The effect of choice set misspecification on welfare measures in random utility models

Random utility models have been widely employed in environmental valuation. But stochastic choice set formation models in the random utility framework are rarely applied in this literature, although previous research has shown that ignoring choice set formation (when it exists) leads to biased parameter estimates and welfare measures. This paper conducts Monte Carlo (MC) experiments to investigate the directionality and magnitude of welfare measure bias arising from ignoring or misspecifying choice set formation. We find that when attribute cutoff-based choice set formation exists in the data generation process, typical RUM models ignoring or misspecifying choice set formation underestimate welfare measures by 30-50%. Models that approximate choice set formation may produce unbiased welfare measures, but constitute a promising area for future research. This paper illustrates the importance of applying choice set formation in environmental valuation and provides practical guidance about the usage of stochastic choice set formation models.