Economist and occasional TOTM guest blogger Steve Salop (Georgetown) recently sent me the following questions spurred by the local debate over Governor McConnell’s proposal to private the retailing of alcoholic beverages:
I have my first antitrust class of the semester tomorrow. Among the issues I teach the first week are (1) the fact that demand curves slope down; (2) restrictions on competition tend to reduce output and consumer welfare; (3) state regulation is often used to restrict output. While procrastinating from class preparation today, I read a Washington Post article about the controversy in Virginia over whether to privatize liquor stores. The article quotes an epidemiologist who sounds like a closet economist: “If you make it easier to drink, people will drink more. … It’s as simple and basic as that. This seemed like a great example to use for my class. (Of course, it’s a little more complicated. The epidemiologist also said, “And if people drink more, we have more alcohol-related problems.” And, then he gave his statement, “it’s as simple and basic as that.” ) Okay, there are externalities, even in Virginia. But, here is where I got confused. The article refers to a study by George Mason economist Don Boudreaux who apparently did a study (available here) that showed that state privatization did not lead to more alcohol-related problems, at least not more alcohol-related deaths.
So, I have two questions: What should I tell my students about the basics of demand theory?, and the overarching benefits of antitrust over regulation? Or, is this just one of those “politics trumps economics” arguments: the demand for limited government outweighs the law of downward-sloping demand.
Here a few thoughts.
First, its good to know that Steve and I teach the same things on the first day of antitrust class. Things are apparently not too different, at least when it comes to antitrust class, on the other side of the Potomac.
Second, to frame the issues for readers, state-imposed restrictions on competition take many forms and are a common problem faced by antitrust authorities, they often involve “boards” appointed by the state granted authority to regulate particular industries, including the imposition of barriers to entry (recall Eric Helland’s post on the monks fighting against restrictions imposed by the Louisiana funeral industry board — indeed, I remember fondly my time at the Federal Trade Commission with the “Dirty Boards” team whose task was to identify these boards). The economic welfare analysis for many of these restrictions in straightforward. The restriction on competition reduces output, raises price, and reduces consumer welfare. What makes the state restrictions on alcoholic beverages more interesting from an economic perspective is that there is at least a plausible claim to be made that reducing output will also reduce the external social costs associated with alcohol consumption, producing benefits that could potentially offset the negative consumer welfare effects. The relative magnitudes of these effects is an empirical question.
Third, as Steve’s question observes, the obvious effect of privatization, lifting the competitive restriction, will be to increase output and reduce prices. The law of demand is pretty easy to follow here. But doesn’t the law of demand also imply that greater competition and reduced prices implies greater social costs in the form of more “problem drinking?”
On the margin, yes! Why? The state restrictions on retail competition as well as those at the wholesaler level, raise price to both marginal consumers with higher demand elasticities at current prices (a 1% increase in price will result in a decrease in consumption greater than 1%) and infra-marginal drinkers with more inelastic demand at current prices (a 1% increase in price will result in a decrease in consumption less than 1%). As Steve notes, the law of demand implies that unless the drinkers who create those social harms have perfectly inelastic demand, that is, there consumption is entirely invariant to changes in prices, there will be “some” effect on consumption from “problem drinkers,” and thus some positive effect in reducing external costs (see my earlier post on the CARE Act’s odd approach to burdens of proof regarding this issue).
But note that the source of externalities in this example come from a concentrated group. As Cook and Moore (2002, p.122) note, “those in the top decile of the drinking distribution consume more than half of all ethanol. Since alcohol problems are also highly concentrated in this group, it seems reasonable to target alcohol-control policies at them.” The impact of regulatory changes on alcohol consumption and behavior in this concentrated group should the margin focused upon for analysis of the magnitude of any “temperance” effect.
Fourth, as a sidenote, several studies have shown a negative relationship between alcohol prices (often measured by excise taxes) and socially harmful behavior. For example, Saffer & Grossman (1987) and Kenkel (1993) report negative relationships between alcohol prices and drunk-driving. Coate & Grossman (1988) find a negative relationship between price and self-reported underage drinking, but this result disappears when religion and other covariates are introduced. More recently, Markowitz & Grossman (1998) find a negative relationship between state beer excise taxes and domestic violence. As Cooper and Wright point out in our paper, one potential reason for this seeming inconsistency between Boudreaux’s results (and the other results referred to in the article finding that deregulation increases consumption without increasing social harms, as well as our own results, discussed below) may be that earlier work on the relationship between alcohol prices and the harms we measure was based on samples from the 1970s and early 1980s and thus unable to include the effect of ZT and BAC08 laws. Consistent with more recent research [e.g., Carpenter (2004); Dee (2001)], it appears that ZT and BAC08 laws are important sources of reductions in drunk-driving and teen drinking. Specifically, our results suggest that ZT and BAC08 laws reduce alcohol-related accidents by 7-8% and 4-5%, respectively.
But there is no reason to believe that this reduction will be large, much less large enough to offset the welfare losses imposed on consumers in the form of higher prices and reduced output. That is an empirical question.
Fifth, the studies finding that reductions in output associated with state-imposed restrictions on competition are not necessarily correlated with significant reductions in the social ills associated with problem drinking (alcohol-related deaths, DUIs, etc.) are consistent with what Cooper and Wright find in our recently study of state post and hold laws. The post and hold laws operate at the wholesale level, but are also restrictions on competition, making price competition between wholesalers more expensive by prohibiting short-term price reductions, and of course, facilitating collusion by requiring that wholesalers share future price information with rivals in advance. To review, we find that post and hold laws result in significant reductions in output — representing consumer welfare losses — without any statistically or practically relevant reduction in measures of alcohol-related social costs. The lack of measurable effect may be because the reduction in consumption is relatively small, leading to only small behavioral changes for those in the top of the alcohol consumption distribution.
So — I’ve written a lot and am not sure I’ve answered Steve’s original question. What should he tell his antitrust class about demand theory, antitrust vs. regulation, and political economy of privatization in light of the Virginia ABC example? Here’s a few answers:
1. The Virginia ABC example does nothing to change the bottom line: (1) the law of demand lives on, (2) restrictions on competition reduce output and raise price and reduce consumer welfare, and (3) states frequently restrict competition with the predicted consequences — I might add the Demsetzian point that these state imposed restrictions are the toughest to get rid of, even for antitrust.
2. Calculating the impact of a regulatory change on consumer welfare can be tough. Here, I think highlighting the tradeoff between welfare reductions of the conventional antitrust sort and reducing negative externalities would be a very useful exercise for the class. On the theoretical end, I think it requires the distinction between marginal and infra-marginal consumers that is tool antitrust lawyers should have in their toolkit (price discrimination, market definition, vertical restraints, etc.). While predicting the output effects are straightforward, the welfare analysis is a bit more complicated in economic terms.
3. For antitrust lawyers, it is worth thinking about how to convince a generalist judge about the relevant economics in this setting. Having students understand the complexities and learn to simultaneously teach, translate and persuade the court of the relevant economic and empirical analysis is something we should be encouraging the students to think about on day one.
4. I can’t think of any great examples for the demand for limited government trumping the law of demand, but the persistence of the state alcohol monopolies despite their negative economic consequences on consumers gives a wonderful opportunity to talk about cartel formation and stability, as well as the political economy of these laws more generally. For politics trumping economics more generally, see “Rent Control” or the proposed antitrust exemption for newspapers.
The disconnect is in the assumed linkage between consumption at the margin (yes, law of demand holds and consumption will go up at the margin as cost decreases) and the negative social externalities attributable to the marginal consumption. As Josh points out, there likely are significant differences in consumption behaviors for the marginal versus inframarginal consumers, and the social ills are more dependent on those consumption behaviors than the simple quantity consumed at the margin.
Another difficulty in these comparative studies is that there are structural differences resulting from the different regulatory regimes that may be contributing factors to the observed social measures (particularly DUI and other alcohol-related offenses). As referenced in the Washington Times articles, the number of ABC stores in Virginia is far smaller than what a license-based market would provide (the article anticipates a 140% increase in the number of locations, from 332 to around 800). This is more generally true: control states have fewer outlets than license states. Thus, the average consumer must drive further to get the product (part of the increased cost), meaning that are also driving further after purchasing it, increasing the likelihood that the driver may drink while driving and therefore have an alcohol-related traffic offense. Consider that many control states restrict the sale of spirits (and sometimes wines) from stores that sell beer, and you potentially compound the problem by requiring that consumers of a range of alcohol products must travel that much more to procure the product. The market structure parameters imposed by most control state regimes actually increase the likelihood of negative consequences among those who do drink.
Another example: At least when I lived in Pennsylvania 12 years ago, one could only buy beer by the case at the state store. (Actually, the way the law was written, it was illegal in the state of PA to purchase three six-packs in one transaction…licensed restaurants could sell no more than two six-packs in one transaction; states stores only sold by the case.) Price-sensitive beer drinkers would (predictably) buy at the state store where per unit prices were lower, meaning larger quantities of alcohol were 1) kept on hand and available for consumption (which in a household with minors, increases the possibility of underage drinking), or 2) consumed at one time. Again, perverse consumption incentives created by the “control” regulatory regime.
So while Josh is correct in pointing out that the welfare calculations are simply very difficult due to the heterogeneous nature of consumer preferences and consumptive behaviors (never mind any arguments of individual liberty and responsibility), the endogeneity of important structural characteristics of the market and their implications for the social welfare outcomes makes even more difficult the task of doing systematic comparisons that assess the net social costs (or benefits) of one regime or the other.
An observation based on what is going on here in the UK. The problem here is off-trade – that is multiple retailers and convinience shops using booze as a loss-leader. I wonder whether this change the distribution between marginal and inframarginal, in that it is not just a matter of price, but also of consumption patterns – at home in private (or in a park with your mates) rather than in public (so called on-trade channel). I should say food for thought…