Many economists advocate a “congestion tax” because it improves allocative efficiency. People who value the use of a road less than the total costs (including external congestion effects on other drivers) “should” put off their use of the road.

But moving from one regime (ration using time) to another (ration using price) has distributional consequences.  It’s tempting to think we are talking Coase Theorem, but in fact we are talking Kaldor-Hicks-Scitovsky, which is another thing entirely.

Should Manhattan charge a congestion tax, or just charge Uber?