The designed economy

-Donal Curtin

Friday 14th November 2008

Market design means paying attention to economic details so that the grand schemes can tick along just fine.

Economists don't have much of a public reputation as successful designers: most people wouldn't expect them to be grease-under-the-fingernails, let's-see-what-I-can-rustle-up-out-in-the-shed kinds of guys.

And yet, every now and then, they turn their minds to the itsy bitsiest details of how to design a market to fix a problem. And when they do, the results can be spectacular.

Probably the market design idea that's had the biggest practical impact is the Vickrey auction. Named for William Vickrey, the American economist who described how it worked in an academic article in 1961, the auction features what looks at first like a distinctly odd design detail: the winner of the auction gets the item at the second highest bidder's price, plus a small increment.

This design tweak has big effects. For bidders, it deals to the "winner's curse": if you're bidding for (say) 3G spectrum licences, your biggest fear is that you will overestimate what they're worth and end up paying far more than anyone else, potentially putting you at a serious or even terminal competitive disadvantage.

In the Vickrey auction, however, if you bid $100 million when the next highest bid was $10 million, you get the licence but only pay $10 million plus a bit.

You might reckon that bidders overpaying would suit the sellers just fine, so why should sellers go the Vickrey route? The answer is that sellers also do better because the
protection from the "curse" encourages more bidders in the first place, improving sellers' prospects of getting a good price.

So it's not surprising that Vickrey-style auctions have swept the world, facilitating markets in everything from US Treasury bonds to the initial Google share issue. There's more detail (in plain English) in Mark Skousen's Econopower: How A New Generation of Economists is Transforming the World.

Much less well known, but becoming increasingly popular in markets where money doesn't change hands, is the "Gale Shapley deferred acceptance algorithm". Dating from 1962, it is designed to handle situations in which there are lots of applicants for something on one side of a market, and lots of providers of something on the other. A classic example is students applying to get into the university of their choice.

It's the detailed design of the algorithm that makes it work. Applicants go first, ranking their true preferences. Providers who get more applications than they can take on reject some, but don't accept anyone for certain at this point - this is the key "deferred" bit.

Applicants then have another go, and so the process rolls on. No provider commits to accept anyone, until there are no more rejected applicants who still want to go on asking. Only then do providers go final with their offers.

It doesn't sound much of an idea: just another way out of many, you might think, to match up people. But doing it this way has proved to be much better than the ways typically used before it.

The classic example is the New York public high school system. The early process was chaotic: 30,000 of the 90,000 students ended up assigned to a school that wasn't on their preferred list. With the new system, that number dropped to 3,000.

One reason the algorithm worked was because it dealt to the incentives on both schools and students to "work the system" - for example, not to show the real number of places you had (for the schools) or not to reveal your true preferences (for the students).

For those interested in how to allocate resources - even matching up kidney donors and recipients, for example - there's more information at Alvin Roth's website at Harvard, and in his handy article (not too heavy going) in March's Economic Journal.

As they've pulled on some overalls and got poking around under the hoods of markets, economists have discovered some general principles. It's important, to use Roth's terminology, that marketplaces provide thickness (plenty of buyers and sellers).

But it's also important they avoid congestion (buyers and sellers getting in the way of each other, in a desirably busy but efficient market). And they have to provide safety (good reasons to use the market and not rip it off, be ripped off, or bypass it).

If you're designing a new market - and we're going to see more and more of them, including some important ones relating to carbon emissions - those overarching grand design principles will take you some of the distance. But to make them really work, whether for organ transplants or the world's climate, it's the design micro-details that matter.

*A free disclosure statement is available on request

Comments from our readers

No comments yet