Does regret matter in public choices for air quality improvement policies? A comparison of regret-based and utility-based discrete choice modelling

Abstract

The choice experiment used to elicit public preferences and willingness to pay is increasingly popular in the studies of air quality improvement policies. However, the traditional random utility maximization (RUM) models underlying the choice experiment approach ignores the important influence of anticipated regret on individual choice behaviour. To explore the importance of regret in public preferences for air quality improvement policies, this study introduces random regret minimization (RRM) models in the analysing of public choices, and further compares the performance of utility-based and regret-based discrete choice modelling based on the multinomial logit and hybrid latent class models. The results suggest that the RRM models outperform the RUM model regarding goodness of fit and prediction, indicating that the RRM models can be used as an alternative and supplement to traditional RUM models. Particularly, 55.9% of the choices from respondents are mainly driven by regret. Our results also reveal that respondents driven by regret-based choice paradigm have a stronger preference for increased clean air days, reduced haze days and lower mortality, whereas utility-based respondents prefer to shorten the years of policy delay. These findings are useful for the design of socially optimal policies on the sustainable management of air pollution and providing theoretical reference for the development of environment-related policies.