Thursday 5 July 2012

Minimum pricing

What would the world have to look like for minimum alcohol pricing to be a reasonable policy solution?

Suppose it is the case that harmful heavy drinkers, the sort that impose the greatest harms on others when they consume alcohol, really don't care about the quality of the alcohol they're drinking; they're buying whatever product provides alcohol at the lowest price per standard drink. Suppose further that this cohort's consumption is reasonably responsive to price measures: if you raise the price of the cheapest form of alcohol, you'll do a lot to curb that cohort's consumption while not doing much to reduce the normal consumption of moderate drinkers. Finally, assume that there's little overlap between the kinds of alcohol consumed by harmful drinkers and that consumed by moderate low-income drinkers.

In that kind of a world, minimum pricing on alcohol makes more sense than broader-based alcohol excise tax increases: you can get more reduction in harmful drinking at lower spillover cost to moderate drinkers than by simply increasing excise. A quick Google search suggests lab grade 95% pure ethanol sells for less than $7 per litre if you buy in bulk, or less than $0.10 per standard drink. In this hypothetical world, only harmful drinkers ever go for that product. Typical moderate drinkers instead choose something like, say, Glenmorangie 10 year old single malt. That scotch currently sells for $77 and includes 280 mL of pure alcohol, so just under $14 is excise. With no excise, Glenmorangie would sell for $2.86 per standard drink. If you wanted a $2 minimum price per standard drink, you'd have to charge $1.90 per standard drink to do it via excise to make sure you're charging enough on the lab grade alcohol. But that would load a lot of costs onto the moderate drinkers who then forgo enjoying a harmless bit of scotch before bed.

But are we in that kind of world?

First, in the real world, lower tier product is consumed not just by harmful drinkers looking for the lowest per-unit cost product. A lot of it also is consumed by moderate drinkers of lower income. The more overlap there is between low income harmless drinkers and harmful drinkers in product choice, the less attractive is minimum price in avoiding harming moderate consumers. As David Farrar likes to point out, under Labour's preferred $2 minimum price per standard drink, you could not buy a bottle of wine for less than $16. Most of the wine I purchase runs between $12-$18 per bottle; there are pretty decent wines available in the $8-10 range too, if we watch for specials. If we were in a lower income bracket, we'd be sticking with the $8 bottles. A $2 minimum price would double the cost of our consumption were we sticking with the bottom end of the drinkable price range. And, harmful drinkers also might be choosing the Glenmorangie too.

Second, where both heavy drinkers and moderate drinkers are choosing the same kinds of products, albeit in different quantities, we have to worry a lot about how each kind of consumer responds to changes in prices. The best meta-study on the topic remains Wagenaar, who found that heavy drinkers are roughly 60% as price responsive as moderate drinkers: the price elasticity of demand among heavy drinkers is -0.28 while it's -0.44 for average drinkers. If we doubled the price of lower cost products, which we'd have to do to get to Labour's preferred $2 minimum price per standard drink, moderate drinkers who currently choose that class of product would cut back their consumption by about 44% while heavy drinkers would reduce their consumption by only about 28%. A new paper in Drug and Alcohol Review confirms this kind of finding using data from the Australian National Drug Strategy Household Surveys. From their abstract:
A 1% increase in the price of alcohol was associated with a statistically significant increase of 6.41 days per year on which no alcohol is consumed (P 0.049), and a statistically significant decrease of 7.30 days on which 1–4 standard drinks are consumed (P 0.021). There was no statistically significant change for high or moderate-intensity drinking.
People respond to prices changes, but changes in average consumption levels are less informative than changes in consumption patterns. The study above finds that, when drinkers cut back on their consumption with price increases, they tend to do it by reducing the number of days in which they have small amounts of alcohol rather than the number of instances of heavier drinking. Wagenaar found that heavy drinkers respond less to price changes on average than do moderate drinkers; if that responsiveness comes from the least harmful parts of heavy drinkers' consumption, then the benefits of price increases in terms of harm forgone are overstated. If the J-curve is right, we're then imposing harms on light drinkers while not doing a whole lot to reduce the harms imposed by heavier drinkers.

If moderate and harmful drinkers consume similar products, albeit in different quantities, and if harmful drinkers are less responsive to price increases than are moderate drinkers, then it's harder to build a case for minimum pricing over excise as mechanism for internalising external harms from consumption; it's an even blunter instrument than is excise.

But, I don't think we can say that minimum pricing just transfers money to the brewers and distillers. Unless there are other bottlenecks in the system, we should expect that competition among producers and retailers would lead to products at the bottom end of the market disappearing unless they can be usefully bundled with things that consumers care more about than marginal increases in drink quality: packaging improvements, promotions, or free complementary goods like t-shirts or shot glasses. There will be deadweight costs where moderate drinkers preferring lower-cost product are forced to purchase products more expensive than they'd prefer, but these are attenuated to the extent that retailers or producers are able to get around the regulations by including, say, lotto tickets with lower quality products. If instead restrictions on liquor permits give some retailers local monopoly powers, minimum pricing will yield rents for those retailers that will be capitalised into the value of the firm. I'd walked through this two years ago.

If we are comfortable in imposing disproportionate consumption harms on moderate drinkers of low income and if harmful drinkers disproportionately consume lower priced product, then a minimum price plus excise can get around one of the stickier problems with excise. A linear excise tax that matches the average external harm from consumption necessarily undercharges harmful drinkers and overcharges moderate drinkers relative to the external harm each imposes: harms from heavy drinking rise nonlinearly in consumption while excise rises only linearly. A minimum price combined with a lower alcohol excise rate lets you impose lower costs on moderate consumers of products of higher inherent cost while still doing something to curb harmful drinkers' consumption; the collateral damage comes from consumption reductions among light to moderate drinkers who prefer cheaper product. I'm not comfortable with the collateral damage, especially in the absence of strong evidence that heavy drinkers disproportionately choose bottom cost-tier products. I'd also expect reasonable substitution into home brewing and home distillation if prices got anywhere near $2/standard drink.

What can we do if minimum price is a poor instrument and excise is blunt? Combine excise with measures directly addressing the harms imposed by harmful drinkers. For example:


  1. I just conducted research into this subject. At my local supermarket I asked who buys the "Smartprice" beer which sells for 25p/pint and the answer back was "oldies" - minimum pricing would hit the elderly poor the most.

  2. They're not exactly a cohort likely to go out on drunken rampages. And while heavy drinking by the elderly can worsen all kinds of medical conditions, moderate drinking has strongest beneficial health effects among the elderly, especially reductions in heart disease.

  3. From memory drinking by youth has actually declined, so the angst about youth drinking to excess is that its more public than it used to be, eg Queen St. That argues the authorities should do more to police *locations* rather than all drinkers.


  4. The 2010 Social Report showed no increase in youth problem drinking since the drop in the purchase age.

    Drunk teenagers can be obnoxious. Setting a minimum age for possession at least helps target that external cost.

  5. This post is win, Eric.

    Looking forward to a similarly robust rebuttal from the authors at Red Alert and The Standard.

  6. I'm interested in knowing how much harm (medical/criminal/employment) is caused by alcohol that currently costs more than $2/drink. Not all harmful drinking is by young, elderly, and poor people.
    Which suggests a twist: The law, in its majestic equality, forbids the rich and poor alike to drink Thunderbird.

  7. Where harmful consumption is spread across the range of product quality, that pushes towards excise rather than minimum price as solution. But excise combined with measures targeted more directly at harmful behaviours seems even better. You could imagine that a person convicted for the nth time for offending while drunk could have release be conditional on compliance with a treatment programme with something like Naltrexerone (shame that Disulfiram seems less effective than once thought).

  8. Ah, I hadn't seen that post you linked to right at the end before. I have been asking myself why I havent anything in the media on studies comparing pre "18" and post "18" New Zealand whilst the drinking age debate has been raging.

    I have been saying to myself (and my wife): "But we have already run this experiment once! What are the results??"

    I now have my answer - the results didn't match up with what Geoffrey Palmer saw on his trip to Courtney Place one Saturday morning.