Power Distribution Tenders and the Winner’s Curse

A few weeks ago, the Chilean government reported the tender results on contracts for regulated customers which will supply energy to distributors between 2021 and 2040.¹ The government auctioned 13,000 GWh/year, which, according to the National Energy Commission (NEC), is equivalent to over 30% of the energy that will be consumed by regulated customers in 2021. Bids were submitted for a total of about 80,000 GWh/year, and the bid prices ranged from USD29 to USD90/MWh, with an average of USD50/MWh. These prices are considerably lower than contracts awarded in past tenders, which in some cases exceeded USD120/MWh.

The main winners were Mainstream, which bid USD41/MWh, on average, and was awarded just under 3,400 GWh/year; and Endesa, which bid a little under USD51/MWh, on average, and was awarded almost 6,000 GWh/year. Mainstream backed its contract with the promise to build new wind farms, whereas Endesa will use energy that can be generated from existing plants. The rest of the energy (a little more than 3,400 GWh/year) was awarded to a number of companies, mostly wind generators, at prices a little higher than the bid from Mainstream—USD 48/MWh, on average. Almost all of this power will come from wind farms that have yet to be built.

Mainstream and the other wind generators that were awarded blocks all placed bids for under USD 55/MWh. This deserves a closer look. If it costs the same to build a wind farm in Chile as in the United States, the cost in Chile should be USD 100/MWh.² Why did Mainstream and the other wind generators bid so low? In this Brief, I argue that they may have been victims of the winner’s curse.

The phenomenon which is known as the winner’s curse was described in 1971 by three engineers, E.C. Capen, R.B. Clapp, and W. M. Campbell, to explain why oil companies that were awarded oil wells in competitive auctions tended to have meager results or even losses.³ Capen, Clapp, and Campbell observed that the systematic losses reflected the fact that even if each company conducts an independent and unbiased estimate of the amount of oil in each well being put to tender, the competitive auction will choose the bid from the company that overestimates the quantity of oil by the greatest magnitude.

Assume that a well containing 100 barrels of oil is being auctioned and that each bidder’s estimates are subject to uncertainty and variance, although they are independent and unbiased. Thus, one bidder estimates that the well contains 120 barrels; the second, 80 barrels; the third, 160 barrels; and the fourth, just 40 barrels. Although the average estimate is around the actual amount in the well, the contract will tend to be awarded to the bidder who overestimated the amount of oil by the largest magnitude—in this case, the bidder who estimated 160 barrels. If the companies do not adjust their bids downward, to take into account the fact that they will only win if they are overly optimistic, then the result will be that the winners lose—hence, the so-called winner’s curse.

At least three facts suggest that Mainstream and the other wind generators that were awarded power blocks were too optimistic. First, numerous companies participated in the auction, and the bids submitted by the wind generators that were awarded contracts were substantially lower than competing bids. Figure 1 graphs the bids entered by the wind generators that were awarded contracts (solid black line) together with the bids received from other wind generators (dotted black line). Most of the other wind generators bid a price of around USD 60/MWh, and some are even higher. In contrast, the companies that were awarded contracts bid less than USD55/MWh, and Mainstream bid as low as USD41/MWh, on average. Critically, as the figure shows, the wind generators that were awarded power blocks submitted significantly lower bids than Endesa, Gener, and Colbún, all of which were competing to sell energy from power plants that are already built and operating.

Second, as shown in table 1, Mainstream estimated a capacity factor of 31 to 49% for its projects. This rate is questionable, however, given that the capacity factor of wind farms built in Chile since 2013 is 26%, as mentioned above. ⁴ Moreover, in the rest of the world, only offshore wind projects achieve a capacity factor of about 40%.⁵

Finally, based on the average capacity factor in Chile (26%), the prices bid by Mainstream and the other wind generators could only cover their costs if the projects cost significantly less than existing projects in the United States and if the cost of capital was very low. Table 2 shows the levelized cost of different combinations of capital costs (rows) and investment costs per MW of capacity (columns).⁶ The shaded area highlights the combinations of capital costs and costs per MW of installed capacity that put the levelized cost within the bid range. To meet the bid price, Mainstream would need capital costs of under 5% and investment costs of less than USD800/KWh, which is around 50% lower than recorded costs in the United States. Even Mainstream estimates its investment costs at around USD 1,675/kW. According to table 2, using the usual capacity factors in Chile, the levelized cost would range between USD 65/MWh and USD 120/MWh, which is substantially higher than the bid submitted in the tender.

Some argue that the wind generators who were awarded contracts were betting on technological progress—that the cost of wind farms will fall. Is that optimism justified?

Figure 2 shows the weighted average cost per MW of wind farms built in the United States between 1998 and 2014. Costs fell substantially from 2008 to 2014—but those 2014 costs were not very different from project costs in the first half of the last decade. Costs rose from 2004 to 2008 before declining thereafter, which is surely due, at least in part, to the commodity price cycle, which peaked just before the 2008 financial crisis. Regardless, there is no evidence of a sustained downward trend in costs for wind projects that would justify the belief that costs will be cut in half by 2018 or 2019.

Other analysts suggest that the wind generators which were awarded contracts could be speculating on an excess supply of wind turbines, due to the wind and solar generation crisis in China. Because a large share of the recently installed wind capacity in China cannot operate due to transmission restrictions, Chinese authorities have cut subsidies. Industry analysts speculate that this could result in an excess supply of turbines in the world market. Another argument is that the new projects could receive very low financing rates thanks to the monetary policies currently being pursued by many central banks. Table 2 shows just how far investment costs and capital costs would have to fall in order for the promised projects to earn a profit at the bid prices. Moné et. al (2015, p.vi) estimate that non-turbine investment costs are around USD 500/kW. This implies that the price of turbines would have to fall to around USD 500/kW and capital costs to around 1% in order to reach the range of bids submitted by wind generators in the tender.

Some time ago, so the story goes, a car dealer decided to auction used cars to the highest bidder. Over the course of the week, bids were received in sealed envelopes. On Sunday afternoon, the envelopes were opened, and the cars were awarded to the highest bidders.

For a couple of weeks, the dealer thought he had hit the nail on the head, because the used cars sold at much higher prices. But before long, the buyers started coming back to protest, disillusioned (and sometimes furious) that they had paid “too much” for a “bad” car. Facing lawsuits, the owners of the dealership discontinued the auctions. They had learned that when the winner of an auction is the victim of a curse, awarding the contract is not the end of the sale, but the beginning of trouble.