Addressing Heterogeneous Preferences Using Parametric Extended Spike Models

Abstract

Some public programs simultaneously provide a mix of non-rejectable public goods and public bads. Consequently, some individuals would pay for the program, while others might instead need to becompensated. In this paper we estimate twoparametric extended spike models that accountfor positive and negative preferences as wellas indifference for the public good. Weillustrate the models using data on valuationof prescribed burning of underbrush in forests,which reduces the risk of catastrophicwildfires but also produces smoke emissions(the public bad). We compare the two empiricalapproaches to estimate willingness to pay (WTP)for the program and contrast these results withthose obtained from modeling specificationsthat only account for non-negative preferences.Substantial differences in public net benefitswere found between the most flexible parametricextended spike model and the simple spike modelwhere negative responses were coded as zero anda standard binary logit of only positivebidders. The results from the extended spikemodels demonstrated that accounting forindifference and negative values towards thepublic good resulted in substantially lowerwillingness to pay estimates.