Principal-Agent Theory Research Paper

II. Classic Views of the Principal–Agent Relationship

III. Political Science

IV. Critiques

V. Possible Application to Recent Events

I. Introduction

Imagine the worst happens, and you are falsely accused of a crime in a city in which you are vacationing and in which you do not know anyone. You are arrested and advised of your rights to an attorney and one phone call. Which lawyer do you call? Which lawyer has the best training in the type of issues for which you were arrested? Which one has the best record on these matters? Which one can you afford? Will the one you pick work hard for you? Or will you choose one that will do only a minimal job, which could land you in prison? How can you make the best and most rational decision with this lack of information?

Few people in this situation would have enough information concerning which attorney has the appropriate training or which is the most dedicated to clients. However, any lawyer contacted would know these things about themselves. This asymmetry of (or differences in access to) information constitutes one of the key elements of the principal–agent problem. It is also called the adverse selection problem.1You have to hire someone with only limited information concerning his or her qualifications, training, and achievements. Moreover, potential agents have an incentive to overstate their abilities and experiences in order to obtain the commission.

In addition, you and your lawyer have some interests in this case that are different. You may want your lawyer to dedicate the next several months of his or her life to your case. Your lawyer, by contrast, may want to spend the minimum amount of time possible on your case, get paid, win if possible, and pursue other interests. This highlights another aspect of the principal– agent problem: moral hazard. Agents and principals often have competing self-interests, despite the fact that the agent is hired specifically to represent the interests of the principal. The agent (the lawyer in this case) could put his or her interests ahead of the principal’s (you in this case) by taking payment and not putting forth a strong effort. However, to be fair, unless the lawyer insists on payment up front, he or she runs the risk of doing a great job and then not being paid appropriately afterward.

Finally, your lawyer may recommend a course of action (e.g., plead innocent, take a plea bargain) that you cannot adequately analyze, given your inexperience with the law. You do not know the judges, the law, or past outcomes in similar cases. Lawyers should, and this is why they are hired. However, an incompetent lawyer will recommend a strategy with the same confidence as a well-qualified lawyer. This is also an aspect of the adverse selection problem. You would have to hire an additional lawyer or law firm to determine fully the quality of the strategy being employed. This asymmetry in knowledge and expertise between the people who could enter into a contract lies at the heart of the principal–agent relationship. (Otherwise, you could represent yourself in court and forgo hiring an agent to handle your business.)

Another common situation in which one can see this paradigm at work arises in the personal housing market. Imagine you want to buy a house in a new city and engage a real estate broker. This person has knowledge that you do not have (e.g., the quality of different neighborhoods, the quality of schools, how quickly homes sell, the quality of the contractors, the process of buying a house in that locality), and for this knowledge you hire the broker to become your agent. Typically, agents get some set percentage of the price of the house once it is sold. So when the agent discovers your top price, he or she may tend to show you the houses that are at the top, or slightly over, your price range. Why? The agent has a personal interest in selling you the most expensive house you can afford because he or she will make more money from that sale than from a house that sells for less; 6% of a higher number is more than 6% of a lower number. Your self-interest is in paying the least you possibly can for a house that gives you the most value. The agent may be better off (in a narrow sense) if you buy a house you cannot afford, as long as the deal goes through. If you have to sell your house because you cannot afford it, the agent loses nothing, and he or she might even be able to resell the house and earn money again. However, if the agent is concerned about reputation, he or she may strive to get you into an affordable house so that you pass on good reports to other potential clients.

That Person A would hire Person B to do the bidding of Person A and that their relationship is bound by a contract are well established in history. So too is the idea that the agent has some normative, legal, or moral obligations to do what is in the best interests of the principal. In legal terms, the agent has a fiduciary obligation to the principal. In fact, the principal–agent relationship is a key concept of British common law and in the field of law generally, where it is called agency. This idea of agency is therefore pervasive in society. According to Ross (1973), “All contractual arrangements, as between employer and employee or the state and the governed . . . contain elements of agency” (p. 134).

To understand the contemporary use of this approach in the social sciences, one must turn to political economy (Ordeshook, 1990). The principal–agent relationship, or agency, exists at the intersection of economics, political science, business, law, finance, and sociology. It is used as a heuristic tool to understand economic, social, business, or political relationships using self-interest as a guiding principal. This approach uses some, but not all, of the assumptions of microeconomics: that human interactions are best understood as a meeting of two self-interested and rational actors with relatively fixed preferences trying to maximize their own utility—but both of whom have less than complete information.

In political science, the principal–agent relationship is usually studied by rational choice scholars. The rational choice paradigm uses economic assumptions of human nature to study political outcomes. As such, rational choice scholars begin with assumptions of rationality as well as the maximization of (relatively) fixed goals. These are the strong assumptions of rational choice. For example, the assumption of wealth maximization often translates to power maximization or reelection for political leaders (Levi, 1997). It also includes several weaker assumptions, including no information costs; no transaction costs; no collective or organizational costs; no transportation costs; and no role for history, institutions, or culture. There are simplifying assumptions that are not true, per se, but they are held to be true for the parsimony of the model. However, some authors do not include all of the assumptions (or they lift or “assume away” one assumption or another) and examine the likely outcomes of no longer having all the simplifying assumptions in the model. However, different scholars have examined political interactions and have lifted one assumption or another. Olson (1965), for example, lifted the assumption of collective action costs to show how by reintroducing these costs, one could predict more realistic political outcomes than before.

Therefore, the principal–agent paradigm is used to describe situations in which information costs, which are normally lifted or assumed away in microeconomics, are reintroduced, along with risk sharing or risk shifting. In the same way that physicists assume away friction to describe “ideal physics,” so do economists sometimes assume away these other costs of exchange. However, the principal–agent paradigm retains the strong rational choice assumptions of self-interest on the parts of both the principal and the agent, as well as their relatively fixed preferences. The inclusion of transaction costs results in both adverse selection and moral hazard (Moe, 1984).

Let us explore in more depth the key elements of adverse selection and moral hazard in the principal–agent paradigm. An illustration of adverse selection may be seen when applicants for a position claim greater skills or work habits than they actually have, believing that these personal attributes are difficult for potential employers to know. Prospective employees (agents) have the incentive to exaggerate their abilities (or pad their résumé) to get their foot in the door, and then they quickly learn the skills after being hired. The firm doing the hiring (the principal) wants to hire someone already qualified and who brings these skills along. However, the principal can determine only imperfectly whether a prospective employee actually has these attributes or is exaggerating.

The moral hazard aspect of the principal–agent relationship usually occurs after a person is hired and when the amount or quality of work performed (or output) is difficult to measure or monitor. For example, if a contractor (i.e., agent) is building a house for you in your town, you can stop by the site and witness the various parts of the house being constructed. If, however, you hire an agent to manage a plant in a distant state, you would have to spend more time and resources to visit that factory. Moreover, if the agent knows when you are to arrive, he or she can look busy during the visit and hide low productivity.

In addition, even with supervision, sometimes output is inherently hard to measure. For example, if a teacher has many students who do poorly on performance tests, who should be held accountable? Should one blame the teacher who was unable to motivate otherwise “good” students? Or should one blame the “underachieving” students who did not appreciate the high quality of the education they were receiving? In addition, should pay be linked to student outcomes? Similarly, should a car salesperson with a slow month be blamed and seen as a slacker? Or could the salesperson have been quite diligent, but the market turned against him or her that month?

II. Classic Views of the Principal-Agent Relationship

According to Miller (2005), the canonical or orthodox principal–agent relationship is marked by several key ideas. First, the actions of the agent affect the wealth or well-being of the principal, so the principal is expecting some payoff (reward or punishment) arising from actions of the agent. Second, information asymmetries exist, and the agent has special knowledge or abilities, the lack of which keeps the principal from doing the job himself or herself. In addition, the principal can learn of the efforts of the agent only with difficulty or at a high cost. Third, the principal and agent are assumed to have some preferences that differ. As mentioned above, the agent usually wants the most pay for the least amount of work, and the principal wants the biggest reward (or smallest penalty) with the least payment to the agent. The employee–employer relationship provides a classic example. Another aspect of different preferences is that the agent is usually risk averse whereas the principal is considered risk neutral. To understand this, imagine someone offered two different jobs selling cars, one that pays purely on commission and the other that pays a base rate plus a much smaller commission. Even if the expected income were higher at the commission- only job, the risk-averse agent would likely pick the job with the base pay. Risk-neutral people would maximize their income by taking the commission job because its potential for earning money is greater.

Fourth, the initiative for creating a contract lies with a unified (single) principal. When the principal engages the agent, the principal sets the terms of the contract, to which the agent agrees or not. Fifth, both agent and principal know the basics of the process in which the agent will be engaged. Although a principal may not be able to monitor directly how much work an agent does, the principal is assumed to know the results of the agent’s work, which can be used to infer the efforts of the agent. This is called backwards induction—assuming high efforts on the basis of good outcomes and low efforts on the basis of poor outcomes. Information about the shape of the game and the outcomes of the work of the agent are inexpensive to obtain. Last, the principal can set the specifics of the contract, which include the incentive structure for the agent. The principal, using logic, can determine the best contract to obtain the most wealth from the agent, given the first five elements of the relationship between principal and agent.

Therefore, the principal–agent dilemma exists when a principal wants to hire an agent, but given the self-interest of the agent, as well as the agent’s unknown qualifications, the principal’s choice could lead to poor performance on the agent’s part, thereby harming the principal. The principal therefore tries to create a contract, or payment system, such that the agent acts on the principal’s behalf. In other words, their economic interests are aligned through a contract that shapes incentives.

As a result of the above elements of the principal–agent model, the rule of thumb for the relationship is that where the costs of monitoring the agent are high (and the agent is risk neutral), the principal creates a contract that links payment chiefly to outcomes rather than making trips to monitor the agent or devising complex systems of accountability. For example, in the case of civil law, plaintiff lawyers are often paid with contingency fees and only when the outcome is a successful verdict. If the lawyer (agent) does not win the case, then he or she is not paid. For many in sales, some commission system is usually found, often on top of a small salary, given the risk-averse aspect of the agent. Finally, CEOs are often paid some mix of salary and stock bonuses that align their interests with the price of stock shares; in other words, the agent’s interests become aligned with the owners’ (stockholders) interests in higher stock prices.

Where it is relatively easy or low in cost to monitor the actions of the agent (or where the agent is particularly risk averse), then payment is tied to efforts and not outcomes. In fact, many people are paid by the hour because their employers share the view that time at work results in desired outcomes. For example, a mechanic or carpenter who hires an assistant can easily monitor the efforts and quality of work put forth by the assistant. Therefore, the mechanic or carpenter can pay the assistant by the hour without significant adverse selection and moral hazard effects.

Much of the literature on principal–agent theory examines the types of contracts that are optimal under different levels of risk and transaction costs. The dependent variable is the optimal or second-best contract designed by the principal and imposed on the agent. Many authors explore how transaction costs affect the type and nature of the contract created. Much of this literature uses complex mathematical models and proofs to arrive at conclusions.

A leitmotif running though these ideas is that a contract creates incentives that can align the interests of the agent with those of the principal. Within economics, the principal– agent approach originated in the context of insurance and credit markets and examinations of how principals can extract rents from agents (Stiglitz, 1987). Since it is assumed that the principal has control over the contract, a rent-extracting agent is not usually modeled.

According to Jensen (1983), there are two basic strands of principal–agent theory. The first is a more empirical, or positive, approach, often called the positive theory of agency. This approach attempts to arrive at testable hypotheses derived from the particular assumptions of the model (such as the six described above). This approach is more often seen in political science. The positive theory of agency can also be used as a case-illustrative approach to explain phenomena unexplainable by other paradigms or models. The second strand of this literature is economic in orientation, with formal modeling and advanced mathematics used to make predictions about which contracts would be the most efficient under various conditions. It is often called, tout court, principal–agent. It is highly formalistic and mathematical, and rarely empirical.

However, especially on the agency side of the approach, one understands that contracts made between agents and principals occur in a broader societal context than a one-time negotiation resulting in a contract. Government institutions, or regulations, create background conditions in which negotiations or contracts are constructed, especially through the regulation of socially recognized agents. For example, each state in the United States regulates lawyers and empowers the state bar to ensure that each lawyer has met minimum qualifications. If, indeed, determining the quality of any particular lawyer involved no information costs, then making lawyers pass the bar examination would be needless—one would know ahead of time which lawyers were prepared and which were not. But since information is costly, and the potentially harmful societal effect of poorly trained lawyers would be great, the government regulates this market. With the requirement of passing the bar, the chances that a lawyer will have little or no training are much lower. These sorts of regulations establish a floor, rather than a ceiling, for these interchanges. Similarly, medical doctors have to pass exams to practice medicine. Finally, teachers are certified in nearly every state for employment in the public schools: They must take a minimum number of hours in education courses from a credentialed school of higher learning. Each state’s requiring potential agents to have met minimum qualifications reduces, but does not eliminate, adverse selection problems facing potential principals. Government can also regulate food quality, water quality, and more to ensure the smooth working of markets under conditions of transaction and information costs.

Similarly, but expanding beyond the narrow principal– agent problem, both government and private institutions can be seen as forms of megacontracts, or the guidelines under which other contracts are written. Some argue that institutions arise precisely because of information and transaction costs (which give rise to the principal– agent dilemma). As discussed above, the solution to the principal–agent problem is found in the contract itself. Some suggest that most organizations “serve as a nexus for a set of contracting relationships among individuals” (Jensen & Meckling, 1976, p. 310).

One of the pioneers in transaction costs, Ronald Coase (1937), held that the rise of the contemporary business firm—which dominates the landscape of U.S. business— makes sense only when one accounts for transaction and information costs. The existence of these firms would be illogical if these costs were truly zero. Why have an economic institution dedicated to the production of various elements for some product when one could costlessly contract out all the subcomponents? Given the lack of complete information on quality, performance, punctuality, and price for each subcontractor, adverse selection and moral hazard necessarily arise, and these problems have been solved by the evolution of production within a firm where relations were based on hierarchy and authority instead of only price.

As such, formal hierarchical organizations can be used to improve monitoring to reduce moral hazard and create incentive structures, or contracts, such that agents perform the duties expected of them by the principals. Where monitoring is difficult, the contract can employ “proxy” measures of actual quality instead of quality itself (Moe, 1984). Some examples are hours worked, products produced, and sales completed. However, where these things are difficult or costly to measure, others are hired to monitor the work of employees. Therefore, supervisors in stores who are not stocking shelves or sweeping floors are there to make sure that those employed to do so, do so. Lower-level management in firms is hired to monitor employee performance. And, in turn, mid-level management monitors lower-level managers.

Similarly, Williamson (1985) examined economic institutions to see how they form economic incentives that shape human behavior. In fact, Williamson suggests that the principal–agent relationship is a subcomponent—along with property rights (e.g., North, 1981)—of the literature on transaction costs, which looks at incentive structures to explain outcomes. The other branch of the literature on transaction costs examines institutions that rely on nonmarket governance and measurement to have economic outcomes. Williamson claims that traditional microeconomic theory falsely views all deviations from “pure market behavior” to be caused by monopoly. Ironically, economic institutions reduce the very transaction costs (prices and premiums for risk, uncertainty, bargaining, and obtaining information) that are assumed away in traditional microeconomic theory. Jensen and Meckling (1976) suggest that organizations themselves are actually multilateral contracts between and among many principals and agents—thus bridging the principal–agent literature with neoinstitutional ideas. Therefore, the incentives found in contracts are also found in institutions.

Although not all principal–agent relations examine institutions and their roles, one can see how, empirically, society has helped to overcome information costs and make markets operate “as if” information costs were low or nonexistent. Institutions can also play a role in shaping incentives between principals and agents, which act as a form of contract. Therefore, contracts can be seen as more concise versions of institutions, and institutions are longer established rules and norms that create incentives—which in turn shape the interchange between principal and agent.

In addition, private companies can help reduce uncertainty and thereby lower information and transaction costs for consumers if doing so helps private companies move product. Imagine that a new car company with no record of accomplishment wants to enter the market. In order to reduce the risk to the customer, the car company can give a warranty that the car will perform well for X number of miles, or X number of years. So even if the car proves to have many mechanical problems, the customer is somewhat protected from the lack of information before the purchase of the car. This should lower customers’ risk of purchasing a new product that does not yet have a record of accomplishment (and thus less information about quality).

Firms also realize that potential customers incur information costs in making decisions. For example, firms that win awards for quality or safety will highlight this in their advertisements as a form of external verification for otherwise ubiquitous claims of quality and performance. Knowledge of the awards reduces the risk, or uncertainty, the client faces, because uncertainty results from gaps in information. Some information, such as the life span of any particular car, is unknown to both seller and buyer, although the car salesperson has a better probabilistic understanding of how well the different models of cars he or she sells will perform.

III. Political Science

In early political science, Max Weber identified a construct similar to the principal–agent perspective when he discussed the state and sources of authority. He argued that there are three key constituencies in a state: (1) the power holders, (2) their servants (the bureaucracy), and (3) the population (Lane, 2008; Weber, 1978). One can apply the principal–agent relationship to the state and think of the population as the principal and the power holders as the agents (Lane, 2008). Alternatively, one could think of the power holders as the principals who have to monitor the bureaucracy, who are the agents. An example would be congressional oversight of bureaucracies (Weingast, 1984).

However, some might argue that thinking of the population of a country as the principals and officeholders as the agents could be problematic for several reasons. First, do all the principals have a unified interest that the agents can understand? In fact, no matter what politicians do, one could argue that they are responding to some societal demands while ignoring others. Therefore, the lack of a clear and unified principal could make the application of this paradigm to politics more difficult. Moe (1984) discusses the difficulty of analysis with multiple principals.

One possible exception to this problem may be the case of corruption. Several scholars have used the principal– agent model to describe corruption (Alam, 1989; Klitgaard, 1988; Quinn, 2008). Since corruption can be difficult to define, Alam (1989) argued that corruption can be best understood as a function of all principal–agent relationships and as such may be defined as “(1) the sacrifice of the principal’s interest for the agent’s, or (2) the violation of norms defining the agent’s behavior” (p. 442). Here, the principals are the people, and the government is the agent. The normal assumptions are that the population would have a united interest in the most public goods at the lowest costs, and corruption would eat away at either the quality or the quantity of the service rendered. Quinn (2008) suggests that when the political elite becomes the agent for the people (principals) through majority state ownership of most capital-intensive industries or the largest export sector, then the principal–agent problem predicts a rise in corruption. This would be especially true within these economic sectors—to the point that the potentially most productive sectors of the economy can become a drain on wealth instead of an engine of growth. The extreme form of betrayal of public interests by the ruling class (agents) could be seen as predatory rule (Levi, 1988).

Some political scientists studying Africa have argued that politicians could be acting “rationally” when they are following “irrational” economic policies. To understand this point of view, one must suspend the assumptions of both no collective action costs and no transaction costs. Since the agents (the politicians) are maximizing their political power (i.e., incumbency and power) and responding to the part of the society that is best organized (e.g., other elites, urban dwellers, the military), they could establish economic policies that provide a return only or primarily to powerful segments while claiming to pursue the public interest. These powerful actors will be parts of society that can overcome their collective action costs. Thus economic policies that enrich a small segment of politically powerful people are enacted, even though they lead to economic performance that undermines the interests of most of the principals (the majority of citizens; see, e.g., Bates, 1981). (Although few such analyses are specifically pitched in principal–agent terms, they can be understood as such.) Again, some have argued that this betrayal of the principals by the agents is greater with majority state ownership of most economic assets (Quinn, 2002, 2008) because the normal economic interests that would lobby government for better policy are no longer separated from government itself.

Weingast (1984) applied this theory to bureaucracies, in which case the agents are the bureaucrats, and the principal is Congress (acting on behalf of the ultimate principals, the voters). Traditional analysis of how tightly Congress monitored bureaucracies found that the monitoring was very loose (Moe, 1984). However, consistent with the literature on principal–agent relations, when costs for monitoring the behavior of agents are high, then agents are held to account through outcomes. Since what members of Congress care about is reelection, if they are regularly reelected, then the bureaucracies are monitored by the desired outcome. Since members of Congress are reelected at very high levels, the need for more monitoring is low—this conclusion is quite consistent with the literature.2 Therefore, if benefits from these bureaucracies flow to constituents’ needs sufficiently to allow reelection of the members of Congress, then the bureaucracies are not as autonomous from Congress as the loose monitoring implies.

Levi (1988) uses transaction costs, principal–agent relations, and discount rates to predict the types of tax systems different countries will use. She uses the assumption of revenue maximization on behalf of rulers who use agents (which have different costs to monitor) and the citizenry to show how each predicts different tax schemes. Different systems adopt different revenue collection schemes, and she illustrates how each was a response to different discount rates, transactions costs, and the self-interest of both principals and agents.

Lane (2008) argues that nearly all aspects of politics (i.e., the rise of states, political parties, levels of development) can be usefully seen through a principal–agent paradigm. Lane holds that the only way to fully constrain agents to do the bidding of principals is through the rule of law.

IV. Critiques

Many critique this approach in a similar fashion as they would rational choice. First, some argue that people do not maximize their returns, per se; rather, the view of bounded rationality suggests that people “satisfice” instead (March & Simon, 1958; Simon, 1947). The expression “It‘s good enough for government work” sums up this sentiment. Therefore, instead of searching for the most efficient contract between the principal and agent, this view holds that the agent does the minimum to satisfy the principal, and the principal does not work for the best contract but only for an acceptable one. However, with the role of information costs introduced into this paradigm, these two approaches are not necessarily at odds empirically, though they would be theoretically. According to the principal– agent paradigm, should the principal learn that the agent is satisficing instead of maximizing, then the principal would likely cancel the contract or seek another agent. However, with high monitoring costs, agents are held to outcomes, and if the outcomes were acceptable, then the relationship would have “worked.”

In addition, one could criticize this approach because it assumes (a) that human behavior is primarily motivated by economic gain or loss and (b) that each party is only pursuing self-interest. Engaging family members as agents could align goals of principal and agent more strongly than could pure economic motivation. Many African dictators and early European monarchs made sure to have family members in charge of the military or other key parts of the bureaucracy.

However, even loyalty for family members could be seen from a self-interested perspective. If a dictator is from a minority ethnic (tribal) group, then he or she often gives disproportionate access to top positions to members from his or her ethnic group or family. This behavior is logical because these agents have self-interest to be loyal to the ruler: They understand that their control over these jobs and resources would likely disappear should a new leader, especially one from a different ethnic group, replace the current one. This arrangement greatly reduces the agents’ returns under alternative arrangements, and their estimated returns under alternative ruling institutions would likely be quite low (see Quinn, 2002).

Others may hold that the assumptions of principal and agent are too simplistic and that a bureaucracy (especially in the U.S. case) has several competing principals exercising authority over the agent. Those who adhere to the principal–agent paradigm would say that this caveat would not destroy the paradigm but would merely predict more autonomy for agents within the paradigm because principals would be aware that the agents have other principals to whom they are held accountable.

Finally, the formal modeling side of the principal–agent paradigm has been criticized as being too formalistic, with few real-life applications. Real salary schemes seldom match the theoretical models developed. Some have also suggested that the approach is tautological. However, if one uses a paradigm to explain behavior that was previously unexplainable, then the paradigm is useful. The same is true if testable implications of the paradigm emerge. The paradigm is not directly tested, but its ability to predict is (Jensen, 1983).

V. Possible Application to Recent Events

One can use a principal–agent approach to illustrate part of the recent international financial problem with “toxic” mortgage debts. At the heart of this issue is the fact that many real estate agents and banks sold houses to clients who ultimately could not afford them. Why? Several important elements were in place to shift risk and information. First, investors had limited information about the risk involved in real estate. For quite a while, housing prices seemed to be impervious to price declines. This “limited” or false information shifted incentives in the market as to how safe people felt real estate investments were. Second, financial instruments (called derivatives or credit default swaps) were created to reduce risk, but those quite low risk ratings when they were, in fact, high-risk instruments. Third, these investment tools were effectively forms of insurance and were not regulated by government, which made information costs about them higher. Since these instruments were effectively insurance, but not regulated as such, the banks lent more money against these assets than would have been allowed in regular insurance markets. This led to “overleveraging.” So instead of lending against assets at 3, 5, or even 10 times their value, the loans were leveraged up to 30 times. Therefore, the effect of each bad debt was multiplied 30 times in the system instead of only 3 to 10 times.

Nevertheless, why did the real estate agents and bankers sell houses to people who could not afford them or who could barely afford them? These agents had no incentives to be careful. The real estate agents, being on commission, were paid only when houses were sold. Also, it is important to note, they were not punished financially for putting people into unaffordable houses as long as the sale went through. They had personal incentives that were partially at odds with the principals with whom they did business.

The risk-averse part of the “normal” process of selling mortgages should have been at the level of mortgage banks. In the past, if people bought homes that went into default, the bank lost money. Therefore, it made financial sense for mortgage banks to pay the information cost to scrutinize borrowers’ abilities to repay these loans. However, since bankers knew their mortgages were to be repackaged and sold as mortgage derivatives and securities, the banks were able to move the risk on to others while locking in their profits. Since there was a widespread belief that real estate prices would not fall, others bought these mortgage-backed securities as “safe” investments. Then very high side bets (derivatives) were made on what should have been safe investments but which turned out actually to be risky bets. (Here again, transaction costs and information costs clearly impacted markets in a negative way, although these costs are not considered in microeconomics.) It is important that those engaged in the selling of homes and primary mortgages shared none of the risk and made lots of money shifting the risk downstream, so they sold as many homes as possible. This circumstance illustrates the moral hazard involved in the issue because some profits were divorced from the risks of their practices. Now the people who bought the risky mortgage-backed debts faced adverse selection because these loans were rated as non-risky investments when they were, in fact, quite risky. Although a full analysis of the recent financial problems is beyond the scope of this research paper, one can see in basic relief some of the key elements of the principal–agent problem. One also sees that the microeconomic assumptions of complete information predicted perfect market behavior (i.e., the market is always right), which proved to be illusory and almost brought down the financial system.

The debate over merit pay for teachers is another current issue that could benefit from a principal–agent explanation. According to Holmstrom and Milgrom (1991), in cases in which the agents perform multiple and complex tasks that are not easily reducible to simple proxy measures, fixed-pay schemes align incentives better than outcome- based pay can. Should outcome-based pay (or merit pay) or promotions be adopted when agents have to achieve several competing goals, and only a few of those goals are directly measurable, then outcome-based rewards would incentivize agents to shift from the hard-to-measure aspects of their jobs to aspects that are easily measured. This is what many critics of “No Child Left Behind” meant when they said teachers would “only teach to the test.”3 That is to say that the other high-quality items normally taught or required in schools (e.g., discipline, abstract thinking, homework, socialization, effective writing, creative writing, affective learning) were replaced with materials that were to be on the test (e.g., reading comprehension, vocabulary, basic math, geometry). Supporters of teacher merit pay suggest that these incentives would increase teacher effort, especially in poorly performing schools. The idea of raising all salaries to reduce average adverse selection may join these two arguments, especially at the worst performing schools.

In sum, the principal–agent paradigm, which is often used in economics, business, and political science, is a powerful tool, especially when the assumption that people act primarily out of self-interest is most appropriate. Although many argue about whether people maximize returns or look only for acceptable returns, this paradigm can illustrate how by adding the variable of cost of information to models, one can sometimes predict outcomes more accurately. It can also explain how institutions aid or hinder the process of holding agents to account or getting them to do the principal’s bidding when agents’ behavior is otherwise costly to monitor. It is likely a rising paradigm in political science, especially where rational choice (which emphasizes purposeful, self-interested behavior) meets institutionalism (which acknowledges that information asymmetries are real and costly; see, e.g., Hall, 1997).

Notes

1. This review is not meant as a complete review of the all studies on this issue. Rather, it intends to open the door of this literature for undergraduates. Citations emphasize early works, classics, illustrations of a few well known works, and recent reviews, not necessarily a representative sample of the most recent literature and its findings. Most readings that require calculus to understand have been ignored, aside from some general findings.
2. Scandals often result in hearings that appear to be monitoring the bureaucracy for high profile, symbolic issues.
3. In fact, reports of teachers cheating to help students pass the exams are rising. For example, see Axtman (2005).

Bibliography:

1. Alam, M. S. (1989). Anatomy of corruption: An approach to the political economy of underdevelopment. American Journal of Economics & Sociology, 48(4), 441-471.
2. Axtman, K. (2005, January 11). When tests’ cheaters are the teachers. Christian Science Monitor.
3. Bates, R. H. (1981). Markets and states in tropical Africa: The political basis of agricultural policies. Los Angeles: University of California Press.
4. Coase, R. H. (1937). The nature of the firm. Economica, 4, 386-405.
5. Eisenhardt, K. M. (1989). Agency theory: An assessment and review. Academy of Management Review, 14(1), 57-74.
6. Geddes, B. (2003). Paradigms and sand castles: Theory building and research design in comparative politics. Ann Arbor: University of Michigan Press.
7. Hall, P. A. (1997). The role of interests, institutions, and ideas in the comparative political economy of advanced industrial nations. In M. I. Lichbach & A. S. Zuckerman (Eds.), Comparative politics: Rationality, culture, and structure (pp. 174-208). Cambridge, UK: Cambridge University Press.
8. Holmstrom, B., & Milgrom, P. (1991). Multitask principal agent analyses: Incentive contracts, asset ownership and job design. Journal of Law, Economics & Organization, 7, 24-52.
9. Jensen, M. C. (1983). Organization theory and methodology. Accounting Review, 63(2), 319-339.
10. Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3, 305-360.
11. Kiser, E. (1999). Comparing varieties of agency theory in economics, political science, and sociology: An illustration from state policy implementation. Sociological Theory, 17(2), 146-170.
12. Klitgaard, R. (1988). Controlling corruption. Berkeley: University of California Press.
13. Laffont, J. J. (Ed.). (2003). The principal agent model: The economic theory of incentives. Northampton, MA: Elgar.
14. Lane, J. E. (2008). Comparative politics: The principal agent perspective. New York: Routledge.
15. Levi, M. (1988). Of rule and revenue. Los Angeles: University of California Press.
16. Levi, M. (1997). A model, a method, and a map: Rational choice in comparative and historical analysis. In M. I. Lichbach & A. S. Zuckerman (Eds.), Comparative politics: Rationality, culture, and structure (pp. 19-41). Cambridge, UK: Cambridge University Press.
17. March, J. C., & Simon, H. (1958). Organizations. New York: Wiley.
18. Miller, G. J. (2005). The political evolution of principal agent models. Annual Review of Political Science, 8, 203-225.
19. Moe, T. M. (1984). The new economics of organization. American Journal of Political Science, 28(4), 739-777.
20. North, D. C. (1981). Structure and change in economic history. New York: Norton.
21. Olson, M. (1965). The logic of collective action. Cambridge, MA: Harvard University Press.
22. Ordeshook, P. C. (1990). The emerging discipline of political economy. In J. E. Alt & K. A. Shepsle (Eds.), Perspectives in Positive Political Economy (pp. 9-30). Cambridge, UK: Cambridge University Press.
23. Quinn, J. J. (2002). The road oft traveled: Development policies and majority state ownership of industry in Africa. Westport, CT: Praeger.
24. Quinn, J. J. (2008). The effects of majority state ownership of significant economic sectors on corruption: Across regional comparison. International Interactions, 34(1), 84-128.
25. Rockman, B. A. (2001). Theory and inference in the study of bureaucracy: Micro and neoinstitutionalist foundations of choice. Journal of Public Administration Research & Theory, 11(1), 3-28.
26. Ross, S. A. (1973). The economic theory of agency: The principal’s problems. American Economic Review, 62, 134-139.
27. Shapiro, S. P. (2005). Agency theory. Annual Review of Sociology, 31, 263-284.
28. Shavell, S. (1979). Risk sharing and incentives in the principal and agent relationships. Bell Journal of Economics, 10, 55-73.
29. Simon, H. A. (1947). Administrative behavior. New York: Macmillan.
30. Spence, M., & Zeckhauser, R. (1971). Insurance, information and individual action. American Economic Review, 61(2), 380-387.
31. Stiglitz, J. E. (1987). Principal agent. In J. Eatwell, M. Milgate, & P. Newman (Eds.), Allocation, information, & markets (pp. 241-253). New York: W. W. Norton.
32. Weber,M. (1978). Economy and society: An outline of interpretive sociology I II. Los Angeles: University of California Press.
33. Weingast, B. R. (1984). The congressional bureaucratic system: A principal agent perspective (with applications to the SEC). Public Choice, 44(1), 147-191.
34. Williamson, O. E. (1985). The economic institutions of capitalism. New York: Free Press.