Rigol and Roth (2016) report results from a lab-in-the-field experiment in India in which they test the viability of two kinds of monetary payment rules used to incentivize truth-telling: a novel payment rule that relies on ex-post verification of reports and peer prediction methods (See Prelec, 2004 and Witkowski and Parkes, 2012), which rely only on contemporaneous peer reports. In the experiment, farmers were asked to give reports about their neighbors and were told that these reports would be used to determine cash prizes. The authors varied whether farmers received incentives for the accuracy of their reports (via the two payment rules) or not. They find that, in the absence of monetary incentives, respondents lie in favor of their family and friends. However, monetary incentives for accuracy improve the quality of reports and both payment rules result in reports of comparable accuracy. This is a reassuring outcome since peer prediction is much easier to implement (though mechanically complex). Importantly, by imposing structure on their data the authors also find evidence that one peer predictive payment rule, the Robust Bayesian Truth Serum of Witkowski and Parkes (2012), is empirically incentive compatible; respondents maximize their subjective expected utility by reporting truthful answers. Given the broad applicability and the ease of implementation of RBTS, the authors hope that this experiment will serve as a catalyst to verify its usefulness in other contexts.
