Estimating Causal Effects of Alcohol Access and Use on a Broad Set of Risky Behaviors: Regression Discontinuity Evidence

Abstract

A growing body of evidence suggests large increases in criminal behavior and mortality coinciding with a young adult's 21st birthday, when alcohol consumption becomes legal. The policy implications from these findings have focused on the need to reduce drinking among young people, potentially by enforcing stricter alcohol controls. However, mortality and arrests are relatively infrequent outcomes and relatively less is known about the intermediate and more prevalent consequences of legal access to alcohol at age 21. This paper uses the Add Health data combined with a regression discontinuity approach to examine the effects of alcohol access on sexual behavior, drunk driving, violence, and other outcomes. The results suggest relatively large effects that appear concentrated in men. The sample also allows some suggestive policy implications on whether changing the minimum drinking age may reduce these consequences. (JEL I12, I18)

Health policy and genetic endowments: Understanding sources of response to Minimum Legal Drinking Age laws

“…Figures A1–A4 plot age‐in‐month cell averages of binge drinking around the age‐21 cutoff as well as super impose a lowess estimator on either side of the cutoff. Figures A1 and A2 mirrors (Fletcher, 2018) and uses the full sample to compare 12 and 24 months windows. Anticipating the need to focus on a European ancestry sub‐sample, Figure A3 shows the known elevation in binge drinking for white respondents (when compared to Figure A2, which has all respondents) and continues to show a visual “jump” at age 21.…”

“…Table 2 reports results from Equation () on alternating sub‐samples of the data. Column 1 provides the estimate for the full sample to match analysis in Fletcher (2018) regardless of ancestry and genotyping. Column 2 shows that the sample restrictions increase the estimate on gaining legal alcohol access at age 21 on binge drinking from ∼5 to 6 percentage points.…”

“…The analysis follows a relatively straightforward empirical setup (Fletcher, 2018, see also for a battery of demonstrations of the appropriateness of the research design and insensitivity to using alternative specifications in estimating MLDA impacts in the Add Health) using a simple linear specification for the running variable (age‐in‐months): …”

“…For example, there is a large literature documenting the average impacts of the Minimum Legal Drinking Age (MLDA) in the United States on a variety of outcomes such as alcohol use, mortality, morbidity, crime, school performance, etc (Carpenter & Dobkin 2009, 2011, 2015, 2017; Fletcher, 2018; Hansen & Waddell, 2018; Lindo et al, 2013). Some of these studies also highlight additional heterogeneities of the effects, which often vary by sex as well as precursors to the outcome of interest.…”

Light, Moderate, and Heavy Drinking

“…They measure heavy drinking with alcohol-related hospitalizations and find a large increase in these hospitalizations among newly eligible teens as compared to slightly older (previously eligible) teens. Studies by Carpenter et al (2016) and Fletcher (2019) use regression discontinuity to show that reaching the legal drinking age in Canada and the USA, respectively, is associated with an increased incidence of binge drinking episodes.…”