Salience theory: Calibration and heterogeneity in probability distortion

Bordalo et al. (2012b) propose salience theory as an alternative to existing descriptive decision theories. However, the authors do not provide a formal calibration of the central local thinking parameter, which measures by how much individuals distort the probabilities of less salient lottery outcomes. Based on an experiment with multiple price lists, we jointly estimate the parameters of salience theory and alternative decision theories as well as the fraction of decisions described by each theory. We obtain three main results: first, 30%–45% of the subjects in our sample behave in line with salience theory, and the local thinking parameter equals about 0.7–0.8, which is roughly consistent with the assumption of Bordalo et al. (2012b). Second, our estimates of the local thinking parameter remain virtually unchanged when non-linear utility is assumed instead of linear utility. Third, our results reveal substantial heterogeneity: the local thinking parameter is significantly smaller when a lottery's downside is more salient than when its upside is most salient.