Labor Market Effects of Urban Riots: An Experimental Assessment

Abstract

We propose to measure the effects of urban riots on the labour market prospects of workers residing in affected areas through the channel of labour market discrimination based on locality. We investigate the case of the French riots of 2007, which were very geographically concentrated. The town of Villiers-le-Bel is selected as the treatment unit because it received a uniquely high degree of unfavourable exposure in the media. Two other towns serve control groups: i) Sarcelles, which is contiguous to Villiers-le-Bel, has a similar socio-economic-demographic profile, and did experience some rioting activity, and ii) Enghien-les-Bains, which is considered to be economically advantaged and did not experience rioting activity. Using the technique of correspondence testing, we are able to discern disparities in call-back rates for fictitious candidates who respond to actual job postings over four dimensions: gender, ethnic origin, locality of residence (advantaged vs. disadvantaged), and the degree of media exposure during the riots. We propose a new empirical approach to measure discrimination across several dimensions that integrates a set of relevant parameters into one unified system of equations with a compact and tidy structure. We decomposed the probability of receiving a callback for any candidate of given characteristics as a function of several parameters that are evaluated through difference-in-differences estimators. We find statistically significant negative effects of a pure media exposure effect. All other factors held constant, people residing in the area which received negative publicity were 3.2 percentage points less likely to receive a callback. The group of workers who tend to be the most associated with the riots, i.e. men of North African origin (at least in terms of perceptions), are the least affected by potential discrimination by region of residence, while women of French origin are the most affected.