Note: Eric Crampton at Offsettingbehaviour is an expert on the economics of alcohol consumption in New Zealand, and has posted on this. I purposely wrote this post first (because I have looked at modelling of minimum pricing), then checked out his comments. This post is especially long — sorry.
The Ministry of Justice released a report calculating the impacts of a mandated minimum price for alcohol. The reporting (in both the Christchurch Press and the Dominion Post) was that the report said minimum pricing would be good, lots of people were in favour of the policy, and the Minister blocked it anyway.
Minister Collins was absolutely correct to stop the policy, but not for the reason stated. The reporting was that
The ministry recommended that a minimum pricing regime should not be considered for five years. It said this would give time for Government to assess the impact of alcohol reforms which passed in late 2012.
And you think, fair enough, let’s see what happens with existing measures before we go adding new ones.
Back up a moment — who were these ‘people’ who favoured the policy? Well, all the same people who have been trying to reduce alcohol consumption for years. Specifically, SHORE can’t just let people drink in peace. They feature prominently in the article. It turns out that they supplied a lot of the data and analysis for the MoJ report, too.
SHORE got wound up a while back because alcohol has become ‘more affordable’. What does this really mean? It means (a) people have become richer, and (b) they have decided to spend some of their new riches on drink. This is clear evidence that people like alcohol. Given their druthers, they would like more of it rather than less. Alcohol is a ‘normal good’. Affordability is generally a good thing — think housing.
Back to the MoJ report — what did the report say, anyway? I’ll give you the high points.
First, a minimum price produces nearly the same reduction in moderate drinking as harmful drinking. The report acknowledges that harmful drinking is less price sensitive, but says that the fact that harmful drinkers consume cheaper alcohol means that the minimum policy has more bite with harmful drinkers. Figure 27 makes the link, showing that those who drink more frequently are more likely to be buying cheaper alcohol. The link is there but fairly weak. For those who drink daily, 25% are buying alcohol in the cheapest 20% of products (with no link, it would be 20% of them). So, 75% of daily drinkers are buying more expensive products.
My main issue with this is the blatant classism: if you work drunk on Bombay gin martinis or ruin your liver with Dom Perignon, that’s all good. The report wants to sort those icky people drinking Chateau Cardboard and growlers. While trying to target the fraction of the fraction of the population who BOTH buy cheap booze AND drink harmfully (roughly, 25% of 11% (Figure 1), or 3% of consumers), they are penalising people who want a moderate drink on a budget.
My second issue is the elasticities that drive the whole analysis. Just to be clear, there are two main elements to the modelling. The elasticities (how price affects consumption) are one main element. One set of elasticities, from SHORE and AC Nielsen, are in table 7 (p. 24). They are, shall we say, inconsistent with the literature. The report even points this out:
It was decided that the significant reductions in consumption estimated using NZ elasticity estimates are not a realistic representation of what is likely to happen in reality and are contrary to all international evidence of the responsiveness of alcohol consumers to changes in price.
And yet, the analysis still uses these elasticities. The report also uses the Sheffield elasticities, provided in Appendix 3, which have harmful drinkers as more price elastic than moderate drinkers (own price elasticity). So, a key element of the modelling is suspect.
Thirdly, the other main element of the modelling is how drinking links to harm. I spent a little time trying to dig up reliable data, and it isn’t available. The report tells us that it’s a problem:
For all the harm models there is the possibility that the functional form and slope of the relative risk functions are mis-specified (for example, most functions are assumed to be linear). The savings in alcohol-related harm generated are highly sensitive to the form and specification of the relative risk function. (p. 8)
Translation: we don’t really know, and it makes a big difference, but we’re gonna just go with what we’ve assumed.
Finally, the report finds:
A minimum price or excise increase would have some impact on low risk drinkers, but the savings to society significantly outweigh the lost benefits to consumers. (p. 7)
I’m not sure how they’ve made the calculation, but I’ll explain what I’ve seen. Chapter 9 has a big graphic in which the net social effect is the benefits in reduced harm less the ‘Costs of the pricing policy (deadweight loss + lost value of industry assets)’. This figure is after Chapter 8 runs through impacts on consumer surplus, industry revenue and Government revenue. The figures in Chapter 8 are pretty standard, but figure 13, the long-run impact of a price increase, splits the lost consumer surplus into ‘Lost consumer surplus after pricing policy’, which is a deadweight loss triangle, and ‘Transfer of consumer surplus to government or industry’, which is a rectangle showing the rent from the increased price.
The problem arises because it isn’t immediately clear how that rent rectangle is being treated. According to the Chapter 9 definition, it isn’t a social cost, because it isn’t a deadweight loss and therefore isn’t a cost. Now, this is technically true. If the rent can be captured by government or industry, then it is not a loss to society; it is a transfer from consumers (who may then gain by, for example, more government spending).
However, it is still a loss to consumers. They still, as consumers, have to pay more, and are losing that consumer surplus. Therefore, if you want to draw a conclusion like ‘the savings to society significantly outweigh the lost benefits to consumers’, you need to include that transfer in the ‘lost benefits to consumers’. It is not clear that the analysis does this.
So, a summary. This report is suspect. I haven’t read the whole thing and investigated every calculation, but what I’ve seen suggests that all the assumptions are spelled out and all the details are included, so that if you work at it you can begin to understand how parameters and assumptions were transformed into a crusading conclusion that alcohol must cost more! But it should not be the reader’s job to sort through those details. It is incumbent on the Ministry to provide accurate and impartial analysis. I do not believe, in this case, that they have done so.