Our goal is to measure individual-level loss aversion. We elicit WTA and WTP for each of our 360 subjects (customers of a car manufacturer). We benchmark the elicited values with a between-subjects design (with further 300 customers) and find that the within-subjects elicitation did not affect valuations. Most customers are loss averse. A further contribution is that we cross-validate our measure with a lottery choice task that also measures loss aversion. We find that the risky and the riskless measures of loss aversion are significantly correlated. We also observe a significant correlation with hypothetical valuations for automobile attributes like gas mileage, comfort, safety, and information systems, where we also find WTA-WTP disparities. |