Tuesday, 22 September 2015

The Institute of Alcohol Studies' response to Alcohol and the Public Purse

The UK Temperance Alliance, which these days prefers to be known as the Institute of Alcohol Studies (IAS), has responded to my IEA report about the cost of alcohol to the English taxpayer. In the report I showed that the financial, out-of-pocket costs to the government of alcohol-related crime, ill health and worklessness were unlikely to be higher than £3.9 billion a year, which is about a third of what drinkers pay in alcohol duty.

The IAS reply seems to have been written in reasonably good faith so I'll try to be equally constructive in my reply. They begin...

The most important point about the IEA paper is that it deliberately addresses only a very limited question – what the impact of alcohol is on the government’s fiscal balance. By design, it does not look at broader social or economic costs.

That's right. It looks at the cost to the government and therefore the taxpayer. Why? There are two reasons. Firstly, because I think that is what people are most interested in. Secondly, because the wider societal cost of £21 billion that includes intangible costs (such as emotional distress) and private costs (such as lost productivity) is routinely portrayed as being a direct cost to the taxpayer. For example, this is how the figure is typically presented in the media...

The government is seeking to reduce irresponsible drinking that costs the taxpayer 21 billion pounds a year in police and medical bills

Here is Theresa May on a government website:

'Alcohol-fuelled harm costs taxpayers £21 billion a year.'

Here is the Alcohol Health Alliance explicitly comparing a figure dominated by private and emotional costs with financial tax revenues:

And here is Katherine Brown, director of the IAS, describing the £21 billion as a cost to the economy while at the same time saying that it is a direct cost to the taxpayer.

“This government’s failure to keep its promise to come down hard on cheap alcohol has let everybody down except the drinks industry. It’s let the NHS down, the police down, but most of all it lets every taxpayer down. We are all footing the £21 billion bill of alcohol harm to our economy each year.”

Back to the IAS response...

This restriction inevitably limits the force of the IEA’s argument. Most people who are concerned about the level of alcohol harm in the UK are not primarily motivated by a desire to balance the government’s budget. Rather, they are concerned by the damage to public health, crime, social disorder and wider economic costs caused by excessive drinking.

I'm not sure that is true, otherwise temperance groups wouldn't go to such lengths to make the public think that wider societal costs are costs to the taxpayer. At the very least, people are more interested in the financial costs to the state in the context of discussions around alcohol taxes which are supposed to compensate for those costs.

The fact remains that somebody needs to pay these costs – whether it’s protecting themselves against crime, replacing goods that are stolen or vandalised, or functioning less productively at work. The IEA argue that it is misleading to balance these costs against revenue from taxation. But in that case, we need to acknowledge that there is no compensatory mechanism for the billions of pounds of harm suffered by private individuals and businesses.

But there is no compensatory mechanism and the IAS is not suggesting one. To be clear, putting more tax on alcohol will not compensate employers for hungover workers, nor will it pay to have someone's garden fence repaired. Tax goes to government. If alcohol duty is about compensation, it should therefore only meet the costs incurred by government.

(This is separate from the argument about Pigouvian taxation which the Alcohol Health Alliance is vaguely alluding to in the tweet above. The temperance lobby seems to think that alcohol duty should be equal to the gross negative externalities associated with alcohol. This, however, is a misunderstanding of what a Pigouvian tax is. Pigou's aim was not to raise an amount of tax equal to the gross externalities but to use tax to increase the private cost of a product to meet the net externalities. The private cost includes the price of the product before tax. People in Britain spend more than £30 billion on alcohol. That is the private cost, not just the £12 billion or so they spend on alcohol duty.)

The report has performed a valuable service in bringing to light the limitations of previous data/analysis and the confusion they have caused

Hear, hear!

The IEA is also correct in identifying the dangers of conflating social costs, economic costs and fiscal costs. ‘Alcohol and the Public Purse’ should encourage all sides of the debate to be more careful in the language they use. It is vitally important to be clear about the difference between different types of costs, and just as important to clarify who bears these costs – the government, the drinker, other people or businesses.

If it achieves that, I will be very happy (and surprised).

Much of the coverage and interpretation of the IEA’s findings has been misleading

While the IEA’s report criticises public health campaigners for confusing costs to the economy with costs to the Treasury, it is disappointing to see newspaper reports of the paper making the very same error:

“Boozers are subsidising teetotallers as research shows how much they pour into the economy”10 (The Mirror)

“Britain’s drinkers are helping to boost the economy by £6.5 billion a year”11 (The Express)

As discussed above, the IEA’s report says nothing about the cost to the economy, but only refers to the cost to the government.

Yes, it's very frustrating, but we can't control the headlines and nothing in the IEA's press release encouraged this sort of misreporting. The newspapers have been reporting broader economic costs as costs to the taxpayer for years (without being chastised by the IAS) so it is perhaps no surprise that they made a similar mistake this time. I commented on this at the time, both on Twitter and on this blog, saying...

Quite a few of the headlines failed to distinguish between the economy and the treasury - to be clear, taxing people does not 'boost the economy' - but Rome wasn't built in a day.

Back to the IAS...

While these headlines are ultimately the responsibility of the newspapers that publish them, it is notable that the IEA’s official twitter account made the same mistake, tweeting: “Britain's drinkers are helping to boost the economy by £6.5 billion a year”.

Ah, come on, they were tweeting coverage of the report all day and that was the headline in the Daily Express. They read the Express story, pressed the little Twitter button and that's what came out. People do this all the time. It doesn't mean they endorse every word of the article.

The idea of a ‘subsidy’ from drinkers to non-drinkers, which was central to both the IEA’s press release and many of the press reports, is misleading in a more subtle way. This implies that non-drinkers are better off for the fact that other people drink, because this (according to the IEA) reduces their tax burden. But if we are comparing the balance of costs and benefits between drinkers and non-drinkers, it is no longer appropriate to ignore private costs. Therefore, it does not follow that drinkers subsidise non-drinkers, unless the IEA can show that the tax ‘subsidy’ (if it exists) is greater than the private costs to non-drinkers.

The point is that non-drinkers are able to pay less tax as a result of other people drinking. I take the point that there may be other private costs (and, indeed, benefits) from other people drinking that are not reflected in the tax system, but - as mentioned above - they would not be compensated by taxation paid to government anyway.

Moreover, even in narrow fiscal terms, this ‘subsidy’ will be significantly smaller than the widely claimed £6.5bn. The IEA does not account for the fact that drinkers also use public services that are not directly related to their drinking – they still send their children to state schools, they are still treated for non-alcohol related ailments on the NHS, they still have their rubbish bins collected. If we accept the IEA’s numbers, alcohol raises £6.5bn for the Treasury, which is presumably spent on public services. But 79% of this £6.5bn will be spent on drinkers, since drinkers comprise 79% of the population (assuming that drinkers and non-drinkers are equally likely to use public services not related to drinking). This would imply that the subsidy is in fact closer to £1.4bn (21% of £6.5bn).

I applaud the creativity of this argument. It raises an interesting question. Let's say you had a tax on women that brought in £10 billion. Would you say that women are subsidising men to the tune of £10 billion or would you say that the subsidy is only £5 billion because women use half of the public services paid for by the tax? It seems to me that the former would be the normal way of describing it whereas the latter has the whiff of sophistry, but I take the point.

However, if you want to chop off 79 per cent of the benefit (the 'subsidy') because the money will ultimately be spent on drinkers then you must equally chop off 79 per cent of the cost (eg. spending on alcohol-related harm) because it would be paid by drinkers in the absence of a specific duty. By this logic, 79 per cent of the 'cost of alcohol' is not an externality at all because it is paid by people who drink.

If the costs of alcohol to public services came out of general taxation, drinkers would pay 79 per cent of them. So, by the IAS's logic, the cost to the taxpayer is not £3.9 billion but just £819 million. You can't have it both ways.

In ‘Alcohol and the Public Purse’, there are a number of costs that are omitted from the IEA’s analysis. Most significantly, the IEA do not attribute any social care costs to alcohol, despite the evidence that alcohol is often related to child neglect and mistreatment. This has been estimated to cost the government between £1bn- £2.5bn, so on its own could add 50% the IEA’s estimate of alcohol costs.

My report was explicitly modelled on the Cabinet Office report and the subsequent (thinner) documents that were the source of the £21 billion figure. None of them mentioned social care and I have not heard them criticised for ignoring it.

Whilst there may be costs to be found in social care, it is hardly surprising that it has been ignored to date. The data are practically non-existent. The reference given by the IAS is a report about the economic and social costs of alcohol in Leeds. It includes a page and a half about social care with some back-of-an-envelope estimates about the costs associated with alcohol. The estimates are not based on evidence from Leeds. Instead, they are based on a survey which found that parental alcohol misuse was a factor in 14 per cent of child care cases in four London boroughs. The authors assume causation and attribute 14 per cent of total child care social work spending in Leeds to 'alcohol-related expenditure'. It then adds in an estimate for adult social work, but because 'it is not clear what proportion of this expenditure was spent in relation to alcohol misuse' they just assume 25-50 per cent.

Presumably, IAS have extrapolated this guesstimate from Leeds across the whole country to reach their estimate of £1-2.5 billion. It is exactly the kind of guesswork that IAS said I 'performed a valuable service in bringing to light' and is nowhere near robust enough for an IEA report. 

Additionally, despite the IEA’s claim that they “err towards generosity rather than conservatism when compiling the estimates”, there are a few assumptions that appear to understate the cost of alcohol. First, the IEA ignore hospital admissions where alcohol is a secondary diagnosis (except where these are classified as an ‘external cause’).

This was explained at length in the report. It is ridiculous to assume that a secondary diagnosis was the primary cause for a hospital visit. Public Health England and the Department of Health both now agree that only primary diagnoses should be counted.

Second, the IEA assume that 14% of A&E visits are attributable to alcohol. While they acknowledge that there is significant uncertainty around this figure, it is worth noting that this is towards the lower end of the 2-40% range that is sometimes cited.

This was also dealt with in depth in the report. It should be obvious that if one estimate says that 2 per cent of A & E admissions are alcohol-related and another says 40 per cent then one of them is totally wrong. IAS are implying that you should split the difference and assume the real figure is at the midpoint between the two but that means using a figure that you know is totally wrong to come up with an average.

A better approach is to look at the estimates and see which one looks totally wrong. As I explain in the report, figures of 30-40 per cent are simply not credible whereas there are quite a few studies - that's studies, not surveys - of A & E admissions which arrive at single figure estimates. I went with 14 per cent because it was at the high end of the credible range. I strongly suspect that the real figure is quite a bit lower than this.

The IAS has promised to publish some of its own research on A & E admissions soon and I believe that government is working on a new economic assessment on the total cost of alcohol. Hopefully the government will at least bear in mind the point on which the IAS and myself are agreed - that the differences between difference types of cost should be made clear.

No comments: