Incorrectly accounting for preference heterogeneity in choice experiments: what are the implications for welfare measurement?

Gains from the incorporation of monetary values for changes in environmental goods and services within cost-benefit analysis depend on how well researchers can estimate these values. One key problem in both stated and revealed preference approaches is how best to model preference heterogeneity. Researchers have implemented several approaches to represent this heterogeneity, and have shown that the choice of approach can have an effect on welfare estimates. However, the question as to the degree of error in welfare measurement from an inappropriate choice of approach has not been addressed. We use Monte Carlo analysis to investigate this issue in the context of choice modelling of coastal water quality changes, when the researcher chooses between a random parameters and latent class model for representing heterogeneity. This allows us to quantify the errors that emerge from using the wrong model in estimating the benefits of water quality improvements. Our overall conclusion is smaller welfare errors are likely to come from use of a latent class model.