Is the Endowment Effect a Reference Effect?

This paper is aimed to assess, with two lab experiments, to what extent Kőszegi and Rabin's (2006) model of expectations-based reference-dependent preferences can explain Knetsch's (1989) endowment effect. Departing from past work, we design an experiment that treats the two goods (a mug and a pen) symmetrically in all but in the probabilities with which they are expected to be owned. Thus, our "endowmentless" endowment effect experiment shuts down all alternative mechanisms while leaving expectations the only difference between treatments. We find no evidence that expectations alone can reproduce any of the original effect.