Inducing risk-neutral preferences: An examination in a controlled market environment

This article reports 15 first-price auction experiments, each with four bidders, designed to test Cedric Smith' (1961) hypothesis that risk-neutral behavior can be induced in subjects' decisions by paying them in lotteries on money that are linear in the outcome probabilities. We choose the first-price auction environment because of its relatively high success in surviving a large number of tests, which contrasts with the widely documented tendency of subjects to violate the expected utility axioms in making choices among gambles. In the first five experiments, subjects were experienced in first-price auctions with monetary rewards. We prescreened these subjects for exceptionally high bidding consistency with the constant relative risk-averse model. The results unyielded only weak support for the risk-neutralizing procedure (3 of 10 risk-averse cases became risk-neutral, but only 1 in 8 that were retested continued to exhibit risk-neutral behavior). We recruited 16 new subjects with no previous experience for four lottery-only auctions. Eight of the 16 subjects bid as if risk-neutral, but in a retest of 12 subjects only 2 remained consistently risk-neutralized. Finally we recruited 12 inexperienced subjects, and each subject bid against 3 robot bidders whose bidding strategies were known to the human bidder. We use this procedure to control for Nash expectations. These 12 subjects were run under both monetary and lottery reward conditions. Two of the 12 subjects bid as if risk-neutral in the lottery auction, but both of these subjects had shown risk-neutral behavior with monetary rewards. In conclusion, we find very weak support for the risk-neutralizing procedure. We caution other researchers to run calibration tests of the procedure in the particular context they are studying to assess its reliability.