SLIDE 1
Democracy, Information, and Audience Costs (Previously circulated as - - PowerPoint PPT Presentation
Democracy, Information, and Audience Costs (Previously circulated as - - PowerPoint PPT Presentation
Democracy, Information, and Audience Costs (Previously circulated as Informational Effects of Audience Costs) Shuhei Kurizaki & Taehee Whang Waseda University Yonsei University American Political Science Association, Philadelphia,
SLIDE 2
SLIDE 3
Research Program on Audience Costs
Audience costs can make the decision to go to war rational (Fearon 1994) A set of conjectures to be substantiated
◮ Audience costs exist ◮ Audience costs ∝ democracy ◮ Audience costs → bargaining power
Tomz 2007, K+W 2015 K+W 2015 “Democratic Advantage”
SLIDE 4
Research Program on Audience Costs
Audience costs can make the decision to go to war rational (Fearon 1994) A set of conjectures to be substantiated
◮ Audience costs exist ◮ Audience costs ∝ democracy ◮ Audience costs → bargaining power
Tomz 2007, K+W 2015 K+W 2015 “Democratic Advantage” But this causal effect depends on a learning mechanism: Audience costs help to send credible signals and learn each other’s resolve
SLIDE 5
Research Program on Audience Costs
Audience costs can make the decision to go to war rational (Fearon 1994) A set of conjectures to be substantiated
◮ Audience costs exist ◮ Audience costs ∝ democracy ◮ Audience costs → bargaining power
Tomz 2007, K+W 2015 K+W 2015 “Democratic Advantage” But this causal effect depends on a learning mechanism: Audience costs help to send credible signals and learn each other’s resolve
◮ Audience costs → information
⇐ This paper
SLIDE 6
What We Do: Objectives
- 1. Test Whether Audience Costs Facilitate Learning
◮ We model learning as belief-updating in a crisis ◮ We measure the prior and posterior beliefs
SLIDE 7
What We Do: Objectives
- 1. Test Whether Audience Costs Facilitate Learning
◮ We model learning as belief-updating in a crisis ◮ We measure the prior and posterior beliefs
This allows us to test another outstanding question in the literature
- n democracy and conflict.
SLIDE 8
What We Do: Objectives
- 1. Test Whether Audience Costs Facilitate Learning
◮ We model learning as belief-updating in a crisis ◮ We measure the prior and posterior beliefs
This allows us to test another outstanding question in the literature
- n democracy and conflict.
- 2. Test Among Informational Mechanisms of Democracy
Democratic Institutions Institutional constraints Democratic Advantage Signaling via audience costs Transparency Information revelation Schultz (1999 IO)
SLIDE 9
How Do We Do This? Structural Approach
◮ We measure learning itself as it is defined in audience costs theory,
rather than its effect.
◮ Signaling and learning are modeled as beliefs and their changes ◮ Belief-updating and audience costs are both estimated based
- n the estimates of underlying payoffs and outcome
probabilities in international conflict data
SLIDE 10
How Do We Do This? Structural Approach
◮ We measure learning itself as it is defined in audience costs theory,
rather than its effect.
◮ Signaling and learning are modeled as beliefs and their changes ◮ Belief-updating and audience costs are both estimated based
- n the estimates of underlying payoffs and outcome
probabilities in international conflict data ↑ These are already done in Shuhei Kurizaki & Taehee Whang (2015) “Detecting Audience Costs in International Disputes” International Organization
SLIDE 11
How Do We Do This? Structural Approach
◮ We measure learning itself as it is defined in audience costs theory,
rather than its effect.
◮ Signaling and learning are modeled as beliefs and their changes ◮ Belief-updating and audience costs are both estimated based
- n the estimates of underlying payoffs and outcome
probabilities in international conflict data ↑ These are already done in Shuhei Kurizaki & Taehee Whang (2015) “Detecting Audience Costs in International Disputes” International Organization
◮ What’s left for this paper to do:
◮ We estimate prior beliefs and posterior beliefs using the
estimates of the payoffs (and audience costs)
◮ We demonstrate that audience costs improve the amount of
belief-updating
SLIDE 12
Common Theoretical Model of Audience Costs
Resist Back Down
1 ) ( ) (
2 1 1
- BD
u a BD u
~Challenge Status Quo
1 ) ( ) (
1 1
- SQ
u SQ u
Stand Firm
2 1 1 1
) ( ) ( w SF u w SF u
- Challenge
~Resist Fight ~Fight State 1 State 2 State 1 Concession
2 2 1
) ( 1 ) ( a CD u CD u
- Definition
Audience costs for State 1 exist iff u1(BD) < u1(SQ)
SLIDE 13
Beliefs and Belief-Updating in a Model of Audience Costs Singling and Learning (Theoretical Definition) Belief updating = S2’s posterior minus prior beliefs.
a1 S1’s audience costs
1
a
1
ˆ a
Prior beliefs (45°) Posterior beliefs (q) Belief updating ()
1
1
~ a S2’s subjective probability that S1 is resolved
SLIDE 14
Beliefs and Belief-Updating in a Model of Audience Costs
Measuring beliefs requires estimating the payoffs in the underlying game.
◮ Prior Belief
Ex ante probability that State 1 fights Pr(SF) = Pr(u1(SF) ≥ u1(BD))
◮ Posterior Belief
Conditional probability that State 1 fights, given the challenge Pr(SF|CH) = Pr
- u1(SF) ≥ u1(BD)
E[u1(CH)] ≥ u1(SQ)
SLIDE 15
Statistical Model of Audience Costs in Kurizaki & Whang (2015)
Resist
Pr(RS|CH)
Back Down
1 1 1 1
1 1
) (
BD BD BD BD
X BD BD u
- 2
2 2 2
2 2
) (
BD BD BD BD
X BD BD u
- ~Challenge
Pr(~CH)
Status Quo
1 1 1 1
1 1
) (
SQ SQ SQ SQ
X SQ SQ u
- Stand Firm
1 1 1 1
1 1
) (
SF SF SF SF
X SF SF u
- 2
2 2 2
2 2
) (
SF SF SF SF
X SF SF u
- Challenge
Pr(CH)
~Resist
Pr(~RS|CH)
Fight
Pr(F|CH)
~Fight
Pr(~F|CH)
State 1 State 2 State 1 Concession
1 1 1 1
1 1
) (
CD CD CD CD
X CD CD u
- 2
2 2 2
2 2
) (
CD CD CD CD
X CD CD u
- Observable payoffs: mean payoffs + unobservable noise
u1(SF) = SF1 + ǫSF1 = XSF 1βSF 1 + ǫSF1 where ǫSF1 ∼ N(0, Var(ǫSF1))
SLIDE 16
Modeling Beliefs: Empirical Specification of Payoffs
Empirical specifications are true to those in theoretical model. War Payoff: u1(SF) = p − c1 p: Prob that State 1 wins in a war
◮ Balance of power: Capabilities ratio
c1: Cost of war
◮ Material cost: Economic development ◮ Political will to incur the cost: Democracy
Specifications of other payoffs are given in Kurizaki & Whang (2015)
◮ Concession payoffs; Status-Quo payoffs; Back-Down payoffs
SLIDE 17
Data - Dependent Variable Coercive Diplomacy Database (Lewis, Schultz, Zucco 2012)
◮ Unit of analysis: a military challenge case, plus SQ cases ◮ 93 dyadic crisis cases ranging from 1919 to 1939 ◮ Integrate both Militarized Interstate Dispute data (MID) and
International Conflict Behavior data (ICB)
◮ N = 2187 with the addition of SQ cases
Outcome ICB MID Total SQ 2094 CD 28 16 44 BD 5 7 12 SF 33 4 37
SLIDE 18
Estimation Results
Main Status Quo Second AC Democracy Payoff Variable Est (SE) Est (SE) Est (SE) Est (SE) u1(SQ) Constant MaxAge 0.58∗∗ (0.14) 0.36∗∗ (0.14) 0.14∗∗ (0.05) Democracy1 Alliance u1(CD) Constant −1.47 (1.11) 0.98 (0.91) 1.76 (1.90) 1.59∗∗ (0.42) Alliance −2.52 (1.37) −3.51∗∗ (1.16) −2.48∗∗ (1.04) −1.00∗∗ (0.30) CivilWar2 4.07 (2.13) 4.46∗∗ (1.45) 1.95 (1.82) 2.06∗∗ (0.60) Contiguity 1.13 (0.78) 3.16∗∗ (0.90) 1.09 (1.02) 0.99∗∗ (0.36) Democracy1 0.82∗∗ (0.19) u2(CD) Constant −0.40∗∗ (0.39) −1.40∗∗ (0.56) −1.31∗∗ (0.67) −1.43∗∗ (0.27) Alliance 0.67∗∗ (0.35) 0.48 (0.33) 0.41 (0.33) −0.07 (0.10) CivilWar2 −1.43 (0.37) 0.18 (0.21) 0.20 (0.32) −0.03 (0.07) Contiguity −0.17 (0.26) −0.37 (0.23) −0.02 (0.20) −0.11∗ (0.06) Democracy2 0.04 (0.04) u1(BD) Constant −5.98∗∗ (1.57) −4.09∗∗ (0.82) −3.65∗∗ (0.99) −4.19∗∗ (0.36) Democracy1 −0.32∗∗ (0.10) −0.41∗∗ (0.10) −0.25∗∗ (0.09) −0.67∗∗ (0.11) u2(BD) Constant u1(SF) Constant −3.33∗∗ (1.25) −4.62∗∗ (0.79) −3.48∗∗ (0.75) −3.78∗∗ (0.24) CapShare1 −1.30 (0.80) 0.95∗∗ (0.47) 0.84 (0.53) 0.69∗∗ (0.17) Democracy1 −0.09∗∗ (0.04) −0.37∗∗ (0.09) −0.19∗∗ (0.08) −0.68∗∗ (0.11) Develop1 0.10 (0.06) 0.09 (0.05) 0.06 (0.05) 0.01 (0.01) u2(SF) Constant −1.06∗∗ (0.39) −2.73∗∗ (0.79) −1.90∗∗ (0.74) −2.70∗∗ (0.33) CapShare1 0.50 (0.34) 0.61 (0.42) 0.41∗ (0.25) 1.00∗∗ (0.21) Democracy2 0.01 (0.01) 0.01 (0.01) 0.06 (0.05) 0.00 (0.00) Develop2 −0.01 (0.02) −0.02 (0.02) −0.01 (0.02) −0.01 (0.01)
∗∗p < 0.05,∗ p < 0.1 (two-tailed)
SLIDE 19
Estimates of the Prior and Posterior Beliefs
−10 −5 5 10 0.0 0.4 0.8
Main Model
−10 −5 5 10 0.0 0.4 0.8
Status Quo Model
−10 −5 5 10 0.0 0.4 0.8
Second AC Model
−10 −5 5 10 0.0 0.4 0.8
Democracy Model
−10 −5 5 10 0.0 0.4 0.8
Sunk Cost Model
Posterior Belief, Pr(SF|CH) Prior Belief, Pr(SF) Belief Updating, Λ
− Legend −
x−Axis: Democracy Level y−Axis: Probability
SLIDE 20
Findings: Prior Beliefs
S2’s Prior Beliefs
◮ Increasing as AC for S1
increases in 3 of 5 models
◮ Democracy Model: Concave in
AC
◮ Sunk-Cost Model: independent
- f AC for S1
S2’s Posterior Beliefs
◮ Increasing as AC for S1
increases in all models
◮ Statistically different than full
separation
◮ Sunk-Cost Model: Posterior
increases in AC S2’s Belief Updating
◮ Learning is statistically significant ◮ Lower bounds of 95% CI don’t include
zero
◮ Who updates? Everybody ◮ Except for the least democratic
regimes (Democracy1 = −10) Effect of S1’s AC on belief-updating
◮ Learning without AC for
Democracy1 < −5
◮ Increasing as AC for S1 increase in all
models
◮ Is the effect significant?
SLIDE 21
Estimates of the Prior and Posterior Beliefs
−10 −5 5 10 0.0 0.4 0.8
Main Model
−10 −5 5 10 0.0 0.4 0.8
Status Quo Model
−10 −5 5 10 0.0 0.4 0.8
Second AC Model
−10 −5 5 10 0.0 0.4 0.8
Democracy Model
−10 −5 5 10 0.0 0.4 0.8
Sunk Cost Model
Posterior Belief, Pr(SF|CH) Prior Belief, Pr(SF) Belief Updating, Λ
− Legend −
x−Axis: Democracy Level y−Axis: Probability
SLIDE 22
Statistical Test of Fully-Separating Signals
Posterior belief Belief Updating
at Democracy1 = 10 at Democracy1 = −10
Models [Lower, Upper] [Lower, Upper] Main [0.589, 0.999] [0.003, 0.361] Status Quo [0.788, 0.962] [-0.066, 0.293] Second AC [0.618, 0.999] [0.040, 0.355] Democracy [0.625, 0.990] [0.000, 0.163] Sunk Cost [0.482, 0.904] [0.002, 0.142] Bootstrapped 95% Confidence Intervals of the Beliefs and Belief-Updating (Two-tail)
SLIDE 23
Findings: Posterior Beliefs
S2’s Prior Beliefs
◮ Increasing as AC for S1
increases in 3 of 5 models
◮ Democracy Model: Concave in
AC
◮ Sunk-Cost Model: independent
- f AC for S1
S2’s Posterior Beliefs
◮ Increasing as AC for S1
increases in all models
◮ Statistically different than full
separation
◮ Sunk-Cost Model: Posterior
increases in AC S2’s Belief Updating
◮ Learning is statistically significant ◮ Lower bounds of 95% CI don’t include
zero
◮ Who updates? Everybody ◮ Except for the least democratic
regimes (Democracy1 = −10) Effect of S1’s AC on belief-updating
◮ Learning without AC for
Democracy1 < −5
◮ Increasing as AC for S1 increase in all
models
◮ Is the effect significant?
SLIDE 24
Estimates of the Prior and Posterior Beliefs
−10 −5 5 10 0.0 0.4 0.8
Main Model
−10 −5 5 10 0.0 0.4 0.8
Status Quo Model
−10 −5 5 10 0.0 0.4 0.8
Second AC Model
−10 −5 5 10 0.0 0.4 0.8
Democracy Model
−10 −5 5 10 0.0 0.4 0.8
Sunk Cost Model
Posterior Belief, Pr(SF|CH) Prior Belief, Pr(SF) Belief Updating, Λ
− Legend −
x−Axis: Democracy Level y−Axis: Probability
SLIDE 25
Statistical Significance of Belief-Updating
Posterior belief Belief Updating
at Democracy1 = 10 at Democracy1 = −10
Models [Lower, Upper] [Lower, Upper] Main [0.589, 0.999] [0.003, 0.361] Status Quo [0.788, 0.962] [-0.066, 0.293] Second AC [0.618, 0.999] [0.040, 0.355] Democracy [0.625, 0.990] [0.000, 0.163] Sunk Cost [0.482, 0.904] [0.002, 0.142] Bootstrapped 95% Confidence Intervals of the Beliefs and Belief-Updating (Two-tail)
SLIDE 26
Findings: Belief-Updating
S2’s Prior Beliefs
◮ Increasing as AC for S1
increases in 3 of 5 models
◮ Democracy Model: Concave in
AC
◮ Sunk-Cost Model: independent
- f AC for S1
S2’s Posterior Beliefs
◮ Increasing as AC for S1
increases in all models
◮ Statistically different than full
separation
◮ Sunk-Cost Model: Posterior
increases in AC S2’s Belief Updating
◮ Learning is statistically significant ◮ Lower bounds of 95% CI don’t include
zero
◮ Who updates? Everybody ◮ Except for the least democratic
regimes (Democracy1 = −10) Effect of S1’s AC on belief-updating
◮ Learning without AC for
Democracy1 < −5
◮ Increasing as AC for S1 increase in all
models
◮ Is the effect significant?
SLIDE 27
Estimates of the Prior and Posterior Beliefs
−10 −5 5 10 0.0 0.4 0.8
Main Model
−10 −5 5 10 0.0 0.4 0.8
Status Quo Model
−10 −5 5 10 0.0 0.4 0.8
Second AC Model
−10 −5 5 10 0.0 0.4 0.8
Democracy Model
−10 −5 5 10 0.0 0.4 0.8
Sunk Cost Model
Posterior Belief, Pr(SF|CH) Prior Belief, Pr(SF) Belief Updating, Λ
− Legend −
x−Axis: Democracy Level y−Axis: Probability
SLIDE 28
Findings: Effect of AC on Belief-Updating
S2’s Prior Beliefs
◮ Increasing as AC for S1
increases in 3 of 5 models
◮ Democracy Model: Concave in
AC
◮ Sunk-Cost Model: independent
- f AC for S1
S2’s Posterior Beliefs
◮ Increasing as AC for S1
increases in all models
◮ Statistically different than full
separation
◮ Sunk-Cost Model: Posterior
increases in AC S2’s Belief Updating
◮ Learning is statistically significant ◮ Lower bounds of 95% CI don’t include
zero
◮ Who updates? Everybody ◮ Except for the least democratic
regimes (Democracy1 = −10) Effect of S1’s AC on belief-updating
◮ Learning without AC for
Democracy1 < −5
◮ Increasing as AC for S1 increase in all
models
◮ Is the effect significant?
SLIDE 29
Findings: Effect of AC on Belief-Updating
S2’s Prior Beliefs
◮ Increasing as AC for S1
increases in 3 of 5 models
◮ Democracy Model: Concave in
AC
◮ Sunk-Cost Model: independent
- f AC for S1
S2’s Posterior Beliefs
◮ Increasing as AC for S1
increases in all models
◮ Statistically different than full
separation
◮ Sunk-Cost Model: Posterior
increases in AC S2’s Belief Updating
◮ Learning is statistically significant ◮ Lower bounds of 95% CI don’t include
zero
◮ Who updates? Everybody ◮ Except for the least democratic
regimes (Democracy1 = −10) Effect of S1’s AC on belief-updating
◮ Learning without AC for
Democracy1 < −5
◮ Increasing as AC for S1 increase in all
models
◮ Is the effect significant?
Need to regress the amount of updating with Democracy1
SLIDE 30
Illustrating the Effects of AC on Belief-Updating
- Prior
Posterior 0.4 0.6 0.8 1.0
- Main Model
Democracy, +0.224 Non−Democracy, +0.182
- Prior
Posterior 0.4 0.6 0.8 1.0
- Status Quo Model
Democracy, +0.324 Non−Democracy, +0.114
- Prior
Posterior 0.4 0.6 0.8 1.0
- Second AC Model
Democracy, +0.178 Non−Democracy, +0.197
- Prior
Posterior 0.4 0.6 0.8 1.0
- Democracy Model
Democracy, +0.245 Non−Democracy, +0.082
- Prior
Posterior 0.4 0.6 0.8 1.0
- Sunk Cost Model
Democracy, +0.096 Non−Democracy, +0.070
Non−Democracy (Democracy1=−10) Democracy (Democracy1=10)
− Legend −
y−Axis: Probability
SLIDE 31
Implications
The results substantiate the causal mechanism of audience costs model
◮ “Audience costs improve crisis communication through signals”
SLIDE 32
Implications
The results substantiate the causal mechanism of audience costs model
◮ “Audience costs improve crisis communication through signals”
Results allow us to test why democracies can reveal information
- 1. Transparency of democratic processes reveals government’s
intentions apart from conflict processes → Common Priors
◮ Bueno de Mesquita & Lalman (1992)
- 2. Audience costs improve government’s ability to reveal intentions
through conflict behavior → Belief Updating
◮ Fearon (1994), Schultz (1999)
SLIDE 33
Do Democracies Inform or Constrain, and How?
“Do Democratic Institutions Inform or Constrain?” (Schultz 1999 IO)
Democratic Peace Institutional constraints Democratic Advantage Democratic Prudence Informational effects Democratic Institutions
SLIDE 34
Do Democracies Inform or Constrain, and How?
“Do Democratic Institutions Inform or Constrain?” (Schultz 1999 IO)
Democratic Peace Institutional constraints Democratic Advantage Democratic Prudence Informational effects Democratic Institutions
This Paper! How do Democratic Institutions Inform?
Democratic Institutions Institutional constraints Democratic Advantage Signaling via audience costs Transparency Information revelation Schultz (1999 IO)
SLIDE 35
Hypotheses on the Informational Effects of Democratic Institutions
Two Mechanisms for Informational Effects of Democratic Institutions Signaling and Learning Institutional Transparency (Fearon 1994, Schultz 1999) (Bueno de Mesquita and Lalman 1992) S2’s Resistance∗ − − Prior Belief + + Posterior Belief + + Belief Updating +
◮ Existing research design suffers from observational equivalence (*) ◮ Hypotheses on the effect of democracy on beliefs avoid this problem
SLIDE 36
Hypotheses on the Informational Effects of Democratic Institutions
◮ Existing research design suffers from observational equivalence (*) ◮ Hypotheses on the effect of democracy on beliefs avoid this problem
Two Mechanisms for Informational Effects of Democratic Institutions Signaling and Learning Institutional Transparency (Fearon 1994, Schultz 1999) (Bueno de Mesquita and Lalman 1992) S2’s Resistance∗ − − Prior Belief + + Posterior Belief + + Belief Updating +
SLIDE 37
Testing the “Institutional Transparency” Mechanism
Least Most Effect of Democratic Democratic Democracy (Democracy1 = −10) (Democracy1 = 10)
Prior belief
40% 53% +13%
Posterior belief
60% 85% +25%
Belief updating
+20% +32% +12%
Effect of Transparency How common prior changes as S1 becomes more democratic
◮ 53% − 40% = 13% increase
SLIDE 38
Belief-Updating without Audience Costs
Least Most Effect of Democratic Democratic Democracy (Democracy1 = −10) (Democracy1 = 10)
Prior belief
40% 53% +13%
Posterior belief
60% 85% +25%
Belief updating
+20% +32% +12%
Effect of “Democratic” Signaling Signaling with AC
◮ 32% − 20% = 12% increase
SLIDE 39
Testing the Informational Effects of Democracy
Least Most Effect of Democratic Democratic Democracy (Democracy1 = −10) (Democracy1 = 10)
Prior belief
40% 53% +13%
Posterior belief
60% 85% +25%
Belief updating
+20% +32% +12%
Effects of a Threat
◮ 60% − 40% = 20% increase (Effect of Signaling w/out AC) ◮ 85% − 53% = 32% increase (Effect of Signaling w/out AC)
SLIDE 40
Robustness Check and Illustration
- Least democratic
Most democratic 0.0 0.2 0.4 0.6
- Main Model
Transparency, +0.047 Signaling, +0.043
- Least democratic
Most democratic 0.0 0.2 0.4 0.6
- Status Quo Model
Transparency, +0.145 Signaling, +0.210
- Least democratic
Most democratic 0.0 0.2 0.4 0.6
- Second AC Model
Transparency, +0.138 Signaling, −0.019
- Least democratic
Most democratic 0.0 0.2 0.4 0.6
- Democracy Model
Transparency, −0.048 Signaling, +0.164
- Least democratic
Most democratic 0.0 0.2 0.4 0.6
- Sunk Cost Model
Transparency, 0 Signaling, +0.025
Signaling (Belief−updating) Transparency (Prior belief)
− Legend −
y−Axis: Increase in Probability
SLIDE 41
Conclusion
- 1. We find that audience costs do enhance learning in crises.
◮ We estimated audience costs ◮ We estimated belief-updating ◮ Then, we show that belief-updating is statistically significant
and increasing in audience costs
- 2. We distinguish and test two mechanisms of informational
effects of democratic institutions in crises.
◮ We find evidence consistent both with the “signaling and
learning” mechanism and the “institutional transparency” mechanism
◮ We also find evidence against the “institutional transparency”
mechanism
SLIDE 42
Appendix
SLIDE 43
Empirical Strategy: Intuition
◮ Theory: mapping from preferences to outcomes.
Preference relations Choices & Outcomes Equilibrium Deduction Given by assumption
◮ Empirics: mapping from outcomes to preferences.
Preference relations Choices & Outcomes Statistical Equilibrium Estimation Given by data
◮ We ask: “given the observation of outcomes, what prefenreces make
these observed outcomes most likely according to the PBE?”
SLIDE 44
Statistical Model of Audience Costs
Estimation 1 of 2
ln L =
N
- i=1
[YSQi ln PSQi + YCDi ln PCDi + YBDi ln PBDi + YSFi ln PSFi] ,
◮ We estimate a log-likelihood function of equilibrium outcome
probabilities, covariates, payoff specification
◮ Maximization of ln L yields the vector of MLE of β’s.
SLIDE 45
Statistical Model of Audience Costs
Estimation 2 of 2
◮ Estimate var-cov matrix to estimate belief updating correctly
◮ Identification ◮ Seven additional parameters
◮ Correct estimation of belief updating ◮ Previous models as special cases (Lewis and Schultz 2003;
Wand 2006; Signorino and Whang 2009)
SLIDE 46
Testing Conjecture about Association with Democracy Audience costs ∝ democracy Audience costs of some form exist: u1(BD) < u1(SQ).
SLIDE 47
Testing Conjecture about Association with Democracy Outcomes Payoffs Variables Est. (SE) Status Quo SQ1 Constant MaxAge 0.575** 0.135 Back Down BD1 Constant
- 4.09**
0.820 Democracy 1
- 0.411**
0.104 ∗∗ = p < .01, ∗ = p < .05 (two-tailed)
◮ Fearon’s conjecture is confirmed
◮ First evidence that audience costs increase with democracy
score
◮ Support for existing applied work that attributes democratic
uniqueness to audience costs.
SLIDE 48
A note on the signaling value of audience costs
In the Sunk Cost model, the coefficient on Democracy1 is positive and
- significant. This also indicates the signaling value of audience costs.
◮ Recall audience costs ∝ democracy. Thus, this result indicates the
states with higher audience costs are less likely to issue a threat.
◮ The signaling value of audience costs stems not only from the
hand-tying effects but also from the fact that leaders with higher audience costs would be unwilling to make an explicit threat (due to
- ther kinds of costs associated with a public commitment).