Political agency and accountability: empirical evidence

1
Lecture 9

• Political agency and accountability empirical
  evidence
• Readings T. Besley, Principled Agents?, chapter
  3

2
Political agency and accountability

• In the previous lecture we have analyzed the
  basic incentives problem in the delegation of
  power from voters to decision-makers
• Using a simple two-period model of political
  agency where
• Voter (principal) cannot observe the state of the
  world and the type of the politician
• Politician (agent) observes the state of the
  world and knows his own type (dissonant,
  congruent)
• The voter updates his prior on the politician
  type using the information he obtains on his own
  payoff ( which depends on the action undertaken
  by the politician and the state of the world)
• The voter uses election as a tool to provide
  incentives to the incumbent legislator

3
Results

• In the second period, there are no elections,
  hence every politician choose his short term most
  preferred action
• Congruent politician always chooses what voter
  want atst
• Dissonant politician choose the opposite 1-st
• In the first period, congruent politician chooses
  what the voter wants, dissonant politician
  chooses what the voter want provided that he
  earns sufficiently small rent from choosing a
  dissonant policy in the first period
• All politician who choose what voter want are
  re-elected

4
Performance and time horizon

• If the politician is congruent, we will not
  observe any difference in performance between
  first and second period
• However, if the politician is dissonant, his
  performance in the first period will be different
  than in the second period
• Politicians that deliver congruent policies are
  re-elected

5
The term limit effect

• Based on the results of our simple agency model,
  comparing politician who can run for elections
  with politicians who cannot run, we should
  observe a term limit effect
• Politicians behave differently when they can and
  cannot run for re-election
• Can we test this prediction using data on
  politicians, electoral rules and policies?

6
Empirical evidence on term limits and
performance US Congress

• Lott and Brontars (1993) analyse congressional
  voting from 1975-1990 and they do not find any
  significant change in the way congressman vote
  over policies during their last term of office
• Problem the congressman that decide to step down
  are not bound by law to do so, hence we cannot
  use their behaviour to draw conclusions on
  legislators that are bound to step down

7
Empirical evidence on term limits and
performance US Congress

• McArthur and Marks (1988) using data of
  congressional behaviour during the lame duck
  sessions of congress (i.e. the session held after
  the elections but before the swearing in of the
  new Congress) find that members who have not been
  re-elected voted differently

8
Empirical evidence on term limits and
performance US Governors

• In several US States the number of terms a
  Governor can serve is mandated by law and hence
  term limits are in place
• Besley and Case (2000) analyze
• the effect of term limits on policy choice
• The determination of re-election chances of
  governors who are eligible to stand for elections

9
Term limits and policy choices

• independent variable
• pst be the policy chosen in state s at time t
• Main explanatory variable
• Tst dummy variable that is equal to 1 when there
  is a binding term limit and zero otherwise
• Other explanatory variables
• Real income percapita
• Demografic variables
• Party affiliation of governor, etc.
• State fixed effects
• Year fixed effects

10
Term limits and policy choices

• Regression
• Given the introduction of fixed effects and the
  limited number of states changing term limits
  laws, the effect of term limit Tst is primarily
  identified from the difference between first and
  second term of office for incumbent who face term
  limits

11
Term limits and policy choices

• Policies
• Real government spending percapita
• Total taxes percapita
• Sales taxes percapita
• Income taxes percapita
• Corporate taxes percapita

12
(No Transcript)
13
Term limits and policy choices

• Results
• Governor spend significantly more when they are
  lame duck as compared to when they can run for
  elections
• The evidence on taxation is not very clear, but
  for some taxes it turns out that lame duck choose
  significantly higher taxes as compare to the term
  where they could run for election

14
Elections and performance

• Our model of agency also predicts that incumbent
  should be rewarded with re-election if they
  choose congruent policies
• Estimation of the re-election probability of
  governors
• We can use a linear probability model to estimate
  the re-election probability (Probit)

15
Voting regression

• Independent variable
• Rgst dummy variable equal to one if the
  governor g of state s in time t is re-elected and
  zero otherwise
• Explanatory variables
• Pst policy chosen during the term t in state s
• Growth in real taxes percapita
• Growth in real expenditures percapita
• Growth in real income percapita
• Personal characteristics of governor
• Years of experience in politics
• Education
• Other work experience
• Age

16
(No Transcript)
17
Results

• Governors who increase taxes are significantly
  less likely to win elections
• State Income growth is positively correlated with
  re-election probability
• Growth in state spending does not affect
  re-election probability of governors

18
Conclusions

• Evidence on gubernatorial elections an policies
  consistent with agency models of politics
• Politicians who cannot run for elections choose
  different policies as compared to non lame duck
• Electors reward (to some extent) politicians
  based on their performance (taxation policy and
  economic growth affect re-election probabilities)

19
Questions

• Consider the first table presenting the
  regression results of Besley-Case (1995). Do
  gubernatorial term limits have significant
  effects on taxes and spending? Show the size and
  significance of the relevant coefficients and
  provide a theoretical argument explaining the
  relationship between policies and term limits
• Repeat the same exercise for the second table of
  regressions presented in the lecture (voting
  regressions)