Advanced Political Economy

(ECON 3029)
Lectures 17 and 18: Representative Democracy
–
Candidates with Identity

Dr. Emilie Sartre

[email protected]

[email protected]
 What did we learn?

1. Do people vote?
2. Why do people vote?
3. For whom do people vote?
4. How does voter behaviour affect the actions of parties/candidates?

We recently studied voter participation models, focusing on the first

two questions.
However, have we fully addressed question 3 so far?
 What did we learn?

• To model representative democracy, we developed:

• The Downsian Model
• The Probabilistic Voting Model

• In these models, political parties do not have intrinsic policy preferences.

Instead, they adjust their positions to maximize electoral success—
converging toward:
• The Median Voter’s preference in the Downsian Model
• The Swing Voter’s preference in the Probabilistic Voting Model

• Once in power, parties implement the policies they committed to during

the election.
 What did we learn?

• What happens if a political party cannot commit to its policy

platform?
• In the post-election period, the party will implement its own policy
preferences rather than its campaign promises.
• This is especially relevant in situations of class conflict, where different
groups have opposing interests.

• Such a scenario is equivalent to an electoral competition

between partisan parties, which can lead to policy cycling and
the nonexistence of equilibrium in policy choices.
 What did we learn?

• Through the Principal-Agent model, we introduced the idea that politicians

may have preferences that differ from those of society.

• However, the model still assumes the existence of a congruent type—a

politician whose preferences align with the voters’.

=> Today: We will explore electoral competition with partisan parties.

Voter Participation Topic 1 - Direct Democracy

Topic 2 – Representative Democracy

a) The Downsian model
b) The Downsian model and its limits for
Representative Democracy
c) Political Agency
d) Deterministic vs. Probabilistic model
e) Voter participation
f) Voter participation and Empirical studies
g) Candidates with identity
h) Failures of the electoral control: Lobbying
and campaign contribution
 Learning Outcomes

1. Develop a clear understanding of partisan politics and its impact on

the electoral process.

2. Analyse how the electoral system shapes the number of political

parties and influences their ideological divergence or convergence.

3. Examine the role of institutions in shaping political dynamics,

particularly how business interests and powerful groups influence
policymaking.
 Outline

A. A model of electoral competition with partisan parties

B. Proof

C. Intuition

D. Readings: The Citizen-Candidates model and Businessman candidates

Recommended readings (Introductions):

Osborne and Slivinski, A Model of Political Competition with Citizen Candidates, Quaterly Journal of Economics (1996)
Gehlbach, Sonin, and Zhuravskaya, Businessman Candidates, American Journal of Political Science (2010)
 Partisan Parties and Electoral competition

• Analyzing Electoral Competition with partisan parties

• Assume a single policy dimension, denoted p

• Policy space: p ∈ P, where P is convex and compact (use for the

maximization of the Utility functions)

• Two competing parties: Party A and Party B

 Partisan Parties and Electoral competition
• Objective function of Party A:

𝒎𝒂𝒙𝒑𝑨 𝑷(𝒑𝑨 , 𝒑𝑩 ) [𝑸 + 𝑾𝑨 (𝒑𝑨 )] + (𝟏 − 𝑷(𝒑𝑨 , 𝒑𝑩 ))𝑾𝑨 (𝒑𝑩 )

• Objective function of Party B:

𝒎𝒂𝒙𝒑𝑩 𝑷(𝒑𝑨 , 𝒑𝑩 ) 𝑾𝑩 (𝒑𝑨 ) + (𝟏 − 𝑷(𝒑𝑨 , 𝒑𝑩 ))[𝑸 + 𝑾𝑩 (𝒑𝑩 )]

• Definitions:
• 𝑾𝑨 (𝒑), 𝑾𝑩 (𝒑): Party utility functions
• Q ≥ 0: Rent from being in office
• 𝑷(𝒑𝑨 , 𝒑𝑩 ) : Probability that Party A wins the election
 Partisan Parties and Electoral competition
Voter behaviour:

• Consider the case where 𝑷(𝒑𝑨 , 𝒑𝑩 ) is determined by voter

preferences:
❖ If 𝑷(𝒑𝑨 , 𝒑𝑩 ) = 1 if 𝑈 𝑀 𝑝𝐴 > 𝑈 𝑀 (𝑝𝐵 )
❖ If 𝑷(𝒑𝑨 , 𝒑𝑩 ) = 0.5 if 𝑈 𝑀 𝑝𝐴 = 𝑈 𝑀 (𝑝𝐵 )
❖ If 𝑷(𝒑𝑨 , 𝒑𝑩 ) = 0 if 𝑈 𝑀 𝑝𝐴 < 𝑈 𝑀 (𝑝𝐵 )

• Note: We assume here that the social preferences are the ones of the
median voter (ie., 𝑼𝑴 𝒑 )
 Single-peaked preferences and no ideological bias in voters
 Partisan Parties and Electoral competition
Party Preferences & Policy Choice:

• Parties have strictly quasi-concave utility functions (single-peaked

preferences)

• Each party has an ideal policy:

• Party A: 𝒑∗𝑨 = 𝒂𝒓𝒈 𝒎𝒂𝒙𝒑𝑨 𝑾𝑨 (𝒑𝑨 )
• Party B: 𝒑∗𝑩 = 𝒂𝒓𝒈 𝒎𝒂𝒙𝒑𝑩 𝑾𝑩 (𝒑𝑩 )
• First-order conditions: 𝑊𝐴′ 𝑝𝐴∗ = 0 and 𝑊𝐵′ 𝑝𝐵∗ = 0
 Partisan Parties and Electoral competition
Nash Equilibrium in Electoral Competition

• Simultaneous policy choice by both parties

• Nash Equilibrium policies satisfy:

𝒑𝑵
𝑨 ∈ 𝒂𝒓𝒈 𝒎𝒂𝒙 𝒑𝑨 𝑷(𝒑 𝑨 , 𝒑𝑵
𝑩 ) [𝑸 + 𝑾 (𝒑
𝑨 𝑨 )] + (𝟏 − 𝑷(𝒑𝑨 , 𝒑𝑵
𝑩 ))𝑾 𝑵
𝑨 𝑩)
(𝒑
𝑵 𝑵 𝑵
𝒑𝑵
𝑩 ∈ 𝒂𝒓𝒈 𝒎𝒂𝒙𝒑𝑩 𝑷(𝒑𝑨 , 𝒑 𝑩 ) 𝑾 (𝒑
𝑩 𝑨 ) + (𝟏 − 𝑷(𝒑 𝑨 , 𝒑𝑩 ))[𝑸 + 𝑾𝑩 (𝒑𝑩 )]

• Intuition: Each party maximizes its expected utility given the other party’s
choice
 Partisan Parties and Electoral competition
Policy Convergence with Partisan Politics

• Calvert (1985) : Even with partisan politics, there can be policy convergence

• Theorem: Consider the partisan politics model described before:

❖ If 𝒑∗𝑨 ≥ 𝒑𝑴 ≥ 𝒑∗𝑩 and if 𝒑∗𝑩 ≥ 𝒑𝑴 ≥ 𝒑∗𝑨 , then the unique equilibrium involves
𝒑𝑨 = 𝒑𝑩 = 𝒑𝑴 and each party wins the election with probability one half

❖ Otherwise, the unique equilibrium is:

❖ If 𝑸 > 𝟎, 𝒑𝑨 = 𝒑𝑩 = 𝒑𝑴
❖ If 𝑸 = 𝟎, 𝐭𝐡𝐞𝐧 𝒑𝑨 = 𝒑𝑩 = 𝒑∗𝑨 when 𝑼𝑴 𝒑∗𝑨 > 𝑼𝑴 (𝒑∗𝑩 ) and 𝒑𝑨 = 𝒑𝑩 =
𝒑∗𝑩 when 𝑼𝑴 𝒑∗𝑩 > 𝑼𝑴 (𝒑∗𝑨 )
 Outline

A. A model of electoral competition with partisan parties

B. Proof

C. Intuitions

D. Readings: The Citizen-Candidates model and Businessman

candidates
 Proof of Policy Convergence
The Intermediate Median Voter Case

• Assumption: The median voter's preferences are intermediate relative to the two
parties' ideal points.

• Consider the situation where 𝒑𝑨 ≠ 𝒑𝑩 but 𝒑𝑨 = 𝒑𝑴

=> Party A wins for sure -- 𝑷(𝒑𝑨 , 𝒑𝑩 )=1
 Party B’s utility: 𝑾𝑩 (𝒑𝑴 )

• Therefore, if Party B deviates to 𝒑𝑩 = 𝒑𝑴 , then 𝑷(𝒑𝑨 , 𝒑𝑩 )=0.5

𝑸
 Party B’s new utility: + 𝑾𝑩 𝒑𝑴 > 𝑾𝑩 𝒑𝑴
𝟐

Conclusion: The deviation is profitable, so 𝒑𝑨 = 𝒑𝑴 ≠ 𝒑𝑩 cannot be an equilibrium.

 Proof of Policy Convergence
The Intermediate Median Voter Case when Q = 0

• If Q = 0, the argument changes:

• Party A can move slightly away from 𝒑𝑴 toward its ideal point 𝒑∗𝑨

• Party A still wins the election.

• Implements a policy closer to its preferences.

=> However, 𝒑𝑨 = 𝒑𝑴 ≠ 𝒑𝑩 still cannot be an equilibrium

 Proof of Policy Convergence
The Intermediate Median Voter Case

• Assume 𝒑𝑨 ≠ 𝒑𝑩 ≠ 𝒑𝑴 and 𝒑∗𝑨 > 𝒑𝑴 > 𝒑∗𝑩 and 𝑼𝑴 𝒑𝑨 > 𝑼𝑴 𝒑𝑩

 𝑷(𝒑𝑨 , 𝒑𝑩 )=1

=> It implies that 𝒑𝑨 ≥ 𝒑𝑴, otherwise, party A could still find 𝒑′𝑨 such that 𝑼𝑴(𝒑′𝑨) > 𝑼𝑴 𝒑𝑩

• Party B’s response:

• If 𝒑𝑨 ≥ 𝒑𝑴, then party B can switch to 𝒑𝑩 ≠ 𝒑𝑴

𝑸
• It gains : 𝑸 + 𝑾𝑩 𝒑𝑴 if 𝒑𝑨 > 𝒑𝑴 or 𝟐 + 𝑾𝑩 𝒑𝑴 if 𝒑𝑨 = 𝒑𝑴

• Conclusion: Any situation where 𝒑𝑨 ≠ 𝒑𝑩 ≠ 𝒑𝑴 is not an equilibrium. So, 𝑝𝐴 = 𝑝𝐵 = 𝑝𝑀 is the equilibrium.

 Proof of Policy Convergence
Alternative Case – 𝒑∗𝑩 > 𝒑∗𝑨 > 𝒑𝑴 (other configurations give similar results)

• Assume 𝒑𝑨 = 𝒑∗𝑨 .

• Party B’s strategic decision:

• Any 𝒑𝑩 > 𝒑∗𝑨 loses.

• If 𝒑𝑩 = 𝒑∗𝑨 , the election is tied and the chances of being elected are equal to ½ for both
parties.

• Party B can do better by choosing 𝒑𝑩 = 𝒑∗𝑨 - ɛ

=> Closer to median voter’s preferences.
=> Wins the election due to single-peaked preferences.
Proof of Policy Convergence

Is the policy 𝒑∗𝑨 - ɛ better or equal to 𝒑∗𝑨 for Party B?

 Proof of Policy Convergence

• The difference in utility is minuscule for party B if ɛ is small enough

• The gains in terms of the rent from coming to power should be way larger.

• However, if Q = 0, the best thing party B can offer is 𝒑𝑩 ≥ 𝒑∗𝑨 as it does not care about
coming to power.

=> 𝒑∗𝑨 is the equilibrium policy

 Outline

A. A model of electoral competition with partisan parties

B. Proof

C. Intuitions

D. Readings: The Citizen-Candidates model and Businessman

candidates
 Strong Forces Toward Policy Convergence
Remember : If 𝑄 = 0, then 𝑝𝐴 = 𝑝𝐵 = 𝑝𝐴∗ when 𝑈 𝑀 𝑝𝐴∗ > 𝑈 𝑀 (𝑝𝐵∗ ) and 𝑝𝐴 = 𝑝𝐵 = 𝑝𝐵∗
when 𝑈 𝑀 𝑝𝐵∗ > 𝑈 𝑀 (𝑝𝐴∗ )

When there are no rents from winning office and both parties have the same
ideological bias, policy convergence is a strong force.

What is the key assumption here?

❖If 𝑃(𝑝𝐴 , 𝑝𝐵 ) = 1 if 𝑈 𝑀 𝑝𝐴 > 𝑈 𝑀 (𝑝𝐵 )
❖ If 𝑃(𝑝𝐴 , 𝑝𝐵 ) = 0.5 if 𝑈 𝑀 𝑝𝐴 = 𝑈 𝑀 (𝑝𝐵 )
❖ If 𝑃(𝑝𝐴 , 𝑝𝐵 ) = 0 if 𝑈 𝑀 𝑝𝐴 < 𝑈 𝑀 (𝑝𝐵 )

=> The society has single-peaked, single-crossing or intermediate preferences.

 Voter and ideology

What changes when voter’s ideological bias matters?

✓ 𝑷(𝒑𝑨 , 𝒑𝑩 ) can become continuous rather than discrete.

✓ If this happens, policy convergence does not hold! (Assume)

If 𝑃(𝑝𝐴 , 𝑝𝐵 ) is a continuous and smooth function because of probabilistic voting (or

for any other reasons). There there can exist an equilibrium where 𝑝𝐴 ≠ 𝑝𝐵 ≠ 𝑝𝑀 ,
even when Q > 0.
 Key takeaways
Equilibrium policy converges to the median voter’s ideal point:

• Any deviation leads to a strategic response from the opposing party.

• Parties adjust their policies to maximize the chance of winning.

• If rent from power exists, parties strategically position themselves closer to the
median voter.

• Special case (Q = 0): The opposition party stops caring about winning and the
convergence forces are stronger !

• Policy convergence breaks down if 𝑃(𝑝𝐴 , 𝑝𝐵 ) is continuous due to ideological

considerations among voters.
 Outline

A. A model of electoral competition with partisan parties

B. Proof

C. Intuitions

D. Readings: The Citizen-Candidates model and Businessman

candidates
 Osborne and Slivinski (1996)

The Citizen-Candidates model :

• Citizens have their own ideological preferences

• Citizens choose whether or not to run as candidates

• A winner implements her favorite policy.

 Their model is a good explanation for the observed diversity in the number of
candidates across the world and the dispersion in the candidates’ policy
positions.
 Osborne and Slivinski (1996)

The number of candidates and their policy positions depend on:

1. The cost of running as a candidate

2. The benefit of winning

3. The nature of the electoral system

• The plurality rule

• The runoff system

 Osborne and Slivinski (1996)
The plurality rule (First-Past-The-Post):

• In this system, the candidate who receives the most votes wins, even if they don't get more
than half of the votes.
• Commonly used in elections like those in the United States for Congress or the UK for
parliamentary elections.

The runoff system:

• In a runoff system, if no candidate gets a majority (more than 50%) of the votes in the first
round, a second round of voting is held.
• Typically, only the top two candidates from the first round are allowed to compete in the
runoff. This ensures that the winner has broad support from the electorate, as they need to
secure a majority in the second round.
• This system is used in countries like France for presidential elections.
 Osborne and Slivinski (1996)
Main results:

• An increase in the number of candidates when there are lower cost of running for
office and higher benefits of winning the election

• Two-candidate elections are more likely under plurality rule than under a runoff
system while multicandidate elections are less likely under plurality rule than under a
runoff system

• The maximal dispersion in the candidates’ positions in two-candidate equilibria tend

to be less under a runoff system than under plurality rule

• It is still possible that candidates with no chance of winner enter an election, simply
to affect the identity of the winner, even though such entry is optional costly
 Outline

A. A model of electoral competition with partisan parties

B. Proof

C. Intuitions

D. Readings: The Citizen-Candidates model and Businessman

candidates
What do they have in common?
 Gehlbach et al. (2010)

• Businessman Candidates

1. “Why and when do businessmen run for public office rather than rely upon other
means of influence?”

2. “What are the implications of their participation for public policy?”

 Gehlbach et al. (2010)

• Leland Stanford founded the Central Pacific Railroad Company in 1861 and won
gubernatorial office in California. He stayed President of that company throughout
his term as governor and later tenure in the U.S. Senate.

Crandall (1950): More than half of the 53 railroad presidents held public office in
the XIXth century

• Roman Abramovich : In 2000, he run for governor of Chukotka, in the Russian Far
East. He won the election but continued working as a businessman during his eight
years in office.

The authors documented 247 businessmen participating as candidates in 259

gubernatorial elections in Russia between 1991 and 2005.

• Today, Donald Trump is another example !

 Gehlbach et al. (2010)

• Main point of the paper: “Public policy and political participation are
best understood as jointly determined by the institutional environment”
(That’s why we will soon start studying how culture, politics and institutions
interact)

• They develop a model of electoral competition with postelection policy

choice similar to the citizen-candidate model that was presented before

=> Entry is endogenous: At some cost, both professional politicians and

businessmen may enter the race
The compare institutional settings where campaign promises are binding and
others where they are not
 Gehlbach et al. (2010)

• Businessmen can influence policy in two ways:

1. By lobbying the election winner for good treatment

(What we will study next!)
2. By running for election

• The key assumption: A businessman’s opportunity cost of running for

public office is higher than a professional politician’s.
 Gehlbach et al. (2010)

• Two main results:

1. Businessmen are less likely to run for elected office if elected officials are
strongly held accountable to voters

❖When democratic institutions are strong and campaign promises matter, the logic of
political competition encourages businessmen and politicians to adopt similar
platforms.

❖As businessmen have a higher opportunity cost, they do not run as candidate / can pay
politicians to run in their place (lobbying)

❖But when campaign promises do not matter, they can set policy in accordance with their
own preferences. In that case, the benefit from running may be larger than the cost of
lobbying, particularly if the politician receives rents thanks to the lobbying of other
businessmen!
 Gehlbach et al. (2010)

• Two main results:

2. When campaign promises are not binding, businessman candidates are less
likely when the returns to businessmen from policy influence are especially
high

❖When the winner can extract rents by granting or denying favours to

businessmen, there is a gain from holding office for professional politicians as
well as businessmen.

❖But given the higher opportunity cost of businessmen, the latter leave the race
when the returns from policy influence are particularly large
 Russian
Gubernatorial
elections
 Recap
• Introducing politician’s ideological preferences, we saw that:

❖ The absence of rent usually lead to stronger political convergence

❖ The number of candidates and political convergence depend on various parameters : the cost
of running as a candidate, the benefit of winning and the nature of the electoral system
✓ A First-Past-The-Post system makes more likely to have two-candidate elections with more dispersion
(Think about political polarization we will study next!)

❖Thanks to the Citizen-Candidate model, we can better understand why some businessmen run
as candidates
✓ The stronger are democratic institutions, the less likely businessmen run as candidates
✓ However, when democratic institutions are particularly weak and the rents from being in power very high,
businessmen are less likely to run for office

❖ Next time, we will study lobbying !