Columbia University’s William Vickery died only three days after the announcement that he had received the 1996 Nobel for economics. But his insights, particularly on the design of auctions, live on and have practical applications in business and government. They might be useful in states like my home state of Minnesota, where there has been controversy over the administration of Medicaid and other low-income health programs.
A revamp of those programs, incorporating more of the Canadian-born economist’s insights on the design of auctions to improve economic efficiency, could benefit us all.
Minnesota’s Department of Human Services put the administration of these programs, covering more than 300,000 residents, out for bids from HMOs and insurers. There is not one statewide contract, but rather several dozen covering individual counties or blocks of counties that have formed coalitions.
In the bidding for 2016, the HMO that had the largest share of business in the last two years suddenly was cut out of nearly all the action. It will have to lay off several hundred workers, and tens of thousands of state residents will have to go through the rigmarole of a change in insurer.
The HMO lost out for a simple reason: The prices it bid in the auction were much higher than other companies bid. This is little different than one paving contractor losing a county or state contract by submitting a higher bid than its competitors.
“Little different” is not “exactly the same.” Bidding by a very small number of near-statewide health insurers for a complex statewide service does differ from a dozen or more contractors vying for one 12-mile pavement job. These distinctions may cause inefficiencies that might be ameliorated by a more sophisticated auction design — one based on Vickery’s work.
Start by thinking about the basics of auctions. Auctions are essential to what economists call “price discovery.” Market prices can indicate the value of some good, such as used farm equipment or a mile of 3-inch asphalt overlay, to society.
Ditto for services like health care administration.
An economy uses resources most efficiently when the prices at which goods sell are closest to their true value to society. Think about a product like No. 2 yellow corn or 87 octane unleaded gas. There are many buyers, many sellers and good information all around. Forces of supply and demand in a free market ensure that the going market price meets all the conditions economists recognize as “perfect competition” — an accurate reflection of the value society places on the item sold.
That isn’t necessarily true when the number of buyers and sellers shrinks, information is very unclear, the product is not standardized or other “market failures” are present.
Absence of perfect competition is not fatal. There are only a few producers of men’s underwear, not thousands. And Hanes T-shirts may differ slightly from BVD’s. Similarly, Gold Medal flour is not King Arthur, and Macy’s is not Wal-Mart. But such products and vendors are so well known to most that markets for common consumer products are very efficient and the prices are at or near a societal optimum.
Highly dissimilar products require both sellers and buyers to put more time into price discovery. Think of the hours spent haggling in a Middle Eastern bazaar. Buyers and sellers vie to find the absolute price limit of the other. But this is efficient only if time is of little value in the equation.
Time is extremely valuable to farmers and small-business people. When a farm or commercial business is liquidated, there typically are hundreds of used tools, machinery or fixtures to be sold. Each is unique in the amount of wear and general condition. Selling each one by private agreement would be extremely expensive for both sides, so liquidations often involve auctions. This brings together several possible buyers, and the auction process takes only a few minutes to find out which one places the greatest value on any one item. Online auctions have further lowered costs, broadened information and extended this to more situations than ever were possible with physical auctions.
In an open-outcry auction, the bidding is iterative. Someone who wants a John Deere or a Picasso can follow the bidding up step-wise and bow out when her reservation price is passed. And in many cases, a bidder knows who she is bidding against and can judge other bidders’ likely behavior.
For government purchases, transparency to avoid the appearance of corruption makes an auction preferable. This works well with uniform items like diesel fuel for a DOT garage or copy paper for a public school system. It also works well for one-off projects like a new bridge for which detailed plans must be drawn in advance.
But as the product being procured becomes more variable or amorphous, auctions are more difficult. So many government bidding processes involve some sort of technical scoring system to consider nonmonetary factors. This is true for the health administration auction that this HMO lost in Minnesota, one which state human services administrators had developed over the past four years. It got high technical scores for many distinct localities and thus retained a few of them, but for most areas, the tech scores were not high enough to offset its substantially higher dollar prices.
Such a sealed single-bid auction works well for vendors bidding to supply homogeneous goods like fuel or paper. But these are less economically efficient when the auction is for a new product, one that is unique, user-specific, and that involves large values. The best option here was to design more complicated auctions, ones that included multiple rounds, and here’s where Vickery and his students come in.
The Federal Communications Commission sales of radio frequencies was just such a case. Typically, potential buyers submitted initial bids. The bids were opened and disclosed to all. Then each bidder could submit a revised bid, up or down. This is repeated several times.
Among other things, because it eliminated the danger of vastly outbidding everyone else and leaving a lot of money on the table, bidders faced less risk and had to be less cautious. The outcome of the frequency of auctions was more money for the Treasury. But it was also greater economic efficiency.
The existing Human Services bidding process was bad for one bidder. But it was also bad for the patients and counties forced to incur the transaction costs of shifting to a new provider. Moving to an iterative auction with two or three rounds of bidding might have reduced this external cost of the existing system.
Also, the HMO’s predicament sends a message to future bidders that the market is one with great risk of being left high and dry. Generating risk that need not exist is a source of wasted resources. And yet the system needs multiple bidders to prevent a monopolization of the market.
Mathematicians and economists specializing in game theory were hired both by the FCC and bidders to tease out all the ramifications of an optimal process. That need not be necessary for bidding design for administration of government health financing programs like Medicaid or the Tricare programs covering our military. But moving beyond a one-shot auction could have benefits for society as a whole as well as the direct players in this health care field.