Opinion: Alcohol policy is always contentious – but let’s start with something that should be uncontroversial.
If the Government wants to reduce alcohol-related harm, it should aim for measures that do more good than harm overall.
If a harm-reducing policy stacks up, it does so whether the overall social cost of alcohol is $10 billion, $1 billion, or $100 million.
If the benefits of the policy, including harm-reduction, exceed the costs of the policy, including impositions on harmless drinkers, the policy is a good idea.
What matters is cost-effectiveness. And the alcohol levy is meant to fund those kinds of cost-effective interventions.
There are a few more things I think we should all be able to agree on. But if I don’t foreshadow the controversial parts, there is always a risk that people will stop reading. So here goes.
I have finally received a set of documents following Official Information Act requests that have been in progress since September 2023. It took the Ombudsman’s intervention.
The documents show that the initial scope for the independent review into the alcohol levy was partially motivated by a measure of the social cost of alcohol developed by the consultants Berl. As one of the consultants producing the report pointed out in an extensive critique, that measure, and the figures it produces, are flawed. But that critique was removed from the final report.
All in all, it looks like the Ministry of Health wants to promote a flawed and inflated measure of the social cost of alcohol to help it lobby for higher alcohol excise and for a higher levy.
This matters. The ministry advises the minister. The minister makes decisions. If the ministry ensured the deletion of material from a commissioned report contrary to the ministry’s preferred view, that is a problem. It would be unbecoming of a ministry presenting itself as a helpful servant to the minister’s policy deliberations.
Alcohol faces two ‘taxes’. The Government collects excise on alcohol that goes into government general revenues. It also collects a levy on alcohol that funds the Alcohol Levy Fund, used for harm-reduction activities by the Ministry of Health’s Health Promotion Agency.
Excise is $64.57 per litre of pure alcohol in spirits, meaning that $18.08 of the $56.99 price of a 700ml bottle of Kraken Rum (40 percent alcohol) is excise. The same $18.08 would apply on any 700ml bottle of 40 percent spirits, whether the cheapest vodka or the most expensive rare whisky: excise depends only on the quantity of pure alcohol in the bottle, but with a higher charge on spirits than on beer or wine. On July 1, excise increases to $67.22 per litre of pure alcohol.
The levy is much lower than excise. On spirits, it is $0.13 per litre of pure alcohol – or about 4c on that bottle of rum.
The two charges serve different purposes.
If the Government wants to reduce overall alcohol consumption, or to raise revenue to offset its costs to the public health system or to police, a price instrument like excise is appropriate.
The best metastudy I have seen suggests that a 10 percent increase in the price of alcohol reduces overall consumption by 4.4 percent, but only reduces heavy drinkers’ consumption by 2.8 percent.
This illustrates what a cumbersome tool excise is for dealing with harmful consumption. The person who drinks a bottle of whisky per day pays the same amount of excise, per bottle, as the person who drinks a bottle of whisky per year.
So, even if there were some way of getting a ‘right’ level of excise, on average, it would always overcharge harmless drinkers while undercharging those who cause harm.
In any case, the revenue raised by alcohol excise is not set aside for extra nighttime shifts for police or for accident and emergency departments.
Policy cannot solely rely on price measures such as excise for addressing harms. It would be like trying to deal with street-racing and ram-raids by increasing petrol excise. In principle, there could be an excise level that would make street-racing unaffordable. But it wouldn’t be a good solution. It would impose too great a cost on those whose consumption is harmless.
The much-smaller levy, by contrast, directly funds harm-reduction programmes. And the Government has been reviewing the levy. The Stage 1 Rapid Review was published in July 2023.
The Stage 1 Rapid Revew
Unfortunately, we do not yet know whether any of the levy’s funded projects are cost-effective.
The Stage 1 review noted it was “not possible to quantify to what extent current levy investments reduce alcohol-related harm in the timeframe and with the material made available in stage 1 of this review”.
From the released documents, this appears to have been a down-scoping from the original work.
Chapter 5 of the Stage 1 Rapid Review was to have been titled “Effectiveness of harm reduction interventions”, with subsections on projects in health promotion, health protection, and community investment.
That was consistent with the original review scope, which included consideration of “the total levy fund and its impact on alcohol harm generally”.
In late February, a Manager for Alcohol Policy & Advice suggested revising the chapter heading to “Evidence for harm reduction interventions”, noting that it “is a slightly broader focus and would not preclude inclusion of evidence for effectiveness where this is available”.
Perhaps more detail on effectiveness of funded work will appear in the Stage 2 review, reportedly due sometime this week. The Stage 1 review prudently recommended against increases in the alcohol levy until better evidence could be mustered.
Nevertheless, new alcohol levy rates were announced last week, taking effect from July 1 – an increase to $0.19 per litre levy on spirits.
It surely would have been better for public evidence of effectiveness to have preceded the levy announcement.
The Stage 1 Review and the Social Cost of Alcohol
The review’s treatment of the overall social cost of alcohol was more worrying.
The initial scope for the review repeatedly highlighted Berl’s measure of the social cost of alcohol: $7.8 billion. Berl’s figure was cited in the purpose statement for the consultancy services order – it helped in motivating the review.
The Stage 1 review was meant to assess the social cost figure.
The published review included just three short paragraphs on the cost of alcohol-related harm. It noted that Berl’s figure, “or rather the methods used to generate it, have been criticised by some commentators” but that “it has been widely cited in the alcohol-harm research and policy space in New Zealand”. No particulars of the criticism were included, but the bibliography cited the work that co-authors and I had undertaken when I was on faculty in economics at the University of Canterbury.
So I dug into what had been included in the earlier drafts.
At first, the ministry stonewalled completely – refusing everything. The Ombudsman intervened.
The released drafts included a five-page discussion of Berl’s cost-of-alcohol figure, about three and a half pages of which summarised our previous co-authored critique of the Berl figure. We had found that perhaps 20 percent of Berl’s number might be defensible using a more standard economic method, and after correcting for some double-counting, but that even that figure would be unreliable.
An early draft of the review reminded readers that Berl’s figure included double-counting, as well as drinkers’ spending on their own alcohol: issues that will be familiar to Newsroom readers.
What was to have been an independent review instead had its conclusions shifted in line with the biases of the ministry
The draft report noted that “Some of these issues are related to the fundamental differences between an economic approach and a public health approach (eg exclusion of private costs), and some are clearly methodological flaws (eg double counting).”
None of those warnings made it into the final Stage 1 review.
Tracked comments from review rounds give some indication as to why.
Feedback from the National Public Health Service (author identified only as A5) suggested deep unfamiliarity with the critique of Berl’s work, and hostility toward alternative approaches.
They misunderstood how Berl had skewed the health cost figure in its report, criticised me personally, and then suggested it was unbalanced to rely solely on my critique.
Feedback from a reviewer at the ministry was about as generous.
I am not sure how many critiques of a consultancy paper they might expect in a small country. But I cannot take sole credit: Matthew Burgess and Brad Taylor were co-authors. In any case, a critique should stand or fall on its merits, not on the identity of its author. One might have hoped that the ministry would want the figures on which they rely to be robust. But that would be to misunderstand the public sector.
Instead, a reasonable summary of the problems in Berl’s cost figure was expunged from the published review.
This does not increase confidence in the levy review process. What was to have been an independent review instead had its conclusions shifted in line with the biases of the ministry.
It took over half a year of ongoing OIA requests, interventions, and a bit of luck to see how the Stage 1 Review was mangled, and what the independent reviewers had really found.
I do not look forward to Stage 2’s release.
The bottom line remains, as always, that interventions are warranted if they prove cost-effective. Tallies of the overall social cost, large or small, are not helpful to that end. But they are helpful in lobbying for more action in an area, with a large number being used as prima facie justification.
It’s a shame that the Public Health Agency seems keener on using large and unsound numbers to whet appetites for its preferred policies than on being a faithful advisor to its minister and servant of the broader public.
If the Alcohol Levy Fund was used to pay for the review, I do not think that anyone who has contributed to that fund through their purchases should be happy about it.
Dr Eric Crampton is Chief Economist with the New Zealand Initiative, a business-funded think-tank. Its members, including Lion Co, are listed on its website here. His work on the Berl cost of alcohol study, joint with Matthew Burgess and Brad Taylor, was undertaken while he lectured with the Department of Economics at the University of Canterbury.
Nest page for links to NZ Initiative OIA correspondence: nzinitiative.org.nz/links/background-material-newsroom
Certainly, “The person who drinks a bottle of whisky per day pays the same amount of excise, per bottle, as the person who drinks a bottle of whisky per year.” However, over the year the latter pays 365 times more which as a proportion of income is commensurately higher. Won’t that material difference impact on consumption?
Just what is Newsroom’s policy on providing a platform for big business to push its case for a free run to sell an addictive and carcinogenic product?
Thanks to your leader comment in yesterdays edition I did look at Dr Crampton’s 2011 paper. What is the status of that “Working Paper”? Was it just a project for the Centre for Independent Studies. There were no acknowledgements so I presume that it was not peer reviewed and so is ‘grey literature’.
As a cyclist I am acutely aware of broken glass, particularly in the Dunedin student quarter. Almost all the shards appear to be alcohol containers. The word litter does not appear in the 2011 paper.
The section on “Rational Addiction” is a mind bending example of neo-lib contortion to designed to befuddle. Just what the alcohol multi-nationals want.
I have a long-runing disgruntlement with the wowserism that dictates most commentary on alcohol issues in New Zealand. Guyon Espiner recently published a report about the Police view on this matter. I commented in response:
“I thought the Police Commissioner heads an organisation that follows all the evidence, but your article suggests that when it comes to setting a public agenda, that is not the case.
Firstly, my conflicts of interest. I am a scientist of some 50 years standing, including a stint as a malting barley breeder, and a researcher in the Research Laboratories at Carlsberg Breweries. I also owned and ran a wine business for six years, and my son is a well-regarded winemaker still.
On the other hand, my father, his brother and his father were alcohol abusers, as was my wife’s father. Both he and my grandfather endured unspeakable privations during their war service, and the impact of what we would now term PTSD was visited on their families causing mental health issues that have reverberated for more than 100 years.
The domination of the alcohol discourse by the wowserishly-inclined led me to go into the background in considerable detail. My findings are, in summary:
• Alcohol is not addictive, at least, as addiction is normally understood (think tobacco or meth). Otherwise 80% of the adult population would be unable to control their drinking. What we see in that almost all adult alcohol abuse is as a result of people self-medicating mental health problems.
• A number of independent observations suggest that around 97% of the adult population do not abuse alcohol:
o About 3% of deaths occur earlier than they should due to alcohol consumption. These deaths range from acute alcohol intoxication through to dying as a passenger of a sober driver, whose vehicle is hit by a drunk;
o During a three-year period of relatively intensive “random” breath testing, around 3.1 million test were taken each year. Of those, around 30,000 were positive, and approximately 26,000 convictions were secured each year.
o Apparently large numbers of individuals were judged to be alcohol-affected when presenting at the Christchurch Hospital ED over show week some years ago, and trainee medicos used this impact to bewail a bothersome alcohol culture in Christchurch. However, even if show week is representative of life in the city generally, and if each presentation was by a unique individual, this would only amount to a little over 1% of the adult population in the ED catchment area.
• On the other hand, large scale populations studies show that the alcohol intake associated with the lowest mortality and morbidity is in the range 2 – 3 standard drinks a day every day. Using the amount of alcohol excise collected annually, one can estimate that this is very close to the average intake of adult New Zealanders. Those same studies show that one needs to habitually consume 6 standard drinks daily before the health impacts rise to the level suffered by teetotallers! This strongly indicates that the health system benefits from the moderate alcohol consumption of the majority of New Zealanders, allowing it to divert resources both to the abstemious and the sots.
• This means that alcohol levies are all able to be spent on harm mitigation as defined by Commissioner Coster. I don’t have current figures, but when I was able to avoid OIA formalities, I found that the police estimated that they spent 11% of their budget on alcohol-related matters (including the apparently futile roadside breath testing campaign). Over the same period government revenue from alcohol (excise, plus the GST on the marked-up cost of the excise at retail) amounted to 18% of the police budget. Go figure!
• Another point of interest. The hop extracts in beer are at least as stupefying as the alcohol, but are not taken into account when assessing impairment. It is notable that in RTA’s involving alcohol, for a given blood or breath alcohol level, the drinker is five times as likely to have achieved that level drinking beer!
Thanks Graeme,
You might be interested in my column in the Post on Monday on that interview with Coster, and the OIAed materials behind it.
Police have been actively lobbying for legislative change.
https://www.thepost.co.nz/nz-news/350309977/how-police-alcohol-activism-risks-overstepping-mark
And there is an alcohol discourse dominated (and I would guess paid for) by those who want to keep the profits from alcohol rolling in. It is characterised by attacking people who arguments they dislike by calling then “wowsers”, a pejorative term meant to demean their argument by demeaning them as people.
Whereas the people leading the charge for safe alcohol consumption are often the ones who are on the frontlines of seeing the damage that alcohol does. And alcohol does a lot of damage to people, well before they would be considered addicted or alcoholics.
We don’t know how many people are alcoholics or addicted unless they admit it or get treated for it. And alcoholism is often hidden by the consumer and those around them because of the social stigma. Addiction is still seen as a moral failure or a failure of character rather than the effect of powerful drug on the brain that changes its functioning.
Greame Coles say it doesn’t really matter that only 1% of the population turned up at ED with an alcohol related illness or injury during show week – that’s obviously a small number. But there are 52 weeks in the years, year after year after year, decade after decade after decade. Some of those in that 1% will be the same people returning, but even so the number of unique individuals who are harming themselves with alcohol and turning up at ED builds up with the relentlessness of time.
Health consequences of consumption shouldn’t be compared against people who do not drink. Some of those are alcoholics who no longer drink, some are people who can’t drink because of the medications they are on, some of those are people who choose other drugs and some of those are people who can’t afford to drink – all who have a very different health profile than the typical drinker who is a middle-class, European NZer.
Men are more likely to drink beer than women and men are more likely to drive than women (individually and as a couple), it’s not surprising that a drink driver is more likely to be a beer drinker. And that’s all an aside from the socio-economic profile of beer drinkers and where that beer is typically drunk (bars and sports clubs) and where wine and spirits are typically drunk (at home, restaurants).
I recommend people read the alcohol fact sheet put out by Health New Zealand. It provides evidence backed up by studies about alcohol use. https://www.health.govt.nz/system/files/documents/publications/alcohol-factsheets.pdf
Interestingly, it says that the Berl used “a methodology endorsed by the World Health Organization”. I guess that is why the Ministry of Health put more weight on the Berl analysis.
Agree that health outcomes for moderate drinkers in establishing a J-curve should be against lifetime non-drinkers, to avoid confounding with ‘sick quitters’.
And studies that do that still find a J-Curve. See, for example, DiCastelnuovo, Donati et al 2006 which excludes former drinkers.
https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/769554
More recently, see Zhao et al 2023. Figure 2 shows a J-curve relative to never-drinkers in column 2, and a slight J-curve in column 3 (with a set of controls that could risk overcontrolling).
https://jamanetwork.com/journals/jamanetworkopen/article-abstract/2802963
I find it remarkably commonplace for NZ anti-alcohol activists to point to ‘sick quitters’ as some kind of magic totem against the j-curve. This has been looked at in the literature at least since 2006 and the J-curve survives it.
Roerecke and Rehm were very much on point when they said “We sense a desire by some in the field to apply tough standards on protective effects and more lenient standards on other effects.”
https://onlinelibrary.wiley.com/doi/full/10.1111/add.12076
BERL’s method was a disgrace. Even if you like the Collins & Lapsley work on which it was modelled, they made a hash of things relative to C&L. And C&L wasn’t great either.
Making alcohol more expensive seems like a cumbersome means of reducing glass bottle litter.
Wouldn’t it be better to spend money addressing the problem by removing class from the streets?
As a person (no middle-class pretend identity necessary) I often see dog pooh on footpaths. Should we increase the price of dogs, or should we address the actual mess left by irresponsible owners?
Likewise with cyclists and red-light running. Fancy paying an extra $67 per kg of bicycle?
Bernadine: consider things at the margin. The next shot of whisky from a bottle that lasts a year will have approximately no social cost; the next shot of whisky from a bottle that lasts an hour will (expectationally) have much higher social cost. Both shots have the same amount of excise. True that the heavier drinker pays more in excise in total, but the it’s the marginal cost that makes things tougher.
Graeme: The paper had nothing to do with CIS, other than that I wrote a short summary of it for them at some point later.
Me and Matt wrote the 2009 version unfunded because we were annoyed at how BERL’s paper was being used in the Law Commission’s process. We had wanted BERL to tell the Commission that it could not be used that way.
Adrian Slack, the lead author, in comments at TVHE, said that they had no responsibility for how their work was used after it was delivered to the client.
http://www.tvhe.co.nz/2009/04/27/alcohol-regulation-economists-would-do-it-better/#comment-19390
So we figured we needed to go through it. I was then discussant on BERL’s paper at the 2009 NZAE meetings. It was one of the better-attended non-plenaries that year.
We were later approached by an Australian industry consortium that wanted us to go through the Collins & Lapsley work that had formed the basis for the BERL work. That bit was funded, but managed through the University’s Research and Consultancy Office. We had independence. The funding meant we could hire a couple of research assistants who went through a pile of international COI studies to put the C&L / BERL stuff in context. It also resulted in our finding an error in our 2009 work that had led us to shave too much off of BERL’s reported cost; the funded work wound up with a higher reported social cost rather than a lower one.
I submitted a summary of the work to the NZ Med Journal for peer review; they rather unfortunately decided to publish it in their commentary section without putting it through peer review. That was not my choice.
Some reports don’t have a natural home. For academics, tearing up the BERL study is like stealing candy from babies. It is just not worth doing, because nobody academically credible in economics takes the BERL work seriously. Why waste time on it? There is a horrible spot where things look sciency enough to convince the credulous, but are so obviously risible to experts that they aren’t worth looking at. That’s where BERL’s report sits.
I do not pretend that any version of our work on the BERL study presents a comprehensive measure of the social cost of alcohol. Most obviously on the cost side, foetal alcohol syndrome isn’t in it. And if you want to claim the number is a net social cost, which BERL does throughout, then there has to be accounting for benefits. BERL generally assumed those to be zero, so that categories of private cost might be considered social. This is … you just can’t do this. You have to flip to page 173 of their initial report, where they wrote, “We assume that it is irrational to drink alcohol to a harmful level and that harmful alcohol use has zero private benefit. As such, the 50 percent of harmful alcohol consumption estimated in this study has no private benefit to match the private cost, resulting in a net social cost. These private decisions that lead to social costs are included in our estimates.”
This is not how economic analysis is supposed to work. You can’t just deem something you don’t like to be irrational and of no benefit to those who are doing it.
We were NOT claiming that heavy harmful drinkers get overall net benefits from their drinking. We were saying that you can’t simply assume gross benefits to be zero, so that gross costs are net costs.
All we did was take the categories of cost that BERL had tallied, go through how they’d done it, correct obvious mistakes like double-counting [value of lost productivity AND value of lost lives, where the latter includes the former], note errors that we could not correct [health costs], then split costs between those borne by the drinker (like spending on their own alcohol, or their own premature mortality) and those borne by others (like the costs of crime, or the costs drink drivers impose on other people on the roads).
Double-counting: the Ministry of Transport’s manual on this stuff, when BERL did its work, said that the Value of a Statistical Life measure that they use is *inclusive* of all costs of a life lost. For road accident injuries, you need to add lost productivity etc. But not for road accident deaths because the cost of that is already counted in the value of the statistical life lost. BERL counted lost productivity from premature mortality and the value of statistical lives lost.
Health costs: BERL was contracted to follow Collins & Lapsley. C&L on health costs built a giant aetiological table. That table listed a whole pile of different disorders and gave their best guesses as to the proportion of each illness attributable to alcohol. For example, alcohol-induced liver cirrhosis is 100% alcohol-caused. Easy. Right in the name. Bit of a giveaway. But other disorders had negative attributable fractions because in C&L’s best read of the literature, alcohol reduced the burden of those disorders. What did BERL do? They erased every cell in the table where alcohol had a protective effect and put in a zero, with a footnote saying that harmful use by definition can’t have benefits and they were contracted to look at harmful use. That’s putting a giant thumb on the scales to get a bigger number – and one we couldn’t do anything about.
Where does the long term cost of domestic violence as a result of alcohol come into Eric calc?
You do not need to be an addict to become violent or aggressive when under influence of alcohol.
Why do we even discuss this as though alcohol is an essential food.
Perhaps Eric could do well to spend a few days in A&E on a Friday or Sat night and talk to the medics instead of academic research. Restrictions on sale of alcohol, sponsorship and advertising as we do with cigarettes, Should be seriously considered as well as cost.
BERL’s figures included crime costs. They were poorly estimated, but they’re hard to estimate well. We didn’t much revise those figures – they are exactly the kind of thing that should be in a proper accounting of the social costs of crime.
We wrote:
“BERL’s tabulation put loss of life as the largest cost of harmful alcohol use in New
Zealand ($1.52 billion), followed by labour costs ($1.48 billion), drug production costs
($699 million), crime costs ($562 million), health care ($290 million), road crashes ($200
million), and lost quality of life ($42 million). The rank order of these cost items changes
substantially when we consider the external, policy relevant costs: crime is most
important ($410 million) followed by health care ($255 million), labour costs ($174
million), loss of life ($67 million), road crashes not included elsewhere ($38 million) and
lost quality of life ($24 million). An economic cost accounting, which emphasises
external costs, produces a sharply different order of priority to the healthist accounting
which combines private and external costs. If the external costs of reduced productivity
were the primary social cost of alcohol, as implied by BERL’s figures, excise options
could be an appropriate policy solution, perhaps alongside policy changes enabling
employers to make alcohol counseling a condition of continued employment for
employees with alcohol dependence issues; if crime costs are the largest external cost of alcohol use, excise policy is a blunt tool for reducing those costs.”