SYNOPSIS: Krugman explains Stanford economist Frank Wolak's model of the California electricity crisis
We're approaching the first anniversary of the sudden, unexpected end of California's energy crisis. I went way out on a limb, at least by journalistic standards, by saying that market manipulation was a key feature of that crisis. I have since been vindicated: arguments that people called leftist nonsense a year ago are now conventional wisdom.
But of course I wasn't a brilliant investigative reporter; I just knew enough to talk to the right people, and to understand what they were saying. Paul Joskow and Severin Borenstein were very helpful. But my most helpful source of all was Frank Wolak, the Stanford professor who also heads the CAISO market surveillance committee. (CAISO is the "system operator").
In a recent paper (which doesn't seem to be on his web site yet) Wolak offers a very nifty model to explain what was going on. However, as they say in the journalistic trade, he buries his lede: the model is in passing, amid a dense discussion of institutions and their reform. So I thought I would lay it out here, to give you an idea of how I think about the whole thing.
Wolak's model starts with a simplified demand curve. We assume that the demand for electricity is totally inelastic at some given quantity - say 900 megawatt-hours - until the price reaches a ceiling, say $1000 per mwh. It doesn't matter for current purposes whether that's a legal ceiling or the price at which utilities simply refuse to buy.
On the supply side, we assume that there are a smallish number of generators, each with limited capacity - let's say 5 generators with a capacity of 200 mwh each. Each generator has a marginal cost of, say, $20 per mwh actually produced.
Wolak assumes that in the market, each generator submits a bid price for its capacity; then the system operator takes the bids in increasing order of price, but pays all producers the highest bid actually taken. This is a stylized version of the PX, or day-ahead, market that actually operated. He also assumes implicitly that the bids are submitted in order - that the generators go one by one, each knowing what the previous bids were. (It's possible to do this with simultaneous bids; in that case it's a mixed-strategy equilibrium, with qualitatively similar results.)
So what's the Nash equilibrium of this game, given total capacity of 1000 and demand of 900? The first four generators submit bids at $20, their marginal cost; the last generator bids $1000, the maximum. It knows that it will sell only 100 mwh, half its capacity - but far better to sell 100 units at $1000 than 200 at $20!
The really striking thing, of course, is that there is excess capacity in the system - yet the price goes sky-high. And with a little realistic friction added, you could easily imagine blackouts and brownouts as part of the picture. Let me also stress that this is a non-cooperative equilibrium - it doesn't involve collusion, let alone conspiracy, among the generators. You don't have to imagine Ken Lay and Dick Cheney sitting in a room, trading sneers, and chortling over the havoc they are wreaking (which isn't to say that this might not have happened!). All it takes is individual firms, acting in their individual self-interest.
The resemblance of this story to the actual crisis in California, with a record number of plants closed for "repair", with shortages and blackouts continuing through the low-demand winter months, is obvious. Yet the whole exercise may seem suspiciously quick. If it's so easy to have a crisis in which market manipulation produces very high prices, why doesn't it happen all the time? And why did the crisis suddenly end?
But that's the beauty of Wolak's model: the price-rigging equilibrium only happens if the numbers are right, and so it can collapse if the numbers change. In fact, Wolak offers a clear story both about why California plunged into crisis, and why it plunged back out again.
First, it is or should be obvious that the high-price equilibrium only happens if there isn't too much excess capacity. Specifically, the excess capacity must be less than the capacity of a single producer. If demand were 800 instead of 900, the last bidder would not be able to drive up the price. Even if demand were 900, there would be no price-rigging if there were 10 producers, each with only 100 units of capacity.
Wolak's explanation of the onset of the crisis, then, is that the necessary margin of excess capacity was lost - partly because of demand growth, partly because a drought cut off the supplies of hydro that California normally counts on.
Why, then, did the crisis end? Three reasons. First, the gap between capacity and demand increased - a few new plants came on line, but more important, consumers engaged in a lot of conservation.
Second, price controls. When price-rigging is the problem, price controls actually increase supply - because they limit the rewards to producers from keeping capacity off line. (This doesn't come out clearly in the sequential-bid story. But it does emerge in the simultaneous-bid, mixed-strategy version of the model.) Sorry, those of you who think that New York rent control is the model for all price controls: Econ 101 is right most of the time, but sometimes you have to be willing to think beyond it.
Finally, when Gray Davis signed up a lot of power under long-term contracts, he fundamentally changed the game. Go back to the original example, but now assume that each generator has already locked in 100 mwh under long-term contracts. In that case, the last bidder has no incentive to bid high - if he does, he will make no sales beyond his contracts.
The stunning thing to me about this story is how utterly wrongly the situation was - and to a large extent, still is - reported. To me it was obvious by early 2001 that you couldn't make sense of the California crisis without talking about market power, and that both the "get shoveling" advice and the glib declaration that it was the fault of the politicians were nonsense. Yet the media, with a few exceptions - well, read Joe Conason on all that.
Ideological blindness? Intellectual laziness? Media whoredom? Yes.
Originally published, 5.27.02