I share much of Becker's skepticism about a "fat tax" (see my article with Tomas J. Philipson, "The Long-Run Growth in Obesity as a Function of Technological Change," Perspectives in Biology and Medicine, Summer 2003 Supplement, p. S87), though I would look favorably on a tax on soft drinks; I would even consider a ban on the sale of soft drinks to children, as I explain later.
The case for a fat tax, as an economist would be inclined to view it, is that a high-calorie diet contributes to obesity, which contributes to bad health, which imposes costs that are borne in part by thin people (thin taxpayers, in particular). I do think, despite skepticism in some circles, that obesity, even mild obesity, has negative health consequences, including diabetes, high blood pressure, joint problems, and certain cancers, and that much of the cost of medical treatment is externalized. But as Philipson and I emphasized in our article and Becker emphasizes too, lack of exercise is also an important factor in obesity. Moreover, the significance of an externality lies in its effect on behavior, and I am dubious that people would consume fewer calories if they had to pay all their own medical costs rather than being able to unload many of those costs on Medicaid, Medicare, or the healthy members of private insurance pools.
Indeed, if as I believe obesity is positively correlated with poverty, reducing transfer payments to people of limited income might result in more obesity. Indeed, high-caloric "junk food" might conceivably though improbably turn out to be the first real-world example of a "Giffen good," a good the demand for which rises when the price rises because the income effect dominates the substitution effect. A heavy tax on high-caloric food might so reduce the disposable income of the poor that they substituted such food for healthful food, since fatty foods tend to be very cheap and satisfying, and often nutritious as well. However, this is unlikely because food constitutes only a small percentage (no more than 20 percent) of even a poor family's budget.
A fat tax would not only be regressive; to the extent it induced the substitution of more healthful foods (as opposed to the Giffen effect), it would as Becker notes reduce the utility (pleasure) of the people who love junk food. This assumes that the junk-food lovers are rational and reasonably well informed, so that they trade off the pleasure gains of eating such food against the health costs. Here I begin to have doubts. I don't think the fact that obesity is correlated with poverty is due entirely to the fact that fatty foods tend to be cheap as well as tasty and satisfying. I suspect that many of the people who become obese as a result of what they eat do not understand how, for example, something as innocuous as a soft drink can produce obesity. I also suspect that producers of soft drinks and other fatty foods are ingenious in setting biological traps--designing foods that trigger intense pleasure reactions caused by brain structures formed in our ancestral environment (the prehistoric environment in which human beings attained approximately their current biological structure), when a taste for fatty foods had significant survival value. (The producers of soft drinks and other junk food also place vending machines in schools, when permitted.) I am doubtful, however, that much can be done about this problem. I do not think, for example, that a campaign of public education would be effective, because it could be neutralized by industry advertising (which, however, would have the indirect effect of a tax--it would increase the food producers' marginal costs) and because the people who most need the education are probably the least able to absorb it.
However, the consumption by children of soft drinks that contain sugar presents a distinct and perhaps soluble social problem. Soft drinks have virtually no nutritional content (unlike foods rich in cream or butter), and recent studies indicate that they are a significant factor in obesity, as well as a source of caffeine dependence and dental problems. They also have good substitutes in the form of drinks sweetened artifically rather than by sugar. And while generally parents know better than government what is good for their children, many parents who permit their children to drink soft drinks do not. Banning the sale of soft drinks to children could not have a Giffen effect and would not be much more costly to enforce than the ban on the sale of cigarettes to children, and might well be a justifiable policy measure.
Now any measure for improving public health has the following limitation: if people are healthier and live longer, this does not necessarily reduce their lifetime expenditures on health care. Most of those expenditures are incurred in the last six months of life, and no matter how long people live, they will eventually enter that terminal phase. However, the longer their healthier lives, the lower their average lifetime health-care expenditures and the greater their productivity, as well as the greater their utility since poor health reduces utility. (Besides its health effects, obesity reduces physical comfort and attractiveness.) I would therefore expect a ban on sale of soft drinks to children to yield a modest net increase in social welfare.