How Malleable are Risk Preferences and Loss Aversion?

We estimate risk aversion and loss aversion for 550 experimental participants using the Multiple Price List elicitation method. On each screen of the experiment, participants make seven binary choices between a prospect that is held …xed and a sequence of seven (sure-thing) alternative outcomes. In a between-subject design, the set of alternative outcomes is varied by holding the lowest and highest outcomes …xed and changing the …ve intermediate outcomes. These manipulations robustly change measured risk preferences. For prospects in the gain domain, as the intermediate alternative outcomes are decreased –holding the endpoints …xed – the participants’choices become signi…cantly more risk averse, both statistically and economically. We …nd analogous e¤ects for prospects that are in the domain of losses and for prospects that are mixed (with gains and losses). In addition, randomly chosen participants are told the expected value of the prospects. This expected-value information does not a¤ect participants’ choices. Keywords: cumulative prospect theory, framing e¤ects, risk preferences, loss aversion JEL Classi…cation: B49, D03, D14, D83, G11 We thank Helga Fehr-Duda, Philipp Koellinger, Kevin McCabe, and Matthew Rabin for helpful comments. We are grateful to Brendan Price and Alexandra Roulet for outstanding research assistance. We are grateful to National Institute of Aging for …nancial support. J. P. Beauchamp thanks the Trudeau Foundation and the Social Sciences and Humanities Research Council of Canada for …nancial support. E-mail: [email protected], [email protected], [email protected], [email protected].