Statistician John C. Duffy and ASI fellow Christopher Snowdon assess the Sheffield Alcohol Policy Model, used as the basis for the British and Scottish governments’ calls for minimum alcohol pricing. They find that the model is deeply flawed, based on faulty premises and used to justify policy far beyond what it actually proves.
1. The Conservative Party and the Scottish National Party have both stated their intention of introducing a minimum floor price for alcohol, levied at around 50p per unit. Advocates of minimum pricing claim that the policy will significantly reduce alcohol consumption and the problems associated with hazardous drinking.
2. Estimates of how minimum pricing will affect health outcomes have overwhelmingly come from a single computer model—the Sheffield Alcohol Policy Model. This paper argues that the model is based on unreasonable assumptions which render its figures meaningless.
3. Amongst the problems with the Sheffield model is its false assumption that heavy drinkers are more likely to reduce their consumption of alcohol as a result of a price rise. Its calculations are based on controversial beliefs about the relationship between per capita alcohol consumption and rates of alcoholrelated harm. Its assumptions about the relationship between price and consumption have frequently been refuted by real world evidence.
4. The Sheffield model provides figures without estimates of error and ignores statistical error in the alcohol-harm relationship. Data is drawn from different populations and applied to England and Scotland as if patterns of consumption and harm are the same in all countries. When data is not available, the model resorts to what is essentially numerology. Insufficient data is provided for the model to be recreated and tested by third parties.
5. The model ignores the likely effects of minimum pricing on the illicit alcohol trade, it disregards the health benefits of moderate drinking and fails to take account of the secondary poverty created by regressive price rises. The decline in alcohol consumption seen in Britain in recent years has not led to the outcomes predicted by the model.
6. We conclude that predictions based on the Sheffield Alcohol Policy Model are entirely speculative and do not deserve the exalted status they have been afforded in the policy debate.