My UK Telegraph column today is on the likely impact of the country’s sugary drinks tax. The levy, as ever, is being justified on the basis it will internalize the “social costs” of obesity.

There’s a ton of obvious problems. Sugary drinks make up less than 2.5 percent of overall adult calorie consumption in the UK—a drop in the dietary ocean. For young kids, fruit juice is a bigger source of sugar. Obesity, of course, is not simply determined by sugar intake, or even diet either. And it’s not even clear what the social costs of obesity are (above and beyond the private costs—such as lost productivity and worse health), though the UK’s socialized healthcare system complicates matters here.

That all to one side, beneath the line comes a great comment from Tim Hammond highlighting a general under-appreciated problem of using Pigouvian taxes to deal with externalities:

The basic premise of the tax is wrong – externalities should only be taxed if they cause external costs at all levels. Otherwise they should be taxed only at the levels when they start to be damaging. It is obvious (to all except the zealots) that drinking one can of Coke a year is not remotely harmful, either to me or to the NHS (even with the most dodgy claims). So why tax that?

Unit taxes are a blunt instrument to deal with the actual social costs politicians purport to care about.

Think of another example: alcohol taxes are designed in part to ameliorate the social costs associated with alcohol-related crime or driving under the influence. But it might be that some consumers generate these kinds of externalities every time they drink, while others who merely enjoy a glass of wine with their dinner never do. Yet both would face the alcohol taxes designed to mitigate the social problems.

Of course, it may be that taxes are the only practical way of at least trying to account for the externalities. Assessing people for driving under the influence or whether they smoke around young children might be much more costly, both for the government and the regulated. But one can imagine in many instances more targeted ways of dealing with the externalities, such as revoking licenses of drunk drivers.

Tim’s central point though is right: Pigouvian taxes are not clean ways of dealing with external effects of activities. Markets are not perfect in the traditional “perfect competition model” sense. But nor are they “perfectable” by governments.