Improving Alternative Designs for Discrete Choice Models in Estimating Willingness-to-Pay for Public Services

Improving Alternative Designs for Discrete Choice Models in Estimating Willingness-to-Pay for Public Services

(Abstract)Discrete choice models have been widely used to estimate non-market benefits of public services such as improved environmental quality or reduced risks of human health as they depend on potential users' stated willingness-to-pay responses in a framework with “real life” choices. However, given that its elicitation method is based on random utility theory, the choice problem is inherently stochastic from the researcher’s viewpoint, which requires a careful design of alternatives to be made to estimate these choice probabilities. This paper highlights the impacts of the two most contentious aspects for discrete choice models, such as the IIA (Irrelevance of Independent Alternatives) assumption and the opt-out alternatives. It provides a thorough review of theoretical and methodological concepts pertaining to the two issues, and then demonstrates the impacts of such alternative design aspects on actual modeling outcomes using a case of water and sanitation services in Indonesia. This effort will help public policy researchers and practitioners to design the alternatives for discrete choice models more effectively and produce unbiased results which inform their strategy for public service planning.