Combinatorial Agency Michal Feldman (Hebrew University)

1
Combinatorial Agency Michal Feldman(Hebrew
University)

• Joint with
• Moshe Babaioff (UC Berkeley)
• Noam Nisan (Hebrew University)

2
Hidden Actions

• Algorithmic Mechanism Design computational
  mechanisms to handle Private Information.
• (Classical) Mechanism Design
• Private Information
• Hidden Actions
• We study hidden actions in multi-agents
  computational settings

3
Example

• Quality of Service (QoS) Routing FCSS05
• We have some value from message delivery.
• Each agent controls an edge
• succeeds with low probability by default.
• succeeds with high probability if exerts costly
  effort
• Message delivered if there is a successful
  source-sink path.
• Effort is not observable, only the final outcome.

source
sink
4
Modeling Principal-Agent Model
exerts effortcost c gt0
Project succeeds with high probability
Project succeeds with low probability
Does not exert effortcost 0
Agent
Principal
Motivating rational agents to exert costly effort
toward the welfare of the principal, when she
cannot contract on the effort level, only on the
final outcome
Success Contingent contract. The agent gets a
high payment if project succeeds, gets a low
payment if project fails
Our focus is on multi-agents technologies
5
Our Model
The Principals input parameter.

• n agents
• Each agent has two actions (binary-action)
• effort (ai1), with cost cgt0 (ci(1)c)
• no effort (ai0), with cost 0 (ci(0)0)
• There are two possible outcomes (binary outcome)
• project succeeds, principal gets value v
• project fails, principal gets value 0
• Monotone technology function t maps an action
  profile to a success probability
• t 0,1n? 0,1 t(a1,,an)success
  probability given (a1,,an)
• ?i t(1, a-i) gt t(0,a-i) (monotonic)
• Principal designs a contract for each agent
• Project succeeds? agent i receives pi (otherwise
  he gets 0)
• Players utilities, under action profile
  a(a1,,an) and value v
• Agent i ui(a) t(a)pi ci(ai)
• Principal u(a,v) t(a)(v Sipi)
• Agents are in a game, reach Nash equilibrium.

The Principals design parameter Used to
induce the desired equilibrium
6
Example Read-Once Networks

• A graph with a given source and sink
• Each agent controls an edge, independently
  succeeds or fails in his individual task
  (delivering on his edge)
• Succeeds with probability ?lt½ with no effort
• Succeeds with probability 1-? (gt½gt?) with effort
• The project succeeds if the successful edges form
  a source-sink path.
• example t(1, 1, 0) Pr x1 ? (x2 ? x3) 1
  a(1,1,0)
  (1- ?) (1- ?(1-?))

a21
a11
Pr x211- ?
sink
source
a30
Pr x111- ?
Pr x31?
7
Nash Equilibrium
Agent is utility
exerts effort
Does not exert effort

• Principals best contract to induce eq.
  a(a1,,an)
• pi c / Di(a-i) for agent i with ai1
• pi 0 for agent i with ai0
• e.g., (1,0) (1,1)

ui( 1,a-i ) pi t( 1,a-i ) c
ui( 0,a-i ) pi t(0,a-i )
8
Optimal Contract

• the principal chooses a profile a(v) that
  maximizes her optimal equilibrium utility

Probability of success
Total payments
9
Research Questions

• How does the technology affect the structure of
  the optimal contracts?
• Several examples (AND, OR, Majority )
• General technologies
• What is the damage to the society due to the
  inability to monitor individual actions?
• price of unaccountability
• What is the complexity of computing the optimal
  contract?
• Can the principal gain utility from mixed
  strategies?
• Can the principal gain utility from a-priory
  removing edges from the graph?

10
Optimal Contracts simple AND technology

• 2 agents, g ¼, c1
• t(0,0) g2 (¼)21/16
• t(1,0) t(0,1) g(1-g) 3/16
• D0 t(1,0)-t(0,0)3/16 - 1/16 1/8
• t(1,1) (1-g)2 9/16
• Principals Utility
• 0 agents exert effort
• u((0,0),v) t(0,0)v v/16
• 1 agent exerts effort
• u((1,0),v) t(1,0)(v-c/D0)
• 3/16(v-1/(1/8))(3/16)v-3/2
• 2 agents exert effort
• u((1,1),v) t(1,1)(v-2c/D1) 9v/16-3

s
t
x1
x2
At value of 6 there is a jump from 0 to 2 agents
11
Optimal Contract Transitions in AND and OR

• OR

• AND

x1
x1
x2
s
t
s
t
x2
v
v
2
g
g
12
Optimal Contract Transitions in AND and OR

• Theorem For any AND technology, there is only
  one transition, from 0 to n agents.
• Theorem For any OR technology, there are always
  n transition (any number of agents is optimal for
  some value).
• We characterize all technologies with 1
  transition and with n transitions.

13
Proofs Idea-ANDs single transition

• Observation (monotonicity) number of contracted
  agents monotonically non-decreasing in v.
• Proof for ANDs single transition
• At the indifference value between 0 and n agents,
  contracting with 0ltkltn agent has lower utility.
• By the above observation, a single transition.

The 0 and n indifference value
14
Transitions in AND and OR

• Proof (AND)k number of contracted agents
• this function has a single minimum point, thus
  maximized at one of the edges 0 or n

15
Proofs Idea ORs n transitions

• Let vk be the indifference point between k and
  k1 agents ( u(k,vk) u(k1,vk) )
• We show that for OR vk1gt vk
• This ensures that k is optimal from vk-1 to vk

v1 The 1 ,2 indifference value.
v0 The 0 ,1 indifference value.
v1gtv0
16
Transitions in AND and OR

• k number of contracted agentssolve for v u(k)
  u(k1), and let v(k) be the solutionwe have
  to show v(k1) gt v(k) ? d
• E.g., n3

v(2)
v(1)
v(0)
d
17
Majority, 5 agents
18
General Technologies

• In general we need to know which agents exert
  effort in the optimal contract
• Examples
• In potential, any subset of agents (out of 2n
  subsets) that exert effort could be optimal for
  some v.
• Which subsets can we get as an optimal contract?

19
And-of-Ors (AOO) Technology

• Example 2x2 AOO technology
• Theorem The optimal contract in any AOO network
  (with identical OR components) has the same
  number of agents in each OR-component
• Proof by induction based on following lemmas
• Decomposition lemma if STUR is optimal on
  fh?g on some v, then T is optimal for h on
  vtg(R) and R is optimal for g on vth(T)
• Component monotonicity lemma the function
  v?th(T) is monotone non-decreasing (same for
  v?tg(R) )

?
A1,B1
A1,B1,A2,B2
v
20
Decomposition Lemma
if STUR is optimal on fh?g on some v, then T
is optimal for h on vtg(R) and R is optimal for
g on vth(T)

• Proof

21
Component Monotonicity Lemma
The function v?th(T) is monotone non-decreasing
(same for v?tg(R) )

• Proof
• S1 T1 U R1 optimal on v1
• S2 T2 U R2 optimal on v2ltv1
• By monotonicity lemma f(S1) f(S2)
• Since fgh, f(S1)h(T1)g(R1) h(T2)g(R2)
  f(S2)
• Assume in contradiction that h(T1) lt h(T2).
  Since h(T1)g(R1) h(T2)g(R2) , we get g(R1) gt
  g(R2).
• By decomposition lemma, T1 is optimal for h on
  v1g(R1), and T2 is optimal for h on v2g(R2)
• As v1 gt v2, and g(R1) gt g(R2), T1 is optimal for
  h on a larger value than T2.
• Thus, by monotonicity lemma, h(T1) h(T2)

f
R1
h
g
R2
T1
T2
22
And-of-Ors

• Theorem The optimal contract in any AOO network,
  composed of nc OR-components (of size nl)
  contracts with the same number of agents in each
  OR-component. Thus, orbit(AOO) nl1
• Proof by induction on nc
• Base nc2assume (k1,k2) is optimal on some v,
  assume by contradiction k1gtk2 (wlog), thus
  h(k1)gth(k2).By decomposition lemma k1 optimal
  for h on vh(k2) k2 optimal for h on
  vh(k1)gtvh(k2)but if k2 optimal for a larger
  value, k2k1. in contradiction.

23
And-of-Ors
h2
k1
k2
k2
k3
k3
k2



h
h
h
h

• assume (induction) that claim holds for any
  number of OR components lt nc
• Assume 1st component has k1 contracted agents
• Let g be the conjunction of the other (nc-1)
  comp.
• By decomposition lemma, contract on g is optimal
  at vh(k1), thus by induction hypothesis has same
  number of agents, k2, on each OR component.
• Let h2 be conjunction of first two comp.
• By decomp. Lemma, contract on h2 is optimal for
  some value and by induction hypothesis has same
  number of agents, k3
• We get k1k3 (in first comp. k1 agents
  contracted), and k2k3 (in second comp. k2 agents
  contracted), thus k1k2

24
The Collection of Optimal Contracts

• Given t we wish to understand how the optimal
  contract changes with v (the orbit).
• Monotonicity Lemma The optimal contract success
  probability t(a(v)) is monotonic non-decreasing
  with v
• So is the utility of the principal, and the total
  payment
• Thus, there are at most 2n-1 changes to the
  optimal contracts (Orbit(t) 2n)

Is there a structure on the collection of optimal
contracts of t?
25
The Collection of Optimal Contracts

• Observation 1 in the observable-actions case,
  only one set of size k can be optimal (set with
  highest probability of success)
• Observation 2 not all 2n subsets can be obtained
• Only a single set of size 1 can be optimal (set
  with highest probability of success)
• Thm There exists a tech. with optimal
  contracts
• Open question 1 is there a read-once network
  with exponential number of optimal contracts?

Can a technology have exponentially many
different optimal contracts?
26
Exponential number of optimal contracts (1)

• Thm There exists a tech. with optimal
  contracts
• Proof sketch
• Lemma 1 all k-size sets in any k-admissible
  collection can be obtained as optimal contracts
  of some t
• Lemma 2 For any k, there exists a k-admissible
  collection of k-size sets of size
• Based on error correcting code
• Lemma 3 for kn/2 we get a k-admissible
  collection of k-size sets of size ,
  as required.

Collection of sets of size k, in which every two
sets in it differ by at least two elements
27
Proof of Lemma 1
n

• marginal contribution of i ? S is t(S)
  t(S\i) eS

t(S) ½ - eS
k
S
k-1
t(S\i) ½ - 2eS

• Claim at vs(ck) / 2eS2, the set S is optimal
• S better than any other set in col. (by
  derivative of u(S,v))
• S better then any other set not in col. (too
  high payments)

1
28

• Let vs be v s.t.

29
Proof (k-orbit)

• ? admissible collection of k-size sets
• Z all S ? ? U all S\i
• For sets T ? Z
• if ? z? Z z ? T
• else t(T) e T

Pick eS?(0.17,0.2
t(S)½ - eS
S3
? 3-size sets
S1
t(S\i)½ - 2eS
i
S2
Pick eS0.2
S4

• Let vs be v s.t. ,S chosen at
  vs

30
Proof (contd)

• Need to show s yields higher utility at vs than
  any other set s
• s k-1
• If s ? s\i t(s)es ? can be arbitrarily
  small
• If s ? s\i t(s) ½ - es gt 0.3, t(s) ½ -
  2es lt 0.16
• s gt k ? at least one agent is paid 1/e , so
  pick e s.t. payment gt vs

u(s,v0.2)ltu(s0.2,v0,2) ? es gt 0.16
u(s0.18)
u(s0.2)
Size k and (k-1)
v
V0.2
V0.18
31
Exponential number of optimal contracts (1)
Collection in which every two sets in it differ
by at least two elements
collection of optimal sets of size exactly k

• Thm There exists a tech. with optimal
  contracts
• Proof sketch constructive
• Lemma 1 any admissible collection can be
  obtained as the k-orbit of some t
• Define t as follows
• for every set in the collection, Pick eS,
• and define t(S)½ - eS and t(S\i)½ - 2eS
• (thus, marginal contribution of i?S is eS)
• for every set not in the collection, define t to
  ensure that the
• marginal contribution of each agent is very small
• Claim at vs(ck) / 2es2, the set S is optimal
• S better than any other set in col. (by
  derivative of U(S,v))
• S better then any other set not in col. (too high
  payments)

32
Exponential number of optimal contracts (2)

• Lemma For any n k, there exists an admissible
  collection of
• k-size sets of size
• Proof take error correcting code that corrects 1
  error.
• Hamming distance 3 ? admissible
• Known ? codes with W(2n/n) code words.
• Construct a code with sufficient of k-weight
  words
• XOR every code word with a random word r. weight
  k w/ prob
• Expected number of k-weight code words
• There exists r such that the expectation is
  achieved or exceeded

33
Research Questions

• How does the technology affect the structure of
  the optimal contracts?
• What is the damage to the society / principal due
  to the inability to monitor individual actions?
• price of unaccountability
• What is the complexity of computing the optimal
  contract?
• Can the principal gain utility from mixed
  strategies?
• Can the principal gain utility from a-priory
  removing edges from the graph?

34
Observable-Actions Benchmark (first best)

• Actions are observable
• Payment an agent that exerts effort is paid his
  cost (c)
• Principals utility u(a,v) vt(a) Siai1 c
• Principals utility social welfare sw(a,v).
• The principal chooses aOA, the profile with
  maximum social welfare.

35
Social Price of Unaccountability

• Definition The Social Price Of Unaccountability
  (POUS) of a technology is the worst ratio (over
  v) between the social welfare in the
  observable-action case, and the social welfare in
  the hidden-action case
• a - optimal contract for v in the
  hidden-action case
• aOA - optimal contract for v in the
  observable-action case
• Example AND of 2 agents

s
t
v
Hidden actions
0
2
Observable actions
0
2
36
Principals Price of Unaccountability

• Definition The Principals Price Of
  Unaccountability (POUP) of a technology is the
  worst ratio (over v) between the principals
  utility in the observable-action case, and the
  principals utility in the hidden-action case
• a - optimal contract for v in the
  hidden-action case
• aOA - optimal contract for v in the
  observable-action case

37
Price of Unaccountability - Results

• Theorem The POU of AND technology is
• unbounded for any fixed n2, when g?0
• unbounded for any fixed glt½ when n??
• Theorem The POU of OR technology is bounded by
  2.5 for any n

38
Research Questions

• How does the technology affect the structure of
  the optimal contracts?
• What is the damage to the society due to the
  inability to monitor individual actions?
• price of unaccountability
• What is the complexity of computing the optimal
  contract?
• Can the principal gain utility from mixed
  strategies?
• Can the principal gain utility from a-priory
  removing edges from the graph?

39
Complexity of Finding the Optimal Contract

• Input value v, description of t
• Output optimal contract (a,p)

• Theorem There exists a polynomial time algorithm
  to compute (a,p), if t is given by a table
  (exponential input).
• Theorem If t is given by a black box,
  exponentially many queries may be required to
  find (a,p).
• Theorem For read-once networks, the optimal
  contract problem is p-hard (under Turing
  reduction)
• (proof reduction from network reliability
  problem)
• Open problem 3 is it polynomial for
  series-parallel networks?
• Open problem 4 does it have a good approximation?

40
Complexity of Finding the Optimal Contract

• Input value v, description of t
• Output optimal contract (a,p)

• Theorem There exists a polynomial time algorithm
  to compute (a,p), if t is given by a table
  (exponential input).
• Theorem If t is given by a black box,
  exponentially many queries may be required to
  find (a,p).
• Proof
• for value v c(k ½), S is optimal
• Any algorithm must query all sets of size kn/2
  to find S in the worst case

41
Complexity of Finding the Optimal Contract

• Input value v, description of t
• Output optimal contract (a,p)

• Theorem For read-once networks, the optimal
  contract problem is p-hard
• Proof reduction from network reliability problem
• Open problem 3 is it polynomial for
  series-parallel networks?
• Open problem 4 does it have a good approximation?

42
Best Contract Computationin Read-Once Networks

• Proof (sketch) an algorithm for this problem can
  be used to compute t(E) (probability of success)
• Player x will enter the contract only for very
  large value of v (only after all other agents are
  contracted), call this value vc
• At vc, principal is indifferent between E and
  EUx

G
t
gx? ½
43
Research Questions

• How does the technology affect the structure of
  the optimal contracts?
• What is the damage to the society due to the
  inability to monitor individual actions?
• price of unaccountability
• What is the complexity of computing the optimal
  contract?
• Can the principal gain utility from mixed
  strategies?
• Can the principal gain utility from a-priory
  removing edges from the graph?

44
Mixed Strategies
Can mixed-strategies help the principal ? What is
the price of purity ?

• In the non-strategic case NO (convex
  combination)
• What about the agency case?
• Extended game
• qi probability that agent i exerts effort
• t( qi,q-i ) qit(1,q-i ) (1-qi )t(0,q-i )
• Marginal contribution Di(q-I ) t(1,q-i ) -
  t(0,q-i ) 0

45
Nash Equilibrium in Mixed Strategies

• Claim agent is best-response is to mix with
  probability q ? (0,1) only if she is indifferent
  between 0 and 1
• Agent is utility
• Principals utility

Agent is utility
High effort
Low effort
ui( 1,q-i ) pi t( 1,q-i ) ci
ui( 0,q-i ) pi t(0,q-i )
46
ExampleOR with two agents

• Optimal contract for v110
• Pure strategies both agents contracted u
  88.12...
• Mixed strategies q1q20.96.. u88.24...
• Two observations
• q1q2 in optimal contract
• Principals utility is improved, but only
  slightly
• How general are these observations?

47
Optimal Contract in OR Technology

• Lemma For any anonymous OR (any g,n,c,v),
  k?0,1,,n agents exert effort with equal
  probabilities q1qk ? (0,1, and n-k agents
  shirk. i.e. optimal profile (0n-k, qk)
• Proof (skecth) suppose by contradiction that
  (qi,qj,q-ij) s.t. qi,qj? (0,1) and qi gt qj is
  optimal

(qi,qj,q-ij)
qj
For a sufficiently small e , success probability
increases, and total payments decrease. In
contradiction to optimality
qi
48
Optimal Contract in OR Technology
Example OR with 2 agents
49
Price of Purity (POP)

• Definition POP is the ratio between principals
  utility in mixed strategies and in pure
  strategies

Optimal mixed contract
Optimal pure contract
50
Price of Purity

• Definition technology t exhibits
• increasing returns to scale (IRS) if for any i
  and any b a t(bi,b-i)-t(ai,b-i)
  t(bi,a-i)-t(ai,a-i)
• decreasing returns to scale (DRS) if for any i
  and any b a t(bi,b-i)-t(ai,b-i)
  t(bi,a-i)-t(ai,a-i)
• Observations AND exhibits IRS, OR exhibits DRS
• Theorem for any technology that exhibits IRS,
  optimal contract is obtained in pure strategies
• e.g., AND

51
Price of Purity

• For any anonymous DRS technology, POP n
• For anonymous OR with n agents, POP 1.154..
• For any anonymous technology with 2 agents, POP
  1.5
• For any technology (not necessarily anonymous,
  but with identical costs) with 2 agents, POP 2
• Observation the payment to each agent in a mixed
  profile is greater than the min payment in a pure
  profile and smaller than the max payment in a
  pure profile

52
Research Questions

• How does the technology affect the structure of
  the optimal contracts?
• What is the damage to the society due to the
  inability to monitor individual actions?
• price of unaccountability
• What is the complexity of computing the optimal
  contract?
• Can the principal gain utility from mixed
  strategies?
• Can the principal gain utility from a-priory
  removing edges from the graph?

53
Free-Labor

• So far, technology was exogenously given
• Now, suppose the principal has control over the
  technology in that he can ex-ante remove some
  agents from the graph
• Example OR with 2 agents
• Action set of agent i ai ? 1,0,?
• 1 exert effort succeed with probability d.
  costc
• 0 do not exert effort - succeed with probability
  glt d. cost0
• ? do not participate succeed with probability
  0. cost0
• Action ? wastes free-labor since action 0
  increases the success probability with no
  additional cost

as before
54
Free-Labor
Are there scenarios in which the principal gains
utility from wasting free-labor?

• The answer is YES
• Example OR technology, n2, g0.2
• Theorem for technologies with increasing
  marginal contribution (e.g., AND), utilizing all
  free-labor is always optimal

v
0
1
2
1 removed
55
Analysis of OR

• Lemma for any OR with n agents and g which is
  small enough, there exists a value for which in
  the optimal contract one agent exerts effort and
  no other agent participates

g0.49
g0.25
g0.01
56
Version of the Braesss Paradox

• A project is composed of 2 essential components
  A and B
• And-of-Ors (AOO) allow interaction between teams
• Or-of-Ands (OOA) dont allow interaction between
  teams
• Obviously, AOO is superior in terms of success
  probability

project succeeds if at least one of the following
pairs succeed (A1,B1) (A1,B2) (A2,B1)
(A2,B2)
project succeeds if at least one of the following
pairs succeed (A1,B1) (A2,B2)
57
Version of the Braesss Paradox
A1
B1
Example g0.2, v110
gi 1
s
t
A2
B2
And-of-Ors
Or-of-Ands
gt
u(2,2) 75.59..
u(1,1) 74.17..
Or-of-Ands wastes free-labor. Could the
principal gain utility from removing middle edge?
Conclusion it may be beneficial for the
principal to isolate the teams
58
Summary

• Combinatorial Agency hidden actions in
  combinatorial settings
• Computing the optimal contract in general is hard
• Natural research directions
• technologies whose contract can be computed in
  polynomial time
• Approximation algorithms
• Many open questions remain

59
Thank You

• mfeldman_at_cs.huji.ac.il

60
Related Literature

• Winter2004 Incentives and discrimination
• The effect of technology on optimal contract
  (full implementation)
• Winter2005 Optimal incentives with information
  about peers
• Ronen2005Smorodinsky and Tennenholtz2004,2005
• Multi-party computation with costly information
• Holmstrom82 Moral hazard in teams
• Budget-balanced sharing rules