The Social Costs of Environmental Goods Provision: A Cautionary Guide to the Revealed-Preference Approach

Empirically estimated behavioral models have emerged as the preferred approach to revealing the social opportunity costs of pollution abatement in many areas of environmental economics. This paper identifies conceptual issues in the implementation of the revealed-preference approach to nonpoint-source pollution and provides methods to overcome them. We focus on the common second-best setting where emissions are not measurable at the source and pollution reduction is incentivized indirectly through payments tied to practice adoption. First, we show through simulation that in discrete choice models estimated on microdata, the use of predicted opportunity costs provides an erroneous estimate of underlying abatement costs. We then focus on two metrics commonly used to represent the marginal social costs of abatement actions, namely, average and marginal program expenditures incurred by the regulating agency. We show theoretically and empirically that these metrics generally fail to reveal underlying social costs.