This single computer model has been responsible for numerous predictions about what would happen if minimum pricing were introduced. Such forecasts are speculative by their nature, but it is only by reading the Sheffield studies in full that one sees how wild the speculation is.
Hardly anyone does read these studies, of course, least of all politicians and journalists, but the statistician John C. Duffy has and today sees the publication of a paper I have co-authored with him which looks at the flaws in the model in detail.
You can download The Minimal Evidence for Minimum Pricing (Adam Smith Institute) here.
There are bits of coverage in The Sun, The Express, The Telegraph and The Independent. I've written articles about it for The Spectator and ConHome. And this is the press release...
Basis for government alcohol policy is bogusAs a consultation on minimum alcohol pricing launches, the Adam Smith Institute (ASI) has released a report showing that the evidence base for minimum alcohol pricing is, to all intents and purposes, non-existent.
Co-authored by John C. Duffy, a statistician with forty years experience in the field of alcohol epidemiology, the report explains that most of the estimated health outcomes, used to justify calls for a minimum alcohol pricing of 40p or 50p per unit, have come from a single, flawed computer model.
This model, the Sheffield Alcohol Policy Model, is used to predict minimum pricing’s effect on everything from NHS expenditure to unemployment, and is based on false assumptions and wild speculation which render any predictions meaningless.
Arguments for minimum alcohol pricing based on this computer model should be ignored and the Sheffield Alcohol Policy Model should not play a role in the debate.
The model is deeply flawed for a number of reasons:
When calculating health outcomes, the model assumes that heavy drinkers are more likely to reduce their alcohol consumption as a result of a price rise. This contrasts with ample evidence that heavy drinkers are less price-sensitive. The majority of alcohol related harm is linked to heavy drinkers who are much less likely to be deterred by price rise than a casual consumer. By claiming that any price rise will lead to bigger drops in consumption amongst heavy drinkers, the model ignores the complex psychological and societal factors leading to alcoholism and alcohol-related violence.
It bases its calculations on controversial beliefs regarding the relationship between per capita consumption and rates of alcohol related harm. A low rate of per capita alcohol consumption is no guarantee of better health outcomes. There is little to be gained from making moderate drinkers reduce their consumption slightly.
The model provides figures without estimates of error and ignores statistical error in the alcohol-harm relationship. Patterns of consumption and harm are not the same in all countries. When Denmark reduced the tax on spirits by 45% in 2003 it did not experience any increase in alcohol consumption, and instead there was a decline in alcohol-related problems. As alcohol has become more affordable as a result of rising incomes we have seen a decline in alcohol consumption across most of Europe and the US. There are, of course, examples where higher prices have reduced alcohol consumption and alcohol related harm, but it is clear that price interventions are highly unpredictable and cannot be easily extrapolated from a computer model.
The model ignores other potential negative social outcomes of minimum pricing, such as a likely increase in the illicit alcohol trade and the greater poverty it may push many consumers into. It also ignores some of the health benefits associated with moderate drinking habits.
Alcohol-related harm may rise, fall or stay the same under a minimum pricing regime. The evidence simply does not exist for reliable forecasts to be made about the consequences of such a far-reaching policy. The Sheffield Alcohol Policy Model is riddled with flaws and wishful thinking. It has no merit as a guide to policy and the government should not base legislation on such speculative and weak statistics.
Christopher Snowdon, co-author of the report, adds: “In the era of evidence-based policy, it seems that speculative statistics are considered superior to no statistics and a wrong answer is better than no answer. We argue that this is a mistake. The aura of scientific certainty, or even mild confidence, in computer-generated numbers based on dubious assumptions is misplaced. Minimum pricing might reduce alcohol harm, or it might increase it, or it might bring about other unexpected consequences, good or bad. An admission that the evidence base is, to all intents and purposes, non-existent is less likely to mislead decision-makers than a spurious prediction. The only certainty is that minimum pricing will transfer large sums of money from the poorest people in society to wealthy industries. This is a deeply regressive leap into the unknown and it should not be taken as a response to wafer-thin ‘evidence’.”
John C. Duffy, co-author of the report adds: “A supporter of the model might ask me ‘If you’re so smart, what’s your model – what do you predict?’ My answer is that I don’t have a model and therefore I won’t make a prediction. There is not enough information around to produce a reliable model and I won’t invent one that is engineered (by undemonstrated assumptions) to fit the prevailing facts and pretend that it is of any use for prediction. As Taleb says in The Black Swan about those who attempt to justify worthless predictions because ‘that’s their job’—get another job.”