Al Roth: Entrepreneurial Economist

“I’ve always been interested in using mathematics to make the world work better.”

That’s Al Roth from a profile in Forbes.  Roth has applied heavy-duty theory to the very practical problems of matching doctors to residency programs, children to schools, economists to departments and kidneys to patients in a way that is stable, incentive-compatible, and maximizes the gains from exchange.  In my view, Roth is the most influential economist working today. Influential among other economists?  Yes.  But what I really mean is influential in the world.

Roth’s homepage has many introductory papers as well as more technical material, here is his blog, Market Design and here is good video, Roth at Google


You don't think that the most influential economists are those macroeconomists who have made essential contributions to the ruin of the advanced economies?

Never was so much owed by so many because of so few.

There's that famous quote... "the best reason to become an economist is to protect yourself from other economists," but I think Roth is a great example of someone actually out there to make things better, and if not better, at least more interesting.

Thanks for the links Alex.

Even if we exclude the leading macro economists, Hal Varian's work with Google surely influences vastly more lives.

Brian, you probably don't understand the problem.

It's like at the end of high school, when everyone is scrambling to go to college. The doctors are making side-deals with a bunch of different hospitals, and everyone is misrepresenting themselves because doctors want to keep their options open as long as possible. This leads to chaos, scrambling, and inefficient matches.

Look at the economic academic job market: A lot of great people don't get jobs because there's only really one round of serious interviews.

The Gale-Shapley algorithm make it at least a dominant strategy for one side to be honest, avoiding much of the chaos and trying to game the system.


You can't "just let the parents decide" about school choice because, well, there's a finite (and small) number of seats in each school. Parents rank the schools they want their children to attend, and the system tries to do the best it can to give people the matches they want. But if you try to do this naively (by, say, trying to maximize the number of people given their first choice school), you end up creating large incentive compatibility problems, forcing parents to "game the system" or face serious disadvantages. Roth's systems are much more just than the systems they replace, because they don't give advantages to parents who understand how to manipulate their choices.

As for the medical match--read the Economist as Engineer paper and tell me that the market was functioning at all well beforehand, when hospitals had to rely on exploding offers and couldn't risk making offers to their top candidates, for fear of losing out entirely. Or read the paper on what happened in the gastroenterology market when a shock caused their central match to collapse, and suddenly a lot more matches were done with local hospitals, on the basis of who you know.

Bottom line: Roth's matching systems are a huge improvement in efficiency AND justice.

You say, "no matter how brilliantly one person (Roth) makes a decision that affects many people, we often find that letting those individuals decide is superior on net." You seem to be missing two key points: (1) Roth's systems all take the individual agents' own rankings as their key inputs, so they're about coordinating those individuals' decisions, not supplanting them; and (2) the markets Roth works on are ones where the decentralized approach has already fallen splat.

Comments for this post are closed