School choice: further developments

1/20
Introduction

The school choice model seems to capture well the real-life

problem of assigning students to schools.

But the implementation of assignment algorithms (Deferred

Acceptance) raised new issues.

2/20
Introduction

The school choice model seems to capture well the real-life

problem of assigning students to schools.

But the implementation of assignment algorithms (Deferred

Acceptance) raised new issues.

2/20
Weak priorities

In the basic school choice model school’s priorities are assumed to

be strict. This assumption is difficult to justify.

Priorities are generally set by policy makers, following simple

criteria, defining broad categories like:
I students with a sibling in the school;
I students living in the “walking zone”
I students from specific socio-economic or ethnic groups

⇒ students in the same tier have the same priority.

3/20
Weak priorities

In the basic school choice model school’s priorities are assumed to

be strict. This assumption is difficult to justify.

Priorities are generally set by policy makers, following simple

criteria, defining broad categories like:
I students with a sibling in the school;
I students living in the “walking zone”
I students from specific socio-economic or ethnic groups

⇒ students in the same tier have the same priority.

3/20
Weak priorities

In the basic school choice model school’s priorities are assumed to

be strict. This assumption is difficult to justify.

Priorities are generally set by policy makers, following simple

criteria, defining broad categories like:
I students with a sibling in the school;
I students living in the “walking zone”
I students from specific socio-economic or ethnic groups

⇒ students in the same tier have the same priority.

3/20
Weak priorities

In the basic school choice model school’s priorities are assumed to

be strict. This assumption is difficult to justify.

Priorities are generally set by policy makers, following simple

criteria, defining broad categories like:
I students with a sibling in the school;
I students living in the “walking zone”
I students from specific socio-economic or ethnic groups

⇒ students in the same tier have the same priority.

3/20
Weak priorities

In the basic school choice model school’s priorities are assumed to

be strict. This assumption is difficult to justify.

Priorities are generally set by policy makers, following simple

criteria, defining broad categories like:
I students with a sibling in the school;
I students living in the “walking zone”
I students from specific socio-economic or ethnic groups

⇒ students in the same tier have the same priority.

3/20
An example of a weak priority for a school s:

Ps
Alice, Bob
Carol
Denis, Erin, Fred
Gilda

I Alice and Bob have a higher priority than any other student.
But Alice (Bob) doesn’t have a higher priority than Bob
(Alice).
I Carol has
I lower priority than Alice and Bob
I higher priority than Denis, Erin, Fred and Gilda.

4/20
An example of a weak priority for a school s:

Ps
Alice, Bob
Carol
Denis, Erin, Fred
Gilda

I Alice and Bob have a higher priority than any other student.
But Alice (Bob) doesn’t have a higher priority than Bob
(Alice).
I Carol has
I lower priority than Alice and Bob
I higher priority than Denis, Erin, Fred and Gilda.

4/20
An example of a weak priority for a school s:

Ps
Alice, Bob
Carol
Denis, Erin, Fred
Gilda

I Alice and Bob have a higher priority than any other student.
But Alice (Bob) doesn’t have a higher priority than Bob
(Alice).
I Carol has
I lower priority than Alice and Bob
I higher priority than Denis, Erin, Fred and Gilda.

4/20
To run an algorithm like DA we need strict priorities. Ties can be
(for instance) broken randomly.

But doing so may be inefficient.

Before explaining this, we need to take into account that stability is

defined with respect to the original priorities (before breaking ties).

π̄s = [Alice, Bob], Carol, [Denis, Erin, Fred], Gilda

School s has one seat, it’s Alice’s most preferred school.

I Alice can block if Denis is assigned to s: she has a strictly

higher priority.
I Alice cannot block if Bob is assigned to s: she has the same
priority as him.

5/20
To run an algorithm like DA we need strict priorities. Ties can be
(for instance) broken randomly.

But doing so may be inefficient.

Before explaining this, we need to take into account that stability is

defined with respect to the original priorities (before breaking ties).

π̄s = [Alice, Bob], Carol, [Denis, Erin, Fred], Gilda

School s has one seat, it’s Alice’s most preferred school.

I Alice can block if Denis is assigned to s: she has a strictly

higher priority.
I Alice cannot block if Bob is assigned to s: she has the same
priority as him.

5/20
To run an algorithm like DA we need strict priorities. Ties can be
(for instance) broken randomly.

But doing so may be inefficient.

Before explaining this, we need to take into account that stability is

defined with respect to the original priorities (before breaking ties).

π̄s = [Alice, Bob], Carol, [Denis, Erin, Fred], Gilda

School s has one seat, it’s Alice’s most preferred school.

I Alice can block if Denis is assigned to s: she has a strictly

higher priority.
I Alice cannot block if Bob is assigned to s: she has the same
priority as him.

5/20
To run an algorithm like DA we need strict priorities. Ties can be
(for instance) broken randomly.

But doing so may be inefficient.

Before explaining this, we need to take into account that stability is

defined with respect to the original priorities (before breaking ties).

π̄s = [Alice, Bob], Carol, [Denis, Erin, Fred], Gilda

School s has one seat, it’s Alice’s most preferred school.

I Alice can block if Denis is assigned to s: she has a strictly

higher priority.
I Alice cannot block if Bob is assigned to s: she has the same
priority as him.

5/20
To run an algorithm like DA we need strict priorities. Ties can be
(for instance) broken randomly.

But doing so may be inefficient.

Before explaining this, we need to take into account that stability is

defined with respect to the original priorities (before breaking ties).

π̄s = [Alice, Bob], Carol, [Denis, Erin, Fred], Gilda

School s has one seat, it’s Alice’s most preferred school.

I Alice can block if Denis is assigned to s: she has a strictly

higher priority.
I Alice cannot block if Bob is assigned to s: she has the same
priority as him.

5/20
Efficiency loss

PAlice PBob PCarol

s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

Suppose we break ties with the following order:

Alice, Bob, Carol

So for s1 after breaking ties Bob has a strictly higher priority than
Carol
6/20
Efficiency loss

PAlice PBob PCarol

s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

Suppose we break ties with the following order:

Alice, Bob, Carol

So for s1 after breaking ties Bob has a strictly higher priority than
Carol
6/20
Efficiency loss

PAlice PBob PCarol

s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

Suppose we break ties with the following order:

Alice, Bob, Carol

So for s1 after breaking ties Bob has a strictly higher priority than
Carol
6/20
 PAlice PBob PCarol πs1 πs2 πs3
s2 s3 s2 Alice Bob Carol
s1 s2 s3 Bob Alice Alice
s3 s1 s1 Carol Carol Bob

Running DA (students proposing) we obtain

µ(Alice) = s1 , µ(Bob) = s2 , and µ(Carol) = s3 .

7/20
 PAlice PBob PCarol
s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

µ(Alice) = s1 , µ(Bob) = s2 , and µ(Carol) = s3 .

But the following is also a stable matching (preferred by Bob and

Carol):

µ0 (Alice) = s1 , µ0 (Bob) = s3 , and µ0 (Carol) = s2 .

8/20
 PAlice PBob PCarol
s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

µ(Alice) = s1 , µ(Bob) = s2 , and µ(Carol) = s3 .

But the following is also a stable matching (preferred by Bob and

Carol):

µ0 (Alice) = s1 , µ0 (Bob) = s3 , and µ0 (Carol) = s2 .

8/20
We now have two sources of inefficiency:
I Due to the conflict between stability and Pareto efficiency (in
general they’re not compatible);
I Weak priorities can generate an assignment that is not the
student-optimal assignment.

One way to restor efficiency (from the second source) is to allow

students trade their enrollements. . .
. . . So we can use the Top Trading Cycle algorithm!

But there’s a caveat: we may loose stability if the pointing is not

defined properly.

9/20
We now have two sources of inefficiency:
I Due to the conflict between stability and Pareto efficiency (in
general they’re not compatible);
I Weak priorities can generate an assignment that is not the
student-optimal assignment.

One way to restor efficiency (from the second source) is to allow

students trade their enrollements. . .
. . . So we can use the Top Trading Cycle algorithm!

But there’s a caveat: we may loose stability if the pointing is not

defined properly.

9/20
We now have two sources of inefficiency:
I Due to the conflict between stability and Pareto efficiency (in
general they’re not compatible);
I Weak priorities can generate an assignment that is not the
student-optimal assignment.

One way to restor efficiency (from the second source) is to allow

students trade their enrollements. . .
. . . So we can use the Top Trading Cycle algorithm!

But there’s a caveat: we may loose stability if the pointing is not

defined properly.

9/20
We now have two sources of inefficiency:
I Due to the conflict between stability and Pareto efficiency (in
general they’re not compatible);
I Weak priorities can generate an assignment that is not the
student-optimal assignment.

One way to restor efficiency (from the second source) is to allow

students trade their enrollements. . .
. . . So we can use the Top Trading Cycle algorithm!

But there’s a caveat: we may loose stability if the pointing is not

defined properly.

9/20
We now have two sources of inefficiency:
I Due to the conflict between stability and Pareto efficiency (in
general they’re not compatible);
I Weak priorities can generate an assignment that is not the
student-optimal assignment.

One way to restor efficiency (from the second source) is to allow

students trade their enrollements. . .
. . . So we can use the Top Trading Cycle algorithm!

But there’s a caveat: we may loose stability if the pointing is not

defined properly.

9/20
We now have two sources of inefficiency:
I Due to the conflict between stability and Pareto efficiency (in
general they’re not compatible);
I Weak priorities can generate an assignment that is not the
student-optimal assignment.

One way to restor efficiency (from the second source) is to allow

students trade their enrollements. . .
. . . So we can use the Top Trading Cycle algorithm!

But there’s a caveat: we may loose stability if the pointing is not

defined properly.

9/20
Improvement cycles

The idea (due to Erdil and Ergin) is to restrict the schools to

which a student can point.

I We start from an assignment.

I A student i can point to a school s only if:
I it is preferred to her assignment.
I Among all students who preferr s to their assignment student i
is among the highest priority students.

I If there’s a cycle, we perform trade and we start again until no

new trades are realized.

10/20
Improvement cycles

The idea (due to Erdil and Ergin) is to restrict the schools to

which a student can point.

I We start from an assignment.

I A student i can point to a school s only if:
I it is preferred to her assignment.
I Among all students who preferr s to their assignment student i
is among the highest priority students.

I If there’s a cycle, we perform trade and we start again until no

new trades are realized.

10/20
Improvement cycles

The idea (due to Erdil and Ergin) is to restrict the schools to

which a student can point.

I We start from an assignment.

I A student i can point to a school s only if:
I it is preferred to her assignment.
I Among all students who preferr s to their assignment student i
is among the highest priority students.

I If there’s a cycle, we perform trade and we start again until no

new trades are realized.

10/20
Improvement cycles

The idea (due to Erdil and Ergin) is to restrict the schools to

which a student can point.

I We start from an assignment.

I A student i can point to a school s only if:
I it is preferred to her assignment.
I Among all students who preferr s to their assignment student i
is among the highest priority students.

I If there’s a cycle, we perform trade and we start again until no

new trades are realized.

10/20
Improvement cycles

The idea (due to Erdil and Ergin) is to restrict the schools to

which a student can point.

I We start from an assignment.

I A student i can point to a school s only if:
I it is preferred to her assignment.
I Among all students who preferr s to their assignment student i
is among the highest priority students.

I If there’s a cycle, we perform trade and we start again until no

new trades are realized.

10/20
Improvement cycles

The idea (due to Erdil and Ergin) is to restrict the schools to

which a student can point.

I We start from an assignment.

I A student i can point to a school s only if:
I it is preferred to her assignment.
I Among all students who preferr s to their assignment student i
is among the highest priority students.

I If there’s a cycle, we perform trade and we start again until no

new trades are realized.

10/20
Example

PAlice PBob PCarol

s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

11/20
Example
PAlice PBob PCarol
s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

µ(Alice) = s1 , µ(Bob) = s2 , and µ(Carol) = s3 .

I Alice and Carol both want s2 . They have the same priority, so
they can point to Bob (enrolled at s2 ).
I Bob is the only one who want s3 , so he points to Carol.
11/20
Example
PAlice PBob PCarol
s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

µ(Alice) = s1 , µ(Bob) = s2 , and µ(Carol) = s3 .

I Alice and Carol both want s2 . They have the same priority, so
they can point to Bob (enrolled at s2 ).
I Bob is the only one who want s3 , so he points to Carol.
11/20
Example
PAlice PBob PCarol
s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

s2
Alice Bob
s3
s2

Carol
11/20
Example

PAlice PBob PCarol

s2 s3 s2
s1 s2 s3
s3 s1 s1

π̄s1 π̄s2 π̄s3

Alice Bob Carol
[Bob, Carol] [Alice, Carol] [Alice, Bob]

Then we get

µ0 (Alice) = s1 , µ0 (Bob) = s3 , and µ0 (Carol) = s2 .

No new trades are realized, we stop.

11/20
Theorem
If µ is a stable assignment and µ is Pareto dominated by another
assignment µ0 then there exists a stable improvement cycle.

Using data from NYC, Abdulkadiroğlu, Pathak and Roth find that
for the years 2003–2007 they can improve on average the
assignment of about 1,700 students (around 2.5% of the students).

12/20
Theorem
If µ is a stable assignment and µ is Pareto dominated by another
assignment µ0 then there exists a stable improvement cycle.

Using data from NYC, Abdulkadiroğlu, Pathak and Roth find that
for the years 2003–2007 they can improve on average the
assignment of about 1,700 students (around 2.5% of the students).

12/20
DA + improvement cycles is not strategyproof.

PAlice PBob PCarol π̄s1 π̄s2 π̄s3

s2 s2 s1 Alice Carol Carol
s3 s3 s2 Bob [Alice, Bob] Bob
s1 s1 s1 Carol Alice

There are two stable assignments:

µ(Alice) = s2 , µ(Bob) = s3 , and µ(Carol) = s1 .

0 0
µ (Alice) = s3 , µ (Bob) = s2 , and µ0 (Carol) = s1 .

Let
0
PAlice = s2 , s1 , s3
0
PBob = s2 , s1 , s3

13/20
DA + improvement cycles is not strategyproof.

PAlice PBob PCarol π̄s1 π̄s2 π̄s3

s2 s2 s1 Alice Carol Carol
s3 s3 s2 Bob [Alice, Bob] Bob
s1 s1 s1 Carol Alice

There are two stable assignments:

µ(Alice) = s2 , µ(Bob) = s3 , and µ(Carol) = s1 .

0 0
µ (Alice) = s3 , µ (Bob) = s2 , and µ0 (Carol) = s1 .

Let
0
PAlice = s2 , s1 , s3
0
PBob = s2 , s1 , s3

13/20
DA + improvement cycles is not strategyproof.

PAlice PBob PCarol π̄s1 π̄s2 π̄s3

s2 s2 s1 Alice Carol Carol
s3 s3 s2 Bob [Alice, Bob] Bob
s1 s1 s1 Carol Alice

There are two stable assignments:

µ(Alice) = s2 , µ(Bob) = s3 , and µ(Carol) = s1 .

0 0
µ (Alice) = s3 , µ (Bob) = s2 , and µ0 (Carol) = s1 .

Let
0
PAlice = s2 , s1 , s3
0
PBob = s2 , s1 , s3

13/20
For the profile
0
(PAlice , PBob , PCarol )
only µ is stable (not µ0 ).

For the profile

0
(PAlice , PBob , PCarol )
only µ0 is stable (not µ).

If the mechanism selects with the true profile:

I µ0 ⇒ Alice is better off lying (submitting PAlice
0 ).
I 0 ).
µ ⇒ Bob is better off lying (submitting PBob

14/20
For the profile
0
(PAlice , PBob , PCarol )
only µ is stable (not µ0 ).

For the profile

0
(PAlice , PBob , PCarol )
only µ0 is stable (not µ).

If the mechanism selects with the true profile:

I µ0 ⇒ Alice is better off lying (submitting PAlice
0 ).
I 0 ).
µ ⇒ Bob is better off lying (submitting PBob

14/20
For the profile
0
(PAlice , PBob , PCarol )
only µ is stable (not µ0 ).

For the profile

0
(PAlice , PBob , PCarol )
only µ0 is stable (not µ).

If the mechanism selects with the true profile:

I µ0 ⇒ Alice is better off lying (submitting PAlice
0 ).
I 0 ).
µ ⇒ Bob is better off lying (submitting PBob

14/20
For the profile
0
(PAlice , PBob , PCarol )
only µ is stable (not µ0 ).

For the profile

0
(PAlice , PBob , PCarol )
only µ0 is stable (not µ).

If the mechanism selects with the true profile:

I µ0 ⇒ Alice is better off lying (submitting PAlice
0 ).
I 0 ).
µ ⇒ Bob is better off lying (submitting PBob

14/20
For the profile
0
(PAlice , PBob , PCarol )
only µ is stable (not µ0 ).

For the profile

0
(PAlice , PBob , PCarol )
only µ0 is stable (not µ).

If the mechanism selects with the true profile:

I µ0 ⇒ Alice is better off lying (submitting PAlice
0 ).
I 0 ).
µ ⇒ Bob is better off lying (submitting PBob

14/20
How to break ties?

There are two options to break ties:

I Multiple tie-breaking:
Each school has its own tie-breaking.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I For s1 Alice ends up with a higher priority than Bob.
I For s2 Bob ends up with a higher priority than Alice.

I Single tie-breaking:
The tie-breaking is the same for all schools.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I If at s1 Alice ends up with a higher priority than Bob,
I then at s2 Alice ends up with a higher priority than Bob.

15/20
How to break ties?

There are two options to break ties:

I Multiple tie-breaking:
Each school has its own tie-breaking.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I For s1 Alice ends up with a higher priority than Bob.
I For s2 Bob ends up with a higher priority than Alice.

I Single tie-breaking:
The tie-breaking is the same for all schools.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I If at s1 Alice ends up with a higher priority than Bob,
I then at s2 Alice ends up with a higher priority than Bob.

15/20
How to break ties?

There are two options to break ties:

I Multiple tie-breaking:
Each school has its own tie-breaking.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I For s1 Alice ends up with a higher priority than Bob.
I For s2 Bob ends up with a higher priority than Alice.

I Single tie-breaking:
The tie-breaking is the same for all schools.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I If at s1 Alice ends up with a higher priority than Bob,
I then at s2 Alice ends up with a higher priority than Bob.

15/20
How to break ties?

There are two options to break ties:

I Multiple tie-breaking:
Each school has its own tie-breaking.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I For s1 Alice ends up with a higher priority than Bob.
I For s2 Bob ends up with a higher priority than Alice.

I Single tie-breaking:
The tie-breaking is the same for all schools.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I If at s1 Alice ends up with a higher priority than Bob,
I then at s2 Alice ends up with a higher priority than Bob.

15/20
How to break ties?

There are two options to break ties:

I Multiple tie-breaking:
Each school has its own tie-breaking.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I For s1 Alice ends up with a higher priority than Bob.
I For s2 Bob ends up with a higher priority than Alice.

I Single tie-breaking:
The tie-breaking is the same for all schools.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I If at s1 Alice ends up with a higher priority than Bob,
I then at s2 Alice ends up with a higher priority than Bob.

15/20
How to break ties?

There are two options to break ties:

I Multiple tie-breaking:
Each school has its own tie-breaking.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I For s1 Alice ends up with a higher priority than Bob.
I For s2 Bob ends up with a higher priority than Alice.

I Single tie-breaking:
The tie-breaking is the same for all schools.
Example: Alice and Bob are in the same tiers for schools s1
and s2 .
I If at s1 Alice ends up with a higher priority than Bob,
I then at s2 Alice ends up with a higher priority than Bob.

15/20
Multiple tie-breaking seems more fair.

π̄s = [Alice], [Bob, Carol, Denis, Erin, Fred], Gilda

and s has only 1 seat.
If breaking ties gives

πs = Alice, Bob, Carol . . .

Then if Bob wants s he only needs that Alice is not assigned to s.

But if breaking ties gives

πs0 = Alice, Carol, Denis, Erin, Fred, Bob, . . .

Then if Bob wants s he needs that Alice, Carol, Denis, Erin and
Fred are not assigned to s .

16/20
Multiple tie-breaking seems more fair.

π̄s = [Alice], [Bob, Carol, Denis, Erin, Fred], Gilda

and s has only 1 seat.
If breaking ties gives

πs = Alice, Bob, Carol . . .

Then if Bob wants s he only needs that Alice is not assigned to s.

But if breaking ties gives

πs0 = Alice, Carol, Denis, Erin, Fred, Bob, . . .

Then if Bob wants s he needs that Alice, Carol, Denis, Erin and
Fred are not assigned to s .

16/20
Multiple tie-breaking seems more fair.

π̄s = [Alice], [Bob, Carol, Denis, Erin, Fred], Gilda

and s has only 1 seat.
If breaking ties gives

πs = Alice, Bob, Carol . . .

Then if Bob wants s he only needs that Alice is not assigned to s.

But if breaking ties gives

πs0 = Alice, Carol, Denis, Erin, Fred, Bob, . . .

Then if Bob wants s he needs that Alice, Carol, Denis, Erin and
Fred are not assigned to s .

16/20
Multiple tie-breaking seems more fair.

π̄s = [Alice], [Bob, Carol, Denis, Erin, Fred], Gilda

and s has only 1 seat.
If breaking ties gives

πs = Alice, Bob, Carol . . .

Then if Bob wants s he only needs that Alice is not assigned to s.

But if breaking ties gives

πs0 = Alice, Carol, Denis, Erin, Fred, Bob, . . .

Then if Bob wants s he needs that Alice, Carol, Denis, Erin and
Fred are not assigned to s .

16/20
So, multiple tie-breaking seems to give equal chances to each
student.

Theorem
If µ is a stable assignment such that:
I µ can be obtained using a multiple tie-breaking rule
I µ cannot be obtained using a single tie-breaking rule
then that assignment is not a student-optimal assignment.

⇒ we better use single tie-breaking rules.

17/20
So, multiple tie-breaking seems to give equal chances to each
student.

Theorem
If µ is a stable assignment such that:
I µ can be obtained using a multiple tie-breaking rule
I µ cannot be obtained using a single tie-breaking rule
then that assignment is not a student-optimal assignment.

⇒ we better use single tie-breaking rules.

17/20
Simulations using data from NYC

Rank Single Multiple Student-optimal

in choice breaking breaking (efficient)
1 32,105.3 29,849.9 32,701.5
2 14,296.0 14,562.3 14,382.6
3 9,279.4 9,859.7 9,208.6
4 6,112.8 6,653.3 5,999.8
5 3,988.2 4,386.8 3,883.4
6 2,628.8 2,919.1 2,519.5
7 1,732.7 1,919.1 1,654.6
8 1,099.1 1,212.2 1,034.8
9 761.9 871.1 716.7
10 526.4 548.4 485.6
11 348.0 353.2 316.3
12 236.0 229.3 211.2
unassigned 5,613.4 5,426.7 5,613.4

18/20
Constrained choice

In many assignment/matching markets participants are

constrained: a maximum number of options for the submitted
preference lists.

I NYC high school match: 12 schools max.

I College admission in France: 20 programs max.
I College admission in Spain: 8 programs max.
Why?
I Legacy: matching originally made manually.
I Limit “no shows”: force people to put only options they care
about.
I Self constraint: difficult to rank too many options.

19/20
Constrained choice

In many assignment/matching markets participants are

constrained: a maximum number of options for the submitted
preference lists.

I NYC high school match: 12 schools max.

I College admission in France: 20 programs max.
I College admission in Spain: 8 programs max.
Why?
I Legacy: matching originally made manually.
I Limit “no shows”: force people to put only options they care
about.
I Self constraint: difficult to rank too many options.

19/20
Constrained choice

In many assignment/matching markets participants are

constrained: a maximum number of options for the submitted
preference lists.

I NYC high school match: 12 schools max.

I College admission in France: 20 programs max.
I College admission in Spain: 8 programs max.
Why?
I Legacy: matching originally made manually.
I Limit “no shows”: force people to put only options they care
about.
I Self constraint: difficult to rank too many options.

19/20
Constrained choice

In many assignment/matching markets participants are

constrained: a maximum number of options for the submitted
preference lists.

I NYC high school match: 12 schools max.

I College admission in France: 20 programs max.
I College admission in Spain: 8 programs max.
Why?
I Legacy: matching originally made manually.
I Limit “no shows”: force people to put only options they care
about.
I Self constraint: difficult to rank too many options.

19/20
If choice is constrained, no assignment mechanism can be
strategyproof.
But if the unconstrained mechanism is strategyproof, we keep
something.
Theorem
If students cannot put more than k schools in their preferences and
the algorithm used is the DA or TTC:

(a) If a student finds at most k schools acceptable, then she can

do not better than submitting her true preferences;

(b) If a student finds more than k schools acceptable, then she

can do no better than employing a strategy that selects k
schools among the acceptable schools and ranking them
according to her true preferences.

20/20
If choice is constrained, no assignment mechanism can be
strategyproof.
But if the unconstrained mechanism is strategyproof, we keep
something.
Theorem
If students cannot put more than k schools in their preferences and
the algorithm used is the DA or TTC:

(a) If a student finds at most k schools acceptable, then she can

do not better than submitting her true preferences;

(b) If a student finds more than k schools acceptable, then she

can do no better than employing a strategy that selects k
schools among the acceptable schools and ranking them
according to her true preferences.

20/20
From very to less manipulable

21/20