Estimating minimum wage effects on alcohol abuse and wages with alternative likely affected samples and models

We estimated effects of minimum wages on workers’ wages and alcohol abuse with samples defined by education and wages and using traditional two-way-fixed-effects (TWFE) and new heterogeneous, Stacked, models that allow for staggered treatment timing. Alcohol abuse was measured with binary indicators of heavy and binge drinking and count data on days of binge drinking. Longitudinal data were drawn from the Panel Study of Income Dynamics, 2005–2019. We investigated two samples of “likely affected” groups: the first—and most popular in the literature—was based on education i.e. people with at most a high school diploma; the second was based on wages, i.e. workers whose wages were ≤ 150% of minimum wages. Models included individual fixed effects. We found strong evidence that minimum wages lifted workers’ actual wages when “likely affected” was defined by wages but no evidence when “likely affected” was defined by education in both two-way-fixed-effects and Stacked models and across a variety of samples. We found most correlations between minimum wages and drinking were insignificant. While we found weak, suggestive evidence that minimum wages might reduce days of binge drinking, overall, we could not conclude that there was strong evidence that minimum wages either encouraged or discouraged alcohol abuse or, for that matter, that there was no relationship. Our results underscored the importance of considering alternative models before reaching conclusions; had we relied solely on TWFE, we may have concluded that minimum wages decreased alcohol abuse. Finally, our findings raise questions about the credibility of the many minimum-wage-and-health studies that rely on educational categories to define “likely affected.”