Public health and the new paternalism

Arguments for taxing alcohol overestimate social costs and ignore personal benefits, argues Eric Crampton.
Eric Crampton's blog is at http://offsettingbehaviour.blogspot.com

Does anybody else miss honest old-time paternalism? When the revivalist preachers of the 1920s and ’30s condemned the demon drink, their assumptions were clear: without careful shepherding by the church and enforcement by the state, sinful drinkers would condemn themselves to perdition. Preachers claimed superior moral authority derived from divine revelation. Those claims are ineffective now: public policy is largely secular and religion is largely left for individuals to decide on their own.

The new paternalism is more insidious. Today’s anti-alcohol proselytisers garb themselves in the mantle of science rather than religion for authority. Unfortunately, it’s just the garb of science with none of its rigour: it’s sciency rather than science. A century ago, we were ill-equipped to argue that a preacher might not have superior knowledge of the mind of the Divine; we’re now just as badly prepared to examine the new paternalists’ sciency method.

Today’s sciency paternalists reside mostly in academic public health departments and frame their recommendations as being based on economic cost-benefit or, worse, cost-only analysis. Economics and epidemiology provide the pretence of authority.

These new paternalists advance what I’ll call healthism: the notion that there should be no public policy objective beyond improving health and increasing lifespan. Is this really a sensible objective for public policy? And if health is only one of many competing social goals, then the method underlying their prescriptions is seriously flawed.

Healthism
None of us holds health as our only goal. Every time we take a slight risk in traffic, or decide to drive at all, we’re trading the risk of accident against the benefits of getting to where we’d like to go. When we decide to go skiing, we trade off fun against the risks of a broken leg or worse. Even where our children are concerned, we make trade-offs. We could always choose to purchase a little more safety for them than we do. We could spend a little more on the slightly safer car or car seat. We could always expend a little more effort in keeping them from harm. But we don’t make our toddlers wear padded helmets: the child wouldn’t like it and neither would we. Even experiments on animals show health is not the only goal. While mice on a calorie-restricted diet lived longer than a control group, they were very angry mice: they bit their handlers and were more agitated.(1) Health is important, but it’s not the only thing in life that matters.

Sensibly, health isn’t the only goal for public policy. Speed limits trade off convenience against accident risks and safety regulations trade off cost against risk reduction. Government spending priorities certainly do not place health as the only goal: it would be difficult to explain government funding of the arts if health were all that mattered.
Public health practitioners forget this when they wade into public policy debates; instead, health becomes the only goal. It’s one thing to show reductions in smoking deaths with an excise tax increase; it’s quite another to show that the health benefits outweigh the costs. But for many in the public health field, the existence of potential health benefits is sufficient for a call to tax, regulate or ban the offending substance.

So when University of Otago at Wellington Department of Public Health researchers Des O’Dea and George Thompson found that tax increases can reduce rates of smoking and associated health costs, this was sufficient reason to call for increases in tobacco taxes of about 40 percent; the only factors mitigating calls for larger increases were the potential for smuggling and public opposition to larger increases.(2) That many smokers seem to enjoy smoking was largely irrelevant. For Tony Blakely and Nick Wilson, also at the Otago Wellington Department of Public Health, the fact that life expectancy increases with income, but at a decreasing rate, is sufficient reason to suggest income redistribution as health policy: transferring money from rich to poor reduces the life expectancy of the rich by less than it increases the life expectancy of the poor.(3) Even leaving aside substantial methodological problems, including ignoring the deadweight costs of taxation and the other factors that may affect both health and income, those health outcomes are not the only goals of public policy.(4) We need other ways of deciding when policymakers should intervene.

Internal and external costs
Economists draw bright lines between ‘external’ and ‘internal’ costs. External costs occur when private activity imposes costs on others. In public policy discussion, these external costs are frequently called ‘social costs.’ Internal costs are incurred by individuals themselves as a result of their own actions.

Typically, economists believe that if actions impose costs on others not accounted for in individual decision-making, there may be room for the government to step in and improve overall outcomes through taxation or regulatory measures designed to internalise the externality. In other words, the person causing the costs is given an incentive to stop or required to compensate the fisc for the harm caused.

Economists usually don’t see such a clear case for policy interventions to reduce internal costs. We generally assume that individuals have strong incentives to weigh heavily their own costs, health or otherwise, in their decision-making. Internal costs may be sufficient to justify paternalistic regulation but are not sufficient for such interventions to improve efficiency. The new paternalists often cite asymmetric information about health risks—in which producers take advantage of consumers’ inadequate knowledge of health consequences or dangers—as justifying taxation or regulation. Often, however, consumers are not ignorant. For example, the best evidence suggests that individuals overestimate the health costs of smoking.(5) If there’s an information market failure, it’s causing folks to smoke too little, not too much. Where information is the problem, public information campaigns are more likely the appropriate solution, not taxes or other regulations.

More recent work by behavioural economists focuses on costs borne internally which individuals have trouble reducing due to self-control problems. A drinker might sincerely wish he could drink only two pints per day, but instead drinks four and he’s not been able to find any way of stopping himself. If taxes were increased to the point where he’d only purchase two pints per day, so the argument goes, he’d be made better off by his own measure of his own wellbeing. As University of California at Northridge economist Glen Whitman puts it:

[T]he old paternalism said, ‘We know what’s best for you, and we’ll make you do it.’ The new paternalism says, ‘You know what’s best for you, and we’ll make you do it.’(6)

These ‘internalities,’ as they are sometimes called, might matter on the blackboard but it’s difficult to see them as the basis for public policy. Unless a regulatory agency can see into our souls and discern that weakness of will is the problem, we can’t tell that a tax really makes drinkers better off by the drinkers’ judgment.(7) In that more realistic case, we’re back to the old paternalism regardless of how it masquerades. Moreover, the internality argument only posits that some portion of consumption might involve costs in excess of benefits rather than that benefits disappear.

Public health practitioners happily blend together internal and external costs as ‘social costs’, though the efficiency case for considering internalities is controversial at best. When newspaper headlines tell us that the social costs of alcohol, drugs or tobacco have risen to billions of dollars, it’s natural but wrong to assume that presented figures provide a measure of external costs or of net costs—especially when, as is often the case, the presented figures are contrasted with the excise tax take. Public health measures of social costs cannot be taken as measures of economic costs.

Cost-only analysis
Social cost analysis undertaken by public health practitioners, and sometimes by economic consultancy firms, differs substantially from an economic cost analysis. Take a recent cost analysis undertaken for the New Zealand Ministry of Health and the New Zealand Accident Compensation Corporation by Business and Economic Research Limited (BERL).(8) The BERL report largely followed the method recommended by Australian academics Professors David Collins and Helen Lapsley, and the World Health Organization, for conducting such analyses and serves as exemplar of the multiple ways that such studies provide an inflated view of the economic costs of drinking, smoking, eating fatty foods, or having any kind of fun at all.

BERL argued that the social costs of alcohol consumption in New Zealand totalled about $4.8 billion in 2005, or more than $1,000 per Kiwi. Sir Geoffrey Palmer, former Prime Minister of New Zealand and now head of the New Zealand Law Commission, was charged with reviewing New Zealand’s alcohol legislation. He compared these massive costs to the excise tax take of about $700 million and found prima facie evidence of the need to raise the excise tax. How did BERL produce such a large number?

Most importantly, BERL counted as social costs of alcohol use costs that fall solely on the drinker but none of the benefits. BERL counted drinking more than 1.8 pints of beer per day as a social cost. While BERL argued, and continues to argue, that benefits of alcohol were outside of the scope of their analysis, bringing in all private costs as social costs required that they assume the gross benefits of private ‘harmful’ consumption are zero.(9) With zero benefits, costs borne by the drinker like reduced wages, intangible personal health costs, and premature mortality are deemed social costs. When Matt Burgess from the University of Victoria at Wellington and I dug into BERL’s numbers, we found that a more conventional economic approach would yield external social costs of about $675 million, roughly approximating the total alcohol excise tax take.(10)

Imagine a parallel case trying to weigh up the social costs of apples, though we well recognise the difference between apples and alco-pops. If you get $1.50 in enjoyment from eating an apple but spend only $1 on it at the store, most economists would say that you’re better off by 50 cents. From gross benefits of $1.50, gross costs of $1 are netted. Evidence that you’ve spent $1 is sufficient to show that you valued consuming the apple by at least $1. There are additional costs incurred in driving to the store, the time and effort involved in choosing just the right apple, and so on. Since you’ve decided to purchase the apple, we need not worry about these costs: you decided to bear the costs and buy the apple because you value the apple more than all the associated costs. If we instead assumed that you received zero gross benefits from your apple, and further assumed that apple-eating had no health benefits but only health costs for those unfortunates who choke on apple cores or crash their cars while distractedly eating an apple, we could quickly come up with a very large number for the aggregate social costs of apple-eating, including not only all spending on apple eating but also the value of time in going to buy apples, reduced job productivity when trying to eat an apple while typing, and so on. And that, in essence, is what BERL did with alcohol. Yes, alcohol is not the same as apples. But individuals do enjoy consuming it even if they do bear some private costs, and in some cases very high private costs, from such consumption. These benefits should count for more than just a line assuming them equal to zero.

Inflated costs
Worse, even internally-borne costs were greatly inflated. Most prominently, BERL decided that if a worker dies or is unemployed due to alcohol abuse, that worker could never be replaced either by another worker or by a more capital-intensive production method. They multiply all forgone wages by 1.87, the ratio of GDP to wages. All of the potential output from an absent worker is gone forever. This move alone added more than half a billion dollars to BERL’s cost estimate. Other discovered sources of cost inflation included:

• assuming that any surveyed prisoner claiming alcohol had contributed to his offending(11) would, in the absence of alcohol, been gainfully employed at the average wage rate, when in fact prisoners have very poor labour market characteristics even leaving alcohol aside;
• assuming that heavy alcoholics would be identical to average Kiwis but for their drinking, and therefore earning average wages, despite strong evidence of co-morbidity between alcoholism and other psychiatric disorders;
• ignoring positive effects of alcohol consumption (for example, on coronary heart disease);(12)
• Disproportionately counting reducing insurance overhead costs in the absence of insurance claims related to harmful drinking.

Finding a policy-relevant number
The method used trivialises a serious social problem. We found that the bulk of the truly external social costs of alcohol abuse come from criminal activity, with costs to the public health system and costs of drink driving accidents coming in next. If most of the social costs of alcohol abuse come from criminal activity, it’s then important to know whether those costs are better reduced by increases in alcohol taxation that punish moderate drinkers as well as than heavy drinkers, or by increased police activity and penalties targeting offenses committed while drunk. When external and internal costs are blended together, it’s difficult to tell that this is the most relevant policy question.

For economists, there are two ways of providing an honest, policy-relevant number. Either count up all the costs, internal and external, and weigh them up against all the benefits, or, if counting internal benefits is too difficult, count only the external portion. While the latter method ignores costs that heavy drinkers may irrationally impose upon themselves from excess drinking, it also ignores the benefits drinkers enjoy from moderate consumption. If the internal costs from excessive consumption, viewed across the set of all drinkers included in the study, roughly match the benefits those drinkers enjoy from the moderate portion of their consumption, worrying only about externalities is not a bad approach. Of course, for any particular heavy drinker who has had a very adverse outcome like a serious drink-driving accident, costs will well exceed benefits. But for every drinker who has that kind of adverse outcome, there will be hundreds of drinkers who have had only a good time out. Counting only the costs to those who suffer serious adverse consequences without counting the benefits to drinkers who do not is like deeming skiing horribly harmful by counting the costs from deaths in avalanches, without considering the benefits to those who’ve had a good day on the slopes.

Consumption generating negative external costs is necessary but not sufficient for proving the need for a tax increase. Heavy drinkers are much less responsive to price increases than moderate drinkers: most estimates say that they’re about half as responsive.(13) So for every 10 percent increase in the cost of alcohol, moderate drinkers will reduce their consumption by about 5 percent while heavy drinkers will reduce their consumption by only about 3 percent. If moderate drinkers were enjoying life before the tax increase, we impose a disproportionately heavy burden on them for a small decrease in heavy drinkers’ consumption. As economist Edward Stringham puts it, raising the alcohol tax to deal with problem drinking makes as much sense as raising the gas tax to deal with speeding.(14)

It is especially worrying when governments commission cost-only sciency reports to provide justification for existing or proposed regulations. The government-commissioned BERL report on alcohol seems to have been designed to come up with a very big number suitable for arguing for further controls on alcohol rather than to come up with an honest assessment useful for sound public policy. Similarly, a previous New Zealand Police commissioned BERL report on the costs of drug use led the police to claim $379 million in social costs avoided by their latest anti-marijuana campaign; however, almost 90 percent of the costs tabulated by BERL in that report are either the costs of producing marijuana or the costs of running police operations against marijuana growers.(15) In other words, the commissioned figures are utterly useless as a measure of the beneficial effects of police drug busts, as the commissioned report itself warns, but are very useful as agitprop. Instead of using a sciency figure, the police would do better by reverting to the old paternalism as justification for their activities.

A plea for an honest paternalism
There is nothing wrong with paternalism per se. I don’t agree with the view that we’re best considered immature wards of the state for our entire adult lives, but at least it’s honest. The discussion then turns either to empirical evidence, looking at whether any particular paternalistic policy is effective by its own lights, or to values, where we can legitimately argue that ‘nanny (state) knows best’ is repugnant regardless of the effects of such regulations.

But couching paternalism as science short-circuits the debate. In its new sciency flavour, the question is begged: social cost measures that can only be derived based on healthist assumptions are presented as conforming to the normal strictures of economic analysis. A proper economic cost-benefit analysis could conclude that taxes or regulations are best; however, by ignoring all values other than health, the method used by public health researchers assumes its conclusion while using a language suggesting otherwise. Unfortunately, few are equipped to argue when sciency people say that the costs of enjoying ourselves run into the tens of billions of dollars.

Endnotes

1. See discussion by Glen Whitman at http://agoraphilia.blogspot.com/2009/07/eat-more-or-starve-yourself.html.
2. Des O’Dea and George Thompson, ‘Report on tobacco taxation in New Zealand,’ Smokefree Coalition and ASH New Zealand (2007).
3. Tony Blakely and Nick Wilson, ‘Shifting dollars, saving lives: What might happen to mortality rates, and socio-economic inequalities in mortality rates, if income was redistributed?’ Social Science & Medicine 62:8 (2005).
4. See Linda Gottfredson and Ian J Dreary, ‘Intelligence predicts health and longevity, but why?’ Current Directions in Psychological Science 13:1 (2004).
5. Kip W Viscusi, ‘Do smokers underestimate risks?’ Journal of Political Economy 98 (1990).
6. Glen Whitman, ‘Against the new paternalism: internalities and the economics of self-control,’ Cato Policy Analysis 563 (2006).
7. James Zuccollo, ‘The problem with procrastination,’ Insight (New Zealand Institute of Economic Research, March 2009).
8. Adrian Slack, Ganesh Nana, et al. ‘Costs of harmful alcohol and other drug use,’ BERL Economics (2009).
9. As above, 173.
10. Eric Crampton and Matt Burgess, ‘The price of everything, the value of nothing: A (truly) external review of BERL’s study of harmful alcohol and other drug use,’ Department of Economics and Finance working paper no. 10/2009 (University of Canterbury, 2009). Read in conjunction with our rejoinder to BERL’s response, www.econ.canterbury.ac.nz/personal_pages/eric_crampton/Reply.pdf.
11. Surveyed prisoners answering that alcohol had contributed ‘some,’ ‘a lot,’ or ‘completely to their offending were assumed to become perfectly average law-abiding citizens in the absence of alcohol.
12. David Collins and Helen Lapsley, ‘The costs of tobacco, alcohol and illicit drug abuse to Australian society in 2004/05,’ Report to the Australian Ministry of Health (2008), note reduced medical costs for some categories of illness. See Corrao et al (2000), ‘Alcohol and coronary heart disease: a meta-analysis,’ Addiction, for evidence on heart disease
13. Alexander Wagenaar, et al. ‘Effects of beverage alcohol price and tax levels on drinking: a meta-analysis of 1003 estimates from 112 studies,’ Addiction 104:2 (2009).
14. Edward Stringham and Adrian Moore, ‘The catastrophe of what passes for alcohol policy analysis,’ Reason Foundation Policy Brief 78(2009).
15. BERL, ‘New Zealand Drug Harm Index,’ Report to New Zealand Police (2008).