To the Editor:

In a letter to the Oct. 1, 2003, , a letter from Lewis Evans and Kevin Counsell claims that a "pay-as-bid" day-ahead market would produce prices comparable to real-time prices even if loads understated their demand day-ahead, eliminating the incentive to underschedule (as allegedly happened in California).

In fact, a uniform-price day-ahead auction should produce the same effect. Evans and Counsell state that in a day-ahead market bidders will attempt to "guess the market-clearing price" and bid close to it. On the other hand, auction theory predicts that in a second-price auction suppliers will bid their costs. The important point is the correct definition of cost.

Suppose that a generator has an operating cost of $20/MWh, and anticipates the real-time price will be $40/MWh. In a second-price real-time market that generator will bid $20/MWh. But for the purposes of bidding into the day-ahead market, which will reliably be followed by an opportunity to sell into the real-time market, that generator's cost is the $40/MWh in real-time revenue it forgos by placing power into the day-ahead auction instead-its opportunity cost. (In reality, the bid would be modified to account for the uncertainty in the prediction of real-time prices.)

This means that bidding behavior in a day-ahead second-price auction should be the same as in a pay-as-bid day-ahead auction, as long as it is not the last opportunity to transact. So why then should purchasers still be able to play a demand understatement game, as they allegedly did in the California ISO markets?