This is the 17th in this year’s series of posts by PhD students on the job market.
Crime is a major concern for citizens around the world (PewResearchCenter,2014); therefore, governments and international organizations must decide how much to invest in improving citizen security. To make these decisions, policy makers need information on how costly and effective policies are at reducing crime and on the benefits of crime reduction, which implies having a comprehensive understanding of the costs of crime. From a citizen perspective, there are costs incurred by victims (ex post costs). Moreover, even if many or most citizens never become victims, people might spend economic resources to prevent or avoid criminal victimization. In the case of labor markets, ex ante costs may arise, for example, when employees avoid crime near the workplace either by (a) adjusting their labor supply or (b) through labor mobility (i.e. switching to jobs located in lower crime areas).
Although there is evidence that citizens avoid crime (Cullen & Levitt, 1999; Chetty et al, 2016), measuring the costs of such behavior has proven challenging. One important reason is that, while ex ante costs take place before the occurrence of a crime, determining whether these costs are caused by crime (i.e., a causal effect) implies constructing a counterfactual of what would have happened in the absence of crime.
In my Job market paper “Avoiding Crime at Work: Homicides and Labor Markets”, I estimate crime avoidance costs in the aftermath of homicides that occur near employees' workplaces. In particular, I combine incident-level data on homicides in São Paulo City, Brazil, with a matched employer-employee dataset that contains information on formal-sector workers. The linked employer-employee data contain information about labor demand and supply sides simultaneously, which allow me to study (a) the crime incidence between firms and employees and (b) the type of jobs that employees switch to after a crime, which is crucial to understanding the role of labor mobility in avoiding crime. By geo-coding the precise location of both establishments and homicides and calculating distances between those coordinates, I am able to define a measure of establishments and employees' exposure to homicides based on geographic proximity. Then, to address the potential endogeneity problem between homicides and labor market outcomes, I use a difference-in-differences design that allows me to exploit hyper-local variation in the location and timing of each homicide and compare changes over time between employees working in the same neighborhood in establishments located closer and farther away from a homicide.
Understanding the Effects of Homicides Near the Workplace
Homicides are shocks that, as I show, (a) are unanticipated (i.e., they are not predicted by previous events), (b) do not predict future homicides and (c) are dissipated within a few blocks. This implies that the level of previous crime exposure and future crime risk between employees who have and have not been exposed to this type of shocks is similar overall (i.e., neighborhood characteristics are the same) and the difference in their crime exposure relies on their proximity to the crime scene. However, the salience of spatial proximity may still affect both employees and firms in several ways. From employees' perspectives, such an event near their workplace may affect, for example, their risk attitudes (e.g., Callen et al, 2014) and crime perceptions (e.g., Mastrorocco & Minale, 2018), triggering a crime avoidance behavior and changing their employment choices. Similarly, customers may also affect their consumption choices and affect firm sales. As a consequence, firms may need to reallocate resources to attract both employees and customers (Besley & Muller, 2018) or to substitute those workers who leave after a crime (Jager, 2016).
Employees and Firms Labor Market Responses to Homicides
Employees working in establishments within five blocks from a homicide, relative to those working within five to ten blocks, experience a significant and persistent reduction in labor earnings. In particular, 5 years after the homicide, employees' weekly labor earnings are reduced by 7.7%. This decrease on labor earnings is driven by an effect, about the same magnitude, on the hourly wage rather than by a decrease in hours worked.
Then, I analyze the crime incidence between firms and employees and do not find evidence consistent with firm labor market responses to homicides. On the contrary, I find that the effects on wages are driven by employees' responses to homicides. In particular, these results are driven by the fact that being exposed to a homicide significantly affects the type of jobs that employees switch to: exposed employees switch to establishments that typically pay lower wages and are located in other municipalities (i.e., outside São Paulo City).
Why Do Employees React to Homicides?
One reason why employees may react to homicides is because, for example, they dislike working at a location where a murder took place (i.e., backward-looking behavior). However, a second reason is that employees may be preventing a future crime exposure (i.e., forward-looking behavior). While the occurrence of a homicide does not predict another one in the future, the salience of the spatial proximity might alter exposed employees’ risk attitudes or perceptions about crime, triggering a crime avoidance behavior. I find that workers move to establishments located farther from the crime scene and to municipalities with lower murder rates, consistent with a forward-looking behavior (crime avoidance). Overall, the costs of crime (i.e., lower earnings) are estimated after the occurrence of a homicide; however, they have an ex ante interpretation in the sense that employees avoid future crime exposure by moving to safer areas (i.e., ex ante costs)
Compensating Wage Differentials?
Given that I find evidence consistent with employees avoiding crime, firms located nearby a crime scene might find it difficult to attract new workers and fill new vacancies. As a result, firms would need to offer, for example, a compensating wage differential. However, I do not find evidence consistent with firm labor market responses to homicides. One possible reason is that, given that my estimates are locally identified, a firm effect may not be detected if, for example, new employees equally dislike all firms within the whole neighborhood that experienced a homicide. However, I find that the effects on exposed employees fade away completely at five blocks from the homicide meaning that it is possible that there is still a sufficient labor market to fill new vacancies and therefore exposed firms are not significantly affected by homicides.
Taken together, these findings highlight the importance of crime avoidance costs. Homicides increase exposed employees' probability of working in establishments that are located in safer areas but at the expense of a significant reduction in labor earnings. Overall, obtaining causal estimates of the costs derived from avoiding crime has proven challenging; however, having a comprehensive understanding of such costs of crime is necessary to inform policies and investments to improve citizen security. These findings provide evidence that, in addition to the costs imposed on victims (ex post costs), crime avoidance costs (ex ante costs) are consequential when designing and evaluating policies that cost-effectively prevent crime.
Camila Navajas-Ahumada is a Ph.D. Candidate at UC San Diego.
Join the Conversation