Imagine you are in charge of purchasing software, coding services, or other equipment your firm needs. Maybe you are hiring vendors offshore. You set up an auction and you award a contract to one of many bidders, but you are unsure about the quality of the service or products you will receive.

“There is the fear of buying lemons,” said Professor Giuseppe “Pino” Lopomo of Duke University’s Fuqua School of Business, referring to a slang term for a low-quality item and the moniker of a famous economic theory. “You buy from the lowest bidders, and they might turn out to be low quality.”

It is the problem of adverse selection, Lopomo said, a decades-old concept that in auctions works like this: the bidder has private information about the quality and cost of their service or product; but that information is hidden from the buyer who is led to pick the cheapest bidder and then bear the consequences if the service or product happens to be poor.

“There is always some guy who needs to win the auction badly, and he's willing to do the job for nothing,” Lopomo said. “And then, after he wins, he might do a very bad job, leaving the buyer to foot the bill.”

In a new paper  “Optimal Procurement with Quality Concerns“ published by the American Economic Review’s journal, Lopomo and colleagues detail an optimal auction format called LoLA (lowball lottery auction) that they believe provides the best possible choice in situations where quality concerns are high.

The scholars added a minimum floor price in the LoLA system, below which bids aren’t accepted. Lopomo explained the minimum price levels the playing field for the higher-quality bidders, whose prices are higher because of the cost of quality. The researchers argue the minimum floor, high enough to reflect quality concerns, weeds out the cheap bidders, often the low-quality ones. The higher the quality concerns, Lopomo said, the higher the floor is set.

The researchers argue using LoLA, the buyer ends up paying a higher price than in a straight lowest-bid auction, but the higher quality eventually saves the buyer money.

Lopomo and colleagues tested their system with real-world data from the Italian government. The researchers analyzed data from public auctions in the 1990s, when Italian agencies solicited bids for highway work. The setting was ideal to test LoLA, Lopomo said, because quality concerns are high in Italian public auctions.

“You say you are going to do it in two weeks, and it takes six months,” Lopomo said. “Or after you won the bid, you start saying that the costs have gone up and ask for a renegotiation.”

By factoring in the official tally of delays and cost overruns recorded by the Italian authorities, the researchers found that the LoLA format would have saved the Italian government between 18 to 22% in costs, simply by setting a higher minimum floor for bids and awarding higher-quality bidders.

Lopomo said that up until now, the optimal auction system was considered the standard second price option, a format in which the suppliers submit a bid and the company picks the lowest, but charges the winner the second-lowest price. The system was discovered by Nobel Laureate Roger Meyerson and incentivizes bidders to bid the lowest possible price, because if they win they would only pay the second-lowest price, still making a profit.

Lopomo argues the second-price system maximizes the buyer’s value only if quality is not a concern, for example when auctioning for products that have been in the market for a long time like pens, or for common services like cleaning. The researchers argue when quality is not predictable, the second-price option leads to unexpected losses. “It induces a race to the bottom,” Lopomo said.

With their findings, the researchers created two free software applications that any business can use to set up the optimal auction. The application asks the user to measure their quality concerns.

“In the coders example, I want to know how much the buyer is willing to pay for a code with no bugs, how much for a code that has 20% bugs, 50% bugs, and so on,” Lopomo said. The application also asks the user for a prediction on how much the service would cost to the provider. With these two parameters, the software suggests the minimal low price for the auction, Lopomo said.

Lopomo said LoLA would be useful for anyone in the private industry in charge of procuring goods and services “where quality is unobservable.” The food industry could use this system too, he said.

“This year tomatoes are great, next year they are lousy,” Lopomo said. “If you’re willing to sell them for very little, that means you have no demand, because they are not good.”

He also said his system is particularly helpful for the market of coding services, where big tech companies often outsource.

 “They are mostly smart young professionals, but sometimes the code is a mess,” Lopomo said. “And you can’t sue them. So you're burned. In these cases, we suggest you use LoLA.”