Comments Comparative Seminar Spring 2021

[rafaelchduran]

Comments on the presentation

• Bring the RQ upfront to the first slide

Comments on the research project

• Neal: 5 papers in 1. Cut it down to a simple story with one outcome (or similar ones), assure conditional independence or identification, and one mechanism. Can have robustness checks but not too many.
• Pablo: make the story clear; simply it.

Comments on the paper -Lucía: where does the money come from when signing an agreement? Is the budget now controlled by the upper-level government?

• Tara: Are you measuring spillovers? is there a change in the likelihood of nearby municipalities signing the agreement if you signed it? This is important in terms of causal inference
• Tara and Amy: Centralization and decentralization are usually referred to as a top-down decision rather than a bottom-up one. It might be better to call this a subcontracting dynamic. They are right on this sense so I should check the literature. Moreover, this might be more related to the proxy wars literature.
• Amy on the story: -- having violence increase complicates the story. It would be better if this didn´t happen. Why would they do this? -- the story might be simpler in that when faced with reelection incentives mayors try not to outsource any public good provision to anyone if possible and grab control. They further want to control the bureaucracy. -- if it's given away, then you cannot use it as an instrument for vote gathering. -- what I mentioned is that there is still shirking from the bureaucrats, sort of a tradeoff between top-down monitoring lost and control from the mayor. The net effect is negative. -Other thoughts: --Can mayors coordinate to push forward reelection? Do they have a say in the timing of the reform?

Things that could be done empirically:

• [ ] Lucía: Exploit the difference between an agreement signed with the governor and with the president. Governor can credit claim but no the military under the president. This is a placebo test. It a bit hard to separate the federal ones from the rest as well as mando unicos only. Mando unicos have the benefit that everything, including expenses is delegated. Could play around with this.
• [ ] Martin: Exploit different crimes and variation in cartel presence. See the paper by Magaloni et al. on favelas in Brazil. The problem I will face is that other crimes face the problem of underreporting.
• [ ] Jaime: There might be different degrees in the level of centralization. I could exploit the variation on the items that make up the contract itself. Here, there can be three types of contracts. (1) decentralized, (2) centralized, and (3) cooperation relation where there is information from the top but the execution is made by the bottom. So public security provision is decentralized but there is cooperation that increases information to do so.
• [ ] Lucía: Exploit heterogeneous effects by pretreatment levels of trust in the police and the military. Its weird to see that the mayor reacts by going local when citizens mistrust the police far more than the military.

[rafaelchduran]

How to simplify the paper?

1. On outcome (decentralization): have solid robustness tests. Exploit the differences in the type of agreements and who are the agreements signed with. -could play around with the constitutional reform and push from the PRI. Like an additional DiD
2. On mechanisms: Cut mechanisms to one. Everything else condition on it. 2.1 Responsiveness: (a) difference by security preferences; (b) party alignment (story of resources); (c) trust in police. --for placebos: use other worries that shouldn´t be related to public security provision. E.g. poverty; increase in prices; education; natural disasters. See question ap4_2. --use worry about corruption as a thing were we should expect the opposite. --maybe construct an dummy of those states that the governor has more trust than the mayor and viceversa. --trust in transit police shouldn´t matter but I´m assuming that people know the difference between police forces. --for placebos it would be cool to show that they are highly correlated to mechanisms but they don't show results. 2.2. Selection: (a) of incumbents; (b) of challengers 2.3. Cartel presence: condition on this. 2.4. Allowing overgrazing of the bribe base: keeping bureaucrats happy.
3. On electoral and welfare consequences: Cut number down. Check which ones are actually affected by the decrease in centralization.

How to do this?

1. Fix controls: multiply pretreatment by year as if it was a time-invariant covariate, or include the time-variant dummies.
2. Do matching on pretreatment covariates.
3. Do (1) of the above
4. Do (2)
5. Do (3)

[rafaelchduran]

On framing:

1. decentralization vs. centralization
2. reelection incentives make mayors bring in the apparatus for electoral purposes (nothing out of his hands)
3. subcontracting dynamic

[rafaelchduran]

What other story could explain subcontracting?

1. local states strategically opt to remain weak, e.g. as a commitment not to expropriate, as in Acemoglu 2005; as a way of preventing rivals having an effective means of taxation, as in Besley and Persson, 2009; or an effort to prevent the formation of a powerful army capable of sharing rents, as in Acemoglu et al., 2011b.
2. local public security provision benefits the party in power by not shifting the political agenda towards the national scope (political agenda Acemoglu paper)
3. parrochial transfers emerge as the most economical way of meeting citizen demands in the absence of state centralization (political agenda Acemoglu paper)
4. reelection incentives leads local mayors to credit claim even at the expense of a crime increase.

Literatures to check

1. delegation
2. indirect rule. E.g. Garfias in Mexico
3. credit claiming

[rafaelchduran]

Simple subcontracting story

• when the responsibility of security is granted to local authorities, they face the decision to rely on upper levels of government to fight crime
• Plus: acquire greater capacity to fight crime; blame upper levels of government for failures
• Cons: monitoring structure that puts pressure on local bureaucrats and stops their overgrazing; cannot credit claim the fight against crime
• Reelection should change this dynamic and increase a handling over local security provision.
• Why? to a) credit clame and b) to please your bureaucracy, important in clientelistic settings.

[rafaelchduran]

Simple incumbency advantage story:

• Klansja and titiunik found that incumbency advantage...
• Problems: Based on party to members assumptions
• Or Reelection should change this party to member dynamic
• Reelection generated an incumbency advantage
• Why? There are three options. Resource based inc. advantage; Productivity based inc. advantge; quality based inc. advantage; and information based inc. advantage. I find that the increase in incumbency advantage is explained by an increase in productivity of local governments.

[rafaelchduran]

Next steps for subcontracting paper:

1. Get Mando Unico final estimates
2. Get Mando Unico robustness
3. Work on het. effects by security preferences and alignment
4. Show mediation effect on security provision and violence

[rafaelchduran]

Robustness:

1. Change reference period. This is important because everything is referred to my period of 2014 that has the acuerdo_gobestatal2
2. trim ends