Block 2.

1 — Government and Accountability (Precise +

Guided Explanation)

Main Objective:
We want to understand how elections can create accountability even when politicians
are imperfectly observable.
We build a simple two-period political agency model to study:
• Adverse Selection: Politicians may be good (congruent) or bad (dissonant),
• Moral Hazard: Even bad politicians may behave well if elections are strong enough
incentives,
• Role of Information: How voters use signals to decide re-election.

Step-by-Step Model Build

Step 1: Players
Why this step? Define the basic elements: who is acting and who is reacting.
• Politicians:
– Congruent (c-type): prefer policies aligned with voter interests,
– Dissonant (d-type): prefer to deviate (private rents).
• Voters:
– Mass 1,
– Care only about policy outcomes,
– Want to re-elect “good” politicians.

Step 2: Time Structure

Why this step? Specify the timeline to track decisions and information revelation.
• Period 1:
– Politician chooses action (effort) e,
– Outcome is observed with noise.
• After Period 1:
– Voters observe signals,
– Election happens: incumbent may be re-elected or replaced.
• Period 2:

1
 – Same politician (if re-elected) or new politician acts,
– No election afterward.

Step 3: Information Structure

Why this step? Voters cannot observe politician’s type directly — must infer based on
noisy outcomes.
• Voters observe:
– With probability χ, the policy outcome (good/bad),
– With probability τ , the true type of politician directly.
• If no information, voters default to re-elect based on prior beliefs.

Step 4: Politician’s Payoff

Why this step? Define what politicians care about — needed for modeling incentives.
• Politicians value:
– Office rents W (value of staying in power),
– Honest types value implementing good policies (matching true state),
– Dissonant types also value private rents (when misbehaving).
Thus, a d-type faces a trade-off:
Choice Benefit
Behave well (match state) Better chance of re-election
Misbehave (deviate) Immediate private rent

Step 5: Voter’s Strategy

Why this step? Understand how voters react based on what they see.
• If voters observe bad outcome or type revelation, they may choose to vote out the
incumbent.
• If they observe good outcome (or no information), they re-elect.

Step 6: Signal Process

Why this step? Introduce noise — key for imperfect accountability.
• If noise σ 2 is large, outcomes are harder to interpret,
• If noise is small, voters can infer politician quality more accurately.
Thus, signal structure affects re-election pressure.

Step 7: Politician’s Incentive Constraint

Why this step? Formalize the decision rule: when does a d-type behave well?
The d-type will behave congruently (match state) if:

r ≤ (1 − τ )χβ(B + µ)
where:

2
 • r = rent from misbehaving,
• (1 − τ )χ = chance that voters will see misbehavior,
• β(B + µ) = discounted future utility from staying in office.
Thus, if the future reward from staying in office is bigger than today’s temptation, the
bad politician behaves.

Step 8: Probability of Good Behavior (π ∗ )

Why this step? Find how likely a bad politician behaves well.
Given distribution r ∼ G(r):

π ∗ = G ((1 − τ )χβ(B + µ))

Thus, higher observability (χ) or lower noise increases discipline.

Step 9: Voter Welfare

Why this step? Summarize how the system’s parameters affect overall voter utility.
Voter expected welfare W (π|χ, τ ) depends on:
• Probability of good behavior π,
• Value of having a congruent politician in period 2,
• Value of catching and replacing corrupt politicians.
Key formula structure:

W = [p + (1 − p)π]∆ + β {p∆ + (1 − p)(1 − (1 − τ )(1 − χ(1 − π)))p∆}

Thus, it combines today’s behavior and future re-election outcomes.

Why Each Element Matters

Model Step Why it is essential
Players Set up who makes decisions
Time structure Determine when elections happen
Information Explain imperfect observability
Politician’s payoff Model incentives correctly
Voter strategy Capture electoral discipline
Signal structure Show role of noise
Incentive constraint Predict behavior under incentives
Probability of good behavior Measure discipline strength
Voter welfare Evaluate system efficiency

Key Insights from Block 2.1

• Elections can discipline bad politicians even when voters observe only noisy signals.
• Stronger observability (χ) improves accountability.

3
• More direct type revelations (τ ) have ambiguous effects:
– Good for selection (replacing bad types),
– Bad for discipline (bad types may give up trying to behave).
Thus, information structure critically shapes accountability.

4
 Block 2.2 — Quality of Government, Polarization, and Accountability (Precise + Guided)

Main Objective:
We want to understand how electoral competition and polarization affect political
accountability — and how voters can discipline politicians when politicians care about
staying in power.
We build a model where:
• Politicians choose policy to maximize votes,
• Voters judge politicians based on both ideology and competence,
• Elections are noisy — shocks can affect outcomes,
• Polarization (extreme voter biases) can weaken accountability.

Step-by-Step Model Build

Step 1: Players and Preferences
Why this step? Define who makes decisions and what they care about.
• Two political parties (A and B),
• Each party fields either a good or bad candidate:
– Good candidate: cares about welfare,
– Bad candidate: selfish or incompetent.
• Voters care about:
– Policy outcome (public good provision),
– Their ideological bias.
Thus, voters are partly ideological, partly policy-oriented.

Step 2: Timeline
Why this step? Clarify when choices and observations happen.
• Politicians simultaneously announce policies before election,
• Election occurs with random shocks,
• Winning candidate implements promised policy.
Thus, pre-election policies are strategic.

Step 3: Voter Utility

Why this step? Specify how voters evaluate politicians — mix of ideology + policy.
Voter i’s utility from party A:

Ui (A) = qi − αxA
and from party B:

5
 Ui (B) = −qi − αxB
where:
• qi = voter’s intrinsic bias toward A,
• xA , xB = policy levels promised,
• α = importance of policy.
Thus:
• Higher α: voters care more about actual policy,
• Lower α: voters care more about party identity.

Step 4: Election Outcome

Why this step? Model how elections are determined.
A random shock θ affects voter evaluations.
Voter votes for A if:

qi + θ > α(xB − xA )
Thus, elections depend on:
• Ideology bias qi ,
• Random shock θ,
• Policy difference (xB − xA ).
Noise θ represents random factors like campaign effects.

Step 5: Politician’s Problem

Why this step? Understand how politicians set policies strategically.
Each politician wants to maximize probability of winning:
• Good candidates care about welfare + winning,
• Bad candidates care mainly about winning.
Thus, each politician chooses x before election, anticipating voter responses.

Step 6: Cutoff Voter and Probability of Winning

Why this step? Find the ”marginal” voter who decides the election.
Cutoff voter q ∗ satisfies:

q ∗ + θ = α(xB − xA )
Thus:
• Voters with qi > q ∗ vote for A,
• Voters with qi < q ∗ vote for B.
Winning probability depends on the distribution of biases and policy difference.

6
Step 7: Polarization
Why this step? Introduce heterogeneity in voters.
Voter bias qi is distributed with dispersion ξ:
• High ξ: voters are very biased (high polarization),
• Low ξ: voters are moderate.
Greater polarization reduces sensitivity to policy.

Step 8: Politician’s First-Order Condition

Why this step? Find the strategic best-response of politicians.
Each politician’s optimal x trades off:
• Providing better policy to attract moderates,
• Staying close to ideological base.
Good politicians lean more toward welfare-maximizing policies.

Step 9: Comparative Statics: Effect of Polarization

Why this step? Understand how polarization affects accountability.
Higher polarization ξ:
• Reduces importance of policy,
• Increases weight of ideology,
• Weakens electoral discipline.
Thus, in polarized societies, elections are less effective at disciplining politicians.

Step 10: Comparative Statics: Effect of Noise

Why this step? Analyze how uncertainty affects accountability.
Higher variance of election shocks θ:
• Makes outcomes less sensitive to policy,
• Reduces incentives to provide good policies.
Thus, more randomness weakens accountability.

Why Each Step Matters

Step Why Important
Players/Preferences Who acts and how they evaluate politicians
Timeline Clarifies strategic nature of policies
Voter utility Mix of ideology and competence
Election outcome Random shocks affecting votes
Cutoff voter Determines winning probability
Polarization Captures real-world voter biases
F.O.C. for politicians Derives strategic choices
Comparative statics Shows how competition works or fails

7
 Key Insights from Block 2.2
• When voters are more ideological (high polarization), policy matters less and politicians
deliver worse policies.
• When election noise is high, voters’ choices become more random and electoral disci-
pline weakens.
• Accountability is strongest when voters are moderately ideological and elections are
clean (low noise).
Thus, voter behavior and information structure critically affect government
quality.