Curb Your Enthusiasm for Pigouvian Taxes

Abstract

Pigouvian (or "corrective") taxes have been proposed or enacted on dozens of products and activities that may be harmful in excess: carbon, gasoline, fat, sugar, guns, cigarettes, alcohol, traffic, zoning, executive pay, and financial transactions, among others. Academics of all political stripes are mystified by the public’s inability to see the merits of using Pigouvian taxes more frequently to address serious social harms. This enthusiasm for Pigouvian taxes should be tempered. A Pigouvian tax is easy to design — as a uniform excise tax — if one assumes that each individual causes the same amount of harm with each incremental increase in activity on the margin. This assumption of uniform marginal social cost pairs well with the limited information and enforcement capacity of tax institutions. But when marginal social cost varies significantly, a Pigouvian tax will not lead to an optimal allocation of economic resources. Focusing on carbon emissions, where the assumption of uniform marginal social cost happens to be reasonable, obscures this common design flaw. Broadly speaking, Pigouvian taxes should be employed only when (1) the harm is (or is properly analogized to) global pollution, and where the harm does not vary based on the source, or (2) the variation in marginal social cost is easily observed and categorized, as with traffic congestion charges. This straightforward insight has broad implications for how we design any targeted tax or subsidy. It explains why a carbon tax would work reasonably well, but some other environmental taxes would not. It explains why most food taxes are doomed to fail. It explains why most sin taxes are designed to raise revenue, not change behavior. It explains why so many tax subsidies are ineffective. And, more constructively, it identifies the limited conditions when Pigouvian taxes or subsidies should be employed.