Introduction to Market Design,

Econ 2056a and Business 4150:

Professors Al Roth and Peter Coles,
Fall 2011
aroth@hbs.edu, pcoles@hbs.edu
TF: Assaf Romm: aromm@hbs.edu

Some useful websites

Course web page: for syllabus, including links to
reading, and for weekly handouts (including
these slides):
Economics 2056a : Market Design
http://isites.harvard.edu/icb/icb.do?keyword=k80599

My market design web page (for general

background and papers):
http://kuznets.fas.harvard.edu/~aroth/alroth.html
Market design blog:
http://marketdesigner.blogspot.com/

Assignment

One final paper. The objective of the final paper is to

study an existing market or an environment with a
potential role for a market, describe the relevant
market design questions, and evaluate how the current
market design works and/or propose improvements on
the current design.
In the past, these have varied widely; some have been
explorations of mathematical models, some have been full
of institutional description

Well ask you for brief descriptions of your preliminary

ideas from time to time.
The Market Design blog is intended in part to help
generate ideas.
From time to time well also pose exercises

Recommended texts
Roth, Alvin E. and Marilda Sotomayor Two-Sided
Matching: A Study in Game-Theoretic Modeling and
Analysis, Econometric Society Monograph Series,
Cambridge University Press, 1990. (get the paperback
edition.)
Milgrom, Paul "Putting Auction Theory to Work" by
Paul Milgrom (Churchill Lectures), Cambridge
University Press, 2004
Klemperer, Paul "Auctions: Theory and Practice"
(Toulouse Lectures), Princeton University Press, 2004.

Occasional lunches
After about half the class sessions (judging
from past experience), Peter and I will
announce that we are free for lunch
afterwards with anyone interested.
In past years weve normally walked over to
Finale in the Holyoke Center

Lightning overview of the course

Design is both a verb and a noun, and well
approach market design both as an activity and
as an aspect of markets that we study.
The course will have both substantive and
methodological themes.
Design also comes with a responsibility for detail.
Designers cant be satisfied with simple models
that explain the general principles underlying a
market; they have to be able to make sure that all
the detailed parts function together.

Methodology
Responsibility for detail requires the ability to
deal with complex institutional features that may
be omitted from simple models.
Game theory, the part of economics that studies
the rules of the game, provides a framework
with which design issues can be addressed.
But dealing with complexity will require new
tools, to supplement the analytical toolbox of the
traditional theorist.

 Game Theory, experimentation, and

computation, together with careful observation
of historical and contemporary markets (with
particular attention to the market rules), are
complementary tools of Design Economics

Computation helps us find answers that are

beyond our current theoretical knowledge.
Experiments play a role
In diagnosing and understanding market failures, and
successes
In designing new markets
In communicating results to policy makers

An analogy
Consider the design of suspension bridges. Their simple
physics, in which the only force is gravity, and all beams are
perfectly rigid, is simple, beautiful and indispensable.
But bridge design also concerns metal fatigue, soil
mechanics, and the sideways forces of waves and wind.
Many questions concerning these complications cant be
answered analytically, but must be explored using physical
or computational models.

These complications, and how they interact with that part

of the physics captured by the simple model, are the
concern of the engineering literature. Some of this is less
elegant than the simple model, but it allows bridges
designed on the same basic model to be built longer and
stronger over time, as the complexities and how to deal
with them become better understood.

 In this class, the simple models will be models of matching,

and of auctions.

In recent years there have been some great advances in the

theory of each of these, that brings them much closer
together.
A lot of these theoretical insights have come from the
difficulties faced in designing complex labor markets and
auctions (e.g. labor markets in which there may be twocareer households, and auctions in which bidders may wish
to purchase packages of goods).
(Paul Milgrom and his colleagues, and former students in
this class, have led the way on much of the recent
theoretical work.)

Substantive lessons from market failures

and successes
To achieve efficient outcomes, marketplaces need make
markets sufficiently
Thick
Enough potential transactions available at one time

Uncongested
Enough time for offers to be made, accepted, rejected

Safe
Safe to act straightforwardly on relevant preferences

Some kinds of transactions are repugnant

This can be an important constraint on market design
11

Some examples
Kidney exchange (thickness, congestion, incentives)
New England and Ohio (2005)
National US (2010??)
Repugnant: monetary markets

Medical labor markets

NRMP in 1995 (thickness, congestion, incentives)

Gastroenterology in 2006 (thickness, incentives)
Is reneging on early acceptances repugnant?

School choice systems:

New York City since Sept. 2004 (congestion & incentives)

Boston since Sept. 2006 (incentives)

Repugnant: exchange of priorities (particularly sibling priorities)

Denverpresently underway (DC, Chicago and/or New

Orleans maybe soon)

American market for new economists

Scramble ((thickness) March 2006
Signaling (congestion) December 2007

12

Kidney exchange--background
There are 89,603 patients on the waiting list for
cadaver kidneys in the U.S. (as of 9/1/11)
In 2010 34,418 patients were added to the waiting list,
and 27,775 patients were removed from the list.
In 2010 there were 10,622 transplants of cadaver
kidneys performed in the U.S.
In the same year, 4,652 patients died while on the
waiting list (and more than 2,110 others were removed
from the list as Too Sick to Transplant.
In 2010 there were also 6,276 transplants of kidneys
from living donors in the US.
Sometimes donors are incompatible with their
intended recipient.
This opens the possibility of exchange .
13

Simple Two Pair Kidney Exchange

Donor 1

Recipient 1

Blood type A

Blood type B

Donor 2

Recipient 2

Blood type B

Blood type A
14
4

Section 301 of the National Organ Transplant Act

(NOTA), 42 U.S.C. 274e 1984 states:
it shall be unlawful for any person
to knowingly acquire, receive or otherwise transfer
any human organ for valuable consideration
for use in human transplantation.
Legal opinion obtained by the transplant community interprets
this has forbidding buying and selling, but allowing exchange.

15

A classic economic problem:

Coincidence of wants
(Money and the Mechanism of Exchange, Jevons 1876)

Chapter 1: "The first difficulty in barter is to find two persons

whose disposable possessions mutually suit each other's
wants. There may be many people wanting, and many
possessing those things wanted; but to allow of an act of
barter, there must be a double coincidence, which will
rarely happen. ... the owner of a house may find it
unsuitable, and may have his eye upon another house
exactly fitted to his needs. But even if the owner of this
second house wishes to part with it at all, it is exceedingly
unlikely that he will exactly reciprocate the feelings of the
first owner, and wish to barter houses. Sellers and
purchasers can only be made to fit by the use of some
commodity... which all are willing to receive for a time, so
that what is obtained by sale in one case, may be used in
purchase in another. This common commodity is called a
medium, of exchange..."
16

Kidney exchange clearinghouse design

Roth, Alvin E., Tayfun Snmez, and M. Utku nver, Kidney
Exchange, Quarterly Journal of Economics, 119, 2, May,
2004, 457-488.
________started talking to docs________
____ Pairwise Kidney Exchange, Journal of Economic
Theory, 125, 2, 2005, 151-188.
___ A Kidney Exchange Clearinghouse in New England,
American Economic Review, Papers and Proceedings, 95,2,
May, 2005, 376-380.
_____ Efficient Kidney Exchange: Coincidence of Wants in
Markets with Compatibility-Based Preferences, American
Economic Review, June 2007, 97, 3, June 2007, 828-851
___multi-hospital exchanges become commonhospitals
become players in a new kidney game________
Ashlagi, Itai and Alvin E. Roth Individual rationality and
participation in large scale, multi-hospital kidney
exchange, working paper, January 2011.
17

And in the medical literature

Saidman, Susan L., Alvin E. Roth, Tayfun Snmez, M. Utku nver, and Francis L.
Delmonico, Increasing the Opportunity of Live Kidney Donation By
Matching for Two and Three Way Exchanges, Transplantation, 81, 5,
March 15, 2006, 773-782.
Roth, Alvin E., Tayfun Snmez, M. Utku nver, Francis L. Delmonico, and Susan
L. Saidman, Utilizing List Exchange and Undirected Donation through
Chain Paired Kidney Donations, American Journal of Transplantation,
6, 11, November 2006, 2694-2705.
Rees, Michael A., Jonathan E. Kopke, Ronald P. Pelletier, Dorry L. Segev,
Matthew E. Rutter, Alfredo J. Fabrega, Jeffrey Rogers, Oleh G. Pankewycz,
Janet Hiller, Alvin E. Roth, Tuomas Sandholm, Utku nver, and Robert A.
Montgomery, A Non-Simultaneous Extended Altruistic Donor Chain, New
England Journal of Medicine , 360;11, March 12, 2009, 1096-1101.
Ashlagi, Itai, Duncan S. Gilchrist, Alvin E. Roth, and Michael A. Rees,
Nonsimultaneous Chains and Dominos in Kidney Paired Donation
Revisited, American Journal of Transplantation, 11, 5, May 2011, 984994
18

Theres also a growing CS literature

Abraham, D., Blum, A., and Sandholm, T. 2007. Clearing
Algorithms for Barter Exchange Markets: Enabling Nationwide
Kidney Exchanges. In Proceedings of the ACM Conference on
Electronic Commerce (EC).
Ashlagi, Itai, Felix Fischer, Ian A. Kash, Ariel D. Procaccia,2010,
Mix and Match, EC10, June 711, 2010, Cambridge, MA.
Ashlagi, Itai, and Alvin E. Roth 2011 Participation (versus free
riding) in large scale, multi-hospital kidney exchange
Biro, Peter, and Katarina Cechlarova (2007), Inapproximability
of the kidney exchange problem, Information Processing
Letters, 101, 5, 16 March 2007, 199-202
Ioannis Caragiannis, Aris Filos-Ratsikas, and Ariel D. Procaccia.
An Improved 2-Agent Kidney Exchange Mechanism, July 2011.
19

Objectives of a kidney exchange

Thick market: assemble database of incompatible

patient-donor pairs
New England Program for Kidney Exchange (2005)
Alliance for Paired Donation (2006)
National matching (pilot now scheduled for Sept 2010)

Find efficient feasible matchings, in an incentive

compatible way (i.e. in a way that makes it safe for
everyone to reveal their private information and go
ahead with the resulting exchanges).
Prime incentive issues:
All transplants in an exchange performed simultaneously
transplant centers that can do a local exchange should
nevertheless submit pairs to the regional/national exchange.
20

Incentives: 2-way exchange involves 4

simultaneous surgeries.

21

Pair 1

Pair 4

Pair 2

Pair 3

The exchange P1-P2 results in two transplantations, but the

exchanges P1-P4 and P2-P3 results in four.
So even if Pairs 1 and 2 are at the same transplant center, it might
be good for them to nevertheless be submitted to a regional
match

22

Kidney overview
For kidney exchange, the big problem initially has
been establishing a thick market, and weve been
trying to do this while solving the new problems
concerning congestion and incentives that arise.
Some of these congestion problems involved the
logistics of surgery on the one hand, and the finding
of ways to reach the efficient frontier without
straining these logistics too much.
The initial strategic (incentive and equilibrium)
considerations we focused on involved individual
surgeons and their patients; now we are seeing
issues involving transplant centers as the players.
(Were scheduled to talk about kidney exchange and
related matters in detail late in the semester.)
23

Matching doctors to first positions in U.S.

and Canada
The redesign in 1995 of the
U.S. National Resident Matching Program (NRMP)
(approx. 23,000 positions, 500 couples)
Canadian Resident Matching Service (CaRMS)
(1,400 Canadian medical grads, including 41
couples, 1,500 positions in 2005)

The redesign in 2005 of the fellowship market

for Gastroenterologists
Contemporary issues in labor markets for
Orthopaedic surgeons, neuropsychologists, and
law clerks for appellate judges.
24

Background to redesign of the medical

clearinghouses
1900-1945 UNRAVELLING OF APPOINTMENT DATES
1945-1950 CHAOTIC RECONTRACTING--Congestion
1950-197x HIGH RATES OF ORDERLY PARTICIPATION
( 95%) in centralized clearinghouse
197x-198x DECLINING RATES OF PARTICIPATION
(85%) particularly among the growing number
of MARRIED COUPLES
1995-98
Market experienced a crisis of confidence with fears
of substantial decline in orderly participation;
Design effort commissionedto design and compare alternative
matching algorithms capable of handling modern requirements:
couples, specialty positions, etc.
Roth-Peranson clearinghouse algorithm adopted, and employed

25

Stages and transitions observed in various markets

Stage 1:
UNRAVELING
Offers are early, dispersed in
time,
explodingno
thick
market

Stage 2: UNIFORM DATES

ENFORCED
Deadlines, congestion

Stage 3:
CENTRALIZED MARKET

CLEARING PROCEDURES
26

What makes a clearinghouse successful or

unsuccessful?
A matching is stable if there arent a doctor and
residency program, not matched to each other, who
would both prefer to be.
Hypothesis: successful clearinghouses produce stable
matchings.
How to test this?

27

Gale, David and Lloyd Shapley [1962], Two-

Sided Matching Model

Men = {m1,..., mn}
Women = {w1,..., wp}
PREFERENCES (complete and transitive):
P(mi) = w3, w2, ... mi ...
P(wj) = m2, m4, ... wj ...

[w3 >mi w2]

Outcomes= matchings: :MW MW

such that w = (m) iff (w)=m,
And either (w) is in M or (w) = w, and
either (m) is in W or (m) = m
28

Gale, David and Lloyd Shapley [1962], Two-

Sided Matching Model

Men = {m1,..., mn}
Women = {w1,..., wp}
PREFERENCES (complete and transitive):
P(mi) = w3, w2, ... mi ...
P(wj) = m2, m4, ... wj ...

[w3 >mi w2]

Outcomes= matchings: :MW MW

such that w = (m) iff (w)=m,
And either (w) is in M or (w) = w, and
either (m) is in W or (m) = m
29

Stable matchings
A matching is
BLOCKED BY AN INDIVIDUAL k if k prefers being single to being
matched with (k), i.e.
k >k (k)
((k) is unacceptable).
BLOCKED BY A PAIR OF AGENTS (m,w) if they each prefer each
other to , i.e.

w >m (m) and m >w (w)

A matching is STABLE if it isn't blocked by any individual or
pair of agents.
NB: A stable matching is efficient, and in the core, and in this
simple model the set of (pairwise) stable matchings equals
the core.
30

Stable matchings
A matching is
BLOCKED BY AN INDIVIDUAL k if k prefers being single to being
matched with (k), i.e.
k >k (k)
((k) is unacceptable).
BLOCKED BY A PAIR OF AGENTS (m,w) if they each prefer each
other to , i.e.

w >m (m) and m >w (w)

A matching is STABLE if it isn't blocked by any individual or
pair of agents.
NB: A stable matching is efficient, and in the core, and in this
simple model the set of (pairwise) stable matchings equals
the core.
31

Market

Stable

NRMP
yes
Edinburgh ('69)
yes
Cardiff
yes
Birmingham
no
Edinburgh ('67)
no
Newcastle
no
Sheffield
no
Cambridge
no
London Hospital
no
Medical Specialties
yes
Canadian Lawyers
yes
Dental Residencies
yes
Osteopaths (< '94)
no
Osteopaths (> '94)
yes
Pharmacists
yes
Reform rabbis
yes (first used in 97-98)
Clinical psych
yes (first used in 99)

Lab experiments
(Kagel&Roth QJE 2000)

Still in use (halted unraveling)

yes (new design in 98)
yes
yes
no
no
no
no
yes
yes
yes (~30 markets, 1 failure)
yes (Alberta, no BC, Ontario)
yes (5 ) (no 2)
no
yes
yes
yes
yes

yes
no

Stability is an important criterion for a successful clearinghouse.

yes.
no
32

GS Deferred Acceptance Algorithm, with men

proposing
0. If some preferences are not strict, arbitrarily break ties
1 a. Each man m proposes to his 1st choice (if he has any
acceptable choices).
b. Each woman rejects any unacceptable proposals and, if
more than one acceptable proposal is received, "holds" the
most preferred and rejects all others.
k a. Any man rejected at step k-1 makes a new proposal to its
most preferred acceptable mate who hasnt yet rejected him.
(If no acceptable choices remain, he makes no proposal.)
b. Each woman holds her most preferred acceptable offer to
date, and rejects the rest.
STOP: when no further proposals are made, and match each
woman to the man (if any) whose proposal she is holding.
33

GSs 2 Remarkable Theorems

Theorem 1 (GS): A stable matching exists for every
marriage market.
Theorem 2 (GS): When all men and women have
strict preferences, there always exists an M-optimal
stable matching (that every man likes at least as well
as any other stable matching), and a W-optimal
stable matching. Furthermore, the matching M
produced by the deferred acceptance algorithm with
men proposing is the M-optimal stable matching.
The W-optimal stable matching is the matching W
produced by the algorithm when the women
propose.
34

IncentivesMany to one matching

The 1952 NIMP algorithm is equivalent to the

hospital-proposing deferred acceptance algorithm,
i.e. it produces the hospital-optimal stable matching
(Roth 84). (see todays homework for a related
problem
Many-to-one matching (Roth, 1985) No stable
matching mechanism exists that makes it a
dominant strategy for all hospitals to state their
true preferences, although the student-proposing
deferred acceptance algorithm makes it a dominant
strategy for all students to state their true
preferences.
Capacity manipulation (Snmez, 1997) No stable
matching mechanism makes it a dominant strategy
for a hospital to always reveal its capacity.
35

Observation and theory

Empirical Observation (Roth and Peranson, 1999):
The set of stable matchings is small, as is the set of
people who can potentially manipulate (about 1 in
1,000).

36

Some new theory

Theorem (Immorlica and Mahdian, SODA 2005:) Consider a
marriage model with n men and n women, in which each
woman has an arbitrary complete preference list, and each
man has a random list of at most k women as his preference
list (chosen uniformly and independently). Then, the expected
number of women who have more than one stable husband is
bounded by a constant that only depends on k (and not on n).
(So, as n gets large, the proportion of such women goes to
zero)
Theorem (Kojima and Pathak, AER 2009): In the limit, as n
goes to infinity in a regular sequence of random (many-toone) markets, the proportion of employers who might profit
from (any combination of) preference or capacity
manipulation goes to zero in the worker proposing deferred
acceptance algorithm.
37

Current NRMP match

(Roth/Peranson algorithm)

Produces student optimal stable matching (as

close as can be given match complications)
Deals with major match complications
Married couples
They can submit preferences over pairs of positions

Applicants can match to pairs of jobs, PGY1&2

They can submit supplementary preference lists

Reversions of positions from one program to

another
The algorithm starts as a student (and couple)- proposing
deferred acceptance algorithm, and then resolves instabilities
with an algorithm modeled on the Roth-Vande Vate (1990)
blocking-pair-satisfying algorithm (which arose in part in
response to an open problem by Knuth)
38

Stable Clearinghouses (those now using the Roth Peranson Algorithm)

NRMP / SMS:
Medical Residencies in the U.S. (NRMP) (1952)
Abdominal Transplant Surgery (2005)
Child & Adolescent Psychiatry (1995)
Colon & Rectal Surgery (1984)
Combined Musculoskeletal Matching Program (CMMP)

Hand Surgery (1990)

Medical Specialties Matching Program (MSMP)

Cardiovascular Disease (1986)

Hematology (2006)
Hematology/Oncology (2006)
Infectious Disease (1986-1990; rejoined in 1994)
Oncology (2006)
Pulmonary and Critical Medicine (1986)
Rheumatology (2005)

Primary Care Sports Medicine (1994)

Radiology

Interventional Radiology (2002)

Neuroradiology (2001)
Pediatric Radiology (2003)

Surgical Critical Care (2004)

Thoracic Surgery (1988)
Vascular Surgery (1988)

Gastroenterology (1986-1999; rejoined Postdoctoral Dental Residencies in the United States

in 2006)

Oral and Maxillofacial Surgery (1985)

Minimally Invasive and Gastrointestinal Surgery (2003)

Obstetrics/Gynecology

Reproductive Endocrinology (1991)

Gynecologic Oncology (1993)
Maternal-Fetal Medicine (1994)
Female Pelvic Medicine & Reconstructive Surgery (2001)

Ophthalmic Plastic & Reconstructive Surgery (1991)

Pediatric Cardiology (1999)
Pediatric Critical Care Medicine (2000)
Pediatric Emergency Medicine (1994)
Pediatric Hematology/Oncology (2001)
Pediatric Rheumatology (2004)
Pediatric Surgery (1992)

General Practice Residency (1986)

Advanced Education in General Dentistry (1986)
Pediatric Dentistry (1989)
Orthodontics (1996)

Psychology Internships in the U.S. and CA (1999)

Neuropsychology Residencies in the U.S. & CA (2001)
Osteopathic Internships in the U.S. (before 1995)
Pharmacy Practice Residencies in the U.S. (1994)
Articling Positions with Law Firms in Alberta, CA(1993)
Medical Residencies in CA (CaRMS) (before 1970)
********************

British (medical) house officer positions

Edinburgh (1969)
Cardiff (197x)

New York City High Schools (2003)

Boston Public Schools (2006)

39

Market Design Manifesto

We need to understand how
markets work well enough to fix
them when theyre broken.

Homework exercise
Here is the web site of the American Association of Colleges
of Podiatric Medicine
http://www.casprcrip.org/html/casprcrip/students.asp
They run a match, and here is the description of their
algorithm:
http://www.casprcrip.org/html/casprcrip/pdf/MatchExpl.pdf
(this is also on the class web page)
Is their algorithm equivalent to the hospital proposing
deferred acceptance procedure?
Does it produce the same matching, when it produces a
matching?
Does it always (for every preference profile) produce a matching?
Is the description of the algorithm complete enough to be sure?

Come prepared to give an answer at the beginning of next

weeks class (note that the podiatry algorithm is a many-toone match like the college admissions problem, not a oneto-one match like the marriage problemyou may want to
look ahead in the notes)