Which Externalities Should Be Internalized?

Robert Frank's NY Times column on using Pigovian taxes as an "efficient" way to deal with the negative externalities of gas consumption has generated some interesting comments in the blogosphere. The article also points to an important question that has been bugging me for some time. But first, the interesting comments. First up is Gabriel Mihalache who points out an important assumption in Frank's analysis:

...A Pareto improvement [from imposing a Pigovian tax on gas consumption] means that afterwards, everyone is at least as well off (subjectively) as before and some are better off. But how can that be the case when we’re talking about taxing externalities, given that some people’s income is tightly tied with those activities?

The implicit, unstated, assumption of Frank’s article is that we could compensate the losers from the new energy policy from the gains of others. By the logic of what a Pareto improvement means, the gain to some is larger than the loss of others, so there exist potential transfers to compensate the losers and still leave the winners better off.

When supporters of free trade point out that the net losers from the full opening of borders could be compensated, with transfers from the net winners, the common criticism is that those transfers are both politically and institutionally unfeasible. There’s no mechanism we can trust that would identify the correct transfers (from whom, to whom, how much?) and make it in a way that’s politically acceptable.

I will unashamedly yield the same critique against Frank. He wants to seduce us with Pareto improvements but he only tells us half the story, less that half really… he mentions introducing the carbon tax but he remains strangely silent on the ways he’d use to compensate the losers.

Gabriel suggests we avoid resorting to the Pareto efficiency argument and say up front there may be net losers. Josh Hendrickson, meanwhile, also questions the usefulness of invoking Pareto efficiency and goes on to stress that the proper use of a Pigouvian tax requires a Herculean ability to properly assess social costs:

The problem inherent in any such analysis is the view of societal benefit and societal loss that is assumed to be easily calculated and dealt with through Pigouvian taxation. The ability to identify the social cost of a particular action is extremely difficult as each individual has his or her own subjective valuation. The problem is communicating each of these preferences in aggregate form to some central authority. This is a distinct problem in terms of both Hayekian knowledge and a neoclassical framework (Arrow’s Impossibility Theorem). In the absence of this ability, setting the tax rate is extremely difficult.

In short, both of these commentators suggest we should be more humble about our ability to (1) rigorously justify and (2) precisely implement a carbon tax. As noted above, Frank's column also points to another important question that I have been wrestling with for some time: exactly which externalities should be internalized? There are so many negative externalities in society so why stop at those created by gas consumption? Frank alludes to this in his article:

Gasoline is one of a host of goods whose production or consumption generates costs that fall on outsiders. Noisy goods, like leaf blowers, for example, can jolt whole neighborhoods from calm. And goods that don’t biodegrade readily, like many plastic bags, can generate costly waste streams. The list goes on.

Okay, then, why not tax noisy leaf blowers (noise pollution) or billboards along the highway (sight pollution) or rancorous, smelly, ugly people (noise, sight, and smell pollution)? Conversely, should we subsidize quiet neighbors, firms that do not advertise on highway billboards, and beautiful, well-kept people?

Now I am not advocating we tax or subsidize the above items. However, this list does illustrate the fact that society does choose to correct only certain externalities. So what is the decision criteria used in this process? Presumably it involves equating some margins; I am just not sure which one they are though. Any thoughts?

In closing, let me refer you to Peter Klein who, in the context of applying a Pigouvian tax to negative externalities, makes the following statement:

But my main beef with today’s Pigouvians is that they cherry-pick a case here and there — taxes on gasoline, primarily — without fully pursuing the implications of the analysis. If increasing gasoline taxes is efficient, why stop there? What other market failures should the state be empowered to remedy? Here’s my question, specifically:

Please name the activities you believe deserve Pigouvian subsidies. For each activity provide the efficient subsidy amount, explain how this was calculated, and say how the revenues should be raised.

Update: Mark Thoma and others provide answers here.