Published on Development Impact

How do Earlier Bar Closings affect Fetal Health in Brazil? Guest post by Romina Tome

This page in:

This is the tenth in this year’s series of posts by PhD students on the job market

 “Have I told you that my uncle was killed in a bar fight?” This may sound like a line from the movie Fight Club, but it is not. Alcohol abuse and violence act as instigators for each other and social gatherings in bars contribute to that. These and other alcohol-related behaviors influence the social and intrahousehold environments surrounding pregnant women and children in their wombs in developing countries where violence is especially prevalent. So, what do you know about how policies designed to address this issue affect families? My job market paper studies how a policy that restricts access to alcohol in public spaces affects pregnant women and children in utero. Specifically, I estimate the unanticipated impact of restricting operation hours for bars and restaurants on fetal health and survival in a large Brazilian metropolitan area.

Combining Adoption of Dry laws and Families in Brazil

Brazil is characterized by high levels of violence and alcohol consumption. Unsurprisingly, these factors are related. Aiming to reduce community violence, 18 municipalities of the Sao Paulo Metropolitan Area have established laws requiring bars and restaurants to close at 11pm, rather than stay open 24hs, and apply penalties for non-compliance. Importantly, these laws affected not only access to alcohol, but also social gatherings that triggered aggressive behaviors. The roll-out of these laws is a great source of time and space variation, as shown in Figure 1, that I exploit in a difference-in-difference approach.

 Adoption of Dry laws

Studying the effects of this policy on health at birth presents two challenges: 1) adoption of dry laws was not random; nor is the location of residence for mothers to be; 2) families respond to the laws by adjusting their fertility and/or migration decisions. Then, if “higher quality” families – or those more likely to have healthier newborns— are those who reside in or move to adopting municipalities, or those who are more likely to have children after the laws are implemented, a simple comparison of newborns’ outcomes in families residing in adopting municipalities to those born in non-adopting municipalities may lead to overestimating the effect of the laws.

To address these challenges, I leverage confidential vital statistics data to identify siblings. My empirical strategy compares outcomes for newborns prenatally exposed to the dry laws to those of a sibling who was born before the laws were implemented. By using data on siblings, none of whom were exposed to dry laws, I am also able to account for events that coincide with the adoption of these laws and for the inherent effects of birth order. To overcome endogenous fertility decisions, I focus on infants conceived before the introduction of these dry laws; while to account for migration decisions I define exposure using mothers’ residential location before the laws were implemented (and conduct an intent to treat analysis).

Overcoming endogeneity comes with a cost in terms of external validity: working with a sample of siblings rather than the whole population of newborns. As expected, based on different fertility behaviors in my sample of analysis, newborns are, on average, less likely to be white and to have less educated mothers than infants in the overall sample.

Earlier Bars Closing increases fetal survival

I find that prenatal exposure to curtailing alcohol trading hours does not significantly affect the incidence of low birth weight (less than 2,500 grams/5.5 pounds) and preterm birth (less than 37 weeks of gestation). Does that mean that the dry laws do not affect fetal health? If the laws also change the composition of live births, and less healthy infants are more likely to survive, the answer is no. In other words, these results are also compatible with a reduction in fetal losses.

To explore this, I examine the effect of dry laws on a proxy for fetal death: the probability of a live birth being male (as discussed in Sanders and Stoecker, 2015). This is based on findings that male fetuses are more at risk than females under poor conditions in-utero, and that improvement in these conditions should prevent more male fetal losses (Trivers and Willard, 1973; Kraemer, 2000). I find that the probability of having a male child is 17% greater among mothers exposed to the laws during the last three months of their pregnancy. In particular, less healthy male fetuses are more likely to survive after the implementation of the policy, which helps explain the null effects on birth outcomes.

Why do infants exposed later in their gestation benefit from dry laws? Two arguments help explain this: 1) maternal stress is related to the risk of stillbirth, and this risk is higher for males (James, 2015; Wisborg et al., 2008; Owen and Matthews, 2003); 2) fetal responses to maternal stress are stronger after week 27 of pregnancy (Van den Bergh et al., 2005), and cortisol shocks produced by maternal stress late in gestation have been linked to shorter pregnancies (Bolt et al., 2002). For example, male late-term stillbirths increase after terrorist attacks, political events, and mass layoffs. Adoption of dry laws may reduce such maternal stress, as discussed below.

How does closing bars earlier impact fetal survival?

I provide suggestive evidence on the mechanisms by which earlier closings increase fetal survival. Since this policy may have repercussions beyond drinking, I leverage several administrative datasets. Compared to places that do not adopt dry laws, reducing access to alcohol and social gatherings lead to six findings:

1.       A 15% reduction in the homicide rate, mostly among men

2.       A 12% decrease in the rate of death related to traffic accidents, mostly among men

3.       A 6% reduction in workplace accidents, mostly among men working in the construction or industry sectors

4.       There is no significant substitution for alcohol consumption of other drugs

5.       A 7% increase in the likelihood that a mother has a partner when her child is born

6.       There is no significant change in layoffs at bars and restaurants


My findings suggest that dry laws are effective in reducing violence (as it was shown by Biderman et al. 2010), drinking and driving, and changing workers’ routines (sleeping earlier or drinking less alcohol) improving their concentration during working hours. Moreover, dry laws, as well as their consequences, affect family composition. All these channels indicate positive outcomes for pregnant women that would reduce their stress level. Yet, as any reduced-form analysis, it is not possible to disentangle the role of each mechanism on my main results, nor to pinpoint all of them. 

It is worth noting that this policy may have a negative effect on families, if it leads to changes in labor market outcomes. Exploring employment data, I find no evidence of changes in layoffs in bars and restaurants after the adoption of dry laws.

Policy implications

Based on the literature documenting the longer-term consequences of fetal experiences on health, education, and labor market outcome (Barker 1995; Almond and Currie, 2011, for example), my findings highlight the importance of including the potential for improving maternal and infants’ well-being as part of the debate about alcohol-related policies. Considering that Brazil’s drinking culture is similar in many aspects to that in Australia, the United States, and some countries in Europe, these finding would be useful to policymakers in developing and developed countries. They also emphasize the importance of exploring both the intensive, and the extensive, margins of fetal outcomes. As this paper shows, examining only fetal health outcomes may lead to the wrong conclusion when there is a change in survival.

Romina Tome is a Ph.D. candidate in Public Policy at Duke University.

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000