Decomposing the structural identification of non-market values

This paper addresses the issue of transparency in structural modeling by investigating how the assumptions of a structural model contribute to its identification of non-market values. Assumptions about the form of the utility function, the depiction of preference heterogeneity, and the definition for the choice set are shown to jointly identify a partition of preference space which explains observed behavior. The borders of the partition reflect substitution patterns. The regions in the partition set-identify preferences. Assumptions about the distribution of preferences within each region point-identify partial equilibrium welfare measures. Together, these observations provide a framework for decomposing the structural identification of non-market values. The new framework allows analysts to explain how their modeling assumptions influence estimates for benefits and costs. This addresses federal guidelines for quantifying uncertainty in the modeling assumptions which enter benefit-cost analysis. The framework is demonstrated by decomposing estimates of the MWTP for air quality.