Eric has generously given me co-blogger status on Offsetting Behaviour so I can weigh in on the subject of irrational consumers and one’s world view.
The background for this was the session of the NZAE meetings last week when the BERL report on the costs of alcohol was presented. There we heard again the suggestion the question of whether the costs drinkers impose on themselves should be included as costs in a cost-benefit calculation is simply a matter of one’s “world view”. That is, the suggestion is that if you believe that consumers are the best judges of their own well-being then, by definition, any costs consumers impose on themselves are offset by equal or greater benefits and so can be excluded, but if you believe that some consumers make decisions that they will come to regret, these consumers incur non-offset costs that should be included.
Now, my world view is for the most part closer to the first than the second, although I don’t hold dogmatically to this view. That is, I believe that although consumers can and do make mistakes, they mostly are better judges of their own well being than are government officials, or at least are entitled from a personal-liberty perspective to make their own mistakes. Call that “world view 1”. But for the purposes of discussion, let us assume “world view 2”, which is the empirical and ethical assumption that people can make mistakes that can and should be averted by government policy. The question I want to address is how, or indeed if, that world view can be incorporated into cost benefit analysis (or just cost analysis).
It is probably useful to review first just what cost-benefit analysis is. Cost-benefit analysis is one of many approaches to answering two valuation questions: How can we compare bundles of different goods for a single individual to get a sense of whether alternative policies would make that individual better or worse off? and, How do we make comparisons across individuals to decide whether a policy that makes some consumers better off and some worse off is socially desirable?
Cost-benefit analysis is an approach that tries to make as much use as possible of the information contained in prices and individual decisions. For an economist with world view 1, the first valuation question is relatively easy: A rational consumer will consume a good up to the point where the marginal benefit equals the price, and so the price represents the value to that consumer. For goods traded in markets, this gives us a quick and easy way of observing that value. For goods that are not so traded, we ask the conceptual question: How much money would the consumer be prepared to pay to have one outcome compared to another? This can be hard to measure, but at least is conceptually clear. But what if you have world view 2? In this case, you cannot assume that price represents value at the margin, and so have to start making subjective assessments about the value the consumer would place on a good were he not making irrational decisions. That in itself is not a fatal problem. As with the example of non-traded goods above, we need to make subjective decisions all the time in cost-benefit analysis. But what does a dollar measure of cost and benefits mean for an irrational consumer in world view 2? For a drinker who consumes a later-to-be-regretted excess, we can only ask the conceptual question of how much he would, in principle, be willing to spend to reduce the amount he had drunk (i.e. a dollar measure of the costs his excess drinking imposes on himself) if dollars have some meaning as a measure of value for that consumer. If an irrational consumer would choose to spend additional income on things other than alcohol that would make him worse off, then, income is not a measure of value for that consumer at all. It would seem that to use dollar values to calculate the costs of alcohol that irrational drinkers impose on themselves, you can’t adopt world view 2; rather, you need something like world view 1.5—that consumers who are not fully rational in their decisions over the consumption of alcohol and drugs are fully rational for all other allocations. Only with this near-rationality are we able to continue to assume that there is useful value information contained in prices.
Now, it can reasonably be argued that addictive substances are quite different from other goods, and so a world view that assumes rationality except in that narrow sphere might be perfectly sensible. So let’s now consider the other valuation question, and ask whether adding up dollar costs across individuals makes sense with less-than-fully-rational consumers. With fully rational consumers, the idea of adding up dollar costs and benefits rests on the practical idea that dollars (unlike some subjective measure of well-being like “utility”) can be transferred between consumers, and on an ethical notion called the compensation principle. This says that a policy can be justified if, in principle, the consumers who gain from the policy could make a monetary compensation to those who lose in such a way that everyone is made better off by the policy. When, for example, Sir Geoffrey Palmer quotes “the headline figures in the Berl report of $5.296 billion in social costs of alcohol (and drugs), versus the alcohol excise tax take of $795 million, as a basis for his preferred policy option of significantly raising excise taxes to cover the shortfall”, he is invoking the compensation principle.
To illustrate, consider the situation of a rational drinker facing a price of $10 per drink, and imposing an external cost of $10 per drink on other people. Further imagine that demand is highly insensitive to price, but this is not known to the policy maker. In this case, we could put a $10 dollar tax per drink on the alcohol. Tax revenue would now equal $10 times the number of drinks purchased, which would exactly equal the external costs. We would now know that the benefit to the consumer was in fact greater than the external costs imposed, shown by the fact that the consumer is willing to pay the price of $20. If demand were sensitive to price, there would be some reduction in alcohol consumption—a reduction in those drinks for which the gross benefit to the drinker were less than the resource plus external cost of the drink, exactly what we would like according to the compensation principle.
In contrast, consider the same situation of an irrational drinker with price-insensitive demand who faces a price of $10 per drink but receives no benefit at all from the alcohol. This drinker is also imposing a cost of $10 per drink on society ($10 of resources used to make the drink with no offsetting benefits to the drinker), but that cost is entirely borne by himself. We could place a tax of $10 per drink on this drinker, but all that would do is increase the costs to the drinker by the same amount that it provided compensation to “society”. The story does not change significantly if demand is price sensitive. In this fanciful example, the appropriate tax would be the one that pushed the price of a drink so high that the drinker would reduce his consumption to zero (since he receives no benefit from the drink). That is, the appropriate tax would be one derived from his (irrational) demand curve for alcohol, not from any calculation of the dollar value of the costs he imposes on himself.
Contrary to Sir Geoffrey’s statement above, therefore, a comparison of the dollar value of the social costs of alcohol to the value of tax revenue received makes no sense if the social costs include costs imposed on himself by the drinker.
In short, the point here is that one is free to take a world view that people are sometimes not fully rational or informed, and to devise paternalistic policies based on that world view. But don’t use a false precision of dollar values, and don't use the technical apparatus of a valuation technique that depends on an assumption of rationality for its internal consistency.