The problem of motivating one party (the agent) to act on behalf of another (the principal) is known as the principal-agent problem, or agency problem for short.
Agency problems arise in a variety of different contexts. For example, a lawyer is meant to act in the best interest of his or her client; managers act on behalf of shareholders; employees work for their employers; politicians represent their voters and so on.
Agency problems arise when the incentives between the agent and the principal are not perfectly aligned and conflicts of interest arise. As a result the agent may be tempted to act in his or her own interest rather the principal’s. Conflicts of interest are almost inevitable. For example, the agent bears the full cost of putting effort into the task delegated by the principal, but usually does not receive the full benefit that results from these efforts. This may create an incentive for the agent to put in less effort into the task than he or she would do if acting on his or her own behalf. Similarly, traders or managers may take on excessive risk if they enjoy the benefits of doing so (a high bonus in case of success), but not the costs (shareholders and lenders losing a lot of money in case of failure). This type of 'moral hazard' is particularly relevant to the banking and insurance industry, and arises because the agent’s actions that lead to the increase in risk are not publicly observable.
Why can the agent get away with not acting in the best interest of the principal? A first possible explanation is that the cost to the principal of removing or punishing the agent is too high relative to the benefit. For example, a politician may get away with corruption during his term in office because in some environments it may be too costly for dispersed voters to undertake actions to remove the politician from office. A second, more widely applicable, explanation is the presence of information asymmetry. Information asymmetry arises when one party (the agent) is better informed than the other (the principal). Information asymmetry makes it difficult or even impossible for principals to know whether the agent acts in their best interest, especially if crucial variables (such as the agent’s effort or competence) are unobservable. For example, if a company reports disappointing earnings figures then it may be difficult for shareholders to judge whether managers are to blame (incompetence or laziness) or whether the poor results are due to adverse factors beyond managers' control (economic recession, bad luck...).
How can agency costs be mitigated? Most mechanisms focus on aligning the incentives between the principal and the agent through carrots and sticks. Some mechanisms are aimed at reducing the degree of information asymmetry.
Examples of aligning the incentives include employee ownership schemes, executive stock options and profit sharing schemes - all carrots - and dismissal, or criminal prosecution of fraud - sticks.
Examples of reducing information assymmetry include the compulsory provision of company accounts; auditing and monitoring; and legal disclosure requirements.
The Enron scandal revealed in 2001 led to the bankruptcy of the Enron Corporation and the dissolution of Arthur Andersen. Many executives at Enron were indicted for a variety of charges and sent to prison. In the wake of the Enron scandal new legislation (such as the Sarbanes-Oxley Act of 2002) was issued to improve investor protection and to increase the accuracy of financial reporting for public companies. 
What is the 'Principal-Agent Problem'
The principal-agent problem occurs when a principal creates an environment in which an agent's incentives don't align with those of the principle. Generally, the onus is on the principal to create incentives for the agent to ensure they act as the principal wants. This includes everything from financial incentives to avoidance of information asymmetry.
Breaking Down 'Principal-Agent Problem'
The principal-agent problem was first written about in the 1970s by theorists from the fields of economics and institutional theory. Michael Jensen of Harvard Business School and William Meckling of the University of Rochester published a paper in 1976 outlining a theory of ownership structure that would be designed in such a way as to avoid what they defined as agency cost and its relationship to the issue of separation and control.
These issues are central to the principal-agent problem. The separation of control occurs when a principal hires an agent, and the costs that the principal incurs while dealing with an agent can be defined as agency costs. These agency costs can come from setting up monetary or moral incentives set up to encourage the agent to act in a particular way.
Principal-Agent Problem Examples
The principal-agent problem is broad enough one that it can be found in a wide variety of contexts.
An example of how the principal-agent problem occurs between ratings agencies and the companies (the principals) that hire them to set a credit rating. Because a low rating will increase the cost of borrowing for the company, it has an incentive to structure its compensation of the rating agency so that the agency gives a higher rating than what may be deserved. The rating agency is less likely to be objective because it fears losing future business by being too strict.
A principal-agent problem could just as easily occur if one person, the principle, asks another, the agent, to buy them some ice-cream without the agent knowing the flavor preferences of the principle. While the two may have discussed the pay-rate for purchase and deliver of the ice cream, the number of scoops, whether in a cup, cone, or waffle cone, and the delivery date and time, preference in taste was left out and the agent can't pick.
A more common example would be when a person (principle) takes in their car to be serviced by a mechanic (agent). That agent knows more than the typical principle, and the agent has the ability to charge at their own discretion.
Principal-Agent Problem and Contract Design
Because so much of the principal-agent problem has to do with information asymmetry and incentives, one of the best ways to protect against an agent acting out of self-interest as opposed to the interest of the principal is to be very intentional about the language of a contract and the types of incentives being laid out.
Principal-Agent Problem and Employee Compensation
A popular view on employment contracts is to connect compensation as closely as possible with performance measurements. Depending on the business, industry, and individuals, this can all vary. According to Peter Doeringer and Michael Piore's 1971 paper "Internal Labor Markets and Manpower Analysis,"the labor market can be divided into "primary" and "secondary" markets, depending on how different workers will be compensated. Those in the primary market tend to be compensated according to skill, while those in the secondary market have their wages determined primarily by market forces.
Tipping as a form of payment can be seen as a way to combat the principle-agent problem. Theoretically tipping aligns the interest of the principle (quality of service) with that of the agent, because presumably that is the metric that the customer uses to determine the tip. However, in the particular case of tipping in restaurants the practice is inexact. The amount one is tipped has been shown not to correlate with quality of service, and can in fact be discriminatory.
Other forms of compensation can help align an employee's interests with that of their employers. For example, performance pay schedules may decrease the quality of an employees work if the employee feels their effort is not being effectively recognized.
Deferred compensation, essentially paying an agent when the task is complete, is another way of designing around the principal-agent problem. Within this system, however, there are still variables: older workers may get paid more and younger ones get paid less due to age discrimination; or quality of performance evaluation might be skewed to the end of the performance period.
Principles of Contract Design
Four principals of contract design as laid out by Milgrom and Roberts in 1992 can help draft better contracts.
- Holmstrom's informativeness principle, introduced in 1979, states that any available measure of performance by the agent should be factored into the level of compensation in the contract.
- The incentive intensity principle argues that the best kind of incentives are created by four factors: profit created, precision, agent's risk, and agent's responsiveness to incentives. This principle posits that the more compensation varies with effort, the better the agent responds to incentives.
- The monitoring intensity principle states that the level of monitoring will correlated with the level of incentives being offered to the agent.
- The equal compensation principle states that compensation for an activity should also match the value a principle puts on the completion of that activity. This helps agents better prioritize the importance of tasks when engaged in multiple projects.
Principal-Agent Problem and Energy Efficiency
Principal-agent problem can also be applied to energy consumption. When trying to catalog barriers to energy efficiency within and without the market, Jaffe and Stavins found a problem that could be defined as a principal-agent one when looking at energy saving appliances. If, for example, the electrical utilities are covered by the landlord, the tenant is less likely to purchase appliances that are energy efficient. This particular example is unique in that it is hard to determine who the principal is and who the agent is in the landlord and tenant relationship. In addition, information asymmetry doesn't play as much of a role in this relationship. Both parties can be aware of the benefits of cost-saving appliances but not use them.