💾 Archived View for gmi.noulin.net › mobileNews › 4289.gmi captured on 2021-12-03 at 14:04:38. Gemini links have been rewritten to link to archived content

View Raw

More Information

➡️ Next capture (2023-01-29)

-=-=-=-=-=-=-

Free exchange - Game, set and match

Alvin Roth and Lloyd Shapley have won this year s Nobel for economics

Oct 20th 2012 | from the print edition

IN MOST countries it is illegal to buy or sell a kidney. If you need a

transplant you join a waiting list until a matching organ becomes available.

This drives economists nuts. Why not allow willing donors to sell spare kidneys

and let patients (or the government, acting on their behalf) bid for them? The

waiting list would disappear overnight.

The reason is that most societies find the concept of mixing kidneys and cash

repugnant. People often exclude financial considerations from their most

important decisions, from the person they marry to the foster child they adopt.

Even some transactions that do involve money are not really about price.

Universities in America do not admit students based on who pays the most, for

example. Rather, they select students based on complex criteria that include

grades, test scores and diversity. Similarly, students choose their university

on more than just financial factors.

Money is not essential to a market. After all, economics is about maximising

welfare, not GDP. But the absence of a price to allocate supply and demand

makes it harder to know whether welfare is being maximised. This year s Nobel

prize in economics went to two scholars Alvin Roth, who has just joined the

economics department at Stanford University, and Lloyd Shapley, a retired

mathematician at the University of California, Los Angeles who have grappled

with that very problem.

In 1962 David Gale (who died in 2008) and Mr Shapley, now 89, published a

playful paper called College Admissions and the Stability of Marriage . They

noted the similarity between college admissions, in which students and

universities are trying to pair up to their mutual satisfaction, and the

marriage market, in which a fixed number of men and women are trying to find a

match. In romantic comedies, each man and woman marries their own true love. In

real life, some people settle for second-best, which can lead to lots of

trouble. If John and Mary love each other but are married to other people, they

will be tempted to leave their current partner and marry each other. But if

John loves Mary, while Mary loves her husband more than John, both will stay

put.

Mr Gale and Mr Shapley devised an algorithm for matching an equal number of men

and women that would guarantee this second, more stable outcome. Each man and

woman ranks their preferred partners. Each man proposes to his highest-ranked

woman. Each woman rejects all the proposals she gets except the highest-ranked

among them. But she does not accept the proposal, in case a man she prefers

even more proposes next time. The algorithm is rerun until all women have a

satisfactory proposal.

Sadly, co-operative game theory has not yet had the opportunity to transform

the marriage market. But Mr Roth spotted practical applications in other areas.

In the 1940s the competition for new doctors sometimes saw hospitals making

offers to students years before they graduated and thus before their

qualifications were truly known. The National Resident Matching Programme was

devised to match doctors to hospitals in a way that maximised their

satisfaction. This programme, Mr Roth noted in a 1984 paper, was a real-life

example of the deferred-acceptance algorithm of Messrs Gale and Shapley. The

tests of a well-designed market are that participants are satisfied enough that

they don t go around it, and that there is little incentive to game the system

by, for example, lying about their preferences. This was true of the

resident-matching programme, Mr Roth said.

Other systems worked far less well. Both the New York and Boston public-school

systems used to assign students according to their preferred choices, but

students often had to decide before knowing all their options. Thousands ended

up at schools for which they had expressed no preference. Mr Roth helped both

design algorithms that significantly reduced these mismatches.

He also applied his expertise to organ donation. A man who would not donate a

kidney in other circumstances may do so if his wife needs one. If their blood

types do not match, they can be paired with a couple in the mirror-image

position. The New England Programme for Kidney Exchange, which was partly

designed by Mr Roth, incorporates much more complex chains of donors and

recipients and raises the supply of kidneys by making a donor more confident

his loved one will find a match.

I love you, subject to the next algorithm

In time the internet could make formal matching systems viable for even more

transactions. Existing systems cannot always be improved upon, however. Utku

nver at Boston College, who helped develop the kidney-exchange programme with

Mr Roth, points to the allocation of law students to federal-judge clerkships.

Judges have complete control over whom they hire, and many students to choose

from, so there are fewer benefits to a formal clearing-house system. When

economics departments hire new PhDs, their preferences are too difficult to

codify in a matching system. And in many cases such systems should only

facilitate transactions, not execute them. Mr Unver and his colleagues are

developing a way of recommending foster children to adoptive parents in

Pennsylvania, but the final decision is left to social workers and the

families.

In their 1962 article Mr Gale and Mr Shapley noted that their algorithm was not

particularly complicated, illustrating a larger point about their discipline:

any argument that is carried out with sufficient precision is mathematical.

The recognition of Mr Shapley s and Mr Roth s work is also a reminder: that for

all the bad press economics has received since the crisis, the discipline still

brims over with insights that can solve real-life problems.