Willingness to Pay: A Welfarist ReassessmentPDF Download
From a welfarist perspective, willingness to pay (WTP) is relevant only as a proxy for individual preferences or utilities. Much of the criticism levied against the WTP criterion can be understood as saying that WTP is a bad proxy for utility, or that WTP contains limited information about preferences. Specifically, critics of WTP claim wealth effects prevent it from serving as a good proxy for utility. I formalize and extend this critique by developing a methodology for quantifying the informational content of WTP.
The informational content of WTP depends on how WTP is measured and applied. First, I distinguish between two types of policies: (i) policies that are not paid for by the individuals they affect and (ii) policies that are paid for by the individuals they affect. Second, I distinguish between two types of WTP measures: (i) individualized WTP and (ii) uniform, average WTP (like the value of a statistical life). When the cost of the policy is not borne by the affected individuals, individualized WTP has low informational content and increases wealth disparity. Uniform, average WTP has higher informational content and reduces wealth disparity, at least in the case of universal benefits. Therefore, when possible, a uniform, average WTP should be preferred in this scenario. When the cost of the policy is borne by the affected individuals, individualized WTP has high informational content but increases wealth disparity. Uniform, average WTP has lower informational content and indeterminate distributional implications. Here, the choice between individualized WTP and uniform, average WTP is more difficult.
I briefly consider two extensions. The first involves time. I present a dynamic extension of the relationship between the informational content of WTP and the wealth distribution. The second extension emphasizes the effect of forward-looking rationality on the WTP measure. The question of rationality raises additional concerns about WTP-based policymaking.