Thursday, 26 June 2014

Tax incidence isn't a subsidy

Suppose alcohol excise doubled. As you walked down the supermarket aisle, you saw that excise tax pass-through rates varied from product to product: low-cost product prices didn't go up quite as much as you'd expected they would.

Does this mean that supermarkets are subsidising lower-priced products? The University of Sheffield / East Anglia alcohol folks think it does.
The findings, published today in the journal Addiction, showed that supermarkets responded to tax increases by subsidising prices of cheaper products. Price rises for cheaper products were up to 15 per cent below the level expected if the tax increase had been passed on fully. 

Although under-shifting affected around one in six of all product lines, these drinks account for a large proportion of total sales: approximately 68 per cent of beer, 38 per cent spirits and 31 per cent of cider sales. 

There is a likely implication on health with previous research showing the heaviest 5 per cent of drinkers in the UK population, classified as higher-risk drinkers according to NHS guidelines, buy 33 per cent of all shop-bought alcohol and favour cheaper supermarket products. Subsidising cheaper alcohol when taxes are increased is likely to lead to smaller reductions in excessive alcohol consumption, and consequently smaller reductions in the harms caused by excessive alcohol than if tax rises were passed on in full.

Paul Dobson, professor of Business Strategy and Public Policy at UEA, said: “Subsidising cheap alcohol might be attractive to supermarkets in their efforts to increase the number and frequency of store visits that shoppers make, but it is socially irresponsible when it encourages excessive consumption. 
Ok, let's go back to the basics on tax incidence again.

First, we rarely expect perfect price pass-through. The burden of any tax increase will be shared between buyers and sellers depending on the relative price elasticity of the two groups. We get perfect pass through where demand is perfectly inelastic (consumption doesn't vary at all with price), or where supply is perfectly elastic. Otherwise, they share the burden. So undershifting just tells us that we don't have perfectly inelastic demand or perfectly elastic supply. If demand is perfectly inelastic, then prices are a dumb policy for trying to curb consumption in the first place. Supply's likely to be pretty elastic, but perfectly?

Now let's make things a bit more complicated.

Suppose that we have two alcoholic product categories in perfectly separated markets. In both cases, supply is pretty elastic. In the first market, customers are moderately price sensitive on the whole, but don't put a lot of effort into price comparison shopping. In the second market, customers are much less price sensitive when it comes to total consumption, but are incredibly price sensitive when it comes to product or outlet selection. So in market A, customers don't flip brands or stores much when prices go up, but they will scale back total purchases. In market B, customers will flip brands or stores immediately for a penny's price difference while not changing their total consumption very much.

We typically say in tax incidence theory that the relatively inelastic side of the market bears the greater part of the tax burden. If customers would flip to other brands or other retailers really quickly in market B, we'd expect that the retailers and producers would bear a greater part of the burden in market B than in market A. It's going to be a bit complicated by that I'd expect total demand at the bottom to be more price inelastic than in the middle ranges, but the main action here should be in sensitivity across brands and retailers if there are some rents going to producers through imperfect competition.

Again, we get differential responsiveness to the excise change without any "subsidy" to lower-cost products.

The most puzzling thing the paper finds is supranormal pass-through on the higher cost products. I expect this is what draws the "subsidy" explanation. For me, it instead upweights something I've heard a lot from smaller brewers but hadn't expected would have huge absolute effects. They've argued that, because excise gets paid at the brewery/distillery at point of production, and because everybody along the way then takes margin on the total price on the product coming out of the plant, you should expect more-than-proportional pass-through. I've not worried about it a lot, because the excise component of an $8 half-litre bottle of something by Three Boys, Yeastie Boys, Emersons, Panhead or otherwise really isn't that huge, so the absolute effects there wouldn't be huge either. But it would show up in these kinds of measures of retail-end pass-through rates.


  1. Eric, you might be interested in some recent changes to the British Columbia liquor rules... one step forwards, two steps backwards.

    It could make an interesting case study though if someone can get their hands on the data. I can see the potential for both interesting health economics and interesting empirical IO papers.

    Some info here

  2. Further evidence (suggesting) the Misuse of Drugs Act is poly drug abuse.