Monday 14 July 2014

Punishing the Majority: a reply to Will Haydock

Somewhat belatedly, I'd like to reply to a post by Will Haydock about the IEA report I wrote with John Duffy last month. Titled Punishing the Majority, the report looked at the claim that per capita consumption of alcohol is the determining factor of levels of alcohol-related harm and that governments should introduce policies to control or reduce average consumption in order to reduce problematic consumption (AKA the Total Consumption Model or Whole Population Approach). Using real world data and citing decades of research, we show that this theory does not stack up and recommend that policy be aimed at problem drinkers rather than the general population.

Will points out that you don't need to believe in the TCM in order to believe in whole-population policies such as tax rises...

There isn’t any necessary link between the Total Consumption Model and whole population approaches to alcohol policy. There are all sorts of reasons for identifying a whole population solution even if you know the problems are caused by individuals. 

If Will means that the policies that are proposed by advocates of the Total Consumption Model (TCM) can be justified on other grounds, I agree. The government could put up alcohol taxes on the basis that it would have a deterrent effect on heavy drinkers even if it did not care whether overall (per capita) consumption went up or down.

Sometimes this might be about the ease of administering a policy. For example, we place fixed age limits on alcohol consumption, even though potential drinkers mature physically and mentally at different rates, and won’t all be equally well prepared to deal with alcohol at the same age.

I agree that there are perfectly good reasons for having an age limit on alcohol purchase, but raising the limit isn't a big issue for advocates of the Total Consumption Model and the existence of a limit can clearly be justified on other grounds (eg. child protection).

Sometimes this might be about equality. There’s something attractive about the idea that all units of alcohol should be treated (and taxed) in the same way, rather than differentiating on the basis that more problems are associated with a particular drink. 

I happen to agree with that, but it's not really the issue here and, again, it's not something that fans of the TCM are particularly vocal about.

Finally, support for a whole population approach might be political (or perhaps more accurately moral). Kettil Bruun supported a population-wide approach partly because he felt it might avoid stigmatising dependent drinkers. 

He did, and we discuss his motives in the report. Bruun didn't like the stigmatisation of alcoholics in Scandinavia at the time and he tended to think that society, rather than the individual, was to blame for alcohol misuse. This drew him towards a dubious but highly influential version of the TCM—the Ledermann hypothesis—which he used like a drunk uses a lamppost: for support rather than illumination. Bruun was well-meaning and the treatment of alcoholics in post-war Scandinavia could certainly be draconian, but he went to the opposite extreme by ignoring personal responsibility and treating people like helpless pawns of industry and institutions (as socialists and 'public health' people tend to do).

That is, population-level approaches needn’t have the Total Consumption Model as their cornerstone.

I think we agree so far. However...

More importantly, though, population-wide policies aren’t the ‘cornerstone’ of the Total Consumption Model, as Chris also seems to suggest on his blog. The TCM might prop those policies up, but it would be back-to-front thinking to have the solutions explaining a problem. 

It's hard to believe that Will really thinks that raising prices and restricting advertising and availability are not the cornerstone policies of the TCM, but he must do because he says the same thing again towards the end of his blogpost:

Whole population policies might or might not be a sensible approach to alcohol, but it’s misleading to focus on the idea that ‘The 'cornerstone policies' of the Total Consumption Model involve raising taxes, restricting advertising and limiting availability’, since these policies can be justified in a number of other ways.

This is a simple fallacy. Just because these policies can be justified on other grounds does not mean that one school of thought does not place great importance on them. Of course these policies are at the cornerstones of the Total Consumption Model. They are mentioned in virtually every document produced by supporters of the model since the early 1970s and they are the cornerstone policies of the neo-temperance lobby in Britain today. 

Don't take my word for it. Here's what James Nicholls has to say in his 2009 book, The Politics of Alcohol. On pages 252-3, he lists various anti-drink campaigns seen in England over the centuries along with their 'core arguments' and 'preferred solutions'. At the bottom of the list is the most  recent breed, termed 'Public health'. Its 'core argument', he says, is 'Per capita increases in consumption lead to increases in all alcohol-related problems' and that 'Drink should be tackled at the population level'.

Quite obviously, this is the total consumption/whole population approach and the 'preferred solutions' are, he says, 'Tax increases' and 'Restrictions on alcohol through licensing controls'. I won't put words in James's mouth, but I'd be surprised if he didn't agree that restrictions on advertising are also part of their arsenal.

Who believes in this approach? Lots of people in 'public health'. On page 236, James writes: 'In 1994, a major report entitled Alcohol and the Public Good had argued that focusing on problem drinkers was less effective than tackling overall consumption across the population... These conclusions had been reiterated in another major international report compiled by public health researchers in 2003. The consensus among these researchers was that population-based approaches were the only sure way to tackle alcohol-related problems, and that the most effective way to reduce overall consumption was through raising prices via taxation and reducing access to alcohol via licensing restrictions.'

And, lest you think that this approach has since been abandoned, he notes—on page 250—that the establishment of the Alcohol Health Alliance in 2007 'marked an important moment in the development of a coordinated campaign, led by public health campaigners, for action to reduce per capita consumption through tax increases and stronger licensing restrictions' (my emphasis). He also mentions that this was the 'new consensus among alcohol campaigners'.

James's complaint about Punishing the Majority was that it overstated how influential the Total Consumption Model has been in terms of policies that have actually been implemented (as opposed to policies that the 'public health' lobby would like to see implemented). I can understand why he says that and I have responded here. He is right about licensing, which has been relaxed in the last decade, but I maintain that the UK's alcohol taxes and advertising restrictions reflect a belief in reducing per capita consumption (indeed, when writing about the large tax rises on alcohol that began in 2008, Nicholls writes that 'by framing the announcement in terms of affordability, and by effectively targeting all drinkers, the Treasury had accepted a key tenet of the population model' although he adds that this is 'something they had resisted forcefully for decades' (p. 246).)

Politicians do not need to be familiar with Kettil Bruun and bell curves in order to implement population-based policies. As Will says in his blog post, justifications for policies such as tax rises can be found without resorting to the Ledermann hypothesis. The IEA paper was intended only to assess the validity of the Total Consumption Model which, as we show, is explicitly supported by many 'public health' organisations. We wanted to test two claims: (a) that alcohol-related harm will inevitably fall when per capita consumption declines, and (b) that the policies favoured by the public health lobby are effective in reducing alcohol consumption and/or alcohol-related harm.

To test these claims, we only need to look at countries which have either seen per capita consumption decline or have introduced tax rises, ad bans and/or restricted licensing. It doesn't matter whether the politicians are ideologically committed to the Ledermann hypothesis or are persuaded by those who are—or have completely different motives (such as raising taxes to raise revenue).

Chris’ problem with the Total Consumption Model is that it (apparently) supports population-wide policies, which he says are likely to be ineffective, but we haven’t even agreed on how that potential efficacy might be judged. 

Haven't we?! I thought the aim was to reduce alcohol-related harm by reducing per capita alcohol consumption. Therefore we surely judge the efficacy by looking at the incidence of alcohol-related disease and mortality. This is what we do in the report and, in line with other studies that we cite, we find the claim that alcohol-related harm is fixed to overall consumption to be false.

I’m sticking my neck out here, because this isn’t quite the reasoning he offers on the blog, but I have a suspicion that the reason he doesn’t like population-wide policies is because they might affect people whose drinking impinges on no-one but themselves. 

No kidding. Why do you think it's called Punishing the Majority? We say on the very first page that population-wide policies 'have significant general welfare costs' which include 'the deadweight costs of taxation, the welfare cost of being unable to drink at chosen times and search costs incurred by limitations on advertising.'
However, as I’ve noted, there are lots of other arguments in favour of population-wide policies other than the TCM.

Fine. Argue for them on different grounds (although I think you'd still lose the argument). All we're saying is that the justification employed by NICE, Alcohol Focus Scotland, the European Commission and many others is bogus and has been known to be bogus by researchers for many years.

Moreover, MUP may affect the majority of drinkers, but it wouldn’t ‘target’ them (as Chris put it on his blog). All drinkers might all be somewhat affected by MUP, but there’s no doubt that people wouldn’t be equally affected by the policy. One good way of seeing this is to watch Nick Sheron’s presentation about the drinking habits of the people he sees with serious liver conditions: they drink a disproportionate amount of cheap alcohol, and would be disproportionately affected by MUP – whether that would reduce their consumption or simply lead to a financial hit. 

The report isn't about minimum pricing (MUP). I would agree that, in some ways, MUP does attempt to target problem drinkers and is therefore a more nuanced policy than traditional TCM policies. However, it is still very clumsy and would have significant financial and welfare costs on people who are not harmful drinkers. In practice, the only people who will be excluded will be a wealthy minority who do not buy cheap and medium priced alcohol, but that is a different discussion.

There’s also something misleading in Chris’ discussion of risk and health in the context of population-wide policies. It’s perfectly correct to point out that an individual won’t be much affected by a small reduction in their consumption, particularly if they’re not at the top end of the consumption spectrum. However, this is to misunderstand how population-level policies work: they don’t aim to make everyone necessarily live longer by a day or so; they aim to make an average population live longer, and affect some individuals significantly. The nature of the prevention paradox is that an individual won’t be noticeably affected by the small reduction in risk their change in consumption habits produces. These small reductions in risk, though, when aggregated across a whole population, can produce a notable reduction in overall mortality.

I don't see how any fair-minded reader of Punishing the Majority could think I don't know this. We write about Geoffrey Rose's theories (which is really what Will is alluding to here) and dismiss them with good reason. Besides, the Total Consumption Model does not merely argue that everybody drinking a little less will be a little better for everybody's health (a dubious assertion in itself). It says that a decline in consumption amongst moderate drinkers—so long as it leads to a fall in per capita consumption—would somehow make harmful drinkers consume less alcohol and suffer less alcohol-related harm. This is patently false and can be shown to be false by looking at countries, such as England in the last decade, where per capita alcohol consumption has fallen considerably.

Of course it can perfectly reasonably be argued that pushing (not quite nudging) people towards certain choices is no business of the state – and that’s fundamentally where the disagreement here lies. The IEA isn’t an expert in the effectiveness of health interventions; it’s a bit more clued up on political philosophy.

If this is meant to be an appeal to authority, I'd point Will towards the CV of my co-author John Duffy. If it is meant to be a 'why would a free market think care about any of this', I'd refer him to the welfare costs mentioned above. Yes, I am interested in the theories and excuses used by the public health lobby to enact policies that have wide-ranging costs on the public and I think people deserve to know whether they are true.

You don't need to look for hidden subtexts with the IEA. Our commitment to individual liberty and free markets could not be more explicit.

The real policy debate should be a clear discussion of what the problem is, and what an appropriate solution might be – which may not necessarily be the most effective solution, as that might not be acceptable for practical or moral reasons.
If this is a debate about liberty and fairness, let’s have it.

We can have that debate any time, but Punishing the Majority is about a specific empirical claim upon which many neo-temperance campaigners have hung their hat. Either it is true or it is not. When 'public health' campaigners hold ludicrous beliefs and tell a dozen lies before getting out of bed in the morning, the IEA would be missing an open goal if it restricted itself to a debate about philosophy. 



4 comments:

Junican said...

"There’s also something misleading in Chris’ discussion of risk and health in the context of population-wide policies. It’s perfectly correct to point out that an individual won’t be much affected by a small reduction in their consumption, particularly if they’re not at the top end of the consumption spectrum. However, this is to misunderstand how population-level policies work: they don’t aim to make everyone necessarily live longer by a day or so; they aim to make an average population live longer, and affect some individuals significantly. The nature of the prevention paradox is that an individual won’t be noticeably affected by the small reduction in risk their change in consumption habits produces. These small reductions in risk, though, when aggregated across a whole population, can produce a notable reduction in overall mortality."

Sorry to quote the whole section - I could not see how to do otherwise.
What bothers me about such sweeping statements is that they never seem to take into account 'timescales'. Let me put it this way: a tiny postponement of the date of death by one thousandth spread over one thousand people does not equate to one whole life saved. In general terms, postponement of death is not the same as lives saved.
Is it not likely that the 'health benefits', population-wide, will never really surface because people will die for other reasons before such benefits can have any effect?

Vova said...

Two brief comments -

The prevention paradox in alcohol research can only 'work' if there are no beneficial effects of consumption and relevant risk functions are strictly increasing from zero consumption upwards. These conditions are not satisfied.

The idea was first introduced in alcohol in a paper by Kreitman
http://onlinelibrary.wiley.com/doi/10.1111/j.1360-0443.1986.tb00342.x/abstract
which was itself a rescue attempt for control policy following the academic (but not advocacy) death of the Ledermann idea. I was there!

The other angle related to control of consumption for health reasons is the loss of utility suffered by consumers. This aspect of every attempt to control use of everyday substances which some epidemiologist has found to be associated with an adverse health outcome and often unjustifiably identified as causal is almost always ignored in the resulting polemic.

Anonymous said...

One (individual) man's meat is another man's poison. How do statistics deal with that?

Jonathan Bagley said...

Haydock writes,

"However, this is to misunderstand how population-level policies work: they don’t aim to make everyone necessarily live longer by a day or so; they aim to make an average population live longer, and affect some individuals significantly. The nature of the prevention paradox is that an individual won’t be noticeably affected by the small reduction in risk their change in consumption habits produces. These small reductions in risk, though, when aggregated across a whole population, can produce a notable reduction in overall mortality."

Now, "they aim to make an average population live longer", doesn't make much sense. I suspect he means, "they aim to extend the average lifetime of members of a population." He then admits that any increase in individuals' lifetimes is likely to be small, but fails to mention that this will result in a similarly small increase in the average lifetime, and so, by his criteria, is pointless. He then goes on to illustrate what I consider to be the fatal flaw in the Public Health mentality in justifying the policy by claiming that the AGGREGATED, not averaged, reductions in risk, can produce a notable reduction in overall (by this he must mean aggregated, not averaged, over the whole population) mortality. But overall mortality is irrelevant. Were a PH policy applied to China, rather than to Iceland, there might be an enormous aggregated gain in life expectancy rather than a modest gain. However, an individual in China would enjoy the same costs and benefits as a similar individual in Iceland. Justifying a policy which impacts on a population only via its impact on the individual members, cannot be justified on grounds dependent on the size of the population. They should be judged by the effect on individuals. An individual would probably choose not to forgo a pleasure in the anticipation of living on average two months longer, so PH should not be justifying a policy by multiplying two months by either 300,000 in the case of Iceland, or 1.whatever billion in the case of China. This is a completely situation to mass vaccination, which PH might quote in defence. Here, one person's behaviour significantly affects others.