Heterogeneity in loss aversion: evidence from field elicitations

• Published in: Agricultural finance review Vol. 77; no. 1; pp. 196 - 216
• Main Authors: Sproul, Thomas, Michaud, Clayton P
• Format: Journal Article
• Language: English
• Published: Bingley Emerald Publishing Limited 01.01.2017; Emerald Group Publishing Limited
• Subjects: Algorithms; Allocations; Behavior; Computer simulation; Crop insurance; Decision making; Estimates; Expected utility; Field tests; Heterogeneity; Insurance coverage; Parameter estimation; Population; Probabilistic models; Probability; Risk aversion; Standard deviation; Vietnam; Behavioural economics; Heterogeneity; Prospect theory; Expectation maximization; G02; C38; Gaussian mixture model; Loss aversion
• DOI: 10.1108/AFR-05-2016-0045
• ISSN: 0002-1466

Purpose Prospect theory is now widely accepted as the dominant model of choice under risk, but has not been fully incorporated into applied research because of uncertainty about how to include population-level parameter estimates. The purpose of this paper is to characterize heterogeneity across people to lay a foundation for future applied research. Design/methodology/approach The paper uses elicitation data from field experiments in Vietnam to fit a finite Gaussian mixture model using the expectation maximization algorithm. Applied results are simulated for investment allocations under myopic loss aversion. Findings The authors find that about 20 percent of the sample is classified as extremely loss averse, while the rest of the population is only mildly loss averse. This implies a bimodal distribution of loss aversion in the population. Research limitations/implications The data set is only moderately sized: 181 subjects. Future research will be needed to extend these results out of sample, and to other regions. Originality/value This paper provides empirical evidence that heterogeneity matters in prospect theory modeling. It highlights how policy makers might be misled by assuming that average prospect theory parameters are typical within the population.