I, the late 1990s, a large auction house, Auction Services International (ASI), employed art experts to manage ASI's business in various schools of artD French Impressionism, American Realism, and the like. Each expert's job was to persuade art owners to use ASIs auction services to sell their art. ASI earned money by charging the art owners a percentage of the final price at auction. ‘The art expert negotiated this percentage rate with the art owners. Art experts were given discretion to negotiate rates from 10% to 30%, depending on the art expert's assessment of the seller's willingness to pay and knowledge of competitors’ offers. Instead, most of these negotiations yielded relatively low rates, much closer to the 10% minimum. Puzzled, ASI's CEO did some investigation and discovered that the art experts were discounting rates in exchange for gifts from the sellersB cases of fine wine, fur coats, even luxury cars. After she found out about these kickbacks, the CEO took away the experts’ discretion to negotiate the rates. The CEO's action ended the exchange of gifts for lower rates, but the experts had become accustomed to the kickbacks, considering them an important part of their compensation, Consequently, many of the art experts quit, leaving to set up their own independent galleries in direct competition with ASL To make matters worse, the CEO decided to set a 17% price by conspir- ng with a rival auction house. When the conspiracy was discovered, the CEO ‘was sentenced to a year in jail, and the judge tacked on a $7.5 million fine, an Jmount calculated as 5% of the $150 million volume of commerce affected by ‘ice-fixing conspiracy. fe pad Faved this chapter, she would have known better how 0 motivate her employees to work in the firm's best interest, and she may have avoid prison. ter a ani the final two chapters, we come back to is asia problem-solving framework of Chapter 1. Our goal is to show you the270 SECTION VI + Organizational Design amework to help you understand why it works in oots of fi EER Principaizagent Relationships When we study the relationship between a firm and its employees, we use ts call principalsagent models. [Aprincipal wants an agent to act on her behalf. But agents often have different goals and preferences than do principals. the firm or the CEO is the principal, and the linguistic convention that the principal is what economist In the ASI story, for example, art expert is the agent. We adopt the female and the agent male, Like the art expert, the agent often has better information than the prin- cipal, The problem the principal faces is that the agent has different incentives than does the principal, which we call an incentive conflict. In our example, ‘ASI's CEO wanted her art experts to negotiate profitable commission rates, whereas the art experts wanted to increase personal income, including kick- backs from customers. In general, incentive conflicts exist between every principal and every agent throughout the management hierarchyD between sharcholders and managers, beeween managers and subordinates, and between a firm and its various divisions. Incentive conflict generates problems that should sound famil ‘The principal has to decide which agent to hire (a problem of adverse selection). Once the agent is hired, the principal bas to figure out how to motivate the agent (moral hazard). ‘We know (from Chapters 19 and 20) that adverse selection and moral hazard problems are costly to control. In fact, the costs associated with moral hazard and adverse selection are called agency costs because we analyze them using principalsagent models. A well-run firm will find ways to reduce agency costss poorly run firms often blindly incur ageney costs or unwittingly make decisions that increase them. We also know that we can reduce the costs of adverse selection or moral hazard by gathering information about the agent: A principal can reduce agency costs if she gathers information about the agent's type (adverse selection) or about the agent's actions (moral hazard). For adverse selection, information gath ns yack- around of agents before theyre hisedy and for poral hensca wieneenen gathering means monitoring agents’ actions after they're hired. This difference has led some to characterize adverse selection as a pre-contractual problem caused by hidden information and moral hazard as « post alee lem caused by hidden actions. ee a At ASI, for example, had the CEO known when agents were reducing rates in exchange for gifts, she might have devised a simple incentive-semspersationHAPTER 21 + Getting Employees to Work inthe Ftm’s Best Interests 274 scheme (a reward or a punishment) to stop it. But even withou mation, she should have anticipated the art experts opportanone beh especially since she was paying them flat salariesD compensation wereld to performance, Because ASI did not reward art experts for seting profiatia rates, the art owners found it easy to bribe them to set unprofitable ones. When the CEO decided to take away rate-setting discretion from she art experts, she compounded her initial mistake. This solution was costly because the CEO lacked information about what rates owners were willing to pay, Instead, she tried her*17% solution,’ the rate set collusively with her rival, A better solution would have been to leave the rate-setting authority with the art expects but change to an incentive-compensation schemeD for exam ple, to one that paid art experts a percentage of the revenue they brought to the firm. This kind of compensition scheme better aligns the agents' incentives with the firm's goals. If the agents set profitable rates, they'll inerease both the firm's profit and their own compensation. If you think of the art experts as salespeople, this incentive-compensation scheme seems like an obvious solutionBm ost salespeople are compensated with sales commissions. This solution does have one drawback: like all incentive-compensation schemes, it exposes the agents to risk, In this case, should the economy decline, the firm would sell ewer art pieces, and the art experts’ compensation would fall chrough no fault oftheir own. If you are the principal, imposing risk on the agent may not seem like your problem, but we know (from Chapter 9) that people must be compensated for bearing risk. This raises the principal’ cost of using an incentive-compensa- tion scheme. Incentive compensation imposes risk on the agent for which he must be compensated. The risk of incentive compensation reminds us that most solutions to the problems of adverse selection and moral hazard involve trade-offs. We adopt incentive compensation only if its benefits (the agent works harder) exceed its costs (we have to compensate the agent for bearing risk). We measure these costs and benefits relative to the status quo or relative to other potential solutions. Bey contron ing Incentive Conflict ‘We don't have any hard and fast rules for the best way to control incentive conflicts between principals and agents, but we can identify the trade-offs associated with various solutions. Once you understand the basic trade-offs, it is easier to identify the costs and benefits of various solutions. In a well- decision-makers have (1) the information necessary to d (2) the incentive to do so. To ensure that decision- formation to make good decisions, there are two run organization, make good decisions an makers have enough in obvious solutions: Either move information to those who are making decisio cision-making authority to those who have information. ys or move de-‘272 SECTION VI» Organizational Design ers a firm from the bottom 80 that subording, magement hierarchy) are better informed j. but not the CEO, k vase of ASI, the art experts, oe ea a ing to pay. When the CEO centralized decision ma Ph etre, he eompany Tost the ability 10 price dsrningy, between high- and low-value customers. fon-making authority, you should also figure the decision-maker. nation et ‘Typically, inform {who are further down in the mm When you centralize decision cout how to transfer information to This is not as easy as it sounds. Information comes from self-interested ‘e an incentive to manipulate the decision-maker. For Example, sales agents often tell their marketing bosses that they have to reduce price in order to make a sale. They have an incentive to lie if they are paid predominantly based on the number of sales of in the case of ASI, kickbacks. The other solution, leaving pricing discretion with the art expert, decen. parties who may hav. tralizes decision-making authority. When you decentralize decision-making authority, you should also strengthen incentive-compensation schemes. The logic is clear, Once you give an'agent authority to make decisions, you want to make sure that he is motivated to make choices in the firm's best interest. At ASI, the weak incentives were obviousDthe art experts were given no financial incentive to set profitable rates. The CEO should have adopted an incentive-compensation scheme to encourage more profitable rate setting. Recall from Chapter 1 that incentives have two parts: before you can reward good behavior, you have to be able to measure i, You can measure performance informally, with some kind of subjective performance evals- tion, or formally, using sales or profitability as performance metrics. Once you have an adequate performance measure, you create incentives by link- ing compensation to the performance metrics, Here, we speak very generally about compensation: compensation can be pay, increased likelihood of pro- motion, bonuses, or anything else that employees value. The link between performance and compensation creates the incentive for agents to act in the firm's best interest. simple eat feed incenive-compensation schemes is challenging. Take # Shera cle Of & uit farmer trying to decide how to pay pickers. Th Shine recs £0 pay workers apiece rate for each piece picked. A com or is thar the rate has to be increased when pickings are slim ensure that the workers earn the pa i minimum wage required by law. Under this system, however, workers so her to discourage fs y a rs sometimes mc . Ce mas 's monitor each ot! i incentive-compens rate based on this test-pick.! The lesson of this sto ‘game Eee EPO is to realize that workers have an incentive © formance will often te ach vy aha vik & teacher rewarded for student est Pe “e test” rather than foster deeper understandirCHAPTER 21 + Getting Employees to Work inthe Fim’ Best interests 273 of the material, employees will often discover ways to maxi make themselves better off that don't improve or even hurt protean To combat gaming, first try to anticipate the more obvious yanes and adjust the compensation scheme to prevent them. Second, moniter euros to ensure that you are getting the behaviors you really want. Monitoring mney the principal a better performance evaluation metric, which allows her hee ter align the incentives of the agent with the goals of the principal, In the case of ASI, it looks like decentralization, but with stronger incen- tives, would have been the better solution. In general, the answer to whether centralization or decentralization is better depends on the relative cost of the two alternatives, If you wane to centralize decision making, how costly will i be to transfer information from agents to principals? If you want to decen- talize, how costly will it be to institute incentives that adequately compensate agents for bearing risk? PIE] Marketing versus Sales The conflict between the art experts and their employer is fairly typical of the general incentive conflict that arises in organizations with separate sales and marketing divisions. The two divisions rarely get along, and this is often due to the different incentives that they are provided. Marketing managers gener- ally receive profitability bonuses as compensation, whereas salespeople receive commissions based on revenue, They disagree about what price to charge be- cause the marketing principal wants to maximize profitD that is, by making sales where MR > MC. In contrast, the sales agent wants to maximize reve- rue by making sales where MR > 0. This means that the salesperson prefers more sales or, equivalently, lower prices. If the marketing managers know’ when salespeople are making unprofit- able sales, they can easily put a stop to it. Without that information, however, controlling the incentive conflict becomes costly. “To see why, put yourself in the place of a marketing manager who is over- seeing a salesperson who tells you that he has to reduce price to make a par- ticularly tough sale even though it will leave the firm with very little profit. Because you don't know how much each customer is willing to pay, you can't tell whether the salesperson wants to reduce ptice to make a particularly tough sale, which would be reasonable from the firm's perspective, or whether he has decided that the extra effort to sell at the higher price is not worth the small increase in commission, despite the big increase in profit for the firm. ‘Since it seems easy to design an incentive-compensation scheme that rewards the salesperson for increasing profitability rather than revenue, we have to wonder why this kind of incentive compensation is not more widely fer performance evaluations based used. Most salespeople will tell you they pret fom revenue because revenue is what they directly control. They may also per ceive a change from a sales commission to profit commission as a sneaky way for the company to cut labor costs. Remember that profi always lower than revenue.274 SECTION Vi « Organizational Design Je to persuade the sales agent to accept the change r, vow design the profit-based compensation scheme i, seen cve-ncutral? to the salesperson. For example, a 20% commission on a revenit valent to 2.10% commission on revenue if the contribution may aioe pon Agents are guaranteed to earn the same under each compensation Maen even if their behavior does not change. But because they can earn nore money if they change behavior (by pricing less aggressively), thr com pensation should increase under a commission basel om profit ‘Companies often try to control incentive conflicts simply by asking sales agents to change their behaviorDbut actions (and paychecks) can speak much louder than words. Sales agents will change behavior when they have incen- You should be abl 1 profit commission tives to do so. ‘Another common solution is to require that sales agents obtain permission to reduce price below some specific threshold. The sales agent could do this by transferring enough information to the marketing principal to convince her that the price reduction is profitable, EEE] Fanchme We can understand the growth of franchising in the United States over the past 50 years as a solution to a particular principalzagent incentive conflict. The principal is the parent company that owns a popular brand, like McDon- ald’s. As the company grows, it has a choiceDit can open up company-owned stores, or it can let independent franchisees open and run stores. The fran- chisees then pay the company a fee for the right to use the parent company's brand, Suppose you are advising the owner of a fast-food restaurant chain. This chain's owner is trying to decide whether to sell one of its company-owned restaurants, currently run by a salaried manager, to a franchisee. If the chain sells the store, the franchisee will manage it and pay the owner a fixed fran- chise fee for permission to use the brand. Should the owner sell the store? Of course, the answer is,* It depends? In this case, it depends on whether the franchise organizational form is more profitable than the company-owned organizational form. With the company-owned structure, managers don't work as hard as they would if they owned the restaurant (moral hazard), and the salaried management job may have attracted a lazy manager (adverse selection). The company must spend resources on monitoring managers’ productivity after paying off his costs including the f eee a 8 the franchise fee and industrious Running a franchised storeCHAPTER 21 + Getting Employees to Work in the Firm's Best interests 278 However, the franchisee faces more risk than does a salaried manager and. as a consequence, will demand higher compensation in the form of a lower franchise fee. Ifthe franchisee’ demands too much for bearing risk, then the restaurant could be more valuable as a company-owned store than itis franchise. ee Jointly, the parties can split a larger profit pie if they can figure out how to balance these concerns. At one extreme, the company-owned store with a salaried manager leads to shirking on the part of the agentD a type of moral hazard. As mentioned earliey, it also leads to adverse selection because salaried jobs are more likely to attract lazy managers. The company may also incur costs t0 monitor the managers’ actions. _At the other extreme, the franchise organizational form is analogous to an incentive-compensation scheme because the franchisee keeps every dollar he earns after paying off his costs. But if factors other than effort affect profit, this kind of incentive compensation also imposes extra risk on the agent for which he must be compensated. Sharing contracts fall between these two extremes. Instead of a fixed fran- chise fee, the franchisor might demand a percentage of the revenue or profit of the restaurant. This arrangement reduces franchisee risk by reducing the amount the franchisee pays to the franchisor when the store does poorly. However, sharing contracts also increase agency costs {moral hazard, adverse selection, and monitoring costs). EEEJ4 Framework for Diagnosing and Solving Problems Understanding the trade-offs between information and incentives is useful, but it still doesn't tell you how to identify and fix specific problems within an organization. For that, you need to be able to find the source of thé incentive conflict and come up with specific alternatives to reduce the associated agency costs. Then choose the alternative that gives you the highest profit. “To analyze principalsagent problems, we return to the problem-solving framework introduced in Chapter 1. First, reduce the problem to a bad deci sion, and then ask three questions: 1, Who is making the (bad) decision? 2) Does the decision-maker have enough information to make a good decision? 3, Does the decision-maker have the incentive to do so? In principalsagent relationships, the source of the problem is almost always either moral hazard or adverse selection. The frst question identifies ‘the source of this problem, The second examines the employee's information roe sthe nature of the information asymmetry. The third identifies how the non maker is evaluated and compensated. Remember that incentives have he performance evaluation measures whether the individual is two parts: i doing a good job; the compensation rewards good performance276 SECTION VI» Organizational Desien ex's answer the three questions for the ASI example at the beginning of this chapter: 1. Who is making the bad decision? The art experts. ‘They are negotiating . low. aan i deiionmaler have enough information to make a god ac oe vai fac, they are the only ones with enough information to set profitable rates, 3, Does the decision-maker received a fla salary, making it rl them with gifs. In general, answers to the three questions will suggest alternatives for reducing agency costs in three general ways: by (1) changing decision rights, (2) transferring information, or (3) changing incentives. Tn this case, we have two obvious solutions: leave rate-setting authority with the art experts, but adopt stronger incentive compensation; or transfer rate-setting authority to a marketing executive, and then transfer crucial infor- mation to her. The first is a decentralization solution, and the second is a cen- tralization solution. To see how well you understand how to use the framework, imagine that you are called in as'a consultant to a large retail chain of *ge neral stores® that target low-income customers in smaller cities. As the company has grown, the CEO and the stock analysts who follow the company have noticed that newly opened stores are not meeting sales projections. The CEO wants you to find out what's causing the problem and fix it. In the course of your investigation, you learn that the company uses * development” agents to find new store locations and negotiate leases with property owners. The company rewards these agents with generous stock options, provided they open 50 new stores in a single year. Although agents are supposed to open new stores only if the sales potential is at least $1 mil- lion per yeas, this is obviously not happening. Newly opened stores earn just half that amount, have the incentive to do so? No. The art experts Jatively easy for art owners to bribe Before continuing, try to identify the problem. Begin your analysis by asking the three questions, 1. Who is making the bad decision? The de opening unprofitable stores. 2, Dots the decision-maker have enough information to make a good ecision? Yes. Development agents probably have access to information about wheth Ww s i aor aa rene Stores would be profitable. This appears to bea 3. Does the decision-maker have received stock options for ope new stores' profitability, ‘velopment agents. They are the incentive to do so? No. The agents ‘ing SO stores each year, regardless of theCHAPTER 21 + Getting Employees to Work in the Firm’s Best interests 277 ‘The problem is not with information but rather with the incentives of the agent, who is rewarded for opening stores regardless of profitability. Before you continue, suggest at least two solutions to the problem and choose the best one. ~ Before continuing, try to fix the problem. You have at least two obvious solutions: 1, (Decentralization) The company could change the incentives of the devel- ‘opment agents by rewarding them for opening only profitable stores. 2, (Centralization) Alternatively, the company could take the decision to ‘open stores away from agents and then gather its own information about the potential profitability of new store sites. ‘The decentralization solution would leave decision-making authority with the agents, who have specialized knowledge about the profitability of loca- tions for new stores. But the agents would have to wait for a year of store operation before receiving compensation (at which point, they know whether the store made $1 million in sales). However, this solution exposes the agents to risk beyond their controlDthe ir compensation would depend on the behav- ior of the store manager, as well as on the state of the economy. The agent would have to be compensated for bearing this risk in the form of higher com- pensation, which is the usual trade-off between incentive compensation and risk. In this case, the general store chain chose the centralization option. It developed a forecasting model to predict the profitability of new stores based fon local demographic information and the locations of rival stores. Agents were allowed to open new store locations only if the model predicted sales exceeding $1 million. If the model is good at predicting which stores ate likely to be.profit- able, this solution will work well. But if the model cannot identify profitable locations, it will be a poor substitute for the agents’ specialized knowledge or intuition about which new store locations are likely to be profitable, It Sill make both Type I errors (open unprofitable stores) and Type Il errors {fail to open profitable stores). As you should recall from our discussion of minimizing expected error costs in Chapter 17, if the error costs are asym- metric (it is more costly to open an unprofitable store), * shade* your predic~ ton threshold to avoid the more costly error (raise the predicted thresholds for opening stores). In this case, the model predicted well, and the problem disappeared.