Dispensa+- 5- Deterrence -New Developments Flashcards
Actions and Types
- Two countries: A and B
- Country B constituted by Leader (L) and Elite (E)
They both may be hawks or doves.
Timing
4steps
- L provokes a crisis (m=1) or not (m=0)
- E decides whether to support L (s = 1) or challenge L (s = 0)
- Country A decides whether to retaliate (r = 1) or not (r = 0)
- Ruler of country B decided whether to escalate (x = 1) or not (x = 0)
Political Instability
Leader needs support of elite to stay in power
- if Elite chooses support, Leader remains in power
- if Elite seizes power, it becomes Ruler
If Elite decides to seize power, domestic payoffs lose fraction (1−δ): internal conflict is inefficient
Stakes of the Conflict
For hawks in country B: - π1 from provocation - π2 from escalation For doves in country B: - provocation and escalation give payoff of 0 For country A: - −λ1 from provocation - −λ2 from escalation
Retaliation
- Costly for country to execute: c > 0
- Costly for members of country B to endure (deterrence by punishment; impose costs): k > 0
- retaliatory actions reduce the disutility associated with an escalated conflict by q ∈ (0, 1), where q represents the effectiveness of sanctions and military operations
• −λ2 becomes −(1−q)λ2
Domestic Payoffs
Leader and Elite disagree about domestic politics:
They care about the ideological type of ruler, θR.
Member i of B receives a domestic political payoff given by
u(θi;θR)=θi ·θR +(1−θi)(1−θR),
if θi = θR (ideological agreement), payoff equal to 1
if θi ̸= θR (ideological disagreement), payoff equal to 0
If Elite seizes power from Leader, domestic payoff becomes δ
Perfect Bayesian Equilibrium:
- L: provocation strategy, given Θ (ideological composition of B)
- E: support strategy, given Θ and L’s decision
- A: retaliation strategy, given decisions of L and E, and system of beliefs considering the ideological composition of B
- Ruler: escalation strategy for the ruler, considering the previous choices.
Model of conventional deterrence scenario
Elite is removed (= fixing exogenously decision to support, s*=1)
L is a hawk (ΘL=1)
q is the effectiveness of sanctions (extent that retaliation reduces the influence of escalation on the status quo)
credibility constraint
When should country A retaliate?
Ua Not retaliate <= Ua retaliate
−λ1 −λ2 ≤ −λ1 −(1−q)λ2 −c
so when c ≤ qλ2 ≡ c∗
capability constraint
when does a threat deters provocation?
UL no prov ≥ UL prov
1 ≥ π1 + 1 + (1 − q)π2 − k
so when k ≥ π1 + (1 − q)π2 ≡ k∗
Key features of the adapted general model:
1- Incomplete information
2- Political instability
3- Ideological disagreement
Successful deterrence means
no provocation
Credibility Constraint: c ≤ c ∗
Capability Constraint: k ≥ k ∗
Necessary and jointly sufficient for deterrence success
Model with 1
incomplete information
- Probability that L is hawk is μ
- Standard costly signalling game: L is sender and A is receiver
- a dove wouldn’t provoke a crisis, so if L provoked a crisis, L is a hawkIf credibility constraint is satisfied → retaliation after provocation
If capability constraint is satisfied → successful deterrence
Model with 2
Political instability
- Political Instability: country B has L and E, they are both hawks
- No incomplete info - common knowledge that L and E are hawks
- E is a hawk, so if L provokes a crisis, even if E takes power, country A might retaliate
Since conflict is inefficient, E supports L → everything just like when E is absentIf credibility constraint is satisfied → retaliation after provocation
If capability constraint is satisfied → successful deterrence
Model with 2 and 3
Political instability and Ideological disagreement
salience constraint
L is a dove and E is a hawk
‘‘the threat of retaliation by country A can alter the relative salience of ideological disagreements within country B’’
- Country A retaliates if and only if L provoked a crisis and a hawkish ruler controls country B
If E’s decision does not affect whether A retaliates, the ideological disagreement between L and E is the only relevant dimension determining E’s choice
If E’s support prevents retaliation, then E will support a
dovish L if and only if
supporting> seizing power
π1 ≥π1 +δ+(1−q)π2 −k
which, after rearranging, holds as long as
k ≥δ+(1−q)π2(1)≡k∗∗
Salience constraint