This is the 15th in this year’s series of posts by PhD students on the job market.
Exposure to violence is a pervasive and growing development challenge. In 2017, 220 million people lived in close proximity to conflict areas, twice as many as ten years before (World Bank 2020). The threat goes beyond armed conflicts. Homicide, rape, and kidnapping are "everyday crimes" in many parts of the world. Global economic shocks, climate change, and rising income inequality can all trigger new surges in violence.
The effects of violence are numerous and complex, and we should not expect them to be gender-neutral. Changes in crime and victimization often have gender-differentiated impacts on labor and marital outcomes. A large body of research has provided substantial evidence showing how shocks that worsens an individual's options outside of marriage affect their intra-household bargaining power. Increases in community violence may therefore exacerbate gender inequalities inside the household.
In my job market paper, I estimate the effects of the large surge in drug-related crime of the late 2000s on household expenditures and intra-household bargaining power in Mexico.
The Mexican Drug War
Mexico experienced a sudden, unanticipated, and huge increase in violent crime in 2007. The homicide rate in the country almost tripled within five years. The increase in homicides per capita was so drastic it surpassed countries in the midst of armed conflicts. The increase in drug-related violence goes beyond homicides. Mexican civilians have been exposed to a much higher prevalence of other crimes as well, including extortion, kidnapping, rape, and human trafficking.
Extensive research has studied the drivers of this rapid spike in violence. The most accepted hypothesis is that the surge in violence was an unintended consequence of the Mexican drug war (Dell 2015, Calderon et al. 2015, Lindo and Romo 2018). Within weeks of President Felipe Calderon’s election in December 2006, the federal government deployed thousands of troops to fight drug trafficking organizations (DTOs). The government combined its militarized approach with a “kingpin strategy” of arresting the leaders of the main drug cartels. The number of DTOs skyrocketed, and violence both escalated and spread geographically as drug leaders fought for territorial control (Coscia and Rios 2012).
Researchers have documented many negative impacts of the Mexican drug war. These include gendered effects on fear of victimization and labor force participation, both among self-employed and blue-collar workers. Qualitative evidence on social norms also suggests the so-called “Narco culture” may be a further setback to women’s empowerment.
The effect on household expenditures
Did the increase in crime weaken women’s control over household budgets? To explore this question, I estimate a system of household demand equations in which the dependent variables are the budget shares spent on non-durable goods. The specification is linear on the logarithm of total expenditure and local violence, measured by the quartic root of the municipality homicide rate. The equations include time-varying controls, household fixed effects, and location-time dummies to control for prices.
The results show that the Mexican Drug War affected the composition of household expenditures. Increases in homicides per capita decreased the budget share of food and other household necessities, while increasing the share of household expenditures allocated to private male clothing, transportation, and gambling. A household living in a non-violent municipality prior to the escalation in homicides, which then experienced the average rise in crime, decreased the share of total expenditure allocated to food by 2.3 percentage points (pp), and to hygiene and other personal care goods by 0.8 pp. In contrast, the budget share of adult male clothing and transportation increased by 0.47 pp and 1.19 pp, respectively. The results are similar when restricting the analysis to nuclear households (mother, father, and children).
The identification strategy relies on comparing the same households before and after the escalation in violence with a household fixed effects methodology. I use the 2005–2006 and 2009–2012 MxFLS longitudinal survey waves. The analytical sample includes married households formed by 2005–2006. I account for selective migration with an intention-to-treat approach assigning the municipality of residence in 2005-2006 to both survey waves. There is also no evidence of non-random attrition. The increase in violence did not affect total household expenditure, which alleviates concerns of misspecification bias. I also address the potential endogeneity of total expenditure. The coefficients on total expenditure indicate that food and personal care goods are necessities.
The main threat to identification is that the heterogeneous geographic and sharp temporal variation in homicides reported in Mexico was actually anticipated, or was correlated with other underlying trends related to households' allocations. To address these concerns, the regressions control for many time-varying characteristics, and I demonstrate that the results are consistent through multiple robustness checks.
A deterioration in women’s bargaining power
What do the changes in household expenditures tell us about changes in bargaining power? Previous research in Mexico and other developing countries has shown that improvements in women's control over the budget reallocates household expenditures toward goods preferred by female decisionmakers such as food or women's private goods. The demand estimations in this paper also display heterogeneity in line with changes in women's outside options and bargaining power. The impacts on food and male clothing are not present among women who increased their labor supply between survey waves. In households where women became more afraid of victimization, the effect on male clothing is twice as large, and there is an increase in the consumption of drinks and tobacco.
But in the context of rising violence, are the expenditure increases in male goods, or the decreases in food enough to conclude that women’s power decreased? Previous empirical work linking changes in women's bargaining power and consumption allocations has mostly relied on distribution factors. These are variables that are assumed to alter bargaining power but not preferences, for example changes in family law or cash transfers. Admittedly, this is a stronger assumption in the context of a large increase in community violence. To address this concern, I provide further, arguably more direct, evidence of a negative impact on women’s intra-household bargaining power.
I compute the effect of violence on intra-household resource shares, defined as the fraction of the total household budget individuals consume, within a structural model. Under standard utility assumptions, resource shares have a one-to-one relationship with Pareto weights. I allow the local homicide rate to affect both bargaining and preference parameters. I adapt the methodology proposed by Dunbar et al. (2013) to control for households’ unobserved heterogeneity within a panel data framework. The results suggest that, in households that experienced the average rise in crime over the period, women's resource shares decreased by approximately 5 percentage points.
Finally, I find that an increase in crime negatively affected standard measures of female decision-making. It lowered the probability that a woman would self-report as decisionmakers for the following household purchases: food eaten in the household, her clothing, her husband’s clothing, and large expenditures. In contrast, women become more likely to say it is their spouse who decides. Importantly, I find that men in such households agree when it comes to their own clothing, consistent with changes in bargaining power driving the expenditure increase in male clothing.
Understanding the range of implications of violence in communities requires looking into, and not only across, households. Specifically, there could be gendered effects of increasing local violence happening in families. Here I show the worsening of women's intra-household bargaining power – with implications for women's and maybe children's well-being. Moreover, in a companion paper, I also find suggestive evidence that women enter worse marriages as a consequence of the increase in violence.
Maria Hernandez-de-Benito is a Ph.D. Candidate at Georgetown University.
Join the Conversation