A new paradigm for valuing non-market goods using referendum data: Maximum likelihood estimation by censored logistic regression

Abstract

This paper challenges the W. M. Hanemann [Amer. J. Agr. Econom.66, 332–341 (1984)] and C. Sellar, J. P. Chavas, and J. R. Stoll [J. Environ. Econom. Management13, 382–390 (1986)] utilizations of logit models to estimate the value of non-market resources from "referendum" survey data. These data are more informative than conventional choice data. The "random utility" interpretation of logit models is therefore too restrictive. Bypassing the utility function entirely, it will be shown that parameters and standard errors for utility-theoretic inverse Hicksian demand functions can be extracted directly and much more simply. Estimated demand functions need not be limited to those corresponding to the linear-in-parameters utility difference specifications which can be handled by packaged logit programs.