Continuing Education... Alvin Roth on Matching Markets

EconTalk Extra
by Amy Willis
PRINT
Alvin Roth on Matching Markets... Lee Ohanian, Arnold Kling, and...

Can economists save lives? It seems Nobel laureate Alvin Roth can...In this week's episode, Roth talks with EconTalk host Russ Roberts about matching markets and social support for markets. The conversation ranges over indirect kidney donations, (public) school choice, financial market trading, and more.

As always, we'd like to take this chance to enrich the experience of this week's conversation. Use the prompts below to respond in the comments, and use them to spark you own conversations offline. We love to hear from you!

1. What were your top takeaways from this week's episode?

2. Roth asserts that in matching markets, prices function differently than they do in commodities markets? How would you describe this difference?

Why Gets What2.jpg

3. How might the market for kidneys work if it were legal to buy and sell a kidney? Who would be better off and who would be worse if compared to a situation where such exchanges are illegal but donations are coordinated and chained the way Roth describes? Do you think rich people would get all the kidneys? Why or why not?

4. At the end of the episode, Roth defended his role as an economic engineer saying, "People have been designing marketplaces forever. It's what we do." Russ ended the episode here. How might he have responded if he had decided to continue?

Comments and Sharing



TWITTER: Follow Russ Roberts @EconTalker


COMMENTS (11 to date)
Partisan writes:

Re: #3, I think the mainstream view that assumes poor people will line up to mutilate themselves if they're given half a chance to get organ money is a bit offensive.

It's also quite strange to think that organ markets would be bad because the rich would benefit disproportionately. Is it really preferable that more people die as long as some of them have lots of money?

I do think that legal organ markets would lead to the rich getting most of the benefits -- at first. I also think that when news type of TVs, smartphones, and many other things first become available the rich get most of the benefit. It would be amazing if the organ surgery didn't get cheaper and safer as it became more common. Should we be encouraging rich people to bear those costs so everyone else can benefit later?

Matt Harmon writes:

1. Social repugnance cannot always be viewed by economists as a barrier that has to be broken. For years I have thought that a cash market for kidney transplants should exist. Its absence seemed to stem from irrational fears over theoretically victimized poor individuals. I thought it crazy that thousands of people die each year on the kidney waiting list when we could incentivize more donors to participate via cash. But social constraints do not exist without reason. Roth brought up the point that doctors are concerned with “first doing no harm,” The Hippocratic Oath and that there would need to be “lots of informed consent” in a market for kidneys. That makes me wonder if a healthy person can make a fully informed decision about donating a kidney for cash before surgery. He/she could be fully informed about the operation, the recovery, and living with one kidney and agree to go along with it; but at some point after the surgery, that same donor may regret the decision. It could lead to a significant reduction in happiness for that donor. How do you value not just a kidney, but also foregoing the remaining years of your life with two kidneys? There may be a lot of time inconsistency when it comes to having two kidneys ten years from now or having $20,000 in cash today. With these issues surrounding consent, regret, and the difficulty valuing kidneys, I can see why the public and health industry are reluctant to develop a kidney-for-cash market. That would only seem to enhance the importance of what Roth et al. have done with paired donations.
2. In matching markets there is more information to a transaction than can be efficiently portrayed by a single price. Moral issues may dictate that monetary exchange is not an option. In other cases, there may be an ongoing value assessment on one or both ends of a transaction. For instance, once a student is admitted to a school is he/she going to thrive at that school and enjoy the experience? Is the school going to get a thoughtful, attentive student that could potentially become a donor later in life? Does the quality of the student merit a scholarship? How does the student compare to the rest of the application field? Commodity markets are more easily priced because they deal with uniform goods that are consistently traded and usually the only issue at hand is the level of supply and demand. Matching markets deal in infrequently traded, quality variable things that are inherently difficult to value.
3. In a cash for kidney market, waiting list patients on dialysis would almost certainly have it better; they would be more likely to receive a transplant and to live longer. It is uncertain if the paid donor would be worse or better in this market. Presumably, the paid donor opted in, so he/she would be better off, but there is the whole issue of informed consent and changing preferences that may change their feelings. Rich people would not get “all the kidneys,” primarily because it is not evident that the waiting list for kidneys is filled with rich patients. Perhaps wealth is correlated (not caused by, but still) with healthy kidneys. And truly desperate patients could always take out loans, tap all of their savings, or crowd source funds to pay for a kidney. Also, the government might step in to subsidize healthy kidney donors for the sake of patients on Medicare and Medicaid.
4. Russ could have asked if Roth thinks paired kidney transactions could possibly take place independent of economist-coordinated efforts. Could moderated internet message boards take the place of economists and health professionals reviewing waiting lists? My guess is that information in this arena is opaque and that the family and friends of patients with renal failure would not coordinate as well as paid professionals. At some level, all of these patients and donors have to be coordinated, and it is far from obvious that could do so emergently in an efficient way.

Michael Byrnes writes:

1. My top takeaway is that a lot of the markets that we all rely on are more like "matching markets" than stock exchanges or commodity markets. Even ones where money is involved, such as employment markets. I'd like to learn more about how matching markets function - there seem to be quite a range of them, from kidneys to public schools to jobs.

2. In a matching market, prices are only one consideration, whereas in a commodity market prices are generally all that matters. Roth pointed out that if you buy a share of stock, you don't care who you are buying it from, whether or not it was properly taken care of, whether it is in better or worse condition than a share of the same stock that you could get from someone else. All that matters is what the price is and whether or not you do receive the share that you paid for. If you are hiring employees, you do not just publically offer $X per year and adjust $X until the correct number of workers comes in to work for you - you need to identify the people you want to hire and they need to choose to work for you. Money is still a key factor, but it is only one of several.

3. I have no idea how a kidney market would work if buying and selling of kidneys were permitted. In theory, there would be a lot more people willing to provide kidneys of they were being paid - I think that is very likely true. But I think I would personally be very nervous about selling a kidney. The nice part of Roth's system is that - to the extent it is possible to do so - it ensures that kidney donors get a fair price. I donate a kidney (and in some sense, a life) and in return I (in this case my loved one) get back a kidney (and a life). It's as fair an exchange as is possible with kidneys. On the other hand, there are fewer kidneys available under Roth's system than there might be if live donor kidneys could be bought and sold. As to whether only rich people would be able to get kidneys - I think that's possible but not certain.

4. Russ might have responded to Roth by saying that markets are an example of emergent order, "of human action but not of human design". But I think that's only partly true - I think a lot of markets, and particularly those that involve matching - have both designed and emergent components. Commodity markets and stock exchanges work rather well and have emergent aspects - but key parts of those markets have obviously been designed (sometimes but not always with good results).

I also think that Russ has sometimes pointed out that "emergent" doesn't always mean "good". The market for clerkships (or for medical residencies before the system was redesigned) are examples of emergent orders that aren't good. The public schools example is sort of an example, too - that was obviously a designed system, but the behavior of principals (only consider studients who made their school their top choice) was an energent response to that system that had bad consequences. Roth's algorithm, designed, fixed some of those problems.

The simultaneous kidney transplantantaion seem to me like an emergent, low-tech solution to a serious problem - what if I donate my kidney first and then the guy who was supposed to donate to my loved one backs out? That's a huge potential risk that would limit peoples' willingness to participate in kidney exchanges, solved simply (but in a way that has its own limitations). It is similar to how we do a lot of our exchanges, after all - usually the store clerk wants me to pay up when I buy my groceries, not some time later.

I think in many cases markets may function poorly because transaction costs are high. I see what Roth is doing as developing ways to reduce those costs and allow exchanges to proceed more efficiently.

BTW, one thing I've been wondering about since the episode: Roth mentioned that occasional live donors who are just donating (ie not looking to also receive a kidney for a loved one) make the system work better, because they allow more matches to be made, allow non-simultaneous transfers, etc. My question for him: do deceased donor kidneys ever enter these exchanges - I would think this could have the same type of benefit.

Ryan Langrill writes:

4. Alvin Roth is right. People design markets all the time. Look at Apple's app store, or iTunes; think of farmers' markets, or as others have mentioned, stock markets. The creation of these markets is an entrepreneurial action similar to the creation of a firm.

The *success* of those markets is not designed. Those "markets" work because they pass the market test. In other words, the market process selects the types of markets that work.

The same distinction holds when looking at products. People design products, but they do not design the success of products.

It highlights, though, the fact that people mean different things when they talk about markets. "A market" is not "the market" writ small.

Andy writes:

It was mentioned briefly on the show that some restaurants are now requiring payment for reservations. That sounds like a high-end NYC scheme, but thought I would share that we experienced it last fall at Disney World. The "best" places, either by quality or popularity (iconic experiences like Cinderella's castle, and some of the places around Epcot which really are very good) tend to fill up very quickly. They take reservations 180 days out from the beginning of your trip, and people get up very very early on their first available day to make their selections. The reservation itself is free, but if you cancel with less than 24 hrs notice, they charge you $80. At first I found it silly because no doubt they will fill every seat. There's always a stand by line. But apparently people were making multiple reservations for the same meal time so they could decide on-the-fly depending on where they were or what they were in the mood for. The late cancelation penalty limits that and more people can make selections in advance. In fact, finding out when we could actually get in to some places to eat pretty well determined our itinerary several days. That place simply prints money...

Nathan Taylor writes:

Re #4. I'm not sure what Russ Roberts would have asked Alvin Roth at the end for question 4, but what I wanted him to ask was how Roth could reconcile quoting Hayek and immediately talk about designing markets. Perhaps Roth intended this to mean design by iterating slowly in a humble way, knowing the emergent effects would not be clear and we should be cautious. That's Hayekian (and where I suspect Russ stands, as do I). Or he could have meant that he only agreed with Hayek insofar as markets need to be efficient, and designing them whole cloth was ok. Which would have been un-Hayekian. More discussion on this would have been great.

Martin Dertz writes:

re: #4, How Russ might have responded (in an alternate universe)...

"Heavens, what a fascinating quotation ! And well chosen given its speaker is a man whose ideas and quotes I often espouse to imply the exact opposite of your own implicit meaning. This is making me re-evaluate my worldview.... Perhaps I've been a noob about human nature. It reminds me of another quote, 'The market is a fickle mistress' - unknown. I will continue to consider these thoughts, and perhaps I'll even disavow some of my most closely held beliefs because they're in fact not truth and that is what humans do." - Russell Roberts

OK, so maybe Russ wouldn't have said noob, but the rest is on point.

Patrick O'Brien writes:

This episode was unique in that it left me unsettled. Alvin's comment at the end rose all sorts of red flags, and then the episode ended.

I think "People have been designing marketplaces forever. It's what we do." needs a lot of explanation, which I didn't feel I got. Alvin did have some interesting things to say leading up to this statement that gave me hints as to what he could mean, but I don't feel that he provided sufficient context.

I think the basis for my agitation was the lack of a definition for both market and design.

If by market he means - "a mechanism that facilitates the transactions between agents, in which parties participate of their own volition."
and by design he means - "innovating to improve efficiency and value"
then yes, I agree with him. People have found ways to make this happen.

However, there seems to be a big jump from emergent market iterations via innovation and the kind of design that was discussed on the podcast (school choice in NYC). He designed a brand new system from scratch.

Alvin and his colleagues designed a very clever way of divvying up public goods. According to Alvin's stats it appears to be an improvement from the last method. But at the end of the day it's a designed system to distribute goods according to arbitrary rules about "what is fair." Also, participants do not have the choice of leaving the system; they are essentially coerced to playing along. This is not a market. This is a distribution scheme.

Perhaps Alvin has improved the situation in NY, but I think he's playing with fire. There will be loopholes in his system, just like there were in the previous one. People's strategy will change to match the new game. There will be benefits and hazards that come with the new design. I hope for the kids of NY that the benefits are worth it.

Michael Byrnes writes:

Patrick O'Brien wrote:

If by market he means - "a mechanism that facilitates the transactions between agents, in which parties participate of their own volition." and by design he means - "innovating to improve efficiency and value" then yes, I agree with him. People have found ways to make this happen.

However, there seems to be a big jump from emergent market iterations via innovation and the kind of design that was discussed on the podcast (school choice in NYC). He designed a brand new system from scratch.

I don't really think "designed from scratch" is really true. There was a system in place before Roth's algorithm. A system in which students (and their parents) submitted their preferences (in the form of a ranked list), there was no competition on price, and the school system assigned students to schools based on their submitted preferences.

For better or worse, Roth's system retained all of those characteristics. All he did was to change the way the school went about assigning students based on their preferences - by implementing a deferred acceptance system via his algorithm. To me, that is an iterative improvement.

I think the behavior of the school principals (pre-Roth) made a certain amount of sense. Focusing on students who put listed their school as a first choice simplified their process to a great extent. But even with that the system was extremely inefficient (one-third of students unmatched at the end). And it incentivized students to not list their true preferences.

Roth's algorithm simply made the existing process function better - it removed penalties to students for listing their true preferences and reduced the number of unmatched students by an order of magnitude.

But at the end of the day it's a designed system to distribute goods according to arbitrary rules about "what is fair." Also, participants do not have the choice of leaving the system; they are essentially coerced to playing along. This is not a market. This is a distribution scheme.

A market is a distribution scheme.

There will be loopholes in his system, just like there were in the previous one. People's strategy will change to match the new game. There will be benefits and hazards that come with the new design.

Roth's innovation here was an improvement on what could be achieved given the (preexisting) constraints he had to work with. That describes a lot of legitimate human innovation, Roth's work included.


Robert Swan writes:

Thoroughly enjoyed this talk. Very thought-provoking -- listened to it three times.

The preceding comments call to mind the intelligent design vs. evolution argument, but I think an evolution vs. genetic engineering analogy is closer. Evolution is not negated by our doing genetic engineering. Variation coming from GE is a mere detail and natural selection will continue. Similarly, Roth can "design" a market in school placements by setting up rules, but the players in that market are free to react. The market, rules and all, will evolve accordingly.

No Hayekian paradox here. If you zoom in on any one player in a market you'll see central control. You have to take a broader picture to see the bottom-up order.

Jeffrey Scornavacca writes:

Enjoyed this show immensely. Curiously, there is article in the NY Times today about the kidney market in Iran:

http://opinionator.blogs.nytimes.com//2015/07/31/need-a-kidney-not-iranian-youll-wait/

Comments for this podcast episode have been closed
Return to top