Fans of the classic Australian political satire The Hollowmen will remember the standard trick for inflating a figure when doing political math: roll together several years' expenditures to get a bigger number.
A smallish-sounding $10 million spending announcement, rolled up over 10 years, becomes $100 million. As an official on The Hollowmen warned, "Treasury hate it. Might get you some headlines, but who needs that?"
The past fortnight brought a lot of headline stories about the social cost of alcohol. Felix Desmarais' article for the Dominion Post warned that alcohol costs $5 billion. The Herald's headline warned of a "$70b social cost", citing Alcohol Action New Zealand's rolling up 10 years' worth of alleged costs.
If you thought those numbers referred primarily to the costs of alcohol-related crime, or to alcohol-related health care costs, you shall be forgiven. Social cost figures should refer to some cost borne by people other than drinkers themselves – external costs, as economists tend to put it. But BERL's number really is not that.
Let's go back through the history of this rather rotten statistic.
In 2009, the Ministry of Health and the Accident Compensation Commission commissioned BERL to count the social costs of alcohol. BERL tallied $5.3b per year for alcohol jointly with other drugs (around $1266 per capita), and $4.8b per year for alcohol alone.
This was never going to be a cost-benefit study as BERL was asked to measure only the costs. But some costs do not really get to be counted as net social costs unless you assume there are no offsetting benefits.
BERL assumed that any drinker consuming more than four standard drinks per day, for men, or two standard drinks per day, for women, received absolutely no benefit from their alcohol consumption.
The 500mL bottle of Panhead RatRod IPA sitting on my shelf contains 2.6 standard drinks, so you can get a sense of where that threshold sits – a woman drinking a pint of it would, by BERL's assumption, enjoy only costs, and no benefits, no matter how delicious it is.
Consequently, every dollar spent on alcohol by those drinkers, including every dollar of excise they paid, was tallied as a social cost – some $700m of BERL's headline figure. Buying alcohol is a cost, but it isn't normal economic practice to consider a drinker's expenditures on beer to be a social cost in need of policy remedy.
There were other substantial problems, too.
When considering health costs, BERL explicitly wiped out any conditions where alcohol reduces net health costs. They worked from the alcohol aetiological tables listing the proportions of every disorder that might be attributed to alcohol. For alcohol-related liver cirrhosis, the proportion is obviously 100 per cent. But for other disorders, like chronic heart disease, the proportion is negative because alcohol reduces heart disease.
For any disorder where alcohol provides those protective effects, BERL assumed away those benefits.
When BERL tried to estimate the costs of alcohol-related crime, they used a survey of prisoners who were asked to what extent alcohol had contributed to their offending. If the prisoner said at least "some", the costs of that offending were entirely attributed to alcohol.
BERL's estimates were also plagued by double-counting. If you follow standard guidelines, you cannot count both the intangible costs of premature mortality and the costs of lost productivity due to premature mortality. New Zealand's standard measure of the cost of premature mortality already includes all associated costs, including lost productivity. BERL added up both.
Matt Burgess, Brad Taylor and I thoroughly fisked the numbers at the time, finding that perhaps $967m of BERL's $4.8b figure might make sense as a measure of the social cost of alcohol.
Our critique received a fair bit of media attention and a standing-room-only session at that year's New Zealand Association of Economists conference. BERL updated its figure shortly afterward to stop tallying things like excise paid by drinkers as social cost. Their revised figure dropped from $5.3b to $4.9b – still about $4b too high, and still included over $300m in drinkers' spending on alcohol.
We thought we had put an end to that rather rotten statistic.
Last year, BERL came back to argue that alcohol costs the nation $7.8b per year, or $1635 per capita. The number appeared to be an inflation adjustment to the old 2009 figure; but I thought it worth checking. BERL's Ganesh Nana confirms that the figure is an update of the old $4.9b, inflated by GDP growth since then. Our best hope of reducing alcohol's social cost, by that measure, is a very sharp recession.
The Hollowmen thought they were teaching us how to build a big number. They had nothing on BERL. We should all know better than to take their new number seriously. But, if I had a drink for every time that number showed up, well, I'd be experiencing some serious alcohol-related costs.