Alcohol and Health

Causal Effects Of Alcohol Consumption On Outcomes

The other area where economists have contributed substantially to the literature on alcohol is in estimating causal effects of alcohol consumption on outcomes. Adverse health events such as mortality, crime, and risky sexual behavior are the most widely studied outcomes, and the pharmacological profile of alcohol consumption makes a causal role for alcohol in determining each of these outcomes eminently plausible. Of course, extreme alcohol consumption can directly lead to respiratory failure and death. But there are many other pharmacological mechanisms as well. By reducing reaction time and peripheral vision, alcohol-involved driving can directly increase motor vehicle fatality risk. By altering perceptions of right and wrong and compromising a person’s ability to reason through the consequences of one’s choices, alcohol consumption can increase risk-taking that could lead to many other types of nonvehicle-related accidents and to the commission of several types of crime. By increasing aggression and exaggerating emotional state, alcohol consumption can increase the likelihood individuals will commit a violent crime. By incapacitating a person, alcohol consumption can increase criminal victimization risk. And the social aspects of drinking can put people in situations that independently increase their risk of an unwanted physical or sexual encounter. All of these channels make it plausible that alcohol use can cause adverse events.

The plurality of research studies in economics examining the effects of alcohol have examined mortality as the outcome of interest. Although mortality is rare, it is very well measured and is an unambiguously negative outcome. Mortality also has the advantage that certain types of deaths are more likely to be alcohol related than others, for example, motor vehicle fatalities are far more likely to be attributable to alcohol than cancer deaths, and studies of the blood alcohol levels of decedents show that very high proportions of deaths from suicide, falls, drowns, burnings, and other ‘external’ causes are alcohol involved. This means that a relationship between alcohol prices and policies and deaths that are more commonly thought to be alcohol related can provide stronger evidence of a causal role for alcohol use in mortality events. Motor vehicle fatalities are by far the most commonly studied mortality outcome; in the US these data provide the additional advantage that accident characteristics such as time (e.g., nighttime vs. daytime) and day (weekend vs. weekday) can strongly correlate with the likely involvement of alcohol as a contributing factor. Morbidity and nonfatal injury share many of these same benefits (to researchers) as mortality, but availability of comparable large-scale morbidity data spanning multiple places and time periods has been much sparser in the past three decades (with a few exceptions such as occupational and workplace injuries, which are tracked administratively).

Many economics studies report that areas with higher alcohol taxes or stricter alcohol availability regimes have lower motor vehicle fatality rates, though as with the alcohol consumption evidence, studies in this literature have not uniformly shown that alcohol excise tax increases lead to significant motor vehicle mortality decreases. Other quasi-experimental approaches, however, have strongly demonstrated that higher full alcohol prices reduce mortality. For example, economics research has used DID approaches to demonstrate that higher (lower) drinking ages reduce (increase) motor vehicle fatalities in the age groups newly illegal (legal) to drink that are likely to have involved alcohol. More recently, RD approaches have also shown that mortality rates for motor vehicle deaths and suicides increase discretely at the MLDA, suggesting a causal role for alcohol in these mortality events. Perhaps not surprisingly, drunk driving laws such as state movements to lower legal blood alcohol-content thresholds have also been shown to directly and significantly reduce motor vehicle deaths likely to have involved alcohol.

Of the other adverse outcomes associated with (and possibly caused by) alcohol consumption, crime and risky sexual behavior have received the most attention from economists. Both of these outcomes have the advantage over mortality that they are very common events routinely associated with alcohol. Indeed, vast public health literatures show that individuals who consume alcohol are more likely to commit crime, more likely to have been arrested for a crime, more likely to be victims of crime, more likely to have engaged in sexual activity, more likely to have engaged in sexual activity at an earlier age, more likely to have had unprotected sex, more likely to have had an unplanned pregnancy, and more likely to have had a complicated birth. To what extent are these relationships causal effects of alcohol use?

Several studies have used the money price of alcohol in an IV framework to try to disentangle alcohol’s causal role in crime and violence. These studies generally find that individuals in places with low alcohol taxes are more likely to drink, more likely to commit intrahousehold violence, more likely to get into physical fights, and more likely to carry weapons, though concern about omitted variables biases from using cross-sectional variation in alcohol taxes and prices to identify these effects is a serious issue. However, multiple economics studies have used DID methods to examine whether alcohol price increases lead to crime decreases, and these studies have found evidence supporting a causal effect of alcohol availability on certain types of crime – especially violent crime. Studies of drinking ages using the similar approach of relying on state policy changes have also provided evidence that alcohol availability is causally related to crime, and more recent research also using the minimum drinking age in an RD framework has shown that arrests increase discretely at the MLDA – further evidence for a causal effect of alcohol use on the commission of crime.

Economists have also studied alcohol’s causal role in sexual activity using quasi-experimental approaches and have found some evidence that alcohol taxes are negatively related to the probability of sexual intercourse and are positively related to the likelihood of using condoms during intercourse. Other economics research has documented a negative relationship between the full price of alcohol and both teen birthrates and rates of sexually transmitted infections such as gonorrhea and syphilis, including in models that rely on changes in alcohol prices and policies for identification of alcohol’s effects. Arguably stronger evidence for such a relationship comes from research designs based on drinking ages, as these studies have shown that youths exposed to relatively more lenient drinking ages were more likely to have births than otherwise similar youths who came of age in the same state but just a few years before or after and who were exposed to relatively less lenient drinking ages. Because these youths are likely to be very similar on observed and unobserved dimensions, omitted variables bias concerns are mitigated.

In summary, much of the economics literature addressing the causal effects of alcohol use on adverse outcomes has used a variety of quasi-experimental approaches to try to overcome the potentially severe omitted variables bias concerns. These studies have had mixed success in relying on tax-induced variation in alcohol consumption, in part because large alcohol tax changes have historically been rare (at least in the US); often this has translated into precision challenges for research designs that rely on alcohol tax variation. Studies employing alternative alcohol control policies such as drinking ages have produced stronger evidence in this respect, both because there are many policy changes to work with and because multiple age-based designs can be used (e.g., DID and RD approaches). Of course, drinking-age-based designs do not necessarily tell much about the effects of alcohol at higher points in the age distribution, so more research is needed on these important questions.

Finally, it is important to note that alcohol may also have causal effects that are positive, not negative. For example, drinkers earn more than abstainers, and part of this may reflect a causal effect of drinking (plausibly related to social interactions in certain types of occupations). Similarly, very large observational studies in public health have shown that moderate alcohol consumption is associated with reduced risk of heart disease mortality, giving rise to the oft-cited benefits of a glass of wine per day. This too may reflect a causal beneficial effect of alcohol on health (biological mechanisms include the possibility that alcohol reduces plaque deposits in the arteries and reduces the risk of blood clots). Economics research on these plausible benefits of drinking is much less complete than on the costs of drinking, in part perhaps because the types of designs that can provide relatively compelling evidence on causality are better suited to well-measured acute events such as deaths and arrests (as opposed to longevity or earnings, which are more likely the product of a series of important decisions and outcomes). Understanding whether and to what extent alcohol has causal effects on beneficial outcomes is an important area for research.

Health at Advanced Ages
Education and Health