This is the 19th in this year’s series of posts by PhD students on the job market.
Political violence, especially assassinations of local politicians, constrains the capacity of local governments to provide public service in many developing countries. Assassinations remove individuals providing essential inputs for key local government tasks such as managing public finances, delivering public services, and recruiting personnel. Thus, assassinations can jeopardize the effectiveness of the local governments and hurt the local economy. More developing countries are expanding their local governments while experiencing rising political violence. Thus, studying the role of politicians in violent settings is crucial for understanding how capable local governments can be established.
In my job market paper, I explore this issue using assassination attempts against mayors in Mexico. They are usually targeted by organized criminal groups seeking to gain political influence and exploit local resources for their operations. Mexican mayors rank among the most targeted local politicians globally. I isolate the effects of the leaders in violent environments by comparing local government effectiveness from municipalities with successful assassinations against those with failed attempts. I examine two aspects of local government effectiveness: their capacity to collect and allocate financial resources, and their ability to retain productive public workers.
Data sources on Mexican municipalities and methods
For the analysis, I created a novel dataset that combines information from several sources. First, I collect data on assassination attempts on mayors since 2001 from online newspapers and databases on conflicts. The outcome of assassination attempts on mayors are categorized into successful attempts (85 cases), failed attempts leaving mayors unharmed (53 cases), and failed attempts leading to injuries that require time off from duty (25 cases) (Figure 1). To measure local government effectiveness, I use the municipal revenue and expenditure from the annual municipality finance report and the information on municipal government personnel from the Mexican National Institute of Statistics (INEGI). Lastly, I also use population census and other sources to obtain data on non-political violence, organized criminal presence, and demographic characteristics at the municipal level to account for other municipal characteristics that may affect local government effectiveness.
To isolate the effects of the presence of mayors in violent environments, I first narrow down the sample to municipalities with assassination attempts that either assassinate mayors (treatment) or leave them unharmed (control). Thus, the counterfactuals of successful assassinations are the failed assassination attempts that do not affect the presence of mayors at all. By comparing these two types of municipalities with similar levels of political violence and other characteristics, I net out the factors other than the presence of mayors that may affect local government effectiveness. Based on this setup, I use an event-study regression to identify the effects of successful assassinations that eliminate mayors on local government effectiveness.
Assassinations of mayors hurt local governments’ capacity to provide key services
Result 1: Municipal governments collect less taxation and allocate less to key public services
I capture the effectiveness of local governments in managing their public finances using municipal tax revenues and expenditures on key public services. Tax revenues finance the operations of local governments and represent their capacity to provide public services. Expenditures on various public services show the types of services provided and how local governments allocate resources. Thus, these variables capture how well local governments can financially support their bureaucratic functions.
The absence of mayors after assassinations can strain the local government’s ability to collect taxes and allocate resources to public services. Taxation and provision of public goods involve significant input from mayors. Management of public finances can become inefficient after sudden changes in mayors following assassinations due to factors such as the substitution of mayors’ duties by unprepared individuals or illegitimate actors. As a result, the capacity to collect taxes and allocate to key public services may deteriorate after assassinations.
The findings confirm that the absence of mayors following assassinations hurts tax collection and the provision of public services. Tax collection falls by 29% after successful assassinations (Figure 2). The shares of municipal expenditure to municipal institutions providing basic services and to non-infrastructural operating costs decline by 1.8 percentage points (hereafter p.p) and 1.7 p.p respectively (Figure 2). As they each accounted for 9% of all municipal expenditure, the size of the expenditure on these services fall by 20%. Thus, assassinations of mayors hamper local governments’ ability to manage public finances.
Result 2: Municipal governments lose their most productive workers
The next results investigate whether assassinations make retaining capable municipal workers more difficult. Assassinations decrease the benefits of working in municipal governments by exposing workers to a dangerous environment. In particular, the most productive workers have better opportunities in other sectors, further reducing the relative advantages of working in the municipal government. Thus, these workers are more likely to quit and are costlier to retain than other workers after assassinations.
I investigate this hypothesis using the municipal workers at a productive age (ages 30-49), who have the highest labor market earnings (Figure 2). I test whether they are more likely to quit than others. I calculate the increase in wages needed to retain these workers based on workforce changes following the assassinations and labor supply elasticity to wages, as estimated by Dal Bó et al. (2013).
I find that these workers are more likely to quit and costlier to retain than others. The share of municipal workers in their productive age declines by 15 p.p (Figure 2). Considering that 56% of the workers are in this age group before the assassinations, this equates to a 27% decrease in the number of these workers. I also compute that a 13% increase in wages is required to retain these workers after assassinations. Those in other age groups quit less and induce smaller retention costs. Thus, assassinations of mayors hurt local governments’ capacity by driving out productive workers.
Figure 2. Preview of main results
I find that the presence of organized criminal groups temporarily increases in municipalities with successful assassinations. I rule out alternative mechanisms that may yield similar outcomes. Non-political violence, a decrease in economic activity, and population flight could decrease tax collection and the supply of public workers. I find that there are no differences in these factors across the sample municipalities in the study. These results suggest that ineffectiveness in local governments can be explained by the influence of illegitimate actors following successful assassination attempts.
What are the implications?
The results provide important takeaways about decentralization in violent settings. Despite their ability to address local demands, decentralized governments in violent environments may fall short of achieving that goal. I show that local governments become less capable of fulfilling their tasks when illegitimate actors successfully target local politicians. Thus, decentralization may fail to fulfill its potential when illegitimate actors successfully incapacitate key elements of the local government. This is relevant for developing countries with internal conflicts and illegitimate actors that can compete with formal subnational governments.
is a Ph.D. Candidate in Economics at Columbia University, specializing in the political economy of development and public economics. You can learn more about him at: https://seunghunlee918.github.io