Testing for Reference Dependence: An Application to the Art Market

This paper tests for reference dependence, using data from Impressionist and Contemporary Art auctions. We distinguish reference dependence based on ‘rule of thumb’ learning from reference dependence based on ‘rational’ learning. Furthermore, we distinguish pure reference dependence from effects due to loss aversion. Thus, we use actual market data to test essential characteristics of Kahneman and Tversky’s Prospect Theory. The main methodological innovations of this paper are firstly, that reference dependence can be identified separately from loss aversion. Secondly, we introduce a consistent non-linear estimator to deal with measurement errors problems involved in testing for loss aversion. In this dataset, we find strong reference dependence but no loss aversion.