Exclusion in all‐pay auctions: An experimental investigation

• DOI: http://doi.org/10.1111/jems.12243
• Published date: 01 June 2018
• Date: 01 June 2018
• Author: Julia Schmid,Dietmar Fehr

Received: 10 July 2015 Revised: 23 November2017 Accepted: 18 December 2017

DOI: 10.1111/jems.12243

ORIGINAL ARTICLE

Exclusion in all-pay auctions: An experimental investigation

Dietmar Fehr1Julia Schmid2

1Universityof Heidelberg, Alfred-Weber-

Institute forEconomics, Heidelberg,

Germany (Email: dietmar.fehr@awi.uni-

heidelberg.de)

2DIW Berlin, Berlin, Germany

(Email: jschmid@diw.de)

Abstract

Contest designers and managers who wish to maximize the overallrevenue of a contest

are frequently concerned with a trade-off between contest homogeneity and inclusion

of contestants with high valuations. In our experimental study, we find that it is not

profitable to exclude the strongest bidder in order to promote greater homogeneity

among the remaining bidders, even though the theoretical exclusion principle predicts

otherwise. This is because the strongest bidders are willing to give up a substantial

portion of their expected rent in order to minimize the chance of losing the contest.

1INTRODUCTION

Relative performance schemes are considered an important tool for motivating the performance of agents in organizational

settings, sports, and many other domains of our society (e.g., Frank & Cook, 1995). For example, firms often make use of

promotion tournaments and sales competitions, lobbyists seek for influence in the political domain, athletes compete formedals,

and researchers strive for research grants. All these examples have in common that rewards are allocated based on relative

rather than absolute performance, that the effort of the losers is lost, and that the contest designer's main focus is on the overall

performance of the bidders. The closeness of competition, and, by extension, the composition of contestants are critical design

parameters for a contest designer, as a heterogeneous contest may have adverse effects on agents' performance.

In recent years, many sports have seen the presence of dominant athletes, such as Roger Federer and Novak Djokovic on the

Tennis ATP Tour, or Tiger Woods on the Golf PGA Tour. These superstars garner extensive publicity and serve as the face of

their sport. However, excessive dominance by one athlete might also lead audiences to lose interest while encouraging a lower

level of competition. For example, due to Michael Schumacher's dominance in Formula One racing, viewing figures dropped

and, consequently, the FIA changed several of their rules to make the races more closely contested (BBC, 2002).1,2

These examples vividly illustrate the trade-off between the inclusion of superstars and contest homogeneity. Baye, Kovenock,

and de Vries (1993) provide a theoretical foundation of this trade-off and show that under specific assumptions the exclusion

of the strongest bidder can lead to higher revenues for the contest designer (exclusion principle). In contests with one prize, the

presence of a strong bidder may decrease the bids of the weaker bidders, which in turn may also reduce the bid of the strongest

bidder.3This can lead to a lower overall performance. The idea behind the exclusion principle is thus to increase the bids of the

remaining bidders by creating a smaller but more homogeneous contest (without a superstar).

This paper presents an experimental test of the exclusion principle. Specifically, we attempt to answerthe question of whether

a heterogeneous group with one strong bidder or a smaller but more homogeneous group maximizes total revenue fort he contest

designer. Weimplement a repeated all-pay auction with three bidders and complete information about bidders' valuations of the

prize. The valuations in a bidding group are heterogeneous, that is, a group consists of one strong bidder and two weakerbidders.

In order to test the exclusion principle, we randomly vary the participation of the strongest bidder in a bidding group and compare

total revenues when there is no exclusion of the strongest bidder with total revenues in the smaller homogeneous contest where

the strongest bidder is excluded.

We thank a coeditor,two anonymous referees, as well as Dorothea Kuebler, Joerg Oechssler, and Christoph Vanbergfor very detailed and useful comments. We

are grateful for financial support from the Deutsche Forschungsgemeinschaft (DFG) through the SFB 649 “Economic Risk.”

326 © 2018 Wiley Periodicals, Inc. J Econ Manage Strat. 2018;27:326–339.wileyonlinelibrary.com/journal/jems