Wednesday 9 January 2013

Tax maximisation, smoking, and the Stalin Gap

If my lifestyle and leisure choices lead me to remit less to the government in taxes than I would have under alternative scenarios, have I imposed an externality on the State?

James Meanwell asks:
Eric, care to weigh in on a debate I'm having about this elsewhere? Specifically, does loss of tax revenue as a result of lower productivity (e.g. due to smoking, eating too much) count as an externality? I've argued that the government would have to be spending efficiently, which is unlikely, for the loss of revenue to count as an externality and so it probably doesn't (count). Not sure if that's right, but the notion (i.e. loss-of-revenue-due-to-lower-productivity-as-externality) seems odd to me.
I started replying to his comment, but reckoned it deserved its own post.

Let's start with the big picture and work down to details.

Consider two people, alike in relevant ways at age 10 and equal in earnings potential, but with different utility functions. So they make different choices.

Mr. A decides that the rat race isn't for him and decides instead to take a part-time job at 20 hours per week instead of 40. Were he to have worked a full time job, his earnings would have doubled and, because of progressive taxation, his tax payments would have more than doubled. But he's not on welfare; he just can transform relatively little income into a fair bit of happiness because he really likes leisure.

Mr. B does not enjoy sitting idle; he consumes his leisure by smoking while working a full-time job. Associated health issues reduce his productivity and, as consequence, he earns a quarter less than he otherwise would; his tax payments are then perhaps a third lower than they otherwise could have been. Other sources have a lower wage penalty for smoking; we'll stick with a big one for present purposes.

If Mr. B's lower tax payments because of his choice to smoke are of policy consequence, then so too are Mr. A's lower tax payments because of his choices. Indeed, we could say that Mr. A is far worse than Mr. B: while Mr. B pays, in New Zealand, about three times as much in tobacco excise as he'll cost the public health system and is so kind as to die before costing the superannuation system very much, Mr A pays zero tax on his leisure and will earn two-thirds of the national average wage in his retirement despite having contributed relatively little to the superannuation system.

If we need to use policy to nudge Mr. B into smoking less, because of reduced tax earnings, then we also need to nudge Mr. A into working harder. And when nudges don't work in tobacco and start becoming shoves, we need to start shoving Mr. A as well.

Mancur Olson argued that Stalinist Russia had the world's most effective extractive tax regime. Workers were effectively compelled to work by near complete inframarginal taxation combined with very low taxation at the margin. If Stalin had had the ability to solve the Socialist Calculation Problem, he could have done slightly better, leaving each worker with an individualised menu of two choices: starvation, or a personalised bundle of very hard work and some goods leaving the worker epsilon better off than under starvation. I suppose that we could consider any deviation from that level of work as a harm imposed on the state. But it's only Stalin who'd really want to push there.

The Okun Gap is the difference between potential and realised GDP due to excess unemployment. I'll call the Stalin Gap the difference between an individual's potential maximal tax payments and his actual tax payments based on his choice to consume leisure over labour - partially because he would choose some leisure even in the absence of taxes, partially because too high of income tax rates yields substitution to leisure.

Most of us could earn more, and consequently submit higher tax payments, by choosing more labour and less leisure. But we'd be less happy.

So that's the big picture: the world in which the smoker's lower income tax payments (ignoring his much much higher excise tax payments) are sufficient basis for taking away his leisure is one where we want to penalize people for taking holidays or failing to work as many hours as they otherwise could: closing the Stalin Gap. Unless we say that leisure via time off relaxing is, by assumption, good, while leisure via smoking is bad - but that remains question-begging.

Smaller picture: is this even an externality?

Short answer: likely, but likely very small, and that small portion mostly because of how the tax system treats poor people.

All right. An individual chooses between smoking, leisure and labour. In a world with private health care and no taxation, the individual optimally balances health harms from smoking against enjoyment of smoking - he earns less because he smokes, but that's just part of the full cost of tobacco that he enters into his optimisation. All you behavioural guys who want to complain about whether he can be rational or informed on this can shut up for the moment - we're trying to figure out whether he's imposing an externality on the government once we move to a tax regime, so the behavioural stuff is entirely beside the point.

Some optimal level of smoking will be found above. What happens when we add taxation and government? The smoker will bear fewer of the wage costs of smoking because the government takes a portion of his earnings and so he should optimally smoke more. But again, the same is true of the individual's choice to consume other forms of leisure in the presence of income taxation. Recall that in economics, we don't care about externalities because of pecuniary effects like "the government gets less in tax". We care about them rather because they distort choices: people move from taxed labour to untaxed leisure in the presence of income taxation. The incremental increase in smoking when the returns to labour drop under a taxation regime can be viewed as a distortion. But, again, think of anything that helps increase the marginal utility of leisure. In a world with taxes and government, people choose more days off playing cheap but excellent video games relative to the zero-tax world. What then is the productivity cost of video games? Or of any other kind of leisure? It's hard for me to see any productivity externality of smoking that is different in kind from any other labour-leisure distortion generated by taxation. And it would take some number crunching to show that the technological portion of any effect* here isn't already over-internalised by current levels of tobacco excise.

Gordon Tullock reminded us in 1998 that government produces externalities as well as solving some externality problems. The vast bulk of reduced income taxation accruing to government due to smoking could only be remedied by imposing harms on smokers that are larger than the potential gains to government. Now, if you want to assume that smokers are irrational and hurting themselves by their choice to smoke, you can do that, but you don't really need the effects on income taxation to make the case.

Recall that smoking is concentrated among poorer cohorts and that marginal tax rates, though not average tax rates, are very high for the poor. If the deadweight costs of high marginal tax rates are making poor people smoke more than they otherwise would because the personal income losses are low, they're probably also screwing up a whole lot of other choices that are of greater consequence as well. If you're going to worry about it, start by trying to fix the tax schedule and abatement rates for the various income-contingent benefits so that the effective marginal tax rates facing poor families are not insanely high.

The Ministry for Social Development noted that 35% of beneficiaries (people receiving benefits other than just Working for Families, NZ's EITC) in paid work in 2008 enjoyed effective marginal tax rates higher than 75%, with some non-beneficiary low income families enduring abatement of the minimum family tax credit facing EMTRs over 100%. MSD concluded dryly, "Work incentives are very low for such families".

Effective marginal tax rates in excess of 75% are very likely to induce all sorts of very real distortions in behaviour; I expect the decision to smoke incrementally more because of reduced returns to wages is pretty trivial in this mess.

* Again, we don't care about externalities that are pecuniary. Imagine that the smoker smokes exactly as much under a tax regime as under a no-tax regime. The government earns less than it would were the smoker a non-smoker, but this is purely pecuniary: the gain to the government by forcing him to quit would be overmatched by the losses experienced by the smoker. The only portion that can matter for welfare is the excess smoking induced by the lower return to labour under the tax regime. And that's unlikely to be large relative to the amount of smoking that's invariant to income tax rates.


  1. Thanks, Eric, that was a great read.

    So if I understand you: the externality comes from the distortion to behaviour (i.e. swapping out labour for various leisures) induced by the government's choice to use distortionary income tax? So the appropriate response to any "oh but behaviour X reduces productivity which decreases tax revenue" is 1) to point out that the problem with that is the change in behaviour, not the loss of revenue and 2) to point out that the concern as well as the real problem can be fixed by reducing the distortions in the tax system?

  2. From an efficiency perspective, the problem is indeed the small amount of increased smoking that might obtain because incentives to avoid wage losses are attenuated by high effective marginal tax rates.

    I think the more rhetorically effective tack is to ask whether taking an extended unpaid holiday, at your own expense, counts as a cost to the state that should be corrected via taxation.

  3. Jim Buchanan has written on this in Externality in Tax Response and in his book Ethics and Economic Progress.

  4. Thanks for that Sinc.

    If I'm reading that correctly, Buchanan also is considering only the externality components of the reduction in labour supply that is consequent to tax distortion: he's assessing things against what people would earn in the absence of taxation, not in the absence of sloth or in the absence of non-tax-related-but-still-income-reducing choices.

    But his last paragraph remains ambiguous on whether he might want to extend it to any choice to fail to live up to his rather rigid work ethic: "The man who sits on his sunlit patio when he could be earning taxable income is levying costs on his fellows. The man who labors and thereby earns taxable income when he could be sitting in the sunlight is providing his fellows with benefits." I'm not sure that he'd advocate mandating lengthy work hours, but from what I'm told about how he ran the economics department at VPI....

  5. Eric,

    Thanks for this, it's extremely interesting.

    The problem is I feel explanations such as this won't affect the debate. While I'm reluctant to attribute bad faith to other people, the fallacy of the "smokers cost us all" argument has been pointed out so often and so clearly by you, Chris Snowdon, and others, that I don't think any reasonably-informed person can still believe it. I am convinced it is simply a means to disguise a paternalistic, illiberal set of aims in the clothing of common-sense prudence and fairness. And the highly counter-intuitive distinction between technical and pecuniary externalities is simply lost on 99% of the populace. This is in no small because while economists (focusing on the positive rather than the normative) tend to be more interested in welfare-maximisation than distributive fairness, what really angers and motivates the public about the idea of smokers/fatties/drinkers sucking up tax dollars is not the allocative inefficiency, but the distributive fairness aspect of it.

    Anyhow, I really liked the way you point out the irony -- the implicit logic of the smoking and drinking studies is that we all have a duty to maximise our contribution to GDP and the tax take. And these studies are uncritically adopted by social democrats and greens who would reject that logic if made explicit and applied more broadly. Try telling them that quitting your job at a hedge fund/sys admin/marketing exec to become an organic orchardist/social worker/anti-GMO activist is cheating the rest of the country, and that raising GDP and tax revenues is more important than their own preferences and welfare.

  6. I think the parallel to vacations is what does it. The ban-vacations thought experiment makes the transfer obvious, and hopefully makes people think about the transfer they'd be mandating on the tobacco side.

  7. You give people way too much credit. Instead of realising that vacations represent a trade-off of leisure for income (which should be a choice left up to the individual), people would counter the "ban vacations" argument by saying "what about all the jobs that would be lost in tourism, holiday rentals, restaurants etc." And then Bastiat would roll further into his grave.