How Do Expectations Influence Labour Supply? Evidence from Framed Field Experiments

Models of reference dependence have improved the connection between economic theory and documented labour supply behaviour. In particular, the Kőszegi and Rabin (2006, 2007, 2009) [hereafter "KR"] theory of expectation based reference dependent preferences appears to be a disciplined way to unify the conflicting wage elasticity estimates, and recent laboratory and natural experiments suggest this theory may work in practice as well. I take this theory to the field in a pair of laboratory-like experiments designed to test if expectations determine the effort of a group of impoverished individuals involved in piece-rate work in Northeast Brazil. I use Abeler, Falk, Goette, and Huffman's (2011) experimental mechanism, which is a clear test of KR preferences in effort provision, in two experiments: first to test if rational expectations act as a reference point that influences effort, and second to test if adaptive expectations act as a reference point that influences effort. In both experiments, I find that although people do not behave in accordance with KR preferences, they do not behave as though they make their decisions following canonical lines either. I then outline a speculative rationale for the observed behaviour in these experiments—the adaptive heuristic of regret matching—where workers are able to evaluate their ex-post feelings of regret, even if they do not know the source of those feelings, to optimize behaviour going forward.