It doesn’t matter which side you tax – the cost of the tax will fall on the same people. It depends on the incidence of tax:

http://en.wikipedia.org/wiki/Tax_incidence

The reason we want to charge a tax on consumption is because the externality itself is on consumption – so it is easier and more transparent to implement.

That is something that would have to be sorted out – which is why the SSCg3d index is measured.

“and where would the money go, other than the “fat cats” in government?”

The money should be used to provide the relevant health care services – as a result it should lead to a corresponding reduction in, say, income taxes.

Thats a hot term 🙂

CPW, if you believe that a tax on alcohol is inappropriate, then it makes sense that you may think such a tax on junk food is inappropriate. I agree that the link between junk-weight-health is not clear, however the link from junk-health is a bit clearer, as rauparaha said, the obesity externality on health is only one of the externalities associated with this sort of food.

In the UK researchers classified the types of goods that would be taxed based on some GE modelling and a ‘SSCg3d’ index. Foods that received a score on this index that were sufficiently low could be taxed and it would lead to a net improvement in the economy. Foods that fall into this category are the type I’m talking about when I talk about junk, and so a tax seems as appropriate as a tax on alcohol and tobacco (although I still think petrol taxes are more appropriate 🙂 )

Ultimately, there are two potential issues I could have with this tax:

1) What happens if the science is wrong
2) The compliance and administration costs might be too high.

The answer to 1 is a so what answer, we are making policies that are optimal ex-ante, if they don’t happen to be ex-post that is unfortunate. Lets hope the science is good.

The answer to 2 is that we should do a costing, if the cost of implementing it is better than the efficiency gains we shouldn’t do it. However, that does not mean that there is not merit in the policy.

If we are going to analogize to alcohol, which is similar in terms of the troublemakers being easy to identify in the larger population, I would go the other way and say we probably tax the basic product too much. It would be more efficient to impose larger penalties on the crimes committed under the influence (although perhaps alcohol is slightly different in that we can’t assume rational behaviour to begin with).

I’m certainly not denying that the basic principle of taxing externalities could be applicable to junk food. But my model of society here is basically a small group of people who consume no junk food, a large group who consume in moderation and create no externality, and a small group that consumes in excess leading to health problems. The Junk food tax gets you more efficiency for the final group but you’re losing efficiency by having a less diverse tax base and hence higher tax rates, and creating a dead-weight loss in the group that consumes junk food with no externality. So in practical terms I think this is a bad idea, and I’ll stress again that the links between junk food and weight gain, and weight gain and poor health, are not nearly as strong as commonly believed.

On practicality grounds I’d make another objection – implementing a junk food tax would be difficult, and over-eating is probably the larger problem. You’d really need a blanket tax on food, but that would be awfully regressive.

Again, from a wider view point does it makes sense to try and stick an externality tax on any kind of behavior that leads to public health costs, or just to implement a health insurance system that can discriminate on risk in the first place?

As for the exercise thing: if people don’t over-consume unhealthy foods then we don’t need to worry about whether they’re exercising enough to burn off the extra calories. If they’re not going to exercise otherwise then they’ll have more time to be hard working, productive members of society 😉

“If obesity-related health issues are a genetic (endowment) issue, we should be taxing the whole population to pay for it. If obesity is a self-control issue, we should be charging extra for the health services consumed”

Yes, definitely, that is the exact normative judgment I’m aiming for. However, at the health level you can’t easily distinguish between those who naturally suffer from obesity and those who choose to be obese. By taking consumption of unhealthy foods you are specifically targeting the means that people use to become obese, thereby extracting the tax from the correct group.

“We can make a distinction for cigarettes because consumers of cigarettes have no way of knowing whether cigarettes will prove detrimental to their health (specifically), so we pool the risk of smoking over the smoking population. Whereas consumers of food should have an excellent idea of the likely impact on their health.”

I disagree, I think that people have equal information about how cigarettes and food will negatively impact on their health.

“In this case, I think we be introducing a large distortion for the majority of non-obese food consumers and a less than optimal distortion for a minority of food consumers.”

The tax would slightly increase the cost of these goods and would allow us to offer a slight income tax cut (as the new tax money can be used to fund the obesity related part of health care). As a result, we have to weight up the efficiency losses and gains from this change in tax structure – gains (lower dwl of income tax, increased efficiency for those who eat to the point that it costs society) vs losses (the loss of surplus from those who do not over-eat).

Now if we believe the loss is greater, then surely we also have an issue with other taxes where there is some ‘tipping point’ where the good becomes evil, eg alcohol and tobacco. Those that do not drink much are made worse off by a tax on alcohol, but we don’t term an alcohol tax inefficient because of it. Our problem here is the externality is exponential in an individuals consumption, but our tax is linear.

“I would be careful of quantifying the health externality unless you can apportion that between genetic and “voluntary” obesity. Plus I’ve heard that the links between body-weight and health are not particularly robust in empirical studies, nor is there the link between junk food consumption and obesity iron-clad.”

I completely agree with all those points, they are issues you would have to take into account when forming any sort of scheme. Focusing on the body-weight to health thing, it is true that they are not necessarily strongly related, the relationship probably has more to do with consumption to health. The more rubbish that you consume, the more unhealthy you will be, ceteris paribus. This seems like a reason to introduce the tax.

“Existing taxes on alcohol and tobacco reflect inelasticity of demand and moral issues, so what we do currently isn’t always an appropriate guideline to what we should do.”

Well, the demand for unhealthy food is inelastic, and judging by current government policy it is seen as immoral (damn them!). As an economist all you can ask is whether there is an externality, if so how can we solve it, given that people are heterogeneous and our tools are linear. If I accept that alcohol and cigarette taxes are right, then I can buy the case for unhealthy food taxes (the type Oxford University and WHO support, not a random tax on things the government feels like taxing).