Incumbency Advantage Revision

[rafaelchduran]

Dissertation Defense comments:

• Motivation
  • Why does this matter? How does this affect (a) voters and (b) politicians vs parties. Need to separate this when thinking on the 9x9 table.
  • Study how the asymmetry affects party and candidate relationship
• On event study design:
  • should I include cohort fixed effects instead of year fixed effects?
  • what is being compared? So what is the control group in each year. The Figure on the rollout is not useful on this. I need to figure this out
  • Need raw data plots and give more info on the actual data. I need to go back and so some basic descriptives stats and understand the underlying comparison
  • remember that there are some issues with the estimation; I need to check it
  • is there a change in partisan incumbency advantage, i.e. did the reform affect somehow how the control group behaved before and after the reform? I should do a robustness test by comparing pre-control municipalities to post-control municipalities partisan incumbency advantage.
• On mechanisms:
  • Difference btw resource and information-based incumbency advantage
  • Resource-based incumbency advantage is incidental to incumbent, information or effort is not
  • Cox Morgensten and Ferejohn vs. governing a city and competence
• Other comments:
  • is there a scare off of party level?
  • what is the percentage of non-term limited mayors contest? I think this question is related to the point by Fowler and Hall (2014) on how there should be a caution on the proportion of the ones that contend again

There are two ways in which I could write this paper (Tara):

1. Short JOP paper: Validation of Fowler and Hall (2014). Center on the decomposition and why it matters. See that there are actually 9 cases, a 3x3, a not 6.
2. Large contribution: Reconcile the variation in incumbency advantage in different places. There are different options: 2.1 So say these findings are seen in the US with a non-party centered system but are also present in Mexico in a party centered system. 2.2 Talk about the impact on voters and the relationships btw parties and candidates.

[rafaelchduran]

Cohort Seminar comments

Summary:

• Create table that summarizes the different implications for partisan and personal incumbency advantage
• Speak about two relationships. Voters and candidates, and candidates and parties. On the parties, do they have less power? are they weakened by the reform? This is related to the type of paper. If they are not weakend by the reform, is there party dealignment? What is at stake is accountability, rather than party dealignment. So I need to figure out what of the two relationships I'm pushing. It seems both are changing. Term limits were imposed to make candidates accountable, but results indicate that term limits behave worse with
• Add a section on what mayors are actually doing, what are there responsibilities and what do they take charge off. This can be added at the beginning of the mechanisms section.
• Introduction
  • Be more precise on the methodological contribution of the diff-in-disc
  • So think about the normative conclusions and relate it better. Recall that Ashworth do speak of normative implications of inc. advantage
• Add a discussion section before the conclusion on the scope conditions of the results. Answer the following questions: What is expected in later terms? What happens at the thresholds? What happens with larger margins? What happens with no incumbent parties? Mexico is a strong party system but what to expect in weak party systems? How much can we generalize to other settings? Do we expect something like this in larger government units like governors or the president?
• Incorporate the 2021 elections
• Add this idea somewhere: term limits lead to predatory behavior
• Figure 2: Event study things. Include t-3 and hollow dots at zero that that’s the reference group.
• There's a compound effect always: effect of the incumbency, separated from the term limits. Think about this
• On mechanisms: create a table with expectations and definitions. Mechanisms have important implications, normative ones. The welfare effects help for this.
  • Resource based inc. advantage: not what they do but what they get
  • Information based inc. advantage: Its about expenditures, not about budget. So what do they do with these resources?
  • Higher tax collection is greater efficiency, and that is what voters are observing.
  • Need to have some measures of welfare, Ashworth say that information is welfare enhancing
  • Quality based inc. advantage: Education is not a very good measure of quality. Better use experience.
  • Not collecting more revenues is a likely mechanism
• Define what parties mean in the Mexican context, strong parties and strong party system. Mexico legacy of strong parties. Reform thought to debilitate party control but not necessarily. Party system vs within party dynamics. See Klansja and Titiunik paper on this. Does party dealignment affect the party system or the within-party dynamics?

Mariana:

• Really like the part of the implications of the differences between partisan and personal, to hav a table better to summarize this.

• Section on conclusion or discussion on the scope conditions of the results

  • What would happen in later terms?
  • What happens at the thresholds? What happens in larger margins?
  • What happens with no incumbent parties?
  • Mexico is a strong party system but what to expect in weak party systems?
  • How much can we generalize to other settings?

• Confusion:

  • Partisan incumbency disadvantage turning to inc. advantage.
  • Term limits lead to predatory behavior of politicians. When remove term limits there are electoral incentives to get reelected.
    • Mechanism:
    • Just the effect of the incumbency, separated from the term limits.
    • Inc. advantage without term limits. Something to discuss.

• Figure 2: Event study things. Include t-3 and hollow dots at zero that that’s the reference group.

• Mechanisms:

  • Resource based: not what they do but what they get
  • Quality:
  • Information: encompasses everything
  • Having revenues is what matters, so what matters is how you are spending.
    • Right now they can collect more revenues
    • So look at what they are doing with those revenues
  • Its about expenditures, not about budget. Why do they want more revenues, to do more things?
    • How they used that
  • Most revenues come from taxes. So they must be doing something positive.
    • No better economic conditions because there are no consumption revenues increase.
  • Higher tax collection is greater efficiency, and that is what voters are observing.
    • What is driving the incumbency advantage
  • Try to look at expenditures and try to assess the information mechanism
    • Questions on efficiency.
  • Increases in not ear marked infrastructure, so spending more money to gain support
  • Not collecting more revenues is a likely mechanism, so what matters is what they are doing
  • For the quality:
    • Education is not a very good measure of quality.
    • Better use experience.

Anne

• Useful to have some measure of welfare, satisfaction or something like that.
  • Ashworth et al talk that information is welfare enhancing
• Useful to have more detail on what mayors are actually doing. What kind of expenditures they do. What are their responsibilities.
  • Good to start the mechanisms section
• Clarify the methodological motivation.
  • Seemed I was the first to disentangle the partisan from the personal advantage. So clarify the approach.
  • Contribution is the diff in diff in rd, its good that there are pretends. Be more precise on the exact contribution
• Here are the predictions:
  • Great but have a table better.
• On Ashworth et al
  • When doing my predictions, I was drawing normative conclusions, but the point of this paper you can’t draw normative conclusions. So think about the normative conclusions and relate it better.
• Talk on scope conditions:
  • Perhaps not so surprising that we don’t see such a large incumbency advantage. So talk about dimensions. Good to discuss this. Do we expect something like this in larger government units like governors or the president
• Mechanism:
  • Cannot do much on the information based mechanism.
  • Tease out very specific things and then say I have proxies for information, with certain caveats.
  • Not draw straight forward normative arguments.
• Confusion:
  • Parties have less power. Weakening
  • Party dealignment
  • Not really important mechanism.
    • Not talking much about party-mayors mechanisms. So a bit confusing about party dealignment.
  • What is at stake is accountability, rather than party dealignment
  • Term limits were imposed to make candidates accountable, but results indicate that term limits behave worse with
    • Be clearer.
    • Parties have predatory behavior or not.
    • 3 x 3. There it makes sense to talk about this implication. When presenting this paper.
    • Good to have in the theory section but not in the introduction.
• Hard what to prioritize in the framing.
  • If I have a more substantive contribution it would be great.
• Talk less about party-candidate relations but then it changes. Its very interesting. Almost like a motivating puzzle.
  • Politicians are predatory. Is it likely that the inc. disadvantage becomes an advantage.
  • Have this on the discussion section.
  • Ashworth is not a bad thing, it could be more nuanced.
  • I could try to get more data and see what happens long term.

Dongil:

• Scope conditions and broad implications of the study
• How to better shape the introduction:
  • Results are surprising because previous studies talk about US or Europe but mine is with na authoritarian public legacy
  • These elections is about individual candidates, not parties
  • How would we see these patterns, and similarities with different contexts produce the same patterns .
  • Disentangling partisan and personal inc. adv.
    • What do they tell us about more important concepts of democracy, like accountability, representation, responsiveness, (e.g. Titiunik Klansja on unfulfilled acountability)
    • What does it tell us about the welfare of voters, representation
  • I have avoided talking about the normative implications. The paper would be richer
• What I meant by a partisan system more deeply.
  • Define what this means better.
  • Parties institutionalization
  • Weak vs strong organization
  • Sartori. How Mexico differ from other settings.
  • The case of Mexico should be described.
• Why would we see these patterns with previous studies?
  • In Mexico with a legacy of strong parties.
  • Introduction of reelection debilitates party control.
  • But may not be consistent with
    • Party system vs. within party dynamics. Klansja and Titiunik discuss this.
    • Does party dealignment affect a strong party system
    • Absence of party luck
• Mechanisms:
  • Transfers in favor of incumbents. Can mayors ask for transfers
  • Conceptualization of different types of mechanisms into a table and more succinct
  • Inc. advantage has important implications, vs mechanisms.
    • The mechanism helps to make a normative implication
• Not just nice to have an incumbency advantage. We need to do more to get a normative implication. Not just doing the RDD. There are things about welfare.

Conclusions:

1. Discussion section with scope conditions
2. Motivate things more clearly
3. Check Mariana’s paper
   • Do this: Not just nice to have an incumbency advantage. We need to do more to get a normative implication. Not just doing the RDD. There are things about welfare.

[rafaelchduran]

Next steps to write a broad type of paper: Data and Empirics:

• [x] See if I can add the 2021 elections
• [x] check the new data on letters of intent of running for reelection in Magar's github. Answer: its only for congressmen.
• [x] The adopt file is wrong in the database.do. Check that the adopt_year is right.
• [x] check the forcing variable mv_incparty. It seems to be wrongly estimated.
• [x] update wild bootstrap do file
• [ ] Check the probability of winning. There are very few instances when someone looses in the next election.
• [ ] Redo main database and rerun main estimates
• [ ] Separate the reelection from the selection effects. I can do the difference of those btw reelection incentives and type, now I have two elections.
• [ ] Get welfare effects data: education, health, income
• [ ] Should I include cohort fixed effects instead of year fixed effects?
• [ ] Get candidates experience, schooling, and other proxies for types data
• [ ] What is being compared? So what is the control group in each year. The Figure on the rollout is not useful on this. I need to figure this out and potentially create a new figure.
  • The total of municipality-election years is 10,588 observations, assuming that all municipalities held elections from 2010 to 2021.
  • There are 8,570 municipality-election years in the dataset I cleaned from Magar. Thus, I'm missing approximately 2,018 municipality-year observations.
• [ ] Need raw data plots and give more info on the actual data. I need to go back and so some basic descriptives stats and understand the underlying comparison
• [ ] is there a change in partisan incumbency advantage, i.e. did the reform affect somehow how the control group behaved before and after the reform? I should do a robustness test by comparing pre-control municipalities to post-control municipalities partisan incumbency advantage.
• [ ] what is the percentage of non-term limited mayors contest? I think this question is related to the point by Fowler and Hall (2014) on how there should be a caution on the proportion of the ones that contend again
• [ ] Figure 2: Event study things. Include t-3 and hollow dots at zero that that’s the reference group.

[rafaelchduran]

Things to read

• [ ] See paper by Chauvin and Tricaud from Lacea 2021, cool papers folder
• [ ] Check out John marshals papers: https://scholar.harvard.edu/jmarshall/research
• [ ] See mayor's education paper sent by Mariana by Carnes and Lupu (2015)

[rafaelchduran]

Next steps to write a broad type of paper:

Theory:

• [ ] Create table that summarizes the different implications for partisan and personal incumbency advantage. Panel A. Implication for voters to candidates and Panel B. implication of candidates to parties.
• [ ] Read more deeply the Ashworth and EDM paper to understand what the information incumbency advantage is
• [ ] Reshuffle the mechanisms: separate the resource and information incumbency advantage.

Writing:

• [ ] Rewrite the introduction of the paper to make it broader.
• [ ] On the introduction. Be more precise on the methodological contribution of the diff-in-disc
• [ ] With Welfare results think how to write about normative implications. Whether this goes in the intro and/or the conclusion.
• [ ] Rewrite the theory speaking about two relationships. Voters and candidates, and candidates and parties. On the parties, do they have less power? are they weakened by the reform? This is related to the type of paper. If they are not weakend by the reform, is there party dealignment? What is at stake is accountability, rather than party dealignment. So I need to figure out what of the two relationships I'm pushing. It seems both are changing. Term limits were imposed to make candidates accountable, but results indicate that term limits behave worse with
• [ ] Add a section on what mayors are actually doing, what are there responsibilities and what do they take charge off. This can be added at the beginning of the mechanisms section.
• [ ] Discussion section after the results with scope conditions. What is expected in later terms? What happens at the thresholds? What happens with larger margins? What happens with no incumbent parties? Mexico is a strong party system but what to expect in weak party systems? How much can we generalize to other settings? Do we expect something like this in larger government units like governors or the president?
• [ ] Define what parties mean in the Mexican context, strong parties and strong party system. Mexico legacy of strong parties. Reform thought to debilitate party control but not necessarily. Party system vs within party dynamics. See Klansja and Titiunik paper on this. Does party dealignment affect the party system or the within-party dynamics?