Economists Saving Lives (and doing other awesome stuff)


This term I am cross-registering at Harvard to take Market Design, taught by Professors Alvin Roth and Peter Coles. The class is offered jointly by the Harvard Economics Department and the Harvard Business School. I have interests in computer science and economics, and I actually first heard about market/mechanism design from the computer science literature. However, the research done in these two fields are quite different. Generally, economists are concerned about efficiency, equilibrium stability, and incentives, while computer scientists try to optimize the algorithms behind the markets.

Before I list off too many vague terms, here’s an example of market design at work:  kidney matching. Every year, more and more Americans are placed on the waitlist for transplant kidneys to replace their own failing or diseased kidneys. Thousands of people either die or become too sick to receive a transplant while waiting for a matching kidney. It is important to note is that people have two kidneys, but can lead normal lives with just one. A lot of people on the waitlist have loving relatives or friends who would donate one of their kidneys, but they unfortunately do not have the same blood types and other characteristics necessary for transplant.

Kidney exchange first came about when two relatives met at a dialysis center waiting room. They realized that the relative of one patient is a match for the other patient, and vice versa — they could swap kidney donations, and both families would be better off! Professor Roth has worked to establish databases of recipient and donor pairs, and they run a stable matching algorithm on the database every two weeks. The frequency at which we run the matching can change the number of total matches we get over time, since new patients are being added to the database and current patients can be removed for various reasons. Two weeks has just worked out well in practice.

[I actually had a problem on my 6.046 take-home exam last fall motivated by kidney matching. We were supposed to find the fastest algorithm possible to match people, given a set of donor parameters and recipient requirements -- computer science and economics working together :-) ]

Theoretically, it is possible to have longer exchange cycles than just two. (Let A1, B1 be a recipient-donor pair, A2, B2 be a second donor pair, etc, for a total of n pairs. Then we can imagine a situation where A1 can donate to B2, A2 can donate to B3, A3 can donate to B4, etc, until we loop back and An donates to Bn.) In practice — and this is also where economic incentives come in — we need to have all the operations simultaneously to make sure that all the donations actually happen. For a 2-way swap, that’s 4 simultaneous operations, requiring 4 surgical teams, 4 operating rooms, the works. Some longer chains and larger swaps have happened, but they are rare.

One of our MIT professors, Parag Pathak, was a graduate student at Harvard and has worked with Al Roth and co-authors to design allocation mechanisms for public school lotteries in New York, Boston, and San Francisco. The goal is to design the system for which it is a dominant strategy for people to submit their true preferences for schools. This removes the confusion from an important decision like choosing schools and makes the school-matching process stable and efficient.

The applications that we have discussed include job market matching, job markets for couples, and dating websites. We will have a guest speaker this week to present his findings in generalized matching and another guest speaker to talk about the shortage of IP addresses. Did you know that IP addresses were in shortage? Luckily, MIT got a lot of IP addresses way back when the internet first came about. I would just be listing off a bunch of terms if I wanted to talk about the theoretical topics in the course, so visit the course site if you’re interested (link below).

For my final project, I’m exploring the market for vacation house swaps. I just came across a news article about the Iraqi government auctions for their gas fields, though, so I might switch topics. Any suggestions for topics I should consider for my final project in market design would be appreciated.

If you’re interested in learning more about market design, here are some useful links:

Market Design blog by Al Roth and Peter Coles: http://marketdesigner.blogspot.com/

2056a course site: http://isites.harvard.edu/icb/icb.do?keyword=k72549

Al Roth’s game theory, experimental economics, and market design page: http://kuznets.fas.harvard.edu/~aroth/alroth.html

The Market Design class that I am taking (2056a in Harvard code) is a graduate course, with game theory as a prerequisite. If you are interested in taking this class, you should take game theory (e.g. 14.12) as soon as possible because 2056a and 14.12 are only offered in the fall (and also because game theory is awesome!).

I will likely blog next about Development Economics and 14.771 next, stay tuned!

  1. No comments yet.
(will not be published)