More fat taxes, and a minor correction

I chatted with Radio New Zealand about fat taxes on Wednesday. I talked about the problems of defining healthy and unhealthy foods, and the lobbying waste that can result around boundary cases if we go the route of taxing specific foods rather than content. And I noted that Denmark dropped its fat tax because of the administrative cost it placed on Danish firms, as well as the cross-border shopping it induced. So the summary here isn't quite accurate. I wouldn't have said that it was abandoned because of the difficulty of defining healthy and unhealthy foods; I gave that as a reason against implementing one here, but that wasn't quite what happened in Denmark.

For more on Denmark's abandoning of fat taxes, see:

• The Economist
• The New York Times

The Times story notes that it's harder to introduce comprehensive taxes like Denmark's than to simply tax soda. It doesn't note that if you only tax soda, people flip to other sugary products. If you're going to do it at all, you probably have to do it comprehensively: excise on sugar or fat following a regime similar to that for alcohol in that it's levied on producers at the point of manufacture. 

A few minor problems that would need to be sorted out:

• Small producers will have problems in paying the tax up front while waiting for payment of product. Not sure there's much that can be done about it other than giving a bit of time to make payment. I know that some of the smaller alcohol producers worry about this when looking at cash-flow.
• Some food products are both input and output - you can either buy sugar or fats as a consumer product, or you can buy them as an input for making other stuff. If any sugar bought as an input turns 100% into something present in the consumer product, then this doesn't matter. You tax it at point of production and the effect flows through the system. But what of food production where the sugar or fat itself disappears but is a necessary intermediate good? If you fry with butter, only some of that butter makes it into the food (compared to, say, just eating a stick of butter). If you turn sugar into alcohol, the sugar's then gone. Any sugar or fat that cooks off or is turned into other things as part of the food production process means that you're going to have to have a tax rebate system for people using sugar or fat as an input.
• You might think this is easily solved by taxing only those who sell directly for final use. But you still need a way of registering and tracking those who buy products for intermediate good use from the same outlets that sell to final consumers. And you'd still have the problem of assessing things properly for home-based production. If we tax based on the average amount consumed, people who eat sticks of butter are relatively under-taxed while those who use it for greasing cast-iron pans are over-taxed. The same problem holds for cooking with alcohol, and is recognized with tax-free salted cooking wines and sherries. Maybe it is a less important problem in cooking - I do not know enough about food chemistry to know. But I do know that some home brewers and distillers use sugar; maybe taxing them on sugar they don't eat is a second best workaround for there being no tax on small-scale home distillation. 
• If you tax sucrose, people will switch to fructose and glucose. You've gotta tax all the sugars to avoid any inefficient switching, but if there's reason to expect that any of the sugars are less bad than the others, you could levy different taxes.
• You'd need to set up a regime for assessing tax on imported food products, with the importer then being liable for any fat/sugar tax. So long as imported products have all the nutritional labelling that would allow accurate and timely excise assessment, great. Delays at the border kinda suck though, especially for fresh fruit that should be subject to any sugar tax. I suppose that we could deem different categories of fruit to contain some average sugar content and tax them appropriately. [I still don't get why folks who want fat and sugar taxes want to subsidise sugar-laden fruits. Fructose doesn't get any less fructosy for being embodied in a strawberry]. 
• There are substitutes for fat that are worse than fat. I expect a reasonable amount of awfulness consequent to any large tax on butter, lard, or other common fats.

More fundamentally, I just do not believe that there is a plausible market failure based case for fat and sugar taxes. 

The best case you can make for it is the fiscal externality where people eat less healthy things and the cost is borne by the public health system, but I don't find it all that compelling an argument. Recall that we impose Pigovean taxes not because we want to punish bad people but rather to get rid of the deadweight cost of some underlying inefficiency. If fat and sugar consumption are not all that elastic, then the bulk of the effect of the public health system is pecuniary rather than technological. That is, if in a world where people paid 100% of their own health care costs, they wouldn't consume that much less sugar, then the public health system does not much screw up our choices over fat and sugar consumption. And the point of a Pigovean tax is to induce people to make the choice that they would have made in a full-cost world.

Still too opaque? Here's an analogy. Suppose that you (and a lot of people like you) love Marmite and Chex. You have to eat them together - whenever you buy Marmite, you have to eat it on a Chex biscuit. And you eat 3 of those cracker-Marmite combos per day, no matter what. Each of them costs $1, with half the cost being Marmite and half the cost being the Chex. But your demand curve is perfectly inelastic: if the cost were $10, you'd still be eating three per day. Now the government comes in and says that, from now on, Marmite is free. You go to the government dispensary every day and get three servings of Marmite to put on your Chex. Then somebody comes out and puts out a "Social Costs of Chex" study that counts as cost to the country everything that the government spends on Marmite because of people who mix Marmite and Chex. You can make a fairness argument for a Chex tax to compensate the government for its Marmite expenditures - in which case you wonder why they made it free in the first place. But you can't make an economic efficiency argument for it - it's just a transfer. And what about all the people who eat their Chex without Marmite and whose consumption of Chex may be more sensitive to price?

More realistically, there's a mix of technological and pecuniary effects. Because people aren't annually confronted with an insurance bill itemising their failures and the things they could do to lower their costs for the subsequent years, people very likely make some different choices around nutrition, exercise and risk than they otherwise would. But how much of the total cost of their health care would have obtained anyway, and how much is due to the distortion in behaviour caused by the existence of the public health system? Suppose that some of our Chex-eaters, above, shift to eating 4 Chex per day instead of 3 after the free-Marmite policy. That last Chex-and-Marmite bit of wonderfulness is inefficiently consumed and could be subject to Pigovean taxation, but everybody seems to think we need to be charging enough to cover all the free-government-Marmite costs of Chex eaters. 

Further, I've yet to see any full life-cycle costing on obesity. You can get big numbers if you add up the costs of obesity-related disorders, but what about savings to the pension system where the obese die early? And what about the costs that the obese would have imposed on the health system in old age were they instead thin people? I suspect that the sign of the effect is still negative, but I'm not massively confident about it. And I doubt that many of those proposing fat and sugar taxes care about the true direction of the sign on the pecuniary effect. I still worry that, someday, we'll figure out that it's the healthy people who live to 100 who wind up costing the government the most money. And then...

The other argument you can make for fat and sugar taxes is based around behavioural economics and internalities. The case there is that while some true version of ourselves would like to constrain our existing selves against eating tasty things that may have long run costs, that true version of ourselves is every day thwarted by the me-of-the-moment who insists on eating too much Marmite-and-Chex.* But that requires, pretty critically, that those individuals, if given the choice, would prefer to be bound against consuming the less healthy products. And even if we can show that for some individuals, that hardly gives us reason to impose population-wide measures that hit those who really prefer to be unbound - you would have to prove that the internality-based gains to those with self-control problems outweigh the losses in surplus imposed on the rest of us. And, you'd have to show that fat and soda taxes are the most efficient means to that end in a world that has Stickk.

Previously on-topic at Offsetting:

• Paternalism: for children and the lower orders.
• Fat taxes and elasticity with respect to public health
• Penn & Teller on Nudges and Fat Taxes
• Paternalism and class
• Fat Freedom
• Obesity Costs

* I can't stop thinking about Marmite and Chex now. I hate you Kiwis for getting me hooked on the stuff just before the earthquake shut down the Marmite factory.