[Paleopsych] FRB, Richmond: Interview: Thomas Schelling
checker at panix.com
Thu May 12 19:17:14 UTC 2005
Interview: Thomas Schelling
Region Focus Spring 2005: Interview - Federal Reserve Bank of Richmond
[Schelling is among the very most imaginative economists and a fabulous writer
as well. I discovered him in graduate school and am delighted that Alice was
able to take an honors course under him when she studied chemical engineering
at the University of Maryland.]
Thomas Schelling's early research was common fare for economists in
the 1950s. The quality of the work may have been higher than most, but
the topics were relatively mundane. His first two books were titled
simply National Income Behavior and International Economics. But his
interests extended beyond the traditional confines of the discipline,
a point that was made clear with the publication of The Strategy of
Conflict in 1960. In it, he used the tools of economics to illuminate
important issues in international relations, while making significant
contributions to game theory and laying the ground-work for later
research in experimental economics.
Schelling has continued to publish on military strategy and arms
control throughout his career, but his work has led him to a number of
other seemingly disparate issues, such as racial segregation,
organized crime, and environmental policy. In each case, he has been
able to generate original insights from ordinary observation. As his
long-time colleague Richard Zeckhauser has written, Schelling "thinks
about the essence of phenomena. In scanning everyday behavior, he sees
patterns and paradoxes that others overlook."
Schelling spent most of his career at Harvard University, before
joining the faculty of the University of Maryland in 1990. He is a
past president of the American Economic Association and recently
worked with other distinguished economists on the Copenhagen
Consensus, a project designed to prioritize the largest social
problems facing the world. Aaron Steelman interviewed Schelling at his
home in Bethesda, Md., on February 7, 2005.
RF: Your early work focused on topics that were fairly conventional. How did
your work progress into areas, such as strategic bargaining, that largely
had been beyond the scope of economists?
Schelling: In 1948, I had just finished my coursework for the Ph.D. at
Harvard, and a friend of mine called from Washington. He was working
on the Marshall Plan and said that he had an opportunity to go to
Paris but he couldn't leave until he had a replacement. So he asked me
if I would like to replace him. I said sure.
Eventually, I went to Europe as part of this assignment and worked
mainly on negotiations for the European Payments Union. Then, Averell
Harriman, who had been head of the Paris office, went to the White
House to be President Truman's foreign policy advisor. Harriman asked
my boss to go with him, who in turn asked me a few months later to
join him. In 1951, the foreign aid program was shifted to the Mutual
Security Program, with Harriman as director, in the Executive Office
of the President. I moved there, and stayed through the first nine
months of the Eisenhower administration. So when I left, I had spent
five years in the foreign aid bureaus, largely working on
negotiations. That, I believe, was what focused my attention on the
type of issues that showed up in The Strategy of Conflict.
RF: One of the more famous bargaining situations that you propose in The
Strategy of Conflict involves a problem in which communication is incomplete
or impossible - the game where two strangers are told to meet in New York
City but have not communicated with each other about the meeting place. What
does this game tell us about bargaining? And what, if any, are the policy
Schelling: That little exercise, which I designed to determine if
people could coordinate without any communication, became fairly
famous and now I am usually identified as the originator of the idea
of "focal points." My argument was that in overt negotiations
something is required to get people to arrive at a common expectation
of an outcome. And the ability to reach such a conclusion without
communication suggested to me that there was a psychological
phenomenon, even in explicit negotiations, which may work to focus
bargainers eventually on that commonly expected outcome. By
understanding that, I thought, we may be able to more easily
facilitate policy negotiations over such matters as what would be an
appropriate division of the spoils, an appropriate division of labor,
and so forth.
RF: What were the responses when you originally posed this question to
Schelling: When I first asked that question, way back in the 1950s, I
was teaching at Yale. A lot of the people to whom I sent the
questionnaire were students, and a large share of them responded:
under the clock at the information desk at Grand Central Station. That
was because in the 1950s most of the male students in New England were
at men's colleges and most of the female students were at women's
colleges. So if you had a date, you needed a place to meet, and
instead of meeting in, say, New Haven, you would meet in New York.
And, of course, all trains went to Grand Central Station, so you would
meet at the information desk. Now when I try it on students, they
almost never give that response.
Some cities have more obvious focal points than others. For instance,
if I asked people where would you meet in Paris, they probably would
have no trouble. Most would go to the Eiffel Tower. But in other
cities, it's not so clear.
The question first occurred to me while I was driving across country
with two college friends. We were going from San Diego to New
Hampshire and back, and camping along the way. We stopped in San
Antonio and one of the other two guys got out and bought some peanut
butter and crackers. While he was gone, a police officer made me move
on, and because of the one-way streets, it took me about 10 minutes to
get back to where I dropped him off, and he wasn't there. I kept
circling around and eventually we found each other. But we realized
that this could happen to us in any city, and we should come up with a
plan about how to meet if we got separated.
We spent the whole afternoon thinking about it individually, but not
talking about it, and that evening around the campfire we compared
notes. We all wound up in the same place. The criteria we used were
the following: Every city had to have this place and there could be
only one of it, you had to be able to find it by asking any police
officer or fireman, and you had to be able to reach it by public
transportation. That narrowed the list down to the town hall or the
main police station or the main post office.
Well, before we left home, we had each given our mothers a list of
cities in which we would look for mail, and the way you get mail when
you are traveling across country is to have the letter sent to your
name, care of general delivery, and it arrives at the main post office
in that city. That occurred to all three of us, and if we had to
choose among the places that shared the criteria we described, the
main post office seemed to be the obvious choice.
RF: You begin many of your papers with examples that are taken from everyday
life. For instance, in "Hockey Helmets, Daylight Saving, and Other Binary
Choices," you use the case of a player for the Boston Bruins who suffered a
severe head injury to demonstrate why some collective action problems can be
so difficult to solve - in this case, getting hockey players to voluntarily
wear helmets. Is this a conscious strategy of yours to engage readers in
what otherwise might seem like an abstract discussion?
Schelling: I always try to find something that I can put in the first
paragraph to make the article sound interesting. It was just a
coincidence that the hockey player had been hit in the head and that I
had noticed it. It was a good example of a scenario in which everyone
might wish to be compelled to do something that they wouldn't do on
their own individually. So I think that has been part of my style. I
wrote a textbook in international economics that had about a dozen
policy chapters. I tried to have the first page of every chapter
present an interesting puzzle or phenomenon that would get the
interest of the readers.
RF: You have written that the "ordinary human being is sometimes ... not a
single rational individual. Some of us for some decisions are more like a
small collectivity than like the textbook consumer." Could you explain what
you mean by this, perhaps through a few examples?
Schelling: I started working on that subject in the 1970s when I was
asked to join a committee of the National Academy of Sciences on
substance abuse and habitual behavior. I was the only economist there.
Everyone else was a specialist on a certain type of addictive
substance such as heroin or some other health problem like obesity. It
seemed to be taken for granted that if you were addicted - whether to
heroin or alcohol or nicotine - there wasn't much you could do for
yourself. I argued that this was not the case, and gave a number of
examples of ways people can help themselves avoid relapse.
For instance, one person tried to show how addictive heroin was by
pointing out that many former users, even those who had avoided heroin
for a long time, would be likely to use the drug again if they were to
hang out with the people they used to shoot up with or even if they
listened to the same music that they played when they used heroin in
the past. I pointed out that there was some instructive material right
there. Don't associate with the same people. Don't listen to the same
music. And if the place where you used to use heroin is on your way to
work, find a different route. So even though those people may be
inclined to use heroin again, there were clearly some ways in which
they could help prevent themselves from having a relapse.
The more I thought about this issue, the more I began to conclude that
a lot of people have something like two selves - one that desperately
wants to drink and one that desperately wants to stay sober because
drinking is ruining his life and his family. It's as if those people
have two different core value systems. Usually only one is prominent
at a given time, and people may try to make sure that the right value
system attains permanence by taking precautions that will avoid
stimulating the other value system.
RF: Some have called you a "dissenter" from mainstream economics. But it
seems to me that this is true only insofar as it concerns topics of inquiry.
On methodological issues, you don't seem as willing to abandon some of the
core assumptions of neoclassical economics as, say, those people who call
themselves "behavioral economists." Do you think that this is a fair
Schelling: This is something that I talk about a lot. I claim that we
couldn't do without rational choice. But we don't expect rational
choice from a child or an Alzheimer's patient or someone suffering
from shock. We will better understand the uses and limits of rational
choice if we better understand those exceptions. I use the example of
the magnetic compass. It's usually a wonderful way to determine which
direction north is. But if you are anywhere near the actual north
magnetic pole, the compass could point in any direction, even south.
The same is true with rational choice. It is a wonderful tool if used
when appropriate, but it may not work all the time. So I consider
myself in the rational-choice school, absolutely. But I am more
interested in the exceptions than many other economists tend to be.
As for the behavioralist critique of neoclassical economics, I would
conjecture that if you walked into a classroom where a behavioralist
is teaching microeconomics, that person would teach it in a straight,
standard fashion. It's something that you have to master - you can't
do without it. For instance, if a student were to ask about the effect
of a gasoline tax on driving behavior, the response would likely be
that such a tax will tend to lower consumption of gasoline and/or
increase the desirability of more fuel efficient cars. That's just
straight neoclassical economics.
More generally, I think that when a new idea develops, it is important
that the enthusiasts are given free rein to explore and perhaps even
exaggerate that idea. Once it catches on and becomes respectable, then
it's time to become more critical. Rational choice has gone through
that process, and the behavioralists have emerged to challenge some of
its assumptions. The behavioralists have probably overstated their
case, but their ideas are relatively new and will be critiqued as
I think that people like Dick Thaler and Bob Frank, who are clearly
two of the most innovative behavioralist economists today, so much
enjoy what they do that I'm not sure if they consciously exaggerate
the role of these exceptional situations. When I read Bob Frank, I get
the sense that he is passionate, almost emotional about his belief
that American consumers are suffering welfare losses because they are
spending their money trying to avoid the discomfort of not being equal
to their neighbors. I think he overdoes it, and I think that I have
told him so. I don't know if his answer today would be, "Of course I
overdo it. I'm trying to get attention paid to something I think is
important." Or if he would say instead, "No, I don't overdo it. I
really do believe that the phenomenon is that important." But even if
the former is true, I would excuse that. I think that the point is
important enough that if exaggeration will help them get it across,
let them exaggerate.
RF: What is your opinion of modern game theory?
Schelling: That's a hard one, because I don't keep up with all the
latest work in that field. But I would like to make the following
broad claims: Economists who know some game theory are much better
equipped to handle a lot of important questions than those who don't.
But economists who are game theorists tend to be more interested in
the mathematics aspect of the discipline than the social sciences
aspect. Some economists of the latter group are good at using their
theoretical work to examine policy issues. Still, many - and I think
this is especially true of young game theorists - tend to think that
what will make them famous is their mathematical sophistication, and
integrating game theory with behavioral observations somehow will
detract from the rigor of their work.
I'll give you an example. I had a student at Harvard named Michael
Spence, who a few years ago won the Nobel Prize. Mike wrote a
fascinating dissertation about market incentives to engage in
excessive competitive expenditure. I was on his committee, and I
argued that he needed to do two things. First, summarize the theory in
40 pages. Second, find six to 10 realistic examples to illustrate how
the theory worked and why it mattered. He spent much of a year doing
that. But in the end, he published the 40-page version of his
dissertation in a top-tier journal, and used that paper as the first
chapter of a book. Both of them got a lot of attention, and led to his
appointment to the Harvard faculty.
The reason that I advised him to take this approach was quite simple:
If he didn't, other people would and they would get credit for his
work because they were able to apply it to real-world questions. I
think that other economists, especially young game theorists, can
learn from this example. Even very technical work often can be used in
an applied manner - and this can benefit the work as well as the
RF: In 1950, few people would have predicted, I think, that the Cold War
would end as peacefully as it did. For example, it is surely notable that
the conflict ended without the use of nuclear weapons. Why do you think both
sides avoided using means that would have had fairly certain, but
Schelling: I have written and lectured about this quite a bit. When I
give a talk on the subject, I begin by stating, "The most important
event of the second half of the 20th century is one that didn't
happen." I think you have to go through the history to understand it
fully. In the early 1950s, it was believed that the likelihood of the
United States using nuclear weapons was so great that the Prime
Minister of Great Britain came to Washington with the express purpose
of persuading the Truman administration not to use them. And because
the British had been partners in the development of nuclear weapons,
their Parliament thought that the Prime Minister had a good right to
share in any decision about how they would be used.
As we know, they were not used, but the Eisenhower administration
repeatedly asserted that nuclear weapons were just like any other type
of weapon, and that they could be used as such. The attitude in the
Kennedy and Johnson administrations was quite different. They believed
that nuclear weapons were fundamentally different, and their
statements helped to build the consensus that their use was taboo - a
consensus that may have dissuaded Nixon from using them in Vietnam.
Also, in the 1960s there was a great fear that dozens of countries
would come to possess nuclear weapons. But the nonproliferation
efforts were vastly more successful than most people expected. It was
thought that Germany was bound to demand them, and that the Japanese
couldn't afford to be without them. And then it would spiral down to
other countries: the Spanish, the Italians, the Swedes, the South
Africans, the Brazilians would all have nuclear weapons. The process
by which these countries would acquire them, it was thought, was
through nuclear electric power - the reactors would produce enough
plutonium to yield weapons. For several reasons, that didn't occur.
Israel's restraint in the 1973 war was also very important, I think.
Everyone knew that Golda Meir had nuclear weapons, and she had perfect
military targets - two Egyptian armies north of the Suez Canal, with
no civilians anywhere near. But she didn't use them. Why? Well, you
could say, quite reasonably, that they didn't want to suffer worldwide
opprobrium. I think, though, that there was probably another reason.
She knew that if she did, the Iranians, the Syrians, and other enemies
of Israel would likely acquire them and would not be reluctant to use
them. In addition, it was not clear in the late 1970s that the Soviets
shared the nuclear taboo. Yet, they didn't use them in their war
against Afghanistan - and this was also very important.
There is a possibility that nuclear weapons will be used in the
India-Pakistan dispute. But I'm not especially worried about that. The
Indians and the Pakistanis have been involved in nuclear strategic
discussions in the West for decades. They have had a long time to
think about this, and have watched the U.S.-Soviet negotiations. I
think they know that if they were to use nuclear weapons it could
easily lead to something beyond their control. So I think that by now
the taboo is so firmly entrenched, that it is very unlikely we will
see nation-states use nuclear weapons. What we don't know is if that
taboo holds for non-state actors. I think that it might, but I don't
hold that opinion with much conviction.
RF: Some policymakers and analysts have argued that diplomacy is much more
difficult in today's world than it was during the Cold War because there are
now multiple non-state players who seem to place less value on stability
than the Soviets did. How does this change the bargaining game? How can
economics inform the current conflict with Islamic terrorists?
Schelling: One big difference is that you simply don't know who the
non-state actors are. We have made a big deal out of Osama bin Laden.
But we don't know if he is alive, and if he is alive, whether he still
controls the money and organization in the way that he did a few years
ago. Also, there are no recognized private channels of communication
with non-state actors. If you want to get a message to bin Laden, you
either hold a press conference and hope that he will hear it, or send
it to him through a secret private channel.
Also, there is a popular notion that deterrence will not work when you
are dealing with non-state actors. But I'm not so sure that this is
the case. Consider the Taliban. I think that if the leaders of the
Taliban had known what type of response the attacks of Sept. 11 would
produce from the United States, they would have tried to prevent the
attacks. So I think that we should consider what we can do to alienate
bin Laden from some of his supporters. You also need to consider what
types of weapons they are likely to use and what types of targets they
are likely to choose. And we need to determine their objectives.
For instance, we still don't know what the objectives were of the
attacks on the World Trade Center, because the effects were so
widespread. It killed a lot of people. It produced the largest media
coverage of a terrorist attack in history. It demonstrated U.S.
vulnerability, while also destroying a symbol of Western capitalism.
And it demonstrated the competence and some would say the bravery of
the terrorists who were willing to sacrifice themselves. Each of those
could have been the principal objective, or there could have been some
combination of objectives. But we don't know for sure.
When we think about weapons, many people seem to think that terrorists
will use whatever weapon they can get their hands on. But consider the
use of, say, smallpox from a cost-benefit analysis. They could release
smallpox in New York, Chicago, and San Francisco. But smallpox is a
very difficult disease to contain in a world of global travel, and the
United States is the country best equipped to deal with an outbreak.
Releasing smallpox in the United States, then, could result in many
more deaths in poor countries with relatively bad health systems like
Indonesia and Pakistan than in the United States. I'm not sure that
would be a result the terrorists would welcome. By unleashing such
widespread death in the developing world - especially in places where
they enjoy support today - they could substantially reduce their
approval and assistance from people who are now their allies. In
contrast, anthrax might be a more attractive option because it is not
contagious, and its effects could be limited to the United States.
Also, there may be a cultural aspect to this. If releasing a
noncontagious toxin in, say, a subway station is considered by large
parts of Islamic culture to be a cowardly way to attack your enemy,
then this could be costly to them. It could damage their support in
the same way that releasing a contagious toxin could, even though the
effects of the actual attack would be much more direct and localized.
RF: What do you think have been the greatest diplomatic successes and
failures of the past 50 years?
Schelling: I think the great diplomatic success of the 20th century
was the way the Marshall Plan morphed into NATO. Essentially, the
cooperative arrangements between the Marshall Plan countries and the
United States were absorbed when NATO was formed, and the good will
was maintained. As a result, the United States was able to maintain
excellent diplomatic relations with the other major Western powers for
roughly 50 years. But recently that has started to unravel, and it's
going to be very hard to get back the sort of camaraderie and mutual
respect that we had built up.
As for other challenges, I think relations with Russia are much more
complicated than they ever were with the Soviet Union. The power
structure within the Kremlin may have been more complex than we
understood at the time, but that didn't really affect the way we
conducted diplomacy. I don't know what to say about the
Israeli-Palestinian issue. I think that one of the greatest tragedies
for diplomacy in the last few decades was the assassination of Prime
Minister Rabin. The prospects for peace would have been much better, I
think, had that not occurred.
There is one additional observation I would like to make. I think that
technological changes have made diplomacy more centralized.
Television, for instance, makes it possible for the U.S. Secretary of
State to speak directly to billions of people around the world when
she holds a press conference. Similarly, improved air travel makes it
much easier for ambassadors to travel home to get instructions from
RF: I would like to talk about your famous checkerboard example as it
applies to racial segregation. You have written, "A moderate urge to avoid
small-minority status may cause a nearly integrated pattern to unravel, and
highly segregated neighborhoods to form." Could you describe how this
Schelling: When I started thinking about this question, many American
neighborhoods were either mostly white or mostly black. One possible
explanation for this, of course, was rampant racism. But I was curious
about how this might emerge in a world where racism was not
particularly acute, where in fact people might prefer racial
The process works basically like this. Let's say the racial
composition of a neighborhood is 55 percent white and 45 percent
black, and that the majority population in the surrounding areas is
utterly without prejudice. Then you may get a case where more and more
members of the majority group move in. This may be fine with the
minority group for a while. They may not mind going from being 45
percent of the population to 35 percent. But at some point - say, when
their part of the population is only 20 percent - then the most
sensitive members of that group will probably evacuate, reducing their
percentage even further. The result is a highly segregated
neighborhood, even though this wasn't the intent of the majority
I wanted to come up with an easily understandable mechanism to explain
this phenomenon that I could use in teaching a class. I spent several
summers at the RAND Corporation, which had a good library. I looked at
several sociological journals, trying to find something I could use,
but I wasn't able to find anything suitable. So I decided I would have
to do something myself.
One day, I was flying home from somewhere and had nothing to read. So
I passed the time by putting little "X"s and "O"s in a line, with one
group representing whites and the other representing blacks, and used
the assumption that there was a moderate desire to avoid becoming part
of a very small minority group. Well, it turned out that this exercise
was very hard to do on paper, because you had to keep erasing and
But my son had a coin collection at the time, and he had a bunch of
copper coins and a bunch of zinc coins. I laid them out, and then I
decided that putting them in a line wasn't good enough. You needed
more dimensions. So I arranged them on a checkerboard. I got my
12-year-old son to sit down at the coffee table with me, and we would
move things around. Soon, we got quite used to how it worked and how
different the results were if one group was more discriminating than
the other or if one group was more numerous than the other.
I published my results, and it got quite a bit of attention at the
time. But it wasn't until 25 years later that I realized that this
game had pioneered some of the work in what is called "agent-based
modeling" and which is used in a variety of disciplines in the social
sciences. At the time I was working out this example I didn't realize
that I was engaged in an area of research that would one day have a
RF: You have done some research on crime. Why do you think some types of
criminal activity become organized while others do not?
Schelling: Part of this is semantic. Let's say you have a group of
automobile thieves. They may be organized, but we don't call that
"organized crime." Instead, when we use that term we are almost always
referring to a small group of activities: gambling, prostitution, and
drugs are the big ones. My question was: What is it that characterizes
those things we call "organized crime"? The answer is that they all
exist as monopolies. There is strong demand for each of the activities
I mentioned before, but each of them is illegal. So the people who
work in those markets are relatively easy to extort because they
cannot turn to the police. As a result, it is possible to gain
something approaching monopoly power in those markets. So the
bookmakers, prostitutes, and drug dealers are not really the
perpetrators of organized crime. They are the victims.
When I looked at the issue more, though, I found that there were some
markets that were legal but which were also characterized by high
levels of extortion. Two were small laundry services and restaurants.
What did they have in common? They were, at that time at least, mostly
cash businesses and without well-documented accounting practices. So
the proprietors of those businesses could pay the money under the
table to the extortionist. And if they didn't, the extortionists would
break their legs. Furthermore, if those businesses knew that their
competitors in that market were being similarly extorted and thus
realized that they were not being placed at a competitive
disadvantage, they had less of an incentive to turn to the police.
RF: How did you become involved with the Copenhagen Consensus and what type
of policy proposals has the group offered?
Schelling: I don't know precisely why I was chosen. Bjorn Lomborg, the
organizer of the project, wanted to gather a group of economists of
some reputation, and he probably knew that I had written about the
greenhouse gas issue. So that was probably the connection.
When the project started we had a United Nations list of global
problems related mostly to development and poverty. We were asked to
look over that list and pick 10 that we thought would be worth
pursuing. We did that, and then we asked a very distinguished person
in that field to write a major paper on the issue, along with two
other people to write critiques of the paper.
Somewhere along the way, we began to emphasize an idea that wasn't
clear to me at the outset and that I think wasn't clear to many other
people - namely, that this was mainly a budget priority exercise. We
were supposed to do cost-benefit analysis. We were told that we had
$50 billion to spend, and we should decide which projects would
provide the most welfare benefit for the money.
Unfortunately, that approach had not governed our choice of projects
and had not governed the way the papers were written. For instance, no
one really had a good idea of what you could do with some part of $50
billion to generate more liberal trade. The same was true with
education. The papers argued that unless you can reform the
educational systems in the big industrialized countries, more money
won't help. Similarly, it wasn't clear to us how more money would help
us prevent the spread of financial crises. So we had about five topics
that really did not fit, and we treated many of them as not
applicable. In retrospect, I think we should have treated climate
change in the same way.
Of those projects where we could see how the expenditure of money
would help, restricting the spread of HIV and AIDS seemed like it
should be at the top of the list. It is just so crucially important
that we advocated spending about half of the money on it. Then there
were some projects, like malnutrition and malaria control, where you
just got so much for your money, that we put them near the top also.
Projects to improve sanitation also were deemed quite worthwhile.
In general, I think that the program was successful in some ways and
less successful in others. And if we had it to do all over again, I
think that we could do an awful lot better.
RF: How did you come to the University of Maryland?
Schelling: In the 1980s, Congress passed a law making it illegal for
most businesses to have a mandatory retirement age for most employees.
But they allowed colleges and universities a seven-year grace period.
Harvard, at the time, had mandatory retirement at 70, and I was going
to be 70 before the grace period expired. Well, I was in good health,
felt that there was more research that I wanted to do, and still
enjoyed teaching. So I let it be known that I could be attracted to
another university. My first preference was a university in Southern
California, where I grew up. But then a former colleague and a very
good friend of mine who was dean of the University of Maryland's
School of Public Affairs called, and I told him about my situation. He
asked me not to accept another offer until I heard from him. It also
turned out that the chairman of the economics department had been my
teaching fellow at Harvard in the 1960s. So I had two very close
connections at Maryland, and I also knew a few other people on the
faculty, like Mancur Olson. Plus, as we have discussed, much of my
work is very policy-oriented, which made the Washington area pretty
desirable to me. Overall, it seemed like this would be a good fit for
me, so when the president of the university made me a very generous
offer, I accepted it. I have been at Maryland since 1990. I still
teach a class or two, but I am now in an emeritus position.
More information about the paleopsych