Agency models have been developed to examine the relationship that exists when one party (the agent) is engaged to act on behalf of another (the principal).1 The principal delegates authority to the agent because of the agent's ability to observe variables that are important in decision making. The existence of asymmetric information provides incentives for agents to use their superior information strategically. This paper extends traditional agency models by including an audit technology. The audit model is used to consider the problem of asymmetric information and to examine the ability of a principal to design incentives to guide the actions of an agent. Auditing is a concept often used in accounting applications, but rarely considered in economic models. By including an audit technology as a method of acquiring information, agency models of decision making can be extended. This paper develops a formal principal-agent model in which a principal is allowed to audit the actions of an agent to acquire information. Auditing is treated as a costly but productive activity, performed by the principal to produce an information signal. The information signal is used by the principal as an input in its decision process. By comparing the costs of auditing to the value of the information produced, the optimal amount of effort to use in auditing can be determined. Consider an outcome (payoff) that is deter mined by the actions of the agent. A random variable also affects the payoff, introducing uncertainty. The payoff is divided between the principal and the agent according to the terms of a prearranged contract. The principal generally does not have the ability to observe directly the actions of the agent nor the outcome of the random variable that affects the payoff. The inclusion of an audit technology allows the principal to acquire information about the actions of the agent and the realized value of the random variable. Thus the principal is given greater latitude in designing and enforcing contracts with the agent. Agency models with signalling have been developed to examine the value of an informa tion signal. Holmstrom and Shavell have shown that as long as an information signal is informative (if it provides some information about the true value of the unknown variable, although possibly imperfect information) it is valuable to include the signal as an argument in the sharing function. The type of signal considered by Holmstrom and Shavell is very simple. No mention is made of the origin of the signal or of its cost. Myerson has developed an incentive compatible model in which perfect information is acquired by the principal by direct payments to the agent. The principal receives a perfect information signal, but the full advan tage of any information asymmetry is captured by the agent.2 By treating auditing as a means of producing an information signal the costs of information are made explicit. These costs can be compared to the value of information to determine the optimal amount of resources to allocate for information acquisition. The optimal auditing solution can then be compared to non-auditing and incentive compatible solutions.
[1]
R. Myerson,et al.
Efficient and Durable Decision Rules with Incomplete Information
,
1983
.
[2]
R. Myerson.
Incentive Compatibility and the Bargaining Problem
,
1979
.
[3]
Jan Stoeckenius,et al.
Auditing: Incentives and Truthful Reporting
,
1979
.
[4]
J. Riley,et al.
The Analytics of Uncertainty and Information- An Expository Survey
,
1979
.
[5]
S. Ross.
The Economic Theory of Agency: The Principal's Problem
,
1973
.
[6]
S. Shavell.
Risk Sharing and Incentives in the Principal and Agent Relationship
,
1979
.
[7]
R. Myerson.
Mechanism Design by an Informed Principal
,
1983
.
[8]
Sanford J. Grossman,et al.
AN ANALYSIS OF THE PRINCIPAL-AGENT PROBLEM
,
1983
.
[9]
William P. Rogerson,et al.
THE FIRST-ORDER APPROACH TO PRINCIPAL-AGENT PROBLEMS
,
1985
.
[10]
Martin L. Weitzman,et al.
Efficient Incentive Contracts
,
1980
.
[11]
Bengt Holmstrom,et al.
Design of incentive schemes and the new Soviet Incentive model
,
1979
.
[12]
Bengt Holmstrom,et al.
Moral Hazard and Observability
,
1979
.
[13]
R. Myerson.
Optimal coordination mechanisms in generalized principal–agent problems
,
1982
.
[14]
M. Harris,et al.
Optimal incentive contracts with imperfect information
,
1979
.