Here is Mankiw:
One argument for specific taxes is that consuming certain products has an adverse impact on bystanders. Economists call these effects negative externalities.
Taxes on gasoline can be justified along these lines. Whenever you go out for a drive, you are to some degree committing an antisocial act. You make the roads more congested, increasing the commuting time of your neighbors. You increase the likelihood that other drivers will end up in accidents. And the gasoline you burn adds to pollution, including the greenhouse gases thought to cause global climate change.
Many economists advocate gasoline taxes so that drivers will internalize these negative externalities. That is, by raising the price of gasoline, a tax would induce consumers to take into account the harm they cause after making their purchases. One prominent study added up all the externalities associated with driving and concluded that the optimal gasoline tax is over $2 a gallon, about five times the current level (combining the federal and a typical state’s levies) and about the tax rate in many European countries.
Applying that logic to other consumer goods, however, is not as straightforward. Consider cigarettes. They are among the economy’s most heavily taxed products, as governments try to discourage people from smoking. Yet the case for such a policy cannot rely on a conventional externality argument.
When a person sits at home and smokes two packs a day, the main adverse impact is on his or her own health. And even if second-hand smoke is a concern, that problem is most naturally addressed within the household, not at the state or federal level.
Sometimes, advocates of “sin” taxes contend that consumers of certain products impose adverse budgetary externalities on the rest of us — that if the consumption induces, say, smoking- or obesity-related illness, it raises health care costs, which we all pay for through higher taxes or insurance premiums.
Yet this argument has a flip side: If consumers of these products die earlier, they will also collect less in pension payments, including Social Security. Economists have run the numbers for smoking and often find that these savings may more than offset the budgetary costs. In other words, smokers have little net financial impact on the rest of us.
It may seem grisly to consider the budgetary savings of an early death as a “benefit” to society. But when analyzing policy, economists are nothing if not cold-blooded. If one uses budgetary costs to justify taxing particular consumption goods, the accounting needs to be honest and complete.
He ends with this thought:
...To what extent should we use the power of the state to protect us from ourselves? If we go down that route, where do we stop?
Taxing soda may encourage better nutrition and benefit our future selves. But so could taxing candy, ice cream and fried foods. Subsidizing broccoli, gym memberships and dental floss comes next. Taxing mindless television shows and subsidizing serious literature cannot be far behind.
David Leonhardt takes issue with the idea that soda taxes are not Pigovian. He argues that the health implications of soda are not like cigarettes, people are not likely to die much younger but are much more to have to live life with diseases with long-term and costly implications:
The link between sugary drinks and obesity is stronger than the link between any other kind of food and obesity, as Kelly Brownell, the Yale obesity researcher says. Calorie intake from these drinks — most of which have no nutritional benefit — has increased almost threefold since the late 1970s. The increase accounts for about half the total per-capita rise in calorie consumption over the same period.
Obesity, in turn, causes a very different pattern of illnesses than tobacco does. Obesity has caused a sharp increase in costly chronic diseases, like diabetes, but is much less likely to cause rapidly fatal diseases, like lung cancer. An article in Health Affairs estimated the annual cost of obesity to be $147 billion and growing. That translates into $1,250 per household, mostly in taxes and insurance premiums.
A soda tax obviously would not solve the obesity epidemic. But it appears to be one of the most promising responses, given the central role that sugary drinks play in the epidemic and the fact that they have no nutritional benefit. A tax would also help reverse the big decline in the price of soda over the last few decades, at the same that the price of fruit and vegetables has been rising. Finally, as with a gasoline tax, a soda tax would help cover the broader costs that the product imposes on taxpayers.
I am of two minds in this: I am weary of too much government involvement in markets to try and correct all market failures and I am enough of a libertarian to believe that people should be able to make their own choices when the costs and benefits are largely private. That is to say that, like Mankiw, I am not a fan of sin taxes. But if childhood obesity really is an epidemic that has serious potential cost consequences then Pigovian taxes are justified. In fact, I think Mankiw would agree with this premise, but he obviously does not agree that soda has a large societal net cost (or is not yet convinced).
Regardless, the recent push for soda taxes is clearly more about balancing budgets than fixing a market failure.