International Institutions: Weak Commitment and Signals

1
International InstitutionsWeak Commitment and
Signals

• Lisa Martin
• Harvard University
• November 2007, IPES

2
Motivation

• Most models of international institutions focus
  on their ability to commit states to the terms of
  agreements
• Commitment happens through enforcement or
  reputational effects
• However, often seems that the commitment
  properties of institutions are fairly weak

3
Motivation

• Another established approach considers the
  signaling properties of institutions
• If institutions act as costly signals, there must
  be an ex ante cost to joining the institution,
  and this cost must be differentiated
• This paper considers the possibility that
  institutions can be both weak commitment and
  signaling devices

4
Commitment model

• Start by considering international institutions
  as commitment devices
• As a commitment device, an institution will
  impose an ex post cost if a state deviates from
  the terms of an agreement
• Consider a game with uncertainty about a
  potential new member of an institution (A) who
  may or may not have incentives to live up to the
  terms of a commitment

5
Commitment model

• Commitment game Start with move by Nature that
  determines whether state A is reliable
  (probability p) reliable A gets a higher payoff
  from cooperation
• A then decides whether to agree to cooperate
  within the institutional framework

6
Commitment model

• State B, a current member of the institution,
  then decides whether to cooperate with A, within
  the institution if A made that choice or outside
  the institution otherwise
• State A then decides whether to cooperate or to
  renege

7
Commitment game payoffs

• In a pure commitment game, A bears no ex ante
  cost for participating in an institution
• If A does participate in the institution and then
  reneges, it pays cost c
• The benefits of cooperating are higher for a
  reliable A than for unreliable
• B benefits from cooperation but pays cost d if A
  reneges

8
Weak commitment game equilibria

• Consider payoff structures in which the
  commitment institution is able to induce
  cooperation by reliable A but not unreliable
• This requires that even reliable A have an
  incentive to renege in the absence of the
  institution, but the institutional penalty c is
  enough to induce reliable A to cooperate

9
Weak commitment game equilibria

• However, for unreliable A the benefits of
  cooperation are not high enough to deter
  defection even in the presence of the institution

10
Weak commitment

• A weak commitment institution satisfies the
  following properties
• Both reliable and unreliable A renege in the
  absence of the institution
• If A chooses the institution, reliable A will
  cooperate but unreliable will renege and bear the
  associated costs
• Reliable A prefers institutionalized cooperation
  to no cooperation

11
Cooperation with weak commitment

• With a weak commitment institution, if A does not
  offer the institution, B will not cooperate,
  because even reliable A do not derive enough
  benefits from uninstitionalized cooperation to
  live up to the terms of an agreement

12
Unreliable A strategies

• There are two types of weak commitment
  equilibria, differentiated by the strategy that
  unreliable A adopts
• If c is less than the benefit that unreliable A
  gets from reneging, unreliable A will bluff and
  choose institution
• If c is greater than the benefit that unreliable
  A gets from reneging, unreliable A will not
  choose institution will accept the
  no-cooperation payoff (0)

13
Restriction on punishment cost

• I restrict attention to the equilibrium in which
  the punishment cost c is not sufficient to deter
  unreliable A from bluffing
• This seems reasonable, since the ability of
  institutions to impose such costs is limited, and
  it is consistent with the concept of weak
  commitment institutions that I wish to explore

14
Cooperation with weak commitment

• Thus, in the weak commitment equilibrium in which
  punishment cost c is not sufficient to deter
  unreliable A from choosing the institution we
  observe
• All A choose institution
• B cannot update beliefs about As type and so, on
  observing institution, cooperates if prior belief
  that A is reliable is above a threshold
• That probability is a function of Bs benefits
  from cooperation and cost if A reneges

15
Weak commitment equilibrium

• No cooperation without institutions
• Reliable A benefits because it gains some
  cooperation with B (probability p)
• Unreliable A also gains because it gains some
  chance to get B to cooperate and then renege
• Bs expected payoff from a weak commitment
  institution is the same as for no institution at
  all and the institutional setting exposes B to
  risk

16
Summary of weak commitment

• Thus, pure commitment institutions that can
  provide only weak commitments have some
  undesirable properties
• They allow unreliable states to sucker others
  into cooperation from which they will renege
• They do not improve the expected payoff to
  existing members of the institution and expose
  them to risk
• Reliable potential members benefit, but even they
  are only able to cooperate probabilistically

17
Adding signaling

• Focus on the parameters of the commitment game
  that give rise to this weak commitment
  equilibrium
• Now add a signaling dynamic how will this change
  equilibrium strategies and outcomes?
• Does adding a signal improve the payoffs for
  reliable A and for B?
• Signaling means that A pays some ex ante cost (z)
  for choosing the institution

18
Signaling and commitment equilibria
19
Unreliable A payoffs, adding signaling cost to
weak commitment
20
Reliable A payoffs, adding signaling cost
21
B payoffs, adding signaling cost
22
Who would prefer signaling institutions?

• Unreliable A always lose from the addition of
  signaling
• Reliable A gain from signaling if p is low (they
  can distinguish themselves) lose if p is high
  (have to bear signaling costs)
• B gains from signaling wants to hit sweet spot
  will err on side of low ex ante costs

23
Implications when will we see reneging?

• With weak commitment institutions, reneging will
  be common
• With signaling institutions, will see less
  reneging. But will still occur if ex ante cost
  of joining institution is set too low.

24
Implications when will we see institutions?

• If institutions are costless to join and only
  offer weak commitment capabilities, will be
  universal
• If institutions have an ex ante cost
• There will be many with low costs to join
• There will be some with moderate costs
• There will be none with prohibitive costs

25
Research agenda

• Empirical applications
• In studies of peacekeeping (with Page Fortna),
  have found support for a model of peacekeeping as
  a pure signal. Now consider pk as a weak
  commitment institution, see if a better fit,
  provides more leverage differentiating pk from
  peace w/o pk.

26
Research agenda

• U.S. international commitments have found
  support for model of treaties as pure signaling
  devices again, consider they also offer weak
  commitments, ask if better fit.
• Institutional design
• Provides clear predictions about state
  preferences in institutional design stage
• EU or other established institutions policies
  toward new members