Image: Jayesh Amin, CC BY-SA 3.0How economists built a market for kidneys without prices
In the United States about 90,000 people are waiting for a kidney, and a few thousand of them die every year before one becomes available. Many of them have a husband, a sister, or a friend willing to give one of their two kidneys. Often the offer is useless, because blood groups or tissue types do not match, and until the early 2000s the story usually ended there. Today a patient in that position can be matched with another incompatible pair, so that each donor gives to the other patient. The procedure is called kidney exchange, and the rules that make it work were written by economists.
Why there is no price
The simplest market for kidneys would be a market with a price, and it is banned nearly everywhere. The United States prohibited the sale of organs in the National Organ Transplant Act of 1984, the Council of Europe did the same in its convention of 1997, and the World Health Organization's guiding principles say so as well. Iran runs the only legal, regulated system of paid living donation in the world. Whatever one thinks of the ban, it is a fact a transplant surgeon has to work with.
Alvin Roth, who received the 2012 Nobel prize in economic sciences for his work on matching, calls the feeling behind such bans repugnance, and treats it as a constraint on markets rather than a mistake to be argued away. Societies have refused to let certain things be bought and sold at many times and places, and the list changes: lending money at interest was repugnant in medieval Europe, life insurance was repugnant in France until the nineteenth century, and buying a place in the army was normal in the United States during the Civil War. An economist who wants more kidneys transplanted can spend a career arguing that the ban is wrong, or can design something that works within it. Roth chose the second route.
The idea
A kidney exchange between two pairs is easy to describe. Pair A consists of a patient with blood group A and a donor with blood group B; pair B is the mirror image. Neither donor can give to their own patient, and each can give to the other's. Two operations, two transplants, and nobody has paid anybody. The first such exchange in the United States took place in 2000 at Rhode Island Hospital, and surgeons had discussed the possibility since the 1980s. What was missing was a way to find the exchanges when there are hundreds of pairs rather than two.
That is a matching problem, and matching problems were what Roth had worked on since the 1980s, first in the allocation of young doctors to hospitals and then in school choice in New York and Boston. In 2004 Roth, Tayfun Sönmez, and Utku Ünver published a paper in the Quarterly Journal of Economics that treated kidney exchange as such a problem. Given a list of incompatible pairs, with blood groups and tissue types, an algorithm looks for cycles: pair A gives to pair B, B gives to C, and C gives back to A. The algorithm picks the set of cycles that produces the most transplants, or the most transplants weighted by medical priority, subject to one hard constraint.
Why the operations happen on the same day
The constraint is that nobody can be forced to give a kidney. If pair A's donor gives on Monday and pair B's donor is due to give on Wednesday, then on Tuesday pair B has what it wanted and can simply walk away, which would leave pair A with one kidney fewer and no transplant. Contracts cannot fix this, because a promise to donate an organ is not enforceable. The only protection is to perform every operation in a cycle at the same time, in adjacent operating theatres. A three-way cycle therefore needs six surgical teams and six theatres on the same morning, and that is why most exchanges involve two or three pairs rather than ten.
The limit was lifted in a different way in 2007, when a donor who wanted to give a kidney to nobody in particular started a chain. An altruistic donor gives to patient 1; patient 1's donor gives to patient 2; patient 2's donor gives to patient 3, and so on. In a chain, the operations no longer have to be simultaneous, because the pair that could walk away has already received its kidney before its donor is asked to give, and if that donor does renege, no patient is left worse off than before the chain began. A chain of ten transplants described in the New England Journal of Medicine in 2009 was done over eight months. Chains of thirty and more have since been completed, and the longest ones have run for over a year.
What the numbers say
The New England Program for Kidney Exchange, which Roth and his colleagues helped set up in 2004, was the first regional clearing house. National programmes followed in the United States, the United Kingdom (the Living Kidney Sharing Scheme, since 2007), the Netherlands, Canada, Australia, and elsewhere. In the United States, kidney paired donation produced a few dozen transplants a year in the mid 2000s and more than a thousand a year by the end of the 2010s, according to the records of the Organ Procurement and Transplantation Network, which is roughly one living-donor kidney transplant in six. Each of those transplants replaces several years of dialysis, which costs the US Medicare system roughly 90,000 dollars a year for every patient, and a living-donor kidney lasts on average about fifteen to twenty years.
Here is the worked number. Suppose a programme has 100 incompatible pairs, and that blood group incompatibility alone is the obstacle in each case. With the blood group frequencies of a European population, about 40 per cent of patients are in group O, and a group O patient can only receive from a group O donor, while group O donors can give to anyone. That asymmetry means that O patients pile up in the pool and O donors are scarce in it, because an O donor who is compatible with their own patient never enters the exchange at all. A simple two-way exchange can serve only a fraction of the pool, and Roth, Sönmez, and Ünver showed with simulations that allowing three-way cycles raises the number of transplants substantially, with little further gain from longer cycles. Letting a few altruistic donors start chains helps most of all, because a chain can end with a donation to a patient on the deceased-donor waiting list, which uses up no scarce pair.
What this teaches about markets
Three things are worth taking away. The first is that a market is a set of rules, and the rules can be designed. A price is one way of deciding who gets what, and it is often the best one, but it is a special case. The second is that the design has to respect incentives that cannot be legislated away: a donor who can renege will sometimes renege, so the schedule of operations has to make reneging harmless. The third is that the designer is constrained by what society allows. Roth has written that economists should regard repugnance the way engineers regard friction, as a fact about the world rather than a failing of the people who hold the view.
Unit 2 of the Economics Masterclass, on supply, demand, and market equilibrium, ends with exactly this topic, after a full treatment of how prices clear markets when prices are allowed. The research spotlight of that unit presents Roth's 2007 article on repugnance, and his Nobel lecture is a readable account of the whole story from the hospital residency match to the kidney chains. The program page explains how the units are delivered.


