Sunday, 9 May 2010

Leviathan models of excise?

As I like to be disagreeable, I'll disagree with one part of Luke Malpass's rather nice piece in The Spectator (Aussie edition). I'll pick on the part I disagree with; I like the rest of it (especially the parts citing me, of course).
Last Tuesday, New Zealand’s Law Commission produced a report arguing for the total overhaul and tightening of New Zealand’s liquor legislation. The next day, governments in Australia and New Zealand announced Draconian new taxes on cigarettes: both raising the excise dramatically, and Australia introducing compulsory plain packaging. Goodbye nanny government, hello matron state.

Predictably, the justification for such restrictions and tax hikes comes under the increasingly applied and vacuous category of ‘social cost’, the cost to society of producing and consuming these unhealthy products. This is a mixture of economic calculations designed to arrive at a headline figure of what these ‘dangerous’ goods ‘cost’ society.

Of course, a lot of these claims are nonsense, because what many public health economists and academics do is calculate the cost of an activity and either dismiss or discount the possible benefits arising from it. For example, you are assumed to get no benefit out of having a drink.


The worrying thing about all this is the passive attitude that Australians and New Zealanders take to these recommendations. People think it reasonable that smokers be hounded out of existence and drinkers confined to some awful homemade bars in a garage, simply because they enjoy something unhealthy that, let’s face it, an awful lot of people like to quietly indulge in. It also grates that for all its rhetoric about health, government doesn’t really care about stopping smoking and drinking at all, only raising revenue — otherwise it would raise excise by 600 per cent and cut out both.
I'm of course totally with Luke on the problems with "social costs" measures. But I'm not sure about the last line here quoted. If government cared only about raising revenue, we wouldn't see tobacco and alcohol priced such that consumers remained on the inelastic part of their demand curves: government could increase tax rates and increase total tax revenues.

But a 600% increase would possibly push beyond the Laffer-maximum and, if done all at once, would significantly reduce tax collection by encouraging the quick establishment of a clandestine market: bootlegging. If the goal were to cut consumption over time, and if there's a risk of consumer exit to a clandestine market, then the government could well implement something like a limit pricing model where price increases are always just below consumers' trigger points for switching to the illegal market. If there are economies of scale in bootlegging, making sure that there are always only a few folks that are willing to switch over with any tax increase probably matters.

I'd also expect that making home distillation illegal would be a somewhat important part of a Leviathan alcohol excise tax package; it's legal in New Zealand to distill alcohol for home use. I'd be very curious to know the elasticity of home distillation with respect to alcohol excise tax increases, but I'd be worried that any such estimates would fuel calls for bans. Recall of course as well that legal home distilling makes illicit supply all the easier: who's to say that the few gallons of 90% pure alcohol you've distilled wasn't for home use instead of for sale to the slightly dodgy bar down the road that uses it as house vodka in mixed drinks.

So, I'm not sure that a government that wanted to stamp out all use would go for an immediate 600% increase, and I'm not sure that current tax rates are commensurate with a revenue-maximization model. You could probably build a model where dynamic effects mean you want to price on the inelastic part of the curve if the new customer participation elasticity is greater than the current customer consumption elasticity, but I'm not sure that's the simplest explanation. I'd tend more to reckon that taxes go as high as they can (barring the Laffer max) given the constraint of voter push-back; lots of folks hate smokers but many are drinkers. And, as smoking becomes more and more concentrated among the lower deciles, it'll get easier and easier to play on middle class abhorrence of lower class behaviours to further increase taxes.

If I had to give advice to the healthists, I'd suggest campaigns for differential excise tax based on whether the drink is consumed by boguns and/or teenagers. If the median voter associates the drink with the kind of people they like to look down on, they'll be more likely to go for it. So differential taxes on cask wine, RTDs ... whatever you see folks drinking in South Auckland or in Aranui. Plus minimum price legislation on beer so that nothing is cheaper than Tui and Export Gold, because only the wrong kind of people would ever consider drinking something cheaper than Tui or Export Gold. I'd personally draw the line a bit north of Tui or Export, but my beer tastes are more upmarket than the median voter would go. When I last lectured the calc stream of micro, I used Tui as prime example of an inferior good (you buy less as you earn more).


  1. You don't hear much 'lets control the quantity' arguments much now-a-days. Its all about price controls.

    When the Govt jumped in there with a huge excise tax increase for tabacco, you posted a link to Sir Roger yahooing in an empty Parliament about how if the Govt was serious about reducing cigerette use, it would increase tax by 600% (sound familiar?).

    If a Govt was serious about reducing tabacco or alcohol use, rather than fiddle with prices, why not control the quantity aspect and not the price. Let the price be set by the market, and have slowly reducing quantity targets. That way, the revenue argument associated with tax drop out, and the 'social harm', which attaches to the quantity of the good consumed, has a politically determined path to zero.

    Much cleaner from an enforcement and policy dimension. Quantity controls can be monitored at productive source. If quantity for sale can't be tracked to source then it must have been produced illegally (your vodka argument).

  2. I liked the other parts of his speech, but I disagree with him (and Luke) on that part.

    An excise tax or a total production quota can have equivalent effects if the tax is set to the right level. However, the quota system transforms excise tax revenue into producer surplus: it basically mandates a cartel arrangement. Now, that could be converted back into excise tax revenue by charging firms for the permits in an auction, but it seems a convoluted way of getting to what could be achieved via a per-unit tax.

    Perhaps Seamus will jump in with a "prices versus quantities" Weitzman discussion, but I'm having a hard time seeing off the top why it makes more sense to have a time path to zero quotas as compared to a time path to $5000/cig tax.

    I'm also not sure why enforcement becomes easier. Alcohol tax is levied on the brewer; the brewer labels his product. Under either form of regulation, the bar could refill the empties from the illicit source...

  3. I'm not going to jump in with a prices vs quantities argument. PvQ is relevant when there is more uncertainty about costs than benefits or vice versa, leading to greater variance around the point estimate of the price or the quantity at which MB=MC. If aiming for total elimination of the activity, the price and quantity are known (infinity and zero).

    What I do think is relevnat to the discussion is psychological evidence that a series of small price changes will have less effect on behaviour than a single large change, when looking at addictive substances. So, a government looking to reduce tobacco consumption will bring in one large tax increase now. A government looking to increase revenue, will bring in a series of tax increases over time.

  4. I'd still think PvQ might come in if we want a path to the zero end though, no?

    On the psych lit: I'd agree, except that's what has me thinking about optimally mitigating the rise of a black market: a small price change won't have folks flip to the illicit market where a large change would. Empirical question though whether substitution into the illicit market is a bigger effect than folks just quitting, and what the longer term consequences are if there is a robust illicit market.

  5. Thanks guys :)

    A time path to zero is a direct description by Govt that the thing which is 'bad' will be removed. A time path to a stupidly outrageous selling price sends a message that the quantity of the thing is less relevant than its price: its seems more indirect way of saying "we're going to get rid of the thing". And its that indirect-ness which opens the policy-makers up to the accusation that their motives are tainted - this is a tax making exercise. By quantity targeting, there's a clear link between the policy and the desired outcome - we're going to get rid of this thing.

    I love these debates :) Prices tax and quantity targets aren't the same thing ;)

  6. @James: Suppose that the govt said "we want there to be no tobacco use by 2025". And, suppose that the data on demand elasticity were good enough that they could have a reasonable guess as to what the price would be for any given quota quantity cap, or what the quantity consumed would be for any excise tax level. If you use a path to a very very high excise tax, the government collects a lot of revenue along the way. If you use a path with an ever-diminishing quota cap on total tobacco for sale in the country, you give a lot of cartel rents to tobacco companies along the way which could partially compensate them for being driven out of business. Both things will have equivalent effects: in the year before hitting zero, the price of cigarettes ought to be the same under either regime. Maybe the optics change a bit, but that would be about it.

  7. @james - One is assuming you can control the quantities and that there is no black market potential.

  8. @MikeE: I'd think that would have to be equally true across both tax and quota regimes.