COSTS AND BENEFITS AND ALCOHOL POLICY
A Note for the Symposium on Cost-benefit Analysis on Alcohol Policy Barcelona, 18-19 October 2010
This note covers three issues involving cost-benefit analysis (and the related cost of illness analysis) . At the heart of each is a neoclassical theory of decision making, for CBA is rigorously underpinned by various assumptions which many of its critics seem unaware of. I do not object to challenging the assumptions providing it is done with an awareness of what is being done, and preferably in a rigorous way.
The Counterfactual Scenario
Economists always measure economic costs (and benefits) as opportunity costs, that is the opportunity forgone relative to some alternative (called ‘the counterfactual scenario’). It is usually
critical to define the counterfactual scenario, since different ones generate different opportunities forgone.
It matters greatly when evaluating a cost of alcohol study. A scenario in which alcohol abuse – perhaps defined in terms of reduction in drinking which generates harm – is eliminated will have quite a different outcome from the (prohibitionist) scenario in which no alcohol is consumed. This is because some alcohol consumption is beneficial to the consumer and benign to the rest of society. It is retained in the abuse elimination scenario but not in the prohibition scenario and therefore generates a (net) benefit in the latter.
Indeed it is possible that alcohol consumption today provides a net benefit to society as a whole despite the great harm that some drinking generates (although settling the possibility is an empirical issue). That does not mean the issue of harm prevention should be ignored; the benefits of alcohol consumption to society as a whole would be even greater were there a reduction of harm.
A difficult particularly issue is that sometimes a harm reduction program will decrease some of the benefits of alcohol consumption – that is the benefits to consumers who are not generating harm. The issue is further complicated by the ‘benign’ drinkers being a different group from the harmful drinkers and objecting to their personal lost. (And example is that raising excise duties on alcohol will reduce harmful drinking, but it will also reduce ‘benign’ drinking.) The latter object to their loss, which is why it is often hard to find the political coalition to raise excise duties on alcohol.
Sometimes it is argued that some cost-benefit analyses ignore the benefits of alcohol. The arguments are invariably so muddled that there can be no general rebuttal of them. However, cost-benefit analysis has been around for about 60 years, during which the neoclassical approach has been rigorously developed by some exceptionally able economists – and reviewed by many more. Those who claim to have found a flaw in the approach might wonder why those a lot smarter than they are – and many critics do not seem that smart – have not identified the alleged fallacy earlier.
(Note that the avoidable costs issue amounts to an alternative scenario. Instead of comparing the actual situation with, say, a counterfactual scenario where there is no alcohol harm because there has never been any harmful drinking, it considers the counterfactual scenario where all harmful drinking ceases now, but there is a carry-forward from past harmful drinking.)
Very often discussions on rationality amount to the rhetorical declension ‘I am rational; you sometimes behave irrationally; he is insane’ with hardly any more analytic content than that. The neoclassical model has a formal notion of rationality, which involves a well defined set of preferences which are reasonably stable and which one pursues in a logically consistent way. People are of course not quite that rational but for many purposes the ideal is close enough to the reality in aggregate for it to be used.
In recent years some economists have been examining the empirical evidence about the deviations from the reality, in particular paying attention to the research of psychologists; the 2002 Nobel Memorial Prize in economics was awarded to a psychologist and to an economist who developed experimental methods. Paradigms develop slowly, and behavioural economics is still largely a set of ad hoc results, lacking any coherent framework.
One exception is ‘time-inconsistency’. For our purposes it involves consuming alcohol, say, but having neither planned to consume before while regretting the decision afterwards. We have to be cautious how to incorporate such phenomena into the neo-classical model, but it seems likely that we should discount time inconsistent consumption. (This notion developed after the International Working Party on Substance Abuse – of which I was a member – completed its deliberations and so was not considered by them.)
Time inconsistent consumption would justify the decision by David Collins and Helen Lapsley to deduct the benefits of some (but not all) alcohol consumption when calculating the social cost of substance abuse – I followed them in my 1995 study.
However, a couple of caveats. There are no empirical estimates of the amount of time-inconsistent drinking. Second, it is not clear what are the policy consequences, although the so-called ‘nudge’ approach is partly based on it while a couple of economists showed that (using a neoclassical framework) that excise taxes which discouraged time-inconsistent drinking were valued by the drinkers.
The Value of Life
Some brief comments about the value of life. It is unquestionable that the impact of the changes on life (that is differences between the actual and counterfactual scenario) has to be incorporated in a cost-benefit or cost-of-illness/abuse analysis. We are well beyond looking only at mortality or longevity, and today the quality of life of the living has to be included – in principle.
In practice the quality of life from an impairment is not very well measured, especially if the impairment is a question of psychological capacity – which is particularly important in the consequences of harmful drinking.
A further problem is the price of the quality of life, ‘price’ meaning here the tradeoff between the value of life and material goods and services. There are various ways of measuring the price – my preference is for a willingness to pay – but it is rare to see this done in a suitably rigorous way.
The evidence suggests that the ‘value of life’ is high relative to material products (so that, say, 90 percent of the net cost of substance abuse seems to be attributable to reductions in the quality of life). This may not be surprising, but it does lead to very large monetary sums which are often misinterpreted by being compared to GDP (which is invalid because it omits any monetary value of lives of the population). It is disappointing, then, that the impact on the quality of life is so poorly measured and the value of life so contentious.
(A useful development has been the extension of the data base to cover harm to others, although again the quality of measurement fro economic purposes is not high.)
Lurking behind this issue is, I think, the ‘happiness research. Economists have traditionally assumed a person’s wellbeing closely correlates with their material consumption; a notion that is central to the neo-classical model. The empirical evidence suggests that the correlation between material consumption and happiness is not high, especially in affluent economies.
What this means is uncertain – again it is a paradigm which is developing. It may eventually modify the neoclassical paradigm, or it may transform it. I have yet to see the happiness research impact on cost benefit analyses (although a project I was involved with NZ’s SHORE on gambling gets damned close to doing so); in the interim we need to be mindful of this development.
The Neoclassical Paradigm
This note has both defended the economist’s neoclassical paradigm, and pointed out some places important to alcohol analysis where it is fraying. (In another context one could draw attention to other parts of the economics discipline where there is also fraying; the Global Financial Crisis has done little to suggest that the paradigm is sufficiently robust to explain financial market behaviour.)
That is how it should be in a science. A paradigm is under continual pressure to prove itself, and to extend its use to other domains. It develops under a critical process to which this paper belongs.
However, as this note’s opening pointed out. that does not mean all criticism of the paradigm, including the cost-benefit analysis based on it, is valid. Valid and useful criticism has to understand its deep structure and engage with that, not some misunderstood and superficial version of it.