Minimum alcohol pricing is policy-based evidence-making

The Scottish government can now legally bring in minimum alcohol pricing, which will require alcoholic drinks to be sold for at least 50p per unit of alcohol. For reference, a can of 5% strength beer has 2.5 units and a 750ml bottle of 13.5% strength wine has 10 units, so their minimum prices will be £1.25 per can and £5 per bottle respectively. It’s only likely to affect drinks sold in shops.

The SNP and other supportive groups say that this will be good for people’s health – the biggest boost since the indoor smoking ban, in fact. Their basis for this is that alcohol consumption tends to fall when the price rises, and people with the worst dependencies on alcohol usually drink the cheapest stuff. 

But there’s a leap here. Do heavy drinkers cut consumption when the price of booze rises? Intuitively, you’d think not. As one Scottish Twitter user pointed out yesterday, “People with problems will get what the need even meaning cutting out necessities to do so.” 

The empirical evidence supports this. This 1995 paper found that the heaviest drinkers’ responsiveness to price changes was statistically indistinguishable from zero, though it was based on very old data from the 1980s. This more recent one found that hazardous and harmful drinkers (people who consume more than 17.5 units per week) had a very low response to price changes. And this 2013 review of 19 studies found only two that found a significant and substantial reduction in drinking rates in response to alcohol price rises – “and even these two showed mixed results”. 

This 2016 PhD thesis, by Dr Robert Pryce, found that heavy drinkers’ price elasticity of demand was only barely distinguishable from zero, and concluded that:

the quantity results show that price-based measures will have little effect in reducing heavy consumption because of their small absolute price elasticity, whilst simultaneously having a large negative effect on consumer surplus for the light drinking majority, because of their large absolute price elasticity

Not only do the advocates of minimum pricing ignore this, their most-used model (the ‘Sheffield Alcohol Policy Model’) assumes that heavy drinkers have the highest responsiveness to price changes of all types of drinkers. 

This was pointed out by John C. Duffy, a leading health statistician, in a 2012 report for the ASI that he coauthored with Christopher Snowdon. Duffy is scathing about this model in general, describing it as “resorting to numerology” when desired data was not available. He shows that the assumption about heavy drinkers’ elasticity is extremely shaky, and crucial to the health claims about the policy.

It comes from the fact that heavy drinkers are the most price sensitive to price rises for a single brand or type of drink – ie, heavy drinkers are the most likely group to switch from drinking wine if the price of wine rises, or Fosters if the price of Fosters rises. 

But this does not show that their overall alcohol consumption falls, just their consumption of that type of drink. And does not imply that a general rise in alcohol prices will cause heavy drinkers to cut their alcohol consumption down – the evidence above suggests otherwise. That evidence shows that heavy drinkers are the least sensitive to price rises of alcohol overall. They are just the most willing to switch between types of alcohol when relative prices change. It’s extremely weird that the Sheffield Alcohol Policy Model interprets the data in this counterintuitive and contrarian way.

This assumption is the basis for minimum pricing advocates’ claims about health benefits from their policy, along with poorly identified and statistically sloppy studies of the effects of minimum pricing in British Columbia, Canada. 

As Chris Snowdon points out the most commonly cited study did not include any control group, allowing it to attribute a fall in crime in British Columbia to minimum pricing even though the rest of the country, which had no minimum pricing, and indeed most of the developed world also saw reductions in these things. The UK, so far free of minimum pricing, saw a greater fall in crime over the same period. 

As well as that, it extrapolates wildly from tiny, noisy changes in alcohol-attributable death rates to make big claims that simply do not make sense. A 1% rise in the price of alcohol taking place alongside a 3% fall in alcohol-attributable deaths is not established to be causal, and even worse, is multiplied by ten to produce a Daily Mail-friendly headline that a 10% rise in price would cut alcohol deaths by 30%. It’s 'evidence' in name only.

My opposition to minimum pricing isn’t just that there is little to no evidence in favour of it. It is that it causes harm to many people. Many non-problem drinkers like cheap beer and wine from Aldi and Lidl – I’m a big fan of Aldi’s 75p/can Reinbacher pilsner and Lidl’s £4.29 Cotes du Rhone, for instance.

All of those people, many of whom will be on tight budgets already, will either be made poorer, if they keep drinking, or unhappier, if they have to cut down or stop. Problem drinkers, since their elasticity is low, will be much poorer and may cut down on things like heating, and street drinkers may resort to bootleg alcohol or worse. Remember, too, that tax revenues go to the government, but minimum pricing ‘revenues’ go to supermarkets.

Once implemented, the policy will probably ratchet upwards too. 50p/unit is on the lower end of minimum pricing advocates’ target – most would prefer to start at 70p/unit and work upwards from there. And the traditional objective of policies like this, pricing in the negative externalities caused by drinking, is much better achieved through tax, since people who drink expensive alcohol to excess are just as likely to be problems to others as people who drink cheaper alcohol to excess. 

Worst of all, I doubt that the Scottish experience, assuming it does fail, will end up giving us evidence to put English politicians off minimum pricing. Minimum pricing advocates have shown a willingness to distort and misrepresent the evidence to further their goals, and will probably hold up Scotland as a success no matter what happens there. ‘Policy-based evidence-making’ has got them this far – why stop now?