The authorities must stop putting their faith in the bad science of “public health”
If it is true, as J. K. Galbraith once said, that the only function of economic forecasting is to make astrology look respectable, economists should be grateful that we have “public health” modelling to make their forecasts look impressive.
For the best part of twenty years, one model in particular has been remarkably influential despite being consistently and awfully wrong. The Sheffield Alcohol Policy Model emerged in 2008 when the Department of Health wanted to put some evidential meat on the bones of minimum pricing, an anti-alcohol policy that was later dropped in England but warmly embraced in Scotland and Wales. Since 2009, the Scottish government has thrown large sums of taxpayers’ money at a team based at Sheffield University — currently called the Sheffield Addictions Research Group — to justify its neo-temperance policies.
When the Scottish government banned multi-buy discounts on alcohol in 2011, the Sheffield model predicted that it would reduce alcohol sales. It didn’t. The model predicted that cutting the drink-drive limit from 80mg to 50mg (as the UK government is currently proposing) would “reduce fatalities by 6.4% and injuries by 1.4% in the first year after its implementation” but when Scotland introduced such a change in 2014, there was no reduction in either. Numerous iterations of the model predicted that minimum pricing would significantly reduce levels of crime (it didn’t) and alcohol-related mortality (it didn’t). When the model was transferred to Wales, it predicted a similar decline in the number of people drinking themselves to death. This also did not happen; Wales has seen the largest increase in alcohol-specific deaths of any part of the UK since minimum pricing was introduced in 2020.
The only thing the model has ever got right was that minimum pricing in Scotland would lead to a modest (3 per cent) reduction in alcohol consumption, although it made the crucial error of assuming that this would come about as a result of the heaviest drinkers cutting their consumption (which didn’t happen). In general, the Sheffield Alcohol Policy Model is only useful as a reverse weather vane. Whatever it says, you can safely assume the opposite.
You might think that a long track record of total failure would have led to the work drying up for the Poundland Mystic Megs of Sheffield University, but the government’s appetite for policy-based evidence has kept the commissions rolling in. Minimum pricing in both Scotland and Wales came with a sunset clause, but pygmy politicians in neither of these fiefdoms had any intention of letting go. On the contrary, they wanted to increase the price from 50p to 65p, and they knew who to turn to for “evidence”.
Last year, the Sheffield team published a report commissioned by the Welsh Government which claimed that increasing Wales’s minimum price to 65p would lead to a 4.4 per cent reduction in alcohol-specific deaths and that regularly increasing the price in line with inflation was “crucial”.
This month saw the publication of a similar study commissioned by the Scottish Government in which the Sheffield team show that they are now fully cocooned from reality. In the introduction, they not only assert that minimum pricing is “a particularly well-targeted pricing policy”, but claim that Scotland has seen a “rapid reduction in deaths from alcohol dependence syndrome and alcohol-related liver disease” since minimum pricing was introduced in 2018. This is demonstrably untrue; the death rate from alcohol-related liver disease is at a 13 year high. They also claim that minimum pricing reduced “wholly alcohol-attributable deaths by 13.4% all else being equal”. They have actually risen by 9 per cent.
It is only possible to believe such obvious nonsense if you ignore the real world and immerse yourself in theoretical models. The new Sheffield study adds yet another layer of fantasy, claiming that increasing the minimum price to 65p will reduce alcohol consumption by 12 per cent and prevent 132 deaths in the first year. The Scottish government has, in fact, already raised the price to 65p. It did so in September 2024 and there is no sign of any of the Sheffield team’s future-gazing aligning with the facts.
A particularly ludicrous aspect of the new study is that it is all based on “the modelled behaviour of the synthetic population in 2019” because the authors were unable to get the “necessary data” for the last six years. The whole thing is based on what would happen if the minimum price was raised in 2019 and ignores not only the large spike in alcohol-specific deaths that occurred across the UK in 2020 but also the large spike in inflation that occurred between 2021 and 2023. Largely thanks to the miracle of quantitative easing, 65p today is worth slightly less than 50p was worth when minimum pricing was introduced in Scotland in 2018.
The failure of minimum pricing in Scotland and Wales is so obvious that it should be beyond debate
The authors do not explain how a minimum price that has not risen in real terms could have such a dramatic effect on consumption and mortality, nor do they explain why their model shows that it would be three times more powerful in Scotland than in Wales. Ultimately, it doesn’t matter. There is no point in critiquing the evidence because, as the minimum pricing episode has vividly shown, evidence-based policy is a mirage. The only people who benefit from the conceit that “public health” policy-making is evidence-based are the people who are paid to conjure up the evidence (in 2014, the Sheffield Alcohol Policy Model was renamed the Sheffield Tobacco and Alcohol Policy Model, presumably to broaden the market).
The failure of minimum pricing in Scotland and Wales is so obvious that it should be beyond debate. Officially, however, it was a success. The policy was introduced off the back of modelling, justified retrospectively with modelling and when the time came to hike up the price there was a model to support that too. Never mind how many people are dying from alcohol abuse. If the model says it works, it works.
As a coda, the Director-General of the World Health Organisation gave a speech yesterday in which he awkwardly pivoted from talking about the humanitarian crisis of Sudan’s civil war to discussing what really interests him: taxes on sugary drinks. He claimed that “when they are done right, they are a powerful tool for health”. His prime example — indeed, only example — of a sugar tax “done right” is the UK where, he claimed, the Soft Drinks Industry Levy “has been associated with lower obesity rates in girls aged 10 and 11”.
An astute reader might wonder what was so special about girls of this age, and why this “powerful tool for health” didn’t work on girls of other ages or on boys in general. But even a cynic would assume that there must have been a reduction in obesity rates among girls aged 10 and 11. Why would he say it otherwise? But there wasn’t. Notwithstanding the fact that the UK’s childhood obesity measure is worthless, the rate among 10-11 year old girls was 18 per cent when the sugar tax was introduced and is currently 19.6 per cent. But those are only the facts. A model says it fell and in the world of “public health”, that’s all that matters.
