Fashionable Nobel Laureate Designs
Markets!
Alvin E. Roth, Craig
and Susan McCaw Professor of Economics at Stanford University
Alvin E. Roth, Craig
and Susan McCaw Professor of Economics at Stanford University and Gund
Professor of Economics and Business Administration Emeritus at Harvard
University, shares his thoughts. A recipient of the 2012 Nobel Memorial Prize
in Economics for his work in game theory, experimental economics, and market
design, Professor Roth has been helping hospitals in the States and other
countries design markets for kidney exchange based on game-theory principles
among many other laudable accomplishments. You can read about Professor Roth’s
ideas in his book Who Get’s What – and
Why: The New Economics of Matching and Market Design, already out as an
eBook and soon to be available in printed form. Nikkei Business will be
publishing a Japanese version.
1. You have made
significant intellectual contributions in the areas of game theory, market
design, and experimental economics. Can you give us an overview of each of
these areas and how you initially became interested in them? How are the areas
interrelated?
Game
theory is a set of mathematical tools for talking about economics, for talking
about how people deal with environments where other people are also vested
involved and their decisions all matter to each other. Game theory is a natural
backbone for marketing design because game theory is about the rules;
alternatively stated, the design of a market is its rules. Game theory helps us
think how different rules might work in the design of markets. Experimental
economics helps us test whether these designs are correct or not.
2. Can you briefly define repugnant
markets for our readers who may not have previously heard the term? How are
they different from other markets?
Repugnant markets are markets that we are inclined not
to allow to exist. I tend to talk more about repugnant transactions than
repugnant markets. A repugnant transaction is one in which some people would
like to engage, but others do not think should be permitted. In addition, in
transactions we refer to as repugnant people often have difficulty giving a
concrete reason why they feel the transaction should not be allowed.
If you wanted to open a night club next to my house, that would produce a negative externality. The club would generate a lot of noise, and I would not be able to sleep. However, that is not normally what we refer to as a repugnant transaction. For instance, in the past ten years, a lot of change has taken place with respect to same-sex marriage. It is a transaction in which some individuals would like to engage. Some same-sex couples would like to marry one another, a desire that is very controversial here in the United States. Yet explaining why people object is difficult. People do not feel threatened that they would be compelled to marry someone of the same sex if same-sex marriage were legalized, but they may still think you should not be able to marry someone of the same sex. In contrast to my night-club example, the idea of ‘repugnant transaction’ is useful for thinking about same-sex marriage would be considered a repugnant transaction.
I became interested in these types of transactions
through my work on kidney transplantation. Selling and buying organs for
transplantation is illegal just about everywhere in the world. Black markets
exist. There are people who would like to buy a kidney and others who would
like to sell one. But the transaction is illegal in every nation of the world
except the Islamic Republic of Iran. Something illegal just about everywhere is
worth studying.
3. Can you explain the Stable Matching
Problem in simple terms that non-economists would understand? How do the
algorithms used to solve the problem operate?
Before I talk about algorithms, let me first explain
what stable matching is. I spend a lot of my time studying matching markets,
which are different from commodity markets. In the latter, prices decide
everything. If you are purchasing a ton of a specific grade of coal, you are
purchasing a commodity. You do not care who the seller is. You just want a good
price. The role of the market is to find the price at which supply equals
demand. But many, many markets do not work that way. Even if they make use of prices,
the price alone does not determine who gets what. You must be chosen. In
addition to choosing what you want, you yourself also have to be chosen.
Labor markets are an example. You cannot just decide to work for Stanford University. You have to be hired. College admissions is similar. You cannot just decide to become a student at Stanford University. You have to be admitted. All of these “transactions” involve money. Professors are paid. Students pay tuition. But the price does not determine who gets what. Stanford does not raise its tuition until just the number of students needed to fill classes remains. That would be the price at which supply equals demand.
In many markets, other institutions are involved in
deciding who gets what. If the market is very competitive, and people can reach
agreements with whomever they want, you can still ask the question What types
of matchings should we expect? This question brings us to the idea of
stability. In a labor market, we would consider a matching stable when all of
the workers have been matched to jobs, and there are no employers and workers
would rather be matched to each other but are not. In a highly competitive,
free market, if two people who are not matched would rather be matched, you
would have to ask the question What prevented them from matching up with each
other? A matching is stable if no such pairs exist. I might not have the job I
like most, but in that case, that must be because the employer prefers to hire
someone else instead of me. A mutual attraction would be an instability.
One way to organize market places to achieve stable matchings is to have a centralized clearinghouse. We have a clearinghouse for American doctors, and in Japan, you recently have implemented a similar system. Doctors submit a rank-ordered list of the hospitals where they would like to work in their residency programs. The residency programs submit analogous lists rank-ordering their physician preferences. An algorithm is then invoked to produce a stable matching based on the two sets of preference lists. The algorithm operates as follows: Physicians apply to their first-choice hospital, where they are ranked ordered. The hospital rejects those that exceed their capacity. If the hospital can accommodate ten residents, physicians whose ranks are eleven or less lower are rejected.
However, the top ten physicians are not immediately accepted at that point. Rather, they are simply not rejected. Meanwhile, the doctors who have been rejected apply to their second choice. When the hospitals receive these new applications, they add them to the list of physicians who have not yet been rejected, then rank order the new, combined list. They once again reject physicians whose rank order is eleven or below. This process continues until no more additional rejections are made, in other words, no one wants to apply to another hospital. Either everyone’s application is currently being held, or they have applied to all the hospitals of interest, but been rejected. At that point, the hospitals send out acceptances to their top candidates. The doctors are matched to the hospitals holding their applications when the algorithm ends, producing a stable matching. Suppose, for example, I am a doctor who has been matched to his third-choice hospital. I know that my first and second choice hospitals did not want to hire me because I had to first be rejected by them before applying to my third choice. The fact that the hospitals rejected me indicates they have filled their openings with candidates they preferred over me.
Follow-Up Question: What is the role of
price in this matching process?
Suppose I would like to work for Google. Of course,
the compensation they offer will affect my preferences. In fact, one of the
major factors people who will receive job offers from Google will be the salary
compared to that of competitors like Facebook. Google does not set its wages
low enough that just the number of applicants the company needs want to work
there. Google jobs are good! Lots of people would enjoy working for the wages
they offer. If Google is not attracting the employees it wants at the salaries
they offer, they can always raise them. Clearly, wages are important. They
enter the matching model I have just described through preferences. Other
factors equal, I would prefer the higher salary, but that is not the only
consideration in selecting an employer. I might be able to earn a higher salary
at Google than at Stanford, but I prefer to work here. Conversely, Google may
not have wanted to hire me but Stanford did. Lots of factors are considered,
and price can be a huge one.
In kidney exchange, price does not play a role due to
legal restrictions. But these exchanges are still different from commodity-type
markets. When you purchase Sony stock, you do not care who the seller is
because all of the shares are the same. When you make a job offer, you do not
offer the position to everyone on the market, but to a specific person. In this
case, you care about who, specifically, you are hiring. Money is not the only
factor, but you do have to offer the person a salary. In the case of kidneys,
the patient cares a lot about which one he or she gets, but cannot offer money
for it. Markets comprise a continuum. They are neither pure commodity markets
on the one end nor like the market for kidneys, where price does not play a
role for legal reasons. The markets for most things are, rather, somewhere in
the middle.
4. To what extent do you believe the
government should be actively involved in market design? In what specific
sectors do you feel government should especially be involved? Can we regard
market design as one type of government intervention or regulation?
Design involves all the rules used to regulate
markets. In this respect, regulation itself is a type of market design.
Governments are one of the players in market design. Lots of markets are developed
partly by governments piecemeal. Thinking about transportation in the San
Francisco Bay Area, we have public options by and large run my municipalities,
that is, by local governments; and taxis, which are private, but regulated by
government. Taxis need a medallion to legally pick up passengers off the
street. Now, we also have access to services such as Uber, which allow us to
call a private car using a cell-phone application. These services are subject
to much less regulation than taxis. Uber is a privately-designed market place
mostly for private cars. Taxis are cars owned by individual companies or
drivers whose conduct is regulated by cities. Finally, public transportation is
owned and operated by the Bay Area transit authority. If I needed to go from
here to San Francisco, I have all of these choices.
They are all designed differently but some are quite
highly regulated. The train system is government owned and operated. Taxis are
privately owned, but thoroughly regulated. For example, if I hail an available
taxi, the driver’s supposed to pick me up. In this sense, the taxi is somewhat
like a public utility. Uber drivers, on the other hand, do not have to pick me
up if they do not feel like doing so. Here you have three types of market design
and three types of regulation, in markets that are partly substitutes and
partly complements.
Governments should set in law rules and regulations
important for many markets. For example, even people opposed to government
intervention favor government enforcement of property rights. I own a house.
That means the title is registered in the county clerk’s office. I have a way
of proving I am the owner if someone claims that he or she is. Property rights
are clearly important. When we buy and sell, we want to be sure that the seller
owns the property and that you will be the owner after purchasing it.
Even property rights are contingent, however. If you
purchase a copy of a book I have written, you own that book. It’s your
property. You can sell it. You can give it away. You can even throw it away.
But you cannot copy it. Due to copyright laws, you cannot make and sell copies
of the book you have purchased. Either I would own the copyright or my
publisher would. In this way, our property rights are defined. When we talk
about repugnant markets, you own your kidneys, and you can give them away. If
you loved me, and I needed a kidney, you could give me one. But you cannot sell
me your kidney. Property rights are part of market design. Because so many
markets depend on property rights, by and large, we allow governments to define
them.
Some, however, are defined by contract. If you
purchase a condominium, you have a contract that defines your ownership. A
cooperative apartment would have a different contract. My house is registered
in a county office, but my ownership of an apartment in a building might be
subject to private rules.
5. What do you feel the significant
developments in the field of economics have been over the past thirty years of
your career? Where do you believe future developments are likely to take place?
I’ll answer the question in two different ways. If you
talk about the profession of economics, many techniques like game theory,
experimental economics and market design have entered. Computers have also
entered the field. When I was young, statistics were difficult to compute, but
now everyone has a computer on his or her desk. Big data has entered as well.
If you think about the economy, Aa lot of transactions are now conducted
through the internet, generating massive amounts of data. Cash registers are
now computerized, making check-out data available. For consumers, there are
auction data and advertising data. More generally, computers are now
widespread, and many more economic transactions linked to the internet has
resulted in much more data. In addition, new ideas about the role of
economists, what we should do and how we should do it, have emerged.
6. You were educated at Stanford, but
taught for years at Harvard before returning to Stanford as a faculty member?
What are the major similarities and differences between the two? How would you
describe your experiences as a faculty member at both?
Both are elite American universities. As such, Harvard
and Stanford are similar to each other in many respects: Both have excellent
students. Both are concerned about the development of economics and the
economics profession. The view from the window, however, is somewhat different.
At Harvard, you tend to see New York City or Washington, D.C., but here at
Stanford, you are inclined to see Silicon Valley. Accordingly, there is greater
focus on entrepreneurial and start-up enterprises here. One reason I moved to
Stanford is the greater interest in market design here due to the activities in
Silicon Valley. Many companies like Amazon are market places. So are Uber and
Google, which is essentially a market for advertisement. More and more
companies are making markets, and many of those companies are here.
