Yesterday the General Accounting Office released a report on the causes of the wholesale electricity price increases in California August-October 2000. U.S. Representatives Inslee and DeFazio requested the study, which had two objectives: determining whether or not wholesale sellers actually exercised market power to raise price above marginal cost, and determining if the design of California’s market enabled those suppliers to exercise market power. In other words, did the state’s rules create the opportunities to raise prices, and did the sellers seize on the opportunities presented to them through the bad set of rules? The GAO study answers “yes” to both questions. One of the most valuable aspects of this report is that it reiterates something that most analysts have been saying all along, but has not been reflected in California’s response to the situation — if the sellers did exercise market power, one reason they could was that the set of dysfunctional rules in California created the opportunity for them to amass it and to exercise it.
The GAO compared the hourly pattern of prices in the August-October 2000 period with the same period in 1998, a period that exhibited price patterns consistent with competitive behavior. Through econometric analysis they find that the hourly pattern of price movements in 2000 differed from the 1998 pattern, a result consistent with supplier exercise of market power. The report acknowledges, though, that they did not completely control for the other acknowledged causes of the price pattern in 2000: higher-than-expected natural gas prices, increases in emissions credit prices, the increasingly narrow gap between supply capacity and inelastic demand, the retail price cap that gave no incentives to consumers to shift load away from peak hours, the drought in the Pacific Northwest, and temperature. Without performing an analysis that more fully controls for these variables, this current GAO analysis is not really additionally persuasive beyond the claims of market power put forth in other studies. That said, though, this study is well-done, and the econometric analyses they performed is consistent with the other analyses of the California energy crisis, some of which are cited in the report. (A pedantic presentation criticism: they report coefficient estimates and standard errors, but not the actual marginal effects of the changes in variables, so it’s hard to assess the magnitudes of the effects of the variables and compare them to see which had the biggest effect on price).
I found the most interesting result in their analysis to be the effects of the various price caps that the CAISO implemented that affected the period in question — $750 from September 1999 to July 1, 2000, $500 from July 1, 2000 to August 7, 2000, and $250 from August 7, 2000 to the end of the period covered in the analysis. They found that the $500 price cap in July and early August, as well as the $250 price cap from August 2000, actually raised prices. While some may find this result surprising, it’s actually quite consistent with economic theory, and with experimental results. They also found that the $750 price cap raised prices, but with less statistical significance (at 10%). Again, though, we have to take these results with a grain of salt, because once you control for the factors mentioned above, this result might not really exist.
In fact, a Northwestern student whose senior honors thesis I supervised (Terri Kandalepas, now a law student at UCLA) performed a very similar analysis in June 2001. In her analysis, the combination of lagged electricity price in that hour, natural gas prices, emissions credit prices, Stage 1 and Stage 2 alerts, reservoir levels, temperature, and hour of the day accounted for 85 percent of the changes in price. That result means that at most the exercise of market power accounted for 15 percent of the price changes. Given the types of trades we’re analyzing here that is still a serious amount of money. Her analysis did not incorporate the net imports and the price caps as the GAO report did, so they are not directly comparable, but it does demonstrate that controlling for these other effects is crucial to performing a thorough analysis that allows us to infer the actual extent of market power.
One crucial omission from their otherwise very thorough analysis of the evolution of the California energy crisis was the importance of the strategic behavior of the utilities. Among those of us who eat, sleep and breathe electricity the phenomenon of buyer bid underscheduling into power pools is well known — in a bifurcated structure like California’s Power Exchange for day-ahead and CAISO for real-time balancing, the utilities buying wholesale power have incentives to understate their day-ahead demand, with a goal of lowering wholesale prices. Then they end up buying more in the real-time balancing market, but unless supply is tight they have overall paid less for their power. Interact that with the well-worn arguments for why suppliers have incentives to withold from day-ahead, and you see how so much of the spot market activity in California got shifted into the ISO’s real-time balancing market, and at astronomical prices. Plus, the ISO’s real-time market priced at what the market would bear, so it was a convenient way to circumvent the price cap. I would like to have seen a discussion of this strategic dynamic in the GAO report.
A good, and very technical, analysis of the exercise of market power in California is by Steve Puller in the Economics Department at Texas A&M University. And I would be remiss if I don’t engage in some shameless self promotion by mentioning the qualitative analysis of California’s failed electricity policy that Adrian Moore and I wrote in January 2001. The GAO report contains a bibilography containing other analyses of market power.