid-Columbia in Washington, north of California), tended to dominate and mask price behavior in other markets.

The prices in Figure 1 show 22 day-moving medians (there are generally 22 trading days in a month). They indicate that near the end of December 2001, prices on half the trading days during the previous month were greater than $475.00/MWh at Palo Verde. A host of factors in California contributed to these high prices-a power market design ill-equipped to handle live traders trying to make money, a shortage of cheap sources for generating power, incompatible legislative compromises, and the California habit of wanting to have its cake and eat it too.¹

Things were much different in summer 2002 in western markets, but still extreme. Wholesale power prices at Mid-Columbia were as low as $0.25/MWh in July 2002 or 0.1 percent of the $250/MWh they had often been in July 2001. That offers evidence that inexpensive hydropower was in excess supply, not in shortage.

By early 2002 there was really no difference in the price in these two western markets. Then, during the spring and early summer of 2002, prices began to move in different directions. By July 2002, the difference in price had grown to more than $30.00/MWh, a significant amount, especially for markets that were considered at one time commercially connectable.

Interestingly enough, when we treat markets near California as the outliers they are and delete them from the picture, it is clear that most major markets were often very similar in terms of their price behavior during the period ( 2).

The exception is the Into Entergy market in summer 2000, where constrained systems could have set the stage for some exercise of market power.

#2-Price Behavior Grows Similar in Several Markets

The behavior of daily price levels over time has proved generally very similar for Into ComEd, Into Cinergy, and PJM since 1999, and for Into Entergy since the fall of 2000. The similarity of the price levels indicates that these markets are well connected much of the time. This similarity is most striking for Into Cinergy and Into Com Ed. The correlation between price in these two markets has been near perfect in the last several years. In 1999 the correlation was equal to 0.96. Since the turn of the new century the correlation has been near 0.99 from one month to the next, only 0.01 from a perfect correlation.

Thus, it is not surprising that in early May 2002 Into ComEd announced it planned to join PJM, noting its major trading partners and suppliers were to the East. On July 1, 2002 Into ComEd also announced it was withdrawing from the Mid-America Interconnected Network (MAIN) in December 2003 (18 months' notice is required to exit regional reliability councils). Interestingly enough, despite recent problems in the marketing and merchant generation parts of the power business, markets are continuing to evolve and grow and in ways consistent with the FERC goal of developing large, increasingly interconnected RTOs.

Since August 2002, all major eastern markets have been better connected when compared to earlier periods. Nonetheless, PJM prices were often at a significant premium over price in other markets. This observation can be seen by noting the difference in price in this market relative to the other markets. ()

#3-How Does the Power Price Compare with Gas?

As systems become even better connected, bases may get smaller, and large values for the basis will persist for shorter time periods. Nonetheless, significant profit opportunities will continue to emerge, because of the huge shifts in power demand that are possible in the short term, capacity constraints that can almost completely bind movements of power, and other factors that make wholesale power in the United States a most volatile price commodity.

An interesting development in the last year is the similarity of power and natural gas prices at Henry Hub natural gas market and its nearby power market Into Entergy. This is due not only to a continuing integration of these markets, but also to a downturn in the economy and to the availability of more marginal power generation capacity.

A glance at Figure 2 reveals that power prices may be well connected in eastern power markets. Yet, natural gas prices in these same markets, which are often highly correlated with the Henry Hub price, move much differently than power prices at times. Differences in natural gas and power markets were greatest in summer 1999, when constraints in moving power emerged, temperatures elevated significantly, and the ability to exert market power improved. Power prices increased significantly, yet natural gas prices hardly budged.

At other times, power and natural gas prices move in opposite directions, as they did in the fall of 2000 when high natural gas prices were sustained by tight supplies and power prices were pushed lower by weak demand. Yet when demand for power began to increase and natural gas supplies continued to be tight, both prices moved together. Then strong demand was followed by weak demand in both markets and both prices plummeted. By early spring of 2001, power and gas prices were again seeking an equilibrium. As is often the case, the search for equilibrium was disrupted by large and variable changes in demand for power during the summer.

Power and natural gas markets in the East are today better integrated then they were in the past. But power price levels in any one power market are still extraordinarily variable when compared to other commodity markets. This variability will continue to be a blessing to companies with astute traders and/or operationally flexible assets. But such movements in price levels will wind up being a bane to others at times, even those that consider themselves expert traders or risk managers.²

#4-Volatility (Price Risk) Is Moving Over Time

Figure 2 shows significant differences in the size and timing of shifts in price level, according to whether you examine the data over months, seasons, or years.

Power prices increase in the summer, but much differently between years. In the late winter and early spring power prices decline, but for how long varies greatly. If changes in the price level could be well explained by daily deviations from normal temperature, humidity, generation outages, natural gas prices, and other factors that influence power demand and supply, then great progress could be made in determining price volatility, pricing, and selling a variety of price risk insurance models. However, models used to explain such behavior are most often very fragile and very imprecise. Naturally, this fact also helps explain why many merchant generators and wholesale power companies were working very hard to set prices.

Yet regulators are now for the first time fully accepting the fact that markets for natural gas and power have unique and extreme features at times, and that some regular surveillance of these markets is necessary. Some even understand that price risk insurance is a good thing, that price risk management is necessary and not just a deadweight cost, and that a variety of large additional costs may follow from price risk that is left totally uncontrolled.

Perhaps the easiest way for anyone to get a view of price risk is to examine a graph of percentage changes in price, such as daily stock returns or a related figure-the natural logarithm of price today to price yesterday. Such a plot for power prices for Into Cinergy is presented in Figure 3. The relative frequency and size of large price changes are revealed in such figures. A change in price of 100 percent in either direction in a day is not unheard of. Moreover, the average size of the change varies greatly over time. Since volatility is an estimate of the average variability of these price changes, the computation of price risk for power is always difficult, especially for a future time period.³

That appears even clearer in Figure 4, which is based on a GARCH (generalized autoregressive conditional heteroscedasticity) estimation. Figure 4 shows that price risk often varies and comes in clusters.⁴ Periods of high price risk follow periods of much lower risk. Thus, companies that are not hedged could be hit by a price tsunami. Moreover, companies may either pay a high premium or receive a huge discount for price risk insurance if it is based on dated estimates of volatility.

There is also seasonality in price risk. Not surprisingly, price risk increases in the summer. However, the degree and timing of the increase in price risk also varies much across summers.

Power price risk can also increase in the winter, as in heating season 2000/2001. Yet, this increase may be attributable only to the exceptionally high and sustained natural gas prices in that heating season. It may also be a first sign that price volatility will increasingly propagate across natural gas markets to power markets, as these markets become increasingly interconnected and as natural gas power generation continues to grow.

Interestingly enough, power price risk was relatively moderate for power markets between September 2001 and early May 2002. ( 4) But, power price volatility did pick up in summer of 2002.

Naturally, an increasing number of utility companies are paying explicit attention to price risk. Keeping tab of the behavior of price variability over time is an activity that will most likely grow in importance for utility companies, whose revenues or costs, or whose customers' bills, are exposed to such large and varying amounts of price risk.

Some Things Stay the Same, Others Change

Although power price risk is still very high and spark spreads are still significant, they are just not the huge fires they were in past summers. Other aspects of the wholesale power business have changed significantly.

• First, trading volume has dropped. According to , the total volume of trade on the daily wholesale market on July 16, 2002 in the major eastern markets considered here were 272,000 MWh over the 16 peak hours. On the same date in 2001, the volume of trade was more than twice as great. Similar results are obtained for other dates. The decline in volume of trading is consistent with a decline in the number of players and reduced risk in the market. It may also be a consequence of a greater effort at quality control by major price reporting services because of pressure from regulators and the industry.⁵
• Yet trading expands in Texas. Again, according to , Texas now has more distinguishable wholesale power markets with a combined volume of trade of almost 100,000 MWh over the 16 peak demand hours on July 16, 2002-more than 4 times as much as the previous year at the same time. Similar results are also obtained for other dates. This growth is explained by the recent restructuring of electricity markets in Texas.
• Futures markets fail badly. Regulated futures contracts for power, which were greeted with much fanfare at their inception, were finally delisted this year, because of a lack of trading. The failure of these markets may be traceable to the futures contract for power being too much like the very successful natural gas futures contract, since the two businesses are very different. This fact occurred because of the great success of the natural gas futures contract market. Yet the gas business has storage, which can put a damper on price increases for one or several days. Unexpected congestion over several days on a major transmission line is also more common for power than for natural gas. Outages of major power plants are also more frequent than the shutdown of a large number of wells. Such differences would seem to argue for standard contracts with different features to include a shorter term for power.
• Banks emerge as traders. Not only are there new markets, but there are also new entrants to the wholesale part of the industry interested in building business in derivatives as a risk management service and in trading. As discussed earlier, Bank of America has hired a former Enron regional director to head up its new trading desk in electricity derivatives. Earlier in the year, UBS took over Enron's trading operations, and Deutsche Bank expanded its energy desk and opened a new office in Calgary, hiring former Enron managers to build a derivatives business. These companies, of course, have excellent credit ratings and broad knowledge of financial markets, unlike the many wholesale power arms of energy companies that are greatly reducing the size of their wholesale and risk management operations.
• Distributed generation serves to manage risk. It also appears as if companies and major consultants to the industry are looking increasingly and more critically at the flexibility afforded by technologies such as distributed generation and by payment methods such as two-part real-time pricing tariffs as responses to significance price risk. Increasingly, an understanding of price risk is being used to understand the value of such technologies and instruments. Distributed generation allows for better timing to bring generation online. Two-part, real-time pricing tariffs can provide price insurance and price instability which, nonetheless, will stabilize bills over time.⁶
• And credit is king. The future structure, organization, and regulation of the power and gas business will be different from what it was in the past. And a new roster of major players will also appear. Yet significant price risk is most likely to continue in the wholesale power business, if only because of the enormous additional capital expenditure that would be required in transmission and other equipment to reduce it.

These expenditures would entail a phenomenal increase in debt, and despite low interest rates, few companies would be interested in doing this. The emphasis in the industry now is on acquiring a better credit rating and on reducing debt, not increasing it.

---

1. Economists at major institutions in California have now come full circle from believing in an ability to create a power market design that would be immune to the special circumstances and conditions in the power business. Now it is understood that such factors as highly inelastic demand and supply in the short run, frequent and significant capacity constraints, extremely expensive storage, and opportunity costs that can exceed production costs because of trading opportunities across markets can set the stage for a regular assertion of market power. One interesting study found that not only do these factors matter but that 59 percent of the $6.94 billion increase in electricity expenditures in California restructured wholesale market between summer 1999 and summer 2000 is attributable to increased market power. Severin Borenstein, James Bushnell and Frank Wolak, "Measuring Market Inefficiencies in California's Restructured Wholesale Electricity Market," University of California Energy Institute, Berkeley, California, www.ucei.org, June 2002.
2. Enron's chief trader in 2001 booked an enormous $750 million in profits by trading natural gas and by correctly predicting price movements and taking positions accordingly. This trader was probably the most successful trader in 2001 in any commodity market. Yet, this same trader lost 27 million in 2000. See , Tuesday, July 9, 2002, pages C1 and C6.
3. Models are used widely to determine the value of physical and financial assets such as options to put a cap on prices. Yet, ordinary use of these models requires that average daily percentage changes in price be relatively constant over time, or if not relatively constant, able to be reliably estimated. The goal is to obtain measures of price volatility that represent price risk around any level. Thus movements in price levels should not influence price volatility. Ordinary applications of these models also may require that changes in price are normally distributed and independent from one day to the next, just like flips of a coin would be independent from one flip to the next. These assumptions are often not satisfied with power prices. This makes pricing options difficult. This is also why the price of an option may often be much greater than the value calculated from a standard model. It is also why option and other price insurance markets are often thin and why markets that join counter-parties with different price exposures (generators vs. large buyers of the commodity) are successful.
4. For many examples of different types of behavior for price and other related economic variables and how to represent them, see Anindya Banerjee, Juan Dolado, John W. Galbraith, and David F. Hendry, , New York, Oxford University Press, 1993.
5. See Notice: Platt's proposes methodology refinements. , July 1, 2002. This notice was also printed in its July 7 issue.
6. For a full discussion of the current state of distributed generation and some indication of how price risk frameworks can aid in determining the value of these assets and instruments, see Jonathan A. Lesser and Charles D. Feinstein, Distributed Generation: Hype vs. Hope. , June 1, 2002, pp. 20-28; Ahmad Faruqui and Melanie Maulden, The Barriers to Real Time: Separating Fact From Fiction, , July 15, 2002, pp. 30-40.

Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.