Household responses to escalating violence in Mexico

ABSTRACT

Despite catastrophic fallout from the Mexican drug war, headlines fail to capture the full cost to civilians. Using three waves of survey data on the same households, conducted between 2002 to 2012, we show that households adjusted both their consumption habits and altered their behaviour in response to increased homicide rates in their municipality. We demonstrate that higher murder rates are associated with households reducing expenditure on visible commodities and with individuals avoiding carrying valuables and going out at night. The spending changes occurred mainly for middle- and upper-income households, while the behavioural adjustments happened mainly in households that were poorer or headed by a female. Assuming that household behaviour was in equilibrium before the escalation in violence, these adjustments represent a non-obvious, but significant, welfare loss which should be considered in any account of the costs and benefits of the drug war.

KEYWORDS:

• Mexico
• violence
• behaviour
• crime
• consumption

JEL CLASSIFICATION:

• O54
• D10

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplementary data for this article can be accessed here.

Notes

¹ See, for example, Gutiérrez-Romero and Oviedo (2017), Carrasco and Durán-Bustamante (2018), Torres-Preciado, Polanco-Gaytán, and Tinoco‐Zermeño (2017), Carreón Guzmán et al. (2015), Brown et al. (2019), Velásquez (2019), and Brown and Velásquez (2017). There are also a large number of papers examining the reasons behind this huge increase in drug-related violence (e.g., Dell (2015), Calderón et al. (2015), Mejía and Restrepo (2016), and Correa-Cabrera, Keck, and Nava (2015)).

² For more on conspicuous consumption and violence, see Mejía and Restrepo (2013) and Hicks and Hicks (2014). Levitt (1999, 88) argues that ‘individuals distort their behaviour in costly ways (for example, by moving to the suburbs, investing in security systems, or not going out after dark) … [and that] … any measure of the burden of crime should incorporate not only the costs of those victimized, but also the investment made to avoid victimization.’

³ There are several papers that share the same empirical strategy as us but investigate different questions. See, for example, Velásquez (2019), Brown and Velásquez (2017), and Brown et al. (2019).

⁴ For zero values, we use a truncation method (we take the natural logarithm of half the minimum value which is equal to 1 in the sample).

⁵ The advantage of using year and municipality fixed effects is that there are many potential sources of endogeneity in the analysis, including time-invariant municipality level characteristics and trends common to all of Mexico. These will be absorbed by the fixed effects. Thus, the results are thus using variation in the level of violence and expenditure within municipalities over time.

⁶ Appendix Table A2 summarizes our various control variables, including the age, gender, marital status, and educational attainment of the head of the household as well as the number of household members, females, income earners, and children.

⁷ Data are from INEGI (Instituto Nacional de Estadística y Geografía). Summary statistics for homicide rates are shown in Appendix Table A2.

⁸ We create three sub-samples based on income: households reporting no income, reporting greater than no income but less than the top quartile, and the top 25% of income-earning households.

⁹ While households may respond to increased violence by migrating out of the affected area, Velásquez (2019) shows that migration rates are not especially high. If those who migrate are more sensitive to violence, our results understate the effect that violence has on visible consumption.

¹⁰ This is consistent with the findings of Di Tella, Galiani, and Schargrodsky (2010), who show that Argentines wear less jewellery during times of increased violence.