The implications of incorrect utility function specification for welfare measurement in choice experiments

Abstract

Despite the vital role of utility functional form in welfare measurement, the implications of working with incorrect utility specifications have not been examined in the choice experiments (CE) literature. This paper addresses the importance of the specification of both non-monetary attributes and the marginal utility of income. Monte Carlo experiments have been conducted wherein different attribute specifications and assumptions for the Cost parameter –that is, different functional forms of utility– have been assumed to generate simulated choices on which Multi-Nomial Logit and Mixed Logit models have been estimated under correct and incorrect assumptions about the true, underlying utility function. The inferred values have been compared with the true ones directly calculated from the true utility specifications. Results show that working with simple experimental designs and continuous-linear specifications makes attribute specification irrelevant for measuring attribute marginal values regardless of the true effects the attribute has on utility.