Chapter 17
Principal–Agent Issues and Adverse Selection: Can Everyone Agree?
In This Chapter
Working together as principals and agents
Eliminating information inequality
Reducing adverse selection’s poor choices
Minimizing moral hazard
It’s a well-known childhood taunt: “I know something you don’t know.” Whenever children use this taunt, you know that they’ve pulled out the heavy artillery in a battle of wits. And as I was once told, don’t engage in a battle of wits unless you come fully armed.
That childhood taunt has special significance in managerial economics. Because individuals tend to be motivated by self interest, yet must interact with one another in mutually beneficial exchange, it’s very important that they have the same information. But frequently everyone participating in an exchange doesn’t have the same information. These situations have asymmetric information, and if I know something you don’t know, I can skew the exchange to my benefit. I say the used car I’m trying to sell you is great; how do you know whether or not it’s great? If one individual possesses better information than another, there is an increased probability that the exchange won’t be mutually beneficial. And if the used car I sell you is a lemon, you’re going to be mad. When this situation exists for a large number of market participants, mutually beneficial exchanges are less likely to take place. This is reflected in the perception that “used car” salesmen are less trustworthy when compared to other occupations. Therefore, it is crucial for successful markets and businesses to develop mechanisms that offset the information inequality or asymmetric information.
In this chapter, I focus on the problems created by differences in individual goals and asymmetric information and how to overcome these problems. You discover how incentives help convince individuals with different goals to work toward a common objective. In situations where somebody doubts the accuracy of your statements — “I only drove this ten-year-old car to church on the months with five Sundays” — I show you how to make a persuasive, true argument. (I’m not teaching you how to lie. I’m teaching you how to convincingly tell the truth.) In situations where you have less information and doubt another’s claims, I explain how you can get the individual to reveal the information he or she possesses. And finally, you discover the hazard of your actions inadvertently changing individual behavior and how to prevent those changes. Indeed, this is a very informative chapter on information.
Principal–Agent Problem #1: When Managers and Owners Disagree
Incentives matter. If you disagree, have you ever been bribed to eat your vegetables with the offer of dessert? If you have and you ate your vegetables, you know incentives matter.
Businesses are comprised of many individuals who possess different goals. Principals are the ultimate recipient of a situation’s outcome. In the case of a business, the firm’s principals are its owners and the ultimate outcome they receive is profit. Agents serve as the principal’s representatives and make decisions on the principal’s behalf. In a business, agents include the managers and workers.
The principal–agent problem exists when the agent’s objectives differ from the principal’s objectives.
Recognizing that incentives matter isn’t enough for owners and managers. Different employees of the firm with different jobs ultimately have to work for the same thing — maximum profit. Therefore, it’s important to establish incentives that result in a coordinated, cooperative effort to maximize profit. Incentives must make objectives complementary or identical and eliminate the principal–agent problem. The challenge is to accomplish this task given individuals have different interests and preferences.
The principal–agent problem is especially evident when the agent’s effort isn’t observable or measurable.
Contributing to the principal–agent problem is the lack of information. It’s difficult to know how much effort an employee makes unless you sit and watch the employee all the time — but that in and of itself is very unproductive. In general, two major issues contribute to the principal–agent problem:
Principals frequently don’t directly observe the agent’s actions.
The agent’s actions aren’t the sole determinant of the ultimate outcome.
When a company loses a million dollars, was it because of poor managers, or did good managers keep the company from losing two million dollars? Simply looking at the firm’s profit doesn’t provide enough information to tell whether the management is good or bad just like looking at a coach’s record in a given season doesn’t indicate whether the individual is a good or bad coach. Winning half the games with less talented players may indicate great coaching while winning half the games with extremely talented players may indicate bad coaching.
Shareholders in a large corporation, the principals, want to maximize their wealth — the return on their investment. Agents, in this case the firm’s managers, may pursue some combination of other goals, including less effort, higher income, greater job security, lower risk of failure, and better reputation. It’s a difficult challenge to coordinate these various objectives and make them consistent with maximizing the firm’s return on investment.
As principals try to get all employees to work toward maximizing the company’s return on investment or profit, they must determine whether to use sticks or carrots. Sticks focus on supervision and negative consequences such as an employee being fired. The advantage of using sticks is they’re inexpensive to implement. On the other hand, sticks only motivate individuals to a point. That point is the minimum effort necessary to avoid the stick.
On the other hand, carrots focus on rewards. Employees are rewarded for good effort, such as with a profit-sharing plan. The advantage of carrots is employees have an incentive or reward, such as a bonus, for continued hard work past the minimum effort necessary to keep their job. The disadvantage is carrots are expensive.
Behaving badly with a flat salary
As an absentee owner, you realize your firm’s profit increases as your manager’s efforts increase. You decide to pay your manager a flat salary. Regardless of the manager’s effort, the salary doesn’t change. On the other hand, extra effort is bad for the manager — it’s hard work. Instead of working hard, the manager would rather talk with employees, surf the web, or play solitaire on the company’s computers. These activities give pleasure or utility, while managerial effort is hard and thus gives disutility. The resulting “compensation” for the manager equals the utility that comes from the flat salary minus the disutility associated with extra effort.
Figure 17-1 describes the relationship between managerial effort — measured on the horizontal axis — and the principal’s goal (profit) and manager’s goal (utility) — measured on the vertical axis. The solid lines represent the situation from the perspective of the company’s owners, or principals. As managerial effort increases, net revenue increases at a decreasing rate. That is, the curve is rising but becoming flatter. Given the manager receives a flat or constant salary, net revenue minus the manager’s salary also increases at a decreasing rate. So the company’s owners continue to get more and more profit — net revenue minus the manager’s salary — as the manager’s efforts increase. Maximum profit requires the manager’s best effort.
Figure 17-1: Optimal effort with a flat salary.
The manager faces a different situation as described by the dash-dot lines in Figure 17-1. The pleasure or utility the manager receives from salary is constant because salary doesn’t change. No matter how much effort the manager makes, the salary stays the same. The manager receives no additional reward for greater effort. However, effort is hard — so more effort leads to disutility. The manager would rather be surfing the web as opposed to expending a lot of energy working. Thus, effort results in disutility — instead of pleasure it generates “pain.”
The manager’s net utility equals the utility gained from salary minus the disutility of effort. In the case of a flat salary, the manager maximizes net utility by expending zero effort. This result isn’t surprising. Effort results only in pain; it doesn’t generate any more income or utility for the manager. Not surprisingly, the manager wants to expend as little effort as possible.
You should note that changing the flat salary doesn’t change this result. A higher flat salary simply shifts the dash-dot curves upward, but the maximum utility is still associated with zero effort.
Flat salaries tend to lead to the minimum level of managerial effort.
Behaving better with profit sharing
Managers increase effort if they have an incentive to do so. One method absentee owners use to increase effort is through profit sharing. Profit sharing indicates that managers receive some share or percent of profit. Thus, as profit increases due to increases in managerial efforts, managerial compensation increases.
Figure 17-2 illustrates the impact of a profit-sharing arrangement for both the firm — solid lines — and manager — dash-dot lines. The firm’s net revenue again increases at a decreasing rate. However, net revenue minus managerial compensation now changes. Initially, net revenue minus compensation increases at a decreasing rate; however, it now reaches a maximum and then begins to decrease. This decrease is the result of the continued increase in managerial compensation. The vertical difference between the net revenue curve and the net revenue minus managerial compensation curve gets larger as the manager receives more compensation for greater effort.
For the manager, compensation increases as effort increases. However, the utility the manager receives from that additional compensation increases at a decreasing rate due to the law of diminishing marginal utility from Chapter 5. At the same time, greater effort continues to generate disutility. As a result, the manager’s net utility — the utility from compensation minus the disutility of effort — initially increases, eventually reaches a maximum, and then begins to decrease. The manager’s situation is represented by the dash-dot lines in Figure 17-2.
The manager chooses the level of effort that maximizes net utility. In Figure 17-2, this corresponds to the effort level E0. Note how this is different from the flat salary. With the flat salary, the manager wants to expend as little effort as possible. With compensation tied to effort, the manager now works harder in order to get more utility or satisfaction.
Figure 17-2: Optimal effort with incentive salary.
The owner needs to develop a profit-sharing plan that results in managers maximizing their net utility at the same level of effort that maximizes net revenue minus the manager’s salary — the owner’s profit. Determining this balance point is a critical task for the owner.
Behaving better with stock options
Stock options are another method to increase managerial effort. A stock option provides its holder a future opportunity to buy the company’s stock at a predetermined price — the call price. Frequently, the option’s call price is its current price, although there is no single method for determining the call price. Typically, a stock option can be exercised only during a specified period of time and at some point the option expires.
If the company’s stock value increases, the holder of the stock option makes a very quick financial gain through the difference between the call price and the higher future price. If the company’s stock price falls below the call price, the holder of the call is under no obligation to buy the stock and the option isn’t exercised. Any holder of a stock option has an incentive to work harder to increase the company’s stock price.
Your company offers its manager a 1,000-share stock option that can be exercised six months from now at the current price of $20.00 per share, the call price. If six months from now the stock price is $25.00 a share, the manager can make a quick $5,000. Here’s how:
1. The manager exercises the option and buys 1,000 shares at $20.00 per share.
The total cost is $20,000.
2. The manager sells the 1,000 shares at the market price of $25.00 per share.
The revenue from the sale is $25,000.
3. The difference between the revenue and cost is the manager’s financial gain, or $5,000.
If instead the company’s stock price decreased to $15.00, the manager wouldn’t exercise the option. Thus, the manager with a stock option has no risk of loss.
Stock options can carry a variety of restrictions. Vesting requirements limit the manager’s ability to sell the shares or options. Typical vesting requirements include requiring the manager to continue to work for the company for a specified number of years; some specific event to occur, such as an initial public offering (IPO); or certain performance targets to be reached, such as sales or profit goals. These vesting requirements help ensure that the owner’s and manager’s interests are the same.
In addition to increasing managerial effort, stock options can be used by start-up companies to reward employees, while keeping wages and salaries low. Thus, a new firm’s employees have a strong incentive to work hard to make the company successful.
One advantage of stock options over profit-sharing is managers focus more on the long-run value of the stock. Profit-sharing may lead to a shorter-term focus because it’s based on current outcomes.
Keeping managers in line
Profit sharing and stock options involve using carrots to increase managerial effort. As an alternative, you can use a stick in the form of a negative consequence for poor performance. The advantage of this method to the firm’s owner is that it’s very cheap to administer. If the owner is present, the owner can monitor the manager’s effort. Thus, the risk of termination insures some minimal level of effort — at least enough effort for the manager to avoid losing his or her job.
Mergers and takeovers can be another stick that increases managerial effort. Poor effort by a manager can make a company an attractive merger or takeover opportunity. The acquiring company recognizes that greater profit can be made with the right managers. Thus, the new company may try to increase profit by replacing the previous manager.
Non-monetary rewards can also be used to increase managerial effort. Managerial effort is linked to reputation. Being recognized as an outstanding manager increases job mobility by making the manager an attractive employee for other firms, and job offers from rival firms may lead to more compensation from the new firm or the manager’s current firm. In this scenario, managers essentially recognize effort as an investment that leads to higher compensation or returns in the future. Perks, such as the so-called corner office, are also used to increase effort.
Principal–Agent Problem #2: When Managers and Workers Disagree
Just as goals for owners and managers may not coincide, the goals of managers and workers may not coincide. Again, owners and managers must choose between using carrots or sticks.
Determining worker compensation
Carrots can also be used with workers. Compensation arrangements increase worker effort if the worker’s compensation is directly tied to the firm’s success. Profit sharing is one example of a compensation arrangement (see the earlier section “Behaving better with profit sharing” for details). Another common form of linking worker compensation to the firm’s success is revenue sharing. In revenue sharing, workers get some amount of additional compensation based upon how much they sell. Examples of revenue sharing are car salespeople receiving a commission after selling a car, waiters and waitresses receiving a tip as a percentage of their sales, and insurance agents receiving a commission.
Another method for increasing worker effort is to use piece rates. In this situation, worker compensation is based upon how much output they produce — workers get paid a specified amount for each piece produced. One caution with piece rates is quality control must be maintained. Producing a lot of poor quality pieces won’t help the company.
Keeping workers in line
And sticks can also be used to ensure some minimal level of effort among workers. Unannounced random spot checks to monitor effort are inexpensive to implement but must include a penalty and be frequent enough to have real risk.
Time clocks generally aren’t a good method to monitor worker effort. Time clocks simply measure presence, not effort — think of Dagwood Bumstead in the comic strip Blondie.
Recognizing Asymmetric Information and The Market for Lemons
Usually the participants in an exchange don’t have the same information. Asymmetric information refers to a situation where some participants have better information than other participants. In such circumstances, rational individuals with less information ultimately choose not to participate.
Asymmetric information isn’t the same as imperfect information. Imperfect information simply means that participants don’t have all relevant information. No farmer can guess what the weather will be next summer, so all farmers have imperfect information. However, the market for agricultural commodities can still operate efficiently because nobody has an advantage. With asymmetric information, at least two individuals are involved. In these situations, the individual with better information can take advantage of the individual with less information. In these situations, markets don’t operate efficiently.
An excellent example of asymmetric information is the market for used cars. The seller of the car knows exactly what kind of shape it’s in. If the car has been well taken care of, the current owner wants the maximum price.
On the other hand, you’re in the market for a used car. You don’t know much about cars, so you ask the owner, “Have you taken good care of the car?” Assume the owner hasn’t taken good care. When asked your question, he’s still likely to respond “yes,” with the idea that he wants to make as much money as possible. Thus, from your perspective as a buyer you get the same answer — yes — from a person who has taken good care of a car and from a person who has taken poor care of a car. The “yes” answer doesn’t differentiate.
If the answer doesn’t differentiate, you’re rational to doubt its accuracy; thus, you discount the “yes” answer. Instead of paying the value of a car well taken care of, you offer something less. The person who has taken good care of the car refuses your offer. Only the person who has taken poor care of the car is likely to accept your offer — he’s selling you a lemon. Anybody who accepts your offer does so knowing he’s getting a good deal, but a good deal for him is likely to be a bad deal for you. This is an exchange you don’t want to make, so you stop offering to buy used cars because the current owner has more information than you do.
Obviously, there is a market for used cars, but note how this develops. One way to get around the lemons problem is to ask the owner for the car’s maintenance records. Accurate, up-to-date records provide you information to support the owner’s claim and overcome the asymmetric information. The lack of this information defaults you into thinking the car is a lemon, so owners have an incentive to keep accurate records.
Another alternative is for used-car dealers to offer warranties. A warranty reduces the risk of asymmetric information by reducing the cost of unexpected repairs. But note the burden on the seller. If the buyer takes poor care of the car, the repair may be due to the buyer’s carelessness. Thus, warranties tend to have exclusions.
The general solution to asymmetric information is for the party with less information to try to collect more. This information has value. Thus, organizations have been established to provide information. Examples of such organizations include the Good Housekeeping Seal of Approval for household goods, Underwriters Laboratory — UL for electrical appliances, and CARFAX for used cars.
Asymmetric information also exists in the hiring process. Applicants know whether they’re good or bad workers, but employers don’t. If the employer asks, “Are you a good worker?” every applicant responds, “Yes.” So the employer needs to somehow separate good workers from bad workers. One method companies use to separate good from bad workers is college grades. Statistically, individuals with high grade point averages are more likely to be good workers; therefore, employers look at course transcripts to find information on grades. This is an example of statistical discrimination where the employer uses group information to make inferences about individual characteristics. Similar statistical discrimination may occur based upon the individual’s major — for example, a bank preferring to hire an applicant with a finance major over an applicant with a philosophy major.
Increasing Insurance Costs with Adverse Selection
One effect of asymmetric information is adverse selection. Adverse selection arises when an individual has hidden characteristics before the transaction takes place. With hidden characteristics, one party knows things about himself that the other party doesn’t know. This leads to a self-selection bias where individuals act in their own self interest and use private information to determine their optimal action, usually at another party’s disadvantage or cost.
Adverse selection is readily apparent in the market for insurance. The insurance company doesn’t know who is a good driver or who is a bad driver. On the other hand, drivers know whether they’re good or bad. Thus, there is asymmetric information with drivers knowing more about their driving habits and risks than the insurance company.
However, the insurance company does know that on average all drivers have a certain number and value of claims. So, you may wonder whether the insurance company can just set the insurance premium equal to the average claim value, plus a little bit more for profit. And, because of adverse selection, the somewhat surprising answer is “no.”
If you assume that drivers aren’t legally required to buy car insurance, what happens if insurance companies set the insurance premium equal to the average claim value? Bad drivers are likely to pay the average premium because the bad drivers know they have a high likelihood of being in an accident. The bad drivers have a great need for insurance. Thus, the insurance company pays a lot of claims. On the other hand, good drivers believe they have a fairly low likelihood of being in an accident. Given the insurance premium partially reflects the high cost of covering bad drivers, the good drivers are likely to decide that the coverage isn’t worth the cost. Thus, the good drivers decline the insurance. Because of adverse selection, only the bad drivers have bought the insurance.
If this situation continues, the end result is no insurance. After good drivers decide not to buy insurance, the company finds that the average claim is higher than expected. The company had originally based its premium on insuring both good and bad drivers, but because good drivers decided not to buy the insurance, the actual claims reflect only bad drivers. Because the average claim is higher than originally estimated, the insurance company has to raise its premium. But once again, drivers with relatively good records decide the insurance isn’t worth the cost, so they don’t buy the policy. Claims again turn out to be higher than expected, and the cycle continues until finally, the insurance company gives up or goes bankrupt.
Dealing with Adverse Selection
One method for dealing with adverse selection is to force everyone to participate. For example, states commonly require drivers to have car insurance. Thus, it’s possible for car insurance companies to charge a premium that reflects the average claim. However, participants who are unlikely to submit a claim may believe it’s unfair for them to be forced to subsidize those likely to file a claim. I’m a good driver, so why should I subsidize somebody who’s not?
An alternative method for dealing with adverse selection is to group individuals through indirect information, such as statistical discrimination. Insurance companies can’t get individuals to admit whether they’re good or bad drivers, so the companies develop statistical profiles of good and bad drivers. By determining who is most likely to be a bad driver, the insurance company can establish different premiums. Thus, young males are likely to pay more for insurance. Somebody living in Los Angeles pays more than someone in Hanover, Indiana. Drivers with speeding tickets and other traffic violations pay higher premiums. The list goes on and on because the better the characteristics, the more accurate the premiums, and competition among insurance companies helps develop better statistical profiles.
But neither of these alternatives — requiring participation or statistical discrimination — has the actual participants share information to overcome the asymmetry. Both direct and indirect methods lead to the revelation of information that resolves, or at least reduces, adverse selection. The direct method is to use an appraisal, while indirect methods include screening and signaling.
Appraising asymmetric information
Appraisal resolves asymmetric information by examining a characteristic that’s objectively verifiable. In the case of a used car, if the buyer decides to take the car to his mechanic, the mechanic can provide a knowledgeable assessment of its condition. Or in the case of art, an appraiser can verify the painting as an original or a fake.
Appraisal directly resolves asymmetric information under two conditions:
The characteristic associated with the asymmetric information must be objectively verifiable.
The appraisal’s benefit must exceed the seller’s cost.
In the case of health insurance, a medical exam becomes an appraisal. The medical exam yields objective information about the individual’s current health.
Signaling with warranties
While appraisal directly conveys information from one party to another, signaling is a method for indirectly conveying information. By using signaling, the individual with better information convincingly communicates that information to the individual who has less information.
To return to the example of the used car, going to your own mechanic provides you an appraisal of the car’s value. Alternatively, the current owner of the car could signal to you that the car is well taken care of by offering a warranty — a guarantee to fix any repairs during a certain period of time or providing a CARFAX or similar report.
To be successful in resolving asymmetric information, signaling must induce self selection among the better-informed participants. Only participants offering well cared for used cars are willing to offer warranties or a CARFAX report. A warranty carries a real cost to the seller if the car breaks down. Thus, while a warranty may convince you, the buyer, that the car is well taken care of, a sign that simply states “Best Used Cars In Town” isn’t likely to be convincing, because it doesn’t cost much for those selling lemons to make the same sign.
Controlling through screening
Another indirect method for resolving information asymmetry is screening. In screening, the participant with less information controls a variable that leads to the participant with better information revealing that information. One screening method with insurance is the use of deductibles. An insurance deductible is an amount the insured must pay before the insurance company pays on a claim. High risk participants are likely to know they’re high risk. Thus, high risk participants are likely to file a claim and want low deductibles — they don’t want to have to pay very much before the insurance company starts paying. On the other hand, low risk participants know they’re unlikely to file a claim — they’re willing to have a higher deductible because they’re less likely to have to pay it. The deductible provides a mechanism that leads the better-informed participant, the insured, to reveal information to the less well-informed participant, the insurance company.
The number of screening options must correspond to the number of choice characteristics. For example, individuals choosing low deductibles can do so because they’re high risk or alternatively, because they’re risk averse. Thus, deductibles are a good, but not perfect, screening device.
Working with Moral Hazard
Moral hazard is another asymmetric information problem. In the case of moral hazard, one party takes hidden actions or actions that are unknown to the other party after the transaction. Moral hazard encourages bad behavior like poorer driving with insurance.
Insurance provides an example of moral hazard. The probability that the insurance company has to pay a claim is based on the hidden actions of the policy owner. For example, if fire insurance covers 100 percent of the replacement cost of my house and contents, I have less incentive to reduce the chance of loss. I’m more likely to take a quick trip to the store while dinner cooks on the stove — a fire hazard. I’m less likely to incur the cost of installing a sprinkler system or buying a fire extinguisher to minimize fire damage. These efforts have costs, and the efforts’ benefits are reduced given my fire insurance.
Health insurance is another example of moral hazard. If my health insurance covers 100 percent of the cost of medical care, I’m more likely to run to the doctor because the cost to me is so low — just the opportunity cost of my time. On the other hand, my trips have a very real cost to the health insurance company — it has to pay the doctor. The fact that I have insurance changes my behavior in a way that adversely affects the health insurance company.
One way to address moral hazard in health insurance is to have a high deductible. A high deductible shifts more of the cost of going to the doctor back on me. Of course, if something major is wrong, the deductible ends up being a very small part of the total cost. So, for potential big problems, I still go to the doctor. On the other hand, I’m more likely to ignore those small little problems, at least for awhile, to avoid paying the deductible. I don’t go to the doctor as often, keeping healthcare costs down. And that’s the point: For markets to work well, all participants must have the same information and incentives.