Tax policy colloquium, week 4: Donald Marron’s "Should We Tax Unhealthy Food and Drinks?"

This week’s colloquium, featuring Donald Marron of the Urban Institute, offered a neck-wrenching change of topic from recent past and future sessions (one of the things I like about our format).  This time around, rather than corporate inversions, the tax treatment of higher education, and public opinion regarding income tax rates, we discussed the question of whether tax instruments should address unhealthy food choices.

The paper (coauthored with Maeve Gearing and John Iselin) is part of a recent flurry of Tax Policy Center publications addressing corrective (often, but not necessarily, Pigouvian) taxes of various kinds.  Other recent entries address, for example, financial transaction taxes, carbon taxes, and how governments might use the revenues from corrective taxes.

As the paper is rooted in recent advances in nutritional science (and what these reveal about continuing uncertainty in the field), I was dismayed to learn of all the now-rejected science that I’ve followed – counseling, for example, avoidance of avocados and eggs (each actually a super-food) based on their fat content.   By reason of all the complexities in how our bodies mediate between inputs and outputs, there are often no straight lines of the sort that were presumed in earlier stages of the science.

For that matter, whom can I sue about the now-refuted claim that teenagers who eat chocolate get way more acne? No empirical relationship whatsoever, and I’ll never get back those lost chocolate-eating years.  But anyway.

The still-provisional nature of nutritional science is only one reason for caution about designing taxes on unhealthy food and drinks.  Even with certainty, one has to assess motivation and, if one wants to do it, issues of design.

Two classic examples of much easier problems pertain to carbon taxes and cigarette taxes.  Carbon taxes, of course, are an externality story.  The great thing about these taxes, from a design standpoint, is that each atom of carbon would have the same marginal effect on global warming as any other atom, if added to the atmosphere at a particular moment.  There are of course problems in measuring the marginal harm, as well as resolving issues of national versus global welfare and (given the immensity of the stakes) arranging multijurisdictional  cooperation, but at least one need not distinguish between, say, “good” carbon atoms and “bad” or neutral ones.

Also much easier than the unhealthy food tax scenario is that of designing cigarette taxes.  Here, despite passive and third-hand smoke, it is probably mainly an internalities story.  Smoking is extremely harmful, and there is evidence that people often want to quit but find this very hard.  Also, despite the Gary Becker-Kevin Murphy “rational addiction” model, there is reason to doubt that smokers should be viewed as having optimized correctly, from their standpoints, via consumption that merely happens to have highly front-loaded benefits.

Marron has elsewhere written about designing taxes to address internalities, and one of the points he makes, in distinguishing them from externalities, could be explained as follows.  Suppose your actions would impose a $5 cost on someone else, and markets plus tort law can’t succeed in making you internalize this cost.  Then a $5 Pigovian tax, designed purely on efficiency grounds, gets your decision metric just right, from the standard of social costs versus benefits.

Now suppose instead that, due to an internality (i.e., a flaw in your capacity to optimize your own choices from the standpoint of your own true self-interest, as you would judge it while being truly reflective), you under-value a cost that a given choice would impose on you by $5.  Assuming one is willing to accept the parallel – which admittedly, raises issues about “true” choice and valuation that are easier to dodge in the externalities setting – the efficient answer is just the same as before.  Even though only you are bearing the costs and benefits of your choice, once again a $5 tax will get things just right.  (BTW, among other problems with this set-up, it works better for consistent undervaluation – e.g., due to hyperbolic discounting – than it does for framing scenarios where we lack a good handle on your “true” preferences or welfare.)

A point that Marron adds is as follows.  Suppose our reason for considering the internalities tax is beneficence, rather than general social efficiency.  That is, we want to make YOU better off.  Then it is noteworthy that the internality tax, even if it improves your marginal decision-making, might leave you worse-off overall.  E.g., suppose you do some of the thing anyway (e.g., because in some instances the true net benefit to you exceeded $5).  Then you may be worse-off overall, despite improving your marginal decision-making, unless we can find a way to return the revenues to you (e.g., via a lump-sum transfer equal to your expected tax costs).

Turning at least to the idea of taxing unhealthy food and drinks, the paper argues that the strongest case offered by current nutritional science for a corrective tax pertains to sugar – and in particular, that which is added to sweetened soft drinks, such as sodas.  Here there may be both an externalities story and an internalities story.  The former mainly pertains to formal and informal health insurance (e.g., emergency room visits as an example of the latter).  For example, diabetes, which is linked to obesity that may result from drinking lots of sweetened soft drinks, may impose large external costs on other consumers and taxpayers.  But the internalities story may be important as well.

BTW, if I may digress, one thing for which I am grateful to my parents is that I never had Coke, Pepsi, etcetera, when growing up.  The first times I tasted such things, I found them disgusting.  I also never had beer until I went to college – the drinking age was 18, in those days – and I also initially found that disgusting.  But while I decided it would be worthwhile to develop a taste for beer, which didn’t take all that bloody long, I also decided NOT to do it for the sodas, which I therefore still find disgusting, and I’m happy to keep it that way.  So I personally have no internalities problem with respect to soft drinks.  (Don’t ask me about extra portions at dinner, however.)  Like Oscar Wilde, I can resist anything except temptation.

Returning to sweetened soft drinks, a big problem that the paper discusses is the lack of a clear dose-response relationship.  Again, for carbon and externalities, it’s the same for all units of carbon.  For cigarettes and internalities, it may be variable, but at least it’s pretty darned high every step of the way.

For sugar – in common with alcohol, another target of corrective taxes that may reflect both externalities and internalities – the variability of the dose-response relationship makes it far harder to figure out how one should implement and design the tax, even assuming good information about average marginal costs (and undervaluation of internal costs).  For some people and at some usage levels, it’s relatively fine; for others and at different usage levels, considerably worse.  But if all possible approaches, including doing nothing, are imperfect, then certainly this instrument belongs on the tastes-good, not-too-filling table of possible revenue options.