Efficient Assignment with Interdependent
Values
Yeon-Koo Che1 Jinwoo Kim2 Fuhito Kojima3
April 23, 2013
1Columbia University 2Seoul National University 3
Assignments without Money
Many resources are assigned without using money
I office assignment, student placement in public schools, university housing allocation, organ allocation, task assignments in organizations
Can we design a mechanism without money that is
efficientandincentive compatible?
Yes, if the agents know their own values of the objects; private values
I Serial dictatorships (Svensson, 99; Abdulkadiro ˘glu & S ¨onmez, 98), top trading cycles mechanisms
(Abdulkadiro ˘glu & S ¨onmez, 03), hierarchical exchanges (Papai, 00), trading cycles (Pycia & ¨Unver 11) ...
Assignments without Money
Many resources are assigned without using money
I office assignment, student placement in public schools, university housing allocation, organ allocation, task assignments in organizations
Can we design a mechanism without money that is
efficientandincentive compatible?
Yes, if the agents know their own values of the objects; private values
I Serial dictatorships (Svensson, 99; Abdulkadiro ˘glu & S ¨onmez, 98), top trading cycles mechanisms
(Abdulkadiro ˘glu & S ¨onmez, 03), hierarchical exchanges (Papai, 00), trading cycles (Pycia & ¨Unver 11) ...
Assignments without Money
Many resources are assigned without using money
I office assignment, student placement in public schools, university housing allocation, organ allocation, task assignments in organizations
Can we design a mechanism without money that is
efficientandincentive compatible?
Yes, if the agents know their own values of the objects; private values
I Serial dictatorships (Svensson, 99; Abdulkadiro ˘glu & S ¨onmez, 98), top trading cycles mechanisms
Interdependent Values
Values areinterdependentwhen one’s valuation of objects
depends on the information/signals possessed by others
E.g. in school choice, many parents have insufficient
information about the fitness of schools for their children
I Seek advice from other parents through word-of-mouth communication, online social networks ...
Are there incentive compatible mechanisms that implement
the efficient assignments in the interdependent values
Interdependent Values
Values areinterdependentwhen one’s valuation of objects
depends on the information/signals possessed by others
E.g. in school choice, many parents have insufficient
information about the fitness of schools for their children
I Seek advice from other parents through word-of-mouth communication, online social networks ...
Are there incentive compatible mechanisms that implement
the efficient assignments in the interdependent values
Interdependent Values
Values areinterdependentwhen one’s valuation of objects
depends on the information/signals possessed by others
E.g. in school choice, many parents have insufficient
information about the fitness of schools for their children
I Seek advice from other parents through word-of-mouth communication, online social networks ...
Are there incentive compatible mechanisms that implement
the efficient assignments in the interdependent values
Ex Post Incentive Compatibility
Recall that a (direct) mechanism is a mapping that maps
each signal profile to an assignment (stochastic
assignments allowed).
A mechanism isex post incentive compatibleif truth-telling is a Nash equilibriumeven after all signals are revealed
I Best alternative to strategy-proofness in the interdependent values environment
Ex Post Incentive Compatibility
Recall that a (direct) mechanism is a mapping that maps
each signal profile to an assignment (stochastic
assignments allowed).
A mechanism isex post incentive compatibleif truth-telling is a Nash equilibriumeven after all signals are revealed
I Best alternative to strategy-proofness in the interdependent values environment
Our Main Results
No ex post IC mechanism can attain a Pareto efficient
assignment whenever there existsnon-trivial preference
interdependence(and a few auxiliary conditions)
Any ex post “group” incentive compatible mechanism can
only attain atrivial assignment, which is constant
irrespective of signal profiles.
In the two-agent and two-object case, there is a
mechanism that is Pareto efficient and Bayesian incentive
compatible under some reasonable conditions
→It may be important to consider mechanisms thatviolate ex
Our Main Results
No ex post IC mechanism can attain a Pareto efficient
assignment whenever there existsnon-trivial preference
interdependence(and a few auxiliary conditions)
Any ex post “group” incentive compatible mechanism can
only attain atrivial assignment, which is constant
irrespective of signal profiles.
In the two-agent and two-object case, there is a
mechanism that is Pareto efficient and Bayesian incentive
compatible under some reasonable conditions
→It may be important to consider mechanisms thatviolate ex
Our Main Results
No ex post IC mechanism can attain a Pareto efficient
assignment whenever there existsnon-trivial preference
interdependence(and a few auxiliary conditions)
Any ex post “group” incentive compatible mechanism can
only attain atrivial assignment, which is constant
irrespective of signal profiles.
In the two-agent and two-object case, there is a
mechanism that is Pareto efficient and Bayesian incentive
compatible under some reasonable conditions
→It may be important to consider mechanisms thatviolate ex
Our Main Results
No ex post IC mechanism can attain a Pareto efficient
assignment whenever there existsnon-trivial preference
interdependence(and a few auxiliary conditions)
Any ex post “group” incentive compatible mechanism can
only attain atrivial assignment, which is constant
irrespective of signal profiles.
In the two-agent and two-object case, there is a
mechanism that is Pareto efficient and Bayesian incentive
compatible under some reasonable conditions
→It may be important to consider mechanisms thatviolate ex
Related Literature
Impossibility results with mechanism designwith money
I Jehiel & Moldovanu (01): Impossibility of efficient and Bayesian incentive compatible mechanism
I Jehiel et al. (06): Impossibility of nontrivial and ex post incentive compatible mechanism
Impossibility result reinforced by our paper but... with some substantial differences
I Utilitarian efficiencyvs.Pareto efficiency I Public decisionvs. private goods allocation I Multidimensionalvs.single dimensional signals Two-sided matching with interdependent values
Related Literature
Impossibility results with mechanism designwith money
I Jehiel & Moldovanu (01): Impossibility of efficient and Bayesian incentive compatible mechanism
I Jehiel et al. (06): Impossibility of nontrivial and ex post incentive compatible mechanism
Impossibility result reinforced by our paper but... with some substantial differences
I Utilitarian efficiencyvs.Pareto efficiency I Public decisionvs. private goods allocation I Multidimensionalvs.single dimensional signals Two-sided matching with interdependent values
Related Literature
Impossibility results with mechanism designwith money
I Jehiel & Moldovanu (01): Impossibility of efficient and Bayesian incentive compatible mechanism
I Jehiel et al. (06): Impossibility of nontrivial and ex post incentive compatible mechanism
Impossibility result reinforced by our paper but... with some substantial differences
I Utilitarian efficiencyvs.Pareto efficiency I Public decisionvs. private goods allocation I Multidimensionalvs.single dimensional signals Two-sided matching with interdependent values
Illustrative Example: Inefficiency of Ex Post IC
mechanisms
We begin with a simple example (generalized later):
Two agents 1,2 are assigned two objectsa,b.
Signals are single-dimensional;s1,s2∈[0,1]
Letui(s)≡vai(s)−vbi(s)bei0utility difference betweena andb, and assume
∂ui(s)
∂si >
∂u−i(s)
Illustrative Example: Inefficiency of Ex Post IC
mechanisms
0 1
1
A B
C D
E
Sbb
Sab Sba
Saa
s1 s2
I2
Illustrative Example: Inefficiency of Ex Post IC
mechanisms
0 1
1
A B
C D
E
Sbb
Sab Sba
Saa
s1 s2
I2
Illustrative Example: Inefficiency of Ex Post IC
mechanisms
0 1
1
A B
C D
E
Sbb
Sab Sba
Saa
s1 s2
I2
Illustrative Example: Inefficiency of Ex Post IC
mechanisms
0 1
1
A B
C
D
E
Sbb
Sab Sba
Saa
s1 s2
I2
Illustrative Example: Inefficiency of Ex Post IC
mechanisms
0 1
1
A B
C D
E
Sbb
Sab Sba
Saa
s1 s2
I2
Illustrative Example: Inefficiency of Ex Post IC
mechanisms
0 1
1
A B
C D
E
Sbb
Sab Sba
Saa
s1 s2
I2
General Setup
The results can be generalized to assignment ofnagents
tonobjects, exactly one object to each agent.
Incentive requirement: (weak) Ex post IC.
Three main assumptions; Interdependence, Richness,
Connectedness.
Assumption 1(Interdependence)For any agents i,j, objects a6=b, whenever vi
a(s) =vbi(s), there is a vector zj such that ∇zjvai(s)6=∇zjvbi(s).
Requires that agentj0s signal influences agenti’s
preferences between any object pairs, at least wheni is
indifferent (doesn’t require value interdependence to be
General Setup
The results can be generalized to assignment ofnagents
tonobjects, exactly one object to each agent.
Incentive requirement: (weak) Ex post IC.
Three main assumptions; Interdependence, Richness,
Connectedness.
Assumption 1(Interdependence)For any agents i,j, objects
a6=b, whenever vi
a(s) =vbi(s), there is a vector zj such that ∇zjvai(s)=6 ∇zjvbi(s).
Requires that agentj0s signal influences agenti’s
preferences between any object pairs, at least wheni is
indifferent (doesn’t require value interdependence to be
General Setup
The results can be generalized to assignment ofnagents
tonobjects, exactly one object to each agent.
Incentive requirement: (weak) Ex post IC.
Three main assumptions; Interdependence, Richness,
Connectedness.
Assumption 1(Interdependence)For any agents i,j, objects
a6=b, whenever vi
a(s) =vbi(s), there is a vector zj such that ∇zjvai(s)=6 ∇zjvbi(s).
Requires that agentj0s signal influences agenti’s
preferences between any object pairs, at least wheni is
indifferent (doesn’t require value interdependence to be
Inefficiency of weakly Ex Post IC mechanisms;
Assumptions
Fix objectsa,b agentsi,j, and signal profiless−ij ∈S−ij.
Fork,k0∈ {a,b}, define
Skkij 0(s−ij)≡the set of signal profiles(si,sj)for which
1. i ranksk first and the other object in{a,b}second, 2. j ranksk0 first and the other object in{a,b}second, and 3. all others rankaandbbelow any other objects.
Assumption 2(Rich Domain)There exist agents i,j, objects a,b, and signals of others s−ij such that Skkij 0(s−ij)is non-empty
for all k,k0 ∈ {a,b}
The Rich Domain assumption can be satisfiedevenwhen
Inefficiency of weakly Ex Post IC mechanisms;
Assumptions
Fix objectsa,b agentsi,j, and signal profiless−ij ∈S−ij.
Fork,k0∈ {a,b}, define
Skkij 0(s−ij)≡the set of signal profiles(si,sj)for which
1. i ranksk first and the other object in{a,b}second, 2. j ranksk0 first and the other object in{a,b}second, and 3. all others rankaandbbelow any other objects.
Assumption 2(Rich Domain)There exist agents i,j, objects
a,b, and signals of others s−ij such that Skkij 0(s−ij)is non-empty
for all k,k0 ∈ {a,b}
The Rich Domain assumption can be satisfiedevenwhen
Inefficiency of weakly Ex Post IC mechanisms;
Assumptions
Fix objectsa,b agentsi,j, and signal profiless−ij ∈S−ij.
Fork,k0∈ {a,b}, define
Skkij 0(s−ij)≡the set of signal profiles(si,sj)for which
1. i ranksk first and the other object in{a,b}second, 2. j ranksk0 first and the other object in{a,b}second, and 3. all others rankaandbbelow any other objects.
Assumption 2(Rich Domain)There exist agents i,j, objects
a,b, and signals of others s−ij such that Skkij 0(s−ij)is non-empty
for all k,k0 ∈ {a,b}
The Rich Domain assumption can be satisfiedevenwhen
Inefficiency of weakly Ex Post IC mechanisms;
Assumptions
Fork ∈ {a,b}, define
Skij
·
(s−ij)≡set of signal profiles(si,sj)for which1. agentiranksk first and the other object in{a,b}second; 2. agentjranksaandbabove any other object;
3. all others ranka,bbelow any other object
Assumption 3(Connectedness)For some i,j, a,b, and s−ij that satisfy the Rich Domain assumption, and for some
k ∈ {a,b}, both Skij
·
(s−ij)and S·
ijk(s−ij)are connected.Weaker than convexity; satisfied if value functions are
Inefficiency of weakly Ex Post IC mechanisms;
Assumptions
Fork ∈ {a,b}, define
S
·
ijk(s−ij)≡set of signal profiles(si,sj)for which1. agentiranksaandbabove any other object;
2. agentjranksk first and the other object in{a,b}second; 3. all others ranka,bbelow any other object
Assumption 3(Connectedness)For some i,j, a,b, and s−ij that satisfy the Rich Domain assumption, and for some
k ∈ {a,b}, both Skij
·
(s−ij)and S·
ijk(s−ij)are connected.Weaker than convexity; satisfied if value functions are
Inefficiency of weakly Ex Post IC mechanisms;
Assumptions
Fork ∈ {a,b}, define
S
·
ijk(s−ij)≡set of signal profiles(si,sj)for which1. agentiranksaandbabove any other object;
2. agentjranksk first and the other object in{a,b}second; 3. all others ranka,bbelow any other object
Assumption 3(Connectedness)For some i,j, a,b, and s−ij that satisfy the Rich Domain assumption, and for some
k ∈ {a,b}, both Skij
·
(s−ij)and S·
ijk(s−ij)are connected.Weaker than convexity; satisfied if value functions are
Inefficiency of weakly Ex Post IC mechanisms;
Assumptions
Fork ∈ {a,b}, define
S
·
ijk(s−ij)≡set of signal profiles(si,sj)for which1. agentiranksaandbabove any other object;
2. agentjranksk first and the other object in{a,b}second; 3. all others ranka,bbelow any other object
Assumption 3(Connectedness)For some i,j, a,b, and s−ij that satisfy the Rich Domain assumption, and for some
k ∈ {a,b}, both Skij
·
(s−ij)and S·
ijk(s−ij)are connected.Weaker than convexity; satisfied if value functions are
Inefficiency of weakly Ex Post IC mechanisms;
The Result
Theorem 1Under the assumptions of Interdependence, Rich
Domain, and Connectedness, there exists no mechanism that
is both Pareto efficient and weakly ex post incentive compatible.
The proof is done by finding a pair of agents and signals
such that a contradiction as in the previous 2-agent
Inefficiency of weakly Ex Post IC mechanisms;
The Result
Theorem 1Under the assumptions of Interdependence, Rich
Domain, and Connectedness, there exists no mechanism that
is both Pareto efficient and weakly ex post incentive compatible.
The proof is done by finding a pair of agents and signals
such that a contradiction as in the previous 2-agent
Impossibility of Ex Post Group Incentive Compatibility
A mechanismϕisex post group incentive compatibleif, for
any signal profiles, there exist no group of agentsN0and
their reported signals such that everyone in the group can
be made weakly better off, with at least one agent strictly.
We slightly modify some of the previous assumptions
(skipped): Rich Domain*, Connectedness*.
Theorem 2Under the assumptions of Interdependence, Rich
Domain*, and Connectedness*, ifϕis ex post group incentive
compatible, thenϕis constant across signals.
The (individual) ex post incentive compatibility is not
Impossibility of Ex Post Group Incentive Compatibility
A mechanismϕisex post group incentive compatibleif, for
any signal profiles, there exist no group of agentsN0and
their reported signals such that everyone in the group can
be made weakly better off, with at least one agent strictly.
We slightly modify some of the previous assumptions
(skipped): Rich Domain*, Connectedness*.
Theorem 2Under the assumptions of Interdependence, Rich
Domain*, and Connectedness*, ifϕis ex post group incentive
compatible, thenϕis constant across signals.
The (individual) ex post incentive compatibility is not
Impossibility of Ex Post Group Incentive Compatibility
A mechanismϕisex post group incentive compatibleif, for
any signal profiles, there exist no group of agentsN0and
their reported signals such that everyone in the group can
be made weakly better off, with at least one agent strictly.
We slightly modify some of the previous assumptions
(skipped): Rich Domain*, Connectedness*.
Theorem 2Under the assumptions of Interdependence, Rich
Domain*, and Connectedness*, ifϕis ex post group incentive
compatible, thenϕis constant across signals.
The (individual) ex post incentive compatibility is not
Impossibility of Ex Post Group Incentive Compatibility
A mechanismϕisex post group incentive compatibleif, for
any signal profiles, there exist no group of agentsN0and
their reported signals such that everyone in the group can
be made weakly better off, with at least one agent strictly.
We slightly modify some of the previous assumptions
(skipped): Rich Domain*, Connectedness*.
Theorem 2Under the assumptions of Interdependence, Rich
Domain*, and Connectedness*, ifϕis ex post group incentive
compatible, thenϕis constant across signals.
The (individual) ex post incentive compatibility is not
Intuition for the result
Given any two signal profilessands˜, we construct a step-wise path,s=s0→s1→ · · · →sm = ˜s, such that each signal profile is associated with strict preferences and for eachk,
(1) sk andsk+1differ in the signal ofonly one agent, sayjk;
(2) betweensk andsk+1, ordinal preferences differ forat most
one agent, sayik 6=jk
Given (1) & (2), the ex post IC forjk implies
ϕjk(s
k) =ϕjk(sk+1)
Given this & (2), for anyi∈ {/ ik,jk}, the ex post group IC applied to{i,jk}impliesϕi(sk) =ϕi(sk+1)
Lastly,ϕik(s
Intuition for the result
Given any two signal profilessands˜, we construct a step-wise path,s=s0→s1→ · · · →sm = ˜s, such that each signal profile is associated with strict preferences and for eachk,
(1) sk andsk+1differ in the signal ofonly one agent, sayjk;
(2) betweensk andsk+1, ordinal preferences differ forat most
one agent, sayik 6=jk
Given (1) & (2), the ex post IC forjk implies
ϕjk(s
k) =ϕjk(sk+1)
Given this & (2), for anyi∈ {/ ik,jk}, the ex post group IC applied to{i,jk}impliesϕi(sk) =ϕi(sk+1)
Lastly,ϕik(s
Intuition for the result
Given any two signal profilessands˜, we construct a step-wise path,s=s0→s1→ · · · →sm = ˜s, such that each signal profile is associated with strict preferences and for eachk,
(1) sk andsk+1differ in the signal ofonly one agent, sayjk;
(2) betweensk andsk+1, ordinal preferences differ forat most
one agent, sayik 6=jk
Given (1) & (2), the ex post IC forjk implies
ϕjk(s
k) =ϕjk(sk+1)
Given this & (2), for anyi∈ {/ ik,jk}, the ex post group IC
applied to{i,jk}impliesϕi(sk) =ϕi(sk+1)
Lastly,ϕik(s
Intuition for the result
Given any two signal profilessands˜, we construct a step-wise path,s=s0→s1→ · · · →sm = ˜s, such that each signal profile is associated with strict preferences and for eachk,
(1) sk andsk+1differ in the signal ofonly one agent, sayjk;
(2) betweensk andsk+1, ordinal preferences differ forat most
one agent, sayik 6=jk
Given (1) & (2), the ex post IC forjk implies
ϕjk(s
k) =ϕjk(sk+1)
Given this & (2), for anyi∈ {/ ik,jk}, the ex post group IC
applied to{i,jk}impliesϕi(sk) =ϕi(sk+1)
Lastly,ϕik(s
Bayesian Incentive Compatible Mechanisms
A mechanism isBayesian incentive compatibleif
truth-telling is a Bayesian Nash equilibrium
I That is, the truth-telling is a mutual best response for each agentiknowingsi only
Focus on the 2×2 case with single dimensional signals
I For eachi=1,2,si is drawn from[0,1]following cdfFi(·)
I Assume that signals are independently distributed (though correlated signals are fine as long as they are not too negatively correlated)
Bayesian Incentive Compatible Mechanisms
A mechanism isBayesian incentive compatibleif
truth-telling is a Bayesian Nash equilibrium
I That is, the truth-telling is a mutual best response for each agentiknowingsi only
Focus on the 2×2 case with single dimensional signals
I For eachi=1,2,si is drawn from[0,1]following cdfFi(·)
I Assume that signals are independently distributed (though correlated signals are fine as long as they are not too negatively correlated)
Consider the mechanism (denotedϕ∗) as in the figure: s1 s2 0 1 1 ¯ s2 ¯ s1 (1,0) (1,0) (1,0) (0,1) (0,1) (0,1)
(p,1−p)
(p0,1−p0)
I2
I1
E.g. (p,1−p)means that agent 1 (resp. 2) receives object
Possibility of Efficiency with Bayesian IC
ϕ∗ is clearly Pareto efficient (if implemented as described)
Sufficient and necessary condition forϕ∗ to be Bayesian
IC:
For eachi=1,2, the “threshold type”¯si is indifferent
between reportingsi >¯si andˆsi <¯si
Theorem 3There exists a pairp,p0 ∈[0,1]that makesϕ∗
Bayesian incentive compatible, if and only if either
Z 1
0
u1(¯s1,s2)dF2(s2)≥0≥ Z 1
0
u2(s1,¯s2)dF1(s1), or
Z 1
0
u1(¯s1,s2)dF2(s2)≤0≤ Z 1
0
u2(s1,¯s2)dF1(s1)
This condition requires that the preferences of the two
Possibility of Efficiency with Bayesian IC
ϕ∗ is clearly Pareto efficient (if implemented as described)
Sufficient and necessary condition forϕ∗ to be Bayesian IC:
For eachi=1,2, the “threshold type”¯si is indifferent
between reportingsi >¯si andˆsi <¯si
Theorem 3There exists a pairp,p0 ∈[0,1]that makesϕ∗
Bayesian incentive compatible, if and only if either
Z 1
0
u1(¯s1,s2)dF2(s2)≥0≥ Z 1
0
u2(s1,¯s2)dF1(s1), or
Z 1
0
u1(¯s1,s2)dF2(s2)≤0≤ Z 1
0
u2(s1,¯s2)dF1(s1)
This condition requires that the preferences of the two
Possibility of Efficiency with Bayesian IC
ϕ∗ is clearly Pareto efficient (if implemented as described)
Sufficient and necessary condition forϕ∗ to be Bayesian IC:
For eachi=1,2, the “threshold type”¯si is indifferent
between reportingsi >¯si andˆsi <¯si
Theorem 3There exists a pairp,p0 ∈[0,1]that makesϕ∗
Bayesian incentive compatible, if and only if either
Z 1
0
u1(¯s1,s2)dF2(s2)≥0≥
Z 1
0
u2(s1,¯s2)dF1(s1), or
Z 1
0
u1(¯s1,s2)dF2(s2)≤0≤
Z 1
0
u2(s1,¯s2)dF1(s1)
This condition requires that the preferences of the two
Possibility of Efficiency with Bayesian IC
ϕ∗ is clearly Pareto efficient (if implemented as described)
Sufficient and necessary condition forϕ∗ to be Bayesian IC:
For eachi=1,2, the “threshold type”¯si is indifferent
between reportingsi >¯si andˆsi <¯si
Theorem 3There exists a pairp,p0 ∈[0,1]that makesϕ∗
Bayesian incentive compatible, if and only if either
Z 1
0
u1(¯s1,s2)dF2(s2)≥0≥
Z 1
0
u2(s1,¯s2)dF1(s1), or
Z 1
0
u1(¯s1,s2)dF2(s2)≤0≤
Z 1
0
u2(s1,¯s2)dF1(s1)
This condition requires that the preferences of the two
This condition is sufficient andnecessaryfor there to be a
Bayesian IC and Pareto efficient mechanism (not
necessarily of the form ofϕ∗ before) if we require the mechanism to beex post monotonicin the sense that
ϕia(·,sj)is non-decreasing for allsj
Open question: Generalization of the possibility result to
This condition is sufficient andnecessaryfor there to be a
Bayesian IC and Pareto efficient mechanism (not
necessarily of the form ofϕ∗ before) if we require the mechanism to beex post monotonicin the sense that
ϕia(·,sj)is non-decreasing for allsj
Open question: Generalization of the possibility result to
Concluding Remarks
This paper is (to our knowledge) the first to study the NTU
assignment problem in the interdependent values setup
I Impossibility results with ex post incentive compatibility I Some possibility result with Bayesian incentive compatibility I It may be important to pay attention to mechanisms that
violate ex post IC but satisfy Bayesian IC if
interdependence of valuations exists; a sharp contrast to private-values setting.
Directions for future research
I Generalization of Bayesian IC mechanisms I Comparative study of some practical assignment
mechanisms
Concluding Remarks
This paper is (to our knowledge) the first to study the NTU
assignment problem in the interdependent values setup
I Impossibility results with ex post incentive compatibility I Some possibility result with Bayesian incentive compatibility I It may be important to pay attention to mechanisms that
violate ex post IC but satisfy Bayesian IC if
interdependence of valuations exists; a sharp contrast to private-values setting.
Directions for future research
I Generalization of Bayesian IC mechanisms I Comparative study of some practical assignment
mechanisms