Presented By: Social, Behavioral, and Experimental Economics (SBEE)
Social, Behavioral, and Experimental Economics (SBEE)
Daniele Nosenzo, University of Nottingham
Preferences for Truth-Telling
Abstract:
Private information is at the heart of many economic activities. It is thus crucial to understand how people report their private information in order to predict outcomes and to improve institutions ranging from tax systems to labour contracts. In this paper, the researchers document the stylized facts of this burgeoning literature, formalize the suggested reasons for the observed behaviour and conduct new experiments to disentangle between these reasons.
They start by collating the data from about 50 studies covering decisions of more than 16000 subjects. They find that patterns of reporting behaviour are very stable. There is surprisingly little lying – subjects realize on average only about a third of the earnings they could obtain by lying maximally. Increasing stakes does not change this. In most studies, some non-maximal numbers are reported more often than their truthful probability, suggesting that there are subjects who lie but do not lie maximally.
The researchers then analyse the suggested reasons why subjects do not lie.Departing from the vast majority of previous studies, they study asymmetric true distributions, e.g., a lottery with 40% chance to win a small prize and a 60% chance to win a large prize. Almost all previous experiments used symmetric distributions (most often uniform). With the new experiments, the researchers can reject all models currently suggested in the literature. In particular, none of the models which depart only in one way from a standard neo-classical utility function can explain the data. This includes all versions of social preferences (e.g., Fehr & Schmidt QJE 1999), convex lying costs (e.g. Kartik et al. JET 2007), fixed lying costs (Ellingsen EJ 2004), and most versions of a desire to appear honest, a desire to appear non-greedy or a desire to conform to the endogenous social norm. In ongoing work, the researchers are testing two-departure models, e.g., a model that combines a desire to appear honest with a direct lying cost, to zone in on a model that can explain the data.
Abstract:
Private information is at the heart of many economic activities. It is thus crucial to understand how people report their private information in order to predict outcomes and to improve institutions ranging from tax systems to labour contracts. In this paper, the researchers document the stylized facts of this burgeoning literature, formalize the suggested reasons for the observed behaviour and conduct new experiments to disentangle between these reasons.
They start by collating the data from about 50 studies covering decisions of more than 16000 subjects. They find that patterns of reporting behaviour are very stable. There is surprisingly little lying – subjects realize on average only about a third of the earnings they could obtain by lying maximally. Increasing stakes does not change this. In most studies, some non-maximal numbers are reported more often than their truthful probability, suggesting that there are subjects who lie but do not lie maximally.
The researchers then analyse the suggested reasons why subjects do not lie.Departing from the vast majority of previous studies, they study asymmetric true distributions, e.g., a lottery with 40% chance to win a small prize and a 60% chance to win a large prize. Almost all previous experiments used symmetric distributions (most often uniform). With the new experiments, the researchers can reject all models currently suggested in the literature. In particular, none of the models which depart only in one way from a standard neo-classical utility function can explain the data. This includes all versions of social preferences (e.g., Fehr & Schmidt QJE 1999), convex lying costs (e.g. Kartik et al. JET 2007), fixed lying costs (Ellingsen EJ 2004), and most versions of a desire to appear honest, a desire to appear non-greedy or a desire to conform to the endogenous social norm. In ongoing work, the researchers are testing two-departure models, e.g., a model that combines a desire to appear honest with a direct lying cost, to zone in on a model that can explain the data.
Co-Sponsored By
Explore Similar Events
-
Loading Similar Events...
