Wednesday 19 September 2012

A hopefully concluding note on the price elasticity of alcohol

A condensed version of the letter I'd sent to the Press appeared in its letters section last week. My shortened letter:
In Monday's Press, Doug Sellman argued that differences in our interpretations of Chris Auld's study on the effects of minimum prices on consumption hinged on that the clause "relative to other drinks" was missing in one journalistic interpretation. This is hardly the case.
The Sellman press release said "A recent Canadian study has shown that a 10% increase in the minimum price of alcohol reduces its consumption by 16% relative to other drinks." As this statement was interposed between sentences noting the effectiveness of minimum prices in reducing consumption of alcoholic beverages, the reader could be forgiven for drawing the conclusion that the study Sellman cites finds that large a reduction in consumption of alcoholic beverages relative to non-alcoholic beverages. However, "other drinks" here did not mean that.
The sixteen percent estimate instead tells us what happens to, for example, beer consumption when the price of beer increases substantially and consumers then instead consume wine. It is highly misleading to present this as the expected effect of an across-the-board increase in prices.
It is especially misleading when the cited study did provide an estimate of the effects of minimum prices: a minimum price that lifts all prices by ten percent would decrease aggregate consumption of alcoholic drinks by only 3.4%. Indeed, Chris Auld, the economist responsible for the empirical analysis in the work cited by Sellman and Connor, writes in comment on my blog, and subsequently confirmed by email: "Eric, for the record, I agree with your interpretation, and I think Sellman and Connor's wording is very misleading." 
Doug Sellman and Jennie Connor's reply in Monday's paper this week:
"Dr Crampton continues his attack on us and does not admit that he made a mistake in quoting from an incorrect secondary source (Sept 11). But he does downgrade the charge from "screamingly wrong" to "very misleading", and if he were to take several more deep breaths he would realize we are essentially in agreement. We have said often over the past three years that there is no magic bullet to change the heavy drinking culture and the harm that results from it.
Although raising the price is probably the most effective and easily enacted measure the Government could put in place, it is not going to achieve the degree of change that is needed on its own.
However, the fact that a 10 per cent rise in minimum price results in a 3.4 per cent decrease in aggregate consumption should not be dismissed as trivial.
A reinforcing set of alcohol reforms, involving marketing, accessibility, purchase age and drink driving, in addition to raising alcohol prices, is likely to be required to bring about substantial change to the out-of-control and damaging state of alcohol use in New Zealand."
I continue to fail to see how there is any relevance in whether one includes "compared to other drinks". If they'd said "compared to other alcoholic drinks", then I'd have been perpetuating a misquoting. Recall that "other drinks" here means other categories of alcohol, not fruit juice. Keeping it in only clarifies things for those readers who knew that Auld et al meant "other alcoholic drinks" but sure doesn't help things for those who didn't know it.

Further, it's Chris Auld that said "very misleading". I'd said, and continue to say, "screamingly wrong". I was just space-constrained in the letters section.

If I had to guess what happened in the Sellman and Connor press release, I'd guess something like the following - I could pretty easily be wrong though and would happily take correction. Sellman and Connor both got mad that John Key said minimum pricing wouldn't do much to curb heavy drinkers' consumption, they ran a Google Scholar search to find some estimate they could use to beat him up with, gave the paper the most cursory skim to find a line they could use, and then ignored the rest of it. They might not have even noticed the 3.4% estimate further down the paper.

But even on this fairly benign story, there's just a shocking underling failure to put the number in context. Again, if you look at Wagenaar's meta-study, you just can't find a single estimate from any of the 100+ papers surveyed that puts alcohol demand as being relatively price elastic (ie absolute elasticity value  > 1.0); it's all degrees of inelasticity. And Sellman and Connor were happy to jump on a number saying alcohol's not just price elastic, it's really really price elastic. You simply cannot have any familiarity with this literature and expect that a price elasticity of -1.6 could possibly be right. It would be like claiming that acceleration due to Earth's gravity is fifty meters per second squared. You can't know anything about physics and think it plausible that acceleration due to gravity on Earth is fifty meters per second squared; you can't know anything about alcohol economics and think it plausible that the price elasticity of demand is -1.6. It's about that level of magnitude of wrong. And it's hardly "dismissing" the effects of gravity to point out that it's only really 9.8 meters per second squared either.

And this pair are the country's go-to experts on the evils of the booze.


  1. In fact the problem is that consumption at normal levels is influenced by price rises in competing sectors, most obviously the price of petrol. But they do not influence the ferocity of the binges, although the frequency of bingeing may alter. The point is that while to the drinker the effects are incremental we the bystanders only notice those we see or hear being objectionable. Once again we have to ask why the lawmakers avoid making our displeasure explicit with new or revised laws relating to behaviour.
    Might I suggest we have a look at the possibility that for too long we have used the presence of alcohol to minimise punishments when we should be making its presence a maximising influence on the sanctions available?

  2. No disagreement. We do it already for dangerous driving where alcohol is a factor.

  3. McGehan says :
    "Might I suggest we have a look at the possibility that for too long we have used the presence of alcohol to minimise punishments when we should be making its presence a maximising influence on the sanctions available? :"

    Get off the price subject, and excuse me but alcohol on the roads is an invited punishment extra , whereas aggression, and stupidity and tiredness which can not be measured but cause most accidents, can not be taken into account .
    100 demerit points McGehan

    Australia saw all this years ago, with big adverts 'tired have a rest now' and limitations on driving distances, and recognition that there drivers are aggressive .
    Why is driving dangerously worse if you can find a way to link alcohol towards it.
    Driving dangerously causes accidents not alcohol. Most Australian truck drivers need more alcohol not less, and they need Prozac as well.
    And I don't need a J curve to prove it