On Pigouvian Taxes: A Reply to Scott Sumner

Responding to two blog posts by myself and Kevin Corcoran on skepticism about Pigouvian taxes, Scott Sumner gives an example of a congestion tax in Orange County that has been a huge success.  Scott writes:

This example shows that not all Pigovian taxes are a failure.  Other successes include congestion charges in cities like Singapore, London and Stockholm. 

Scott could have added Fairfax County, Virginia, to his list.  I-66 runs from Washington DC through Fairfax County and out toward West Virginia.  66 was a restricted highway: during rush hour, only people in carpools could use it (and there were huge fines for violators).  Yet, 66 was always backed up.  In 2017, the Department of Transportation removed the HOV restriction and opted to switch to dynamic tolling during rush hour.  The tolls would go up or down depending on the volume with the goal of keeping the average speed on I-66 at 55 (the speed limit).  This has been a huge success.  The implementation of dynamic tolling virtually eliminated traffic jams.  This dynamic tolling is a form of tolling is a Pigouvian tax.

It is a Pigouvian tax and one I celebrate wholeheartedly.  In grad school, I traveled 66 a lot. I wasted God knows how many hours sitting in traffic.  When the dynamic tolling was implemented, I could easily decide if the value of my time was worth the toll.  

In our posts, Kevin and I talk about skepticism.  I am skeptical of market failure corrections, but I am not categorically opposed to them.  There are instances where they can work.  As usual, the devil is in the details.

One of the reasons I am skeptical of Pigouvian taxes is that governments can be slow to adjust the tax (either up or down).  There are numerous political factors that get in the way, special interests start to get involved, and just the political process in general is slow.  Thus, a tax may be too high or too low for too long, resulting in suboptimal outcomes.  One of the great things about congestion taxes, especially like the I-66 toll, is that they can be easily adjusted.  The I-66 toll is entirely automated.  It goes up and down fairly quickly to adjust the level of traffic.  No need for votes, lobbying, or other costly procedures.  The adjustment costs are low.

Furthermore, the negative (or positive) effects of the congestion tax are quite quickly realized.  One can instantly see traffic rise or fall with the level of the tax.  The analytical costs of the tax are low.  Conversely, a carbon tax has effects that are very slow to manifest.  Negative (or positive) effects of carbon emissions can take years to emerge.  The analytical costs of a carbon tax are quite high.

Finally, at least in the I-66 case, the tax is collected via a transponder that many vehicles have (or a bill is mailed according to the registration linked to the license plate of the car if no transponder is present).  The tax is instantly collected; there is no need for measurements or audits.  The administrative costs of the tax are low.

I do support congestion taxes.  Having spent many hours sitting in Boston traffic (as I am sure Scott did, too), I wish my home state would do something similar on those highways.  A congestion tax is able to overcome my skepticism.

The point of my posts on market failure corrections is not that such actions should never, ever be done.  It’s not that they are doomed to failure.  Rather, it’s to remind economists and readers of our most basic lessons: there ain’t no such thing as a free lunch.  Market interventions are not costless.  Yet most economists treat them as if they are.  Your standard econ textbook will give just a glib overview of market failure along the lines of: “Markets work great.  But once transaction costs get high, markets fail and government can/should intervene.”  I am trying to inject the natural skepticism of the economist into these decidedly non-economic lines of thinking.  We have to compare real-world alternatives.  Most market interventionists fail to do so: they just see a market failure and assume government intervention will make it better.  We, as economists and scientists, must be skeptical of all plans.  We must look at them realistically.  We must examine the details, for that is where the devil lies.  That skepticism can be overcome.  But it needs to exist.

Interestingly, this is where public choice analysis comes into play.  Without public choice, the advantages of dynamic or automatic tolling isn’t obvious.  But public choice teaches us to examine politics without romance, that the incentives faced by politicians are not the same as incentives faced by people interacting directly in the market with prices guiding them.  To the standard economist, if all else is held equal, having a tax set by a committee or a tax set by automatic dynamic tolling would be equally preferable.  A public choice economist realizes those two methods are not equally preferable.  One would lead to a much worse outcome than the other.