Examining Agency Conflict in Horse Racing

Article excerpt

[Author Affiliation]

Alasdair Brown, , , , alasdair.brown@uea.ac.uk

[Acknowledgment]

I would like to thank the Editor, Laura Razzolini, and two anonymous referees for their comments on the article. This work also benefited from suggestions from Fabrizio Adriani, Pasquale Scaramozzino, Sasha Talavera, and audiences at SOAS and UEA. All remaining errors are my own.

1. Introduction

We employ agents to undertake a number of frequent tasks on our behalf. Agents are entrusted to find us a job, manage our investments, and school our children. As the effort and performance of agents is imperfectly observed, this creates ample opportunities for conflict and often welfare losses for the principal. For example, Rutherford, Springer, and Yavas (2005) and Levitt and Syverson (2008) find that estate agents sell their client's houses for less than their own houses after controlling for observable house characteristics. Similarly, Ang, Cole, and Wuh Lin (2000) find that managers who have little stake in a small firm spend more on expenses and run the firm less efficiently than firms with a single owner-manager.

In this article, we examine agency conflict in horse racing. Racehorse trainers are responsible for the welfare of horses in their stable, supervising how the animal is fed, how hard the horse is run prior to a race, and often deciding in which races the horse will run. Many trainers (agents) divide their time between the preparation of their own horses and those of outsiders (principals). While trainers receive all of the win purse if their horse is successful, they receive only a fraction (typically 10%)1 for training an outsider's horse to victory. In addition--and crucially for any study of agency conflict--much of the trainer's effort is unobservable; moreover, the asymmetric information problem is exacerbated by there being an animal at the center of the conflict. To illustrate the opacity of the trainer's actions, Fox (2005) likens the racehorse trainer to a tribal shaman or witch doctor, whose success is ascribed to impressive work and whose failure is due to factors outside of their control.

There are a number of elements to horse racing that make it an interesting arena for the study of agency conflict. First, in contrast to the infrequent sale of a house, horses run on a regular basis over the course of their careers, yielding a large sample of performance data. We use data on every horse race in the United Kingdom between 2005 and 2010. Second, the interaction of horse trainer and owner is dynamic, as the owner can withdraw a horse from a trainer's stable if performance is deemed unsatisfactory; the estate agent is unlikely to fret to the same degree over repeat business from a house seller. Third, there is no need to infer the intrinsic ability of a horse from its observable characteristics; there is a betting market from which to gauge the prerace expected performance of a horse. Finally, the horse owner is relatively empowered compared to the owner of a small firm; whereas removing a firm manager can be problematic and time consuming, the horse can often be removed from a trainer's stable on short notice and without compensation.

Utilizing betting market data to infer the expected performance of a horse, we find that owner-trainer horses outperform when compared with outsider-trainer horses. This suggests that this particular agency conflict is characterized by the agent shirking. Alternatively, it is plausible that trainers--who undoubtedly possess private information on the racing game--are better able to identify the good horses unappreciated by the betting public. To control for this, we examine the performance of older, higher-class horses for which the private information set should be smaller. As in the broader study, owner-trainer horses outperform outsider-trainer horses in this subset of animals.

One way in which an owner can mitigate the shirking effect is to entrust the preparation of a number of horses to the same trainer. …

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.