Spite vs. risk: explaining overbidding in the second-price all-pay auction. A theoretical and experimental investigation (Q2667283)

The authors study bidding behaviour in second-price all-pay auctions with one prize and two risk neutral bidders to attribute overbidding and underbidding relative to risk neutral Bayesian Nash equilibrium (RNBNE) to risk-aversion and spite. The bidders \(k \in \{i,j\}\) have utility function \(u\), private valuations \(v_{k}\), and submit a bid \(b_{k} = \beta_{k}(v_{k})\) where \(\beta_{k}\) is monotone increasing. Risk preferences are described by constant absolute risk aversion \(u(x) =-r\cdot e^{-x/r}\). Spite is modelled as being linear and independent of the valuation: a spiteful loser \(i\) experiences a utility \(u(-b_{i}-\alpha \cdot (v_{j} - b_{i}))\) where \(\alpha \in [0,1]\) describes the amount of spite; in particular \(\alpha = 0\) corresponds to a non-spiteful bidder. By solving for the symmetric equilibrium bidding function, the authors observe that for uniformly distributed valuation, equilibrium bids of \begin{itemize} \item risk neutral -- that is, \(r=\infty\) -- bidders increase in spite for low valuations and decrease in spite for high valuations; \item non-spiteful -- that is, \(\alpha=0\) -- bidders decrease as risk aversion increases; \item risk averse and spiteful bidders deviate more from the equilibrium bids of non-spiteful risk neutral bidders for more spite and less risk aversion. \end{itemize} The authors measured the connection between bidding behaviour, spite, and risk aversion in an experiment where 138 participants played second-price all-pay auctions for 15 rounds with stranger matching. Risk aversion of the participants were measured by a method of \textit{C. A. Holt} and \textit{S. K. Laury} [``Risk aversion and incentive effects'', Am. Econ. Rev. 92, No. 5, 1644--1655 (2002; \url{doi:10.1257/000282802762024700})], while their spite was quantified by taking an average of three standardised spite measures. The authors also controlled for social value orientation, inequality aversion, and rivalry. The authors found that \begin{itemize} \item aggregate behaviour in the second-price all-pay auctions is better described by a theory that allows for spite alone than by a theory that allows for risk aversion alone; \item for individual bids, in line with spiteful preferences, bidders bid more than the RNBNE for low valuations and bid less than the RNBNE for high valuations; \item bids increase in spite for low valuations, bids decrease in spite for high valuations; \item for low valuations, increased risk aversion is associated with lower bids. \end{itemize} The authors conclude that spite is a relevant and important motive in auctions, and it could be as relevant and important as risk aversion in some competitive situations.