Economics Perspectives On Alcohol Use And Alcohol Regulation: Distinguishing Factors
Economics may not be the first discipline that comes to mind as relevant for studying alcohol. As such, it is useful to clarify the distinguishing characteristics of the economics way of thinking that are relevant for understanding this important topic. Arguably one of the most important distinctions is that economists put value not only on the costs of alcohol consumption in terms of productivity losses, health impairments, and criminality but also on the benefits of alcohol consumption. That is, a great deal of alcohol consumption is utility increasing, and these benefits of drinking must be taken into account when considering tighter restrictions on alcohol availability. The public health tradition, in contrast, generally calls for stricter alcohol control to reduce alcohol-related harms without consideration for the benefits of drinking that accrue to most moderate drinkers. Economics recognizes that adoption of stricter alcohol control policies for the purposes of harm reduction imposes deadweight loss on moderate, responsible consumers. Higher taxes, for example, may reduce alcohol consumption by people whose drinking causes them to be at risk for adverse health events or to commit crime but may also reduce the consumption by law-abiding drinkers. Because a large share of the population consumes alcohol and does so in a responsible way, the foregone value of alcohol consumption by this group cannot be easily dismissed.
This does not mean that economists oppose any move to tighten alcohol restrictions. But the discipline does provide a unified framework for thinking about the conditions under which government intervention in the form of alcohol control may be justified. Specifically, if drinkers impose costs on other members of society (e.g., an alcohol-involved driver may kill or injure someone, or a drinker may commit a crime against someone), it is said that the marginal social costs of alcohol are greater than the private costs (i.e., there is a negative externality), leading unregulated private markets to result in too much alcohol consumption and resulting in alcohol-related harms. In this case, economics theory justifies correcting this behavior in a variety of ways. Next, a host of alcohol control regulations are described that have been proposed and adopted across many places and that deal with the negative externality problem in very different ways. It is important to remember, however, that because economists value both the benefits of drinking and the harms, the socially optimal level of alcohol consumption and alcohol-related harms will be lower than in a completely unregulated environment but will be strictly positive.
A final distinguishing feature of the economics tradition with respect to research on alcohol use and alcohol control is that the discipline of economics has been a leader in the social and public health sciences in advancing methodologies regarding causal inference. In many cases, including alcohol consumption, researchers are faced with the problem that observed associations between a treatment (here, drinking) and an outcome (e.g., death, illness, productivity, crime, etc.) may be simultaneously determined. That is, factors that affect the treatment may independently affect the outcome. In the case of drinking and adverse health outcomes, for example, one might worry about population heterogeneity in risk attitudes and discount rates (i.e., how much people trade off utility today against utility at a later date). It could be that heavily discounting the future causes people to both consume alcohol and engage in other risky behavior that puts them at risk of an adverse health event. If so, one might observe that people who drink are at an increased risk for adverse health events even if there is no direct causal effect of alcohol. Put differently, those same people might have experienced the adverse health event even in the absence of their drinking; alcohol consumption and adverse health events may both simply reflect their high discount rate. To see the importance of disentangling correlation from causation, note that alcohol availability can be (and is) regulated by local, state, and federal governments. If the correlations between alcohol use and adverse events are not causal, then tighter alcohol control will not be an effective means to improve population health; if, in contrast, alcohol use does cause adverse events, then stricter alcohol policies can be expected to reduce not only drinking but also subsequent adverse outcomes. The relative importance of distinguishing correlation from causation varies dramatically across disciplines, with economics very much at the end of the spectrum that cares deeply about this distinction. Public health, health services research, and sociology do not place as much of a premium on this component of research; in these traditions, detailed descriptive analyses of associations between alcohol use and individual-level factors are more common.
How do economists deal with the evaluation problem (sometimes referred to as ‘omitted variables bias,’ ‘unobserved heterogeneity bias,’ ‘endogeneity bias,’ ‘simultaneity bias,’ and others) when treatment assignment is nonrandom? First, note that the ideal solution to nonrandom treatment assignment commonly used in the natural sciences is to randomize treatment and compare outcomes between the treated and untreated; because the treatment assignment was manipulated to be random, the difference in outcomes can be causally attributable to the treatment. In the real world, however, researchers cannot randomize alcohol consumption, and so social scientists have had to take different approaches. One is to try to control for as many of these omitted factors as possible in regression models either directly or through the use of single indices such as propensity scores; these approaches are common in health services and some economics research. In the past few decades, however, economists have pushed for stronger research designs that mimic the experimental variation in the natural sciences. This class of methods, commonly referred to as ‘quasi-experimental’ approaches, includes difference-in-differences (DID), instrumental variables (IV), and regression discontinuity (RD) approaches, among others. When applied appropriately, each of these designs isolates variation in the treatment that is thought to be ‘exogenous to outcomes’ or to create variation in treatment that is ‘as good as random’ for some subpopulation of interest, thus overcoming the omitted variables bias problem. An example with respect to alcohol availability, alcohol consumption, and outcomes is research that has capitalized on labor strikes for workers at government-run liquor stores in Scandinavia (where the government owns a liquor monopoly), which exogenously reduced alcohol availability, alcohol consumption, and subsequent alcohol-related problems. These rigorous standards for identification of treatment effects also distinguish the economics approach to studying alcohol consumption and alcohol control from other disciplinary traditions.