Consider an entrepreneur who starts a small restaurant.
First, she works alone, cooking the food and serving customers. Then she hires a few employees to work with her, and they move into a larger space. The business grows and eventually the entrepreneur decides to open more locations and hire a manager to run the individual restaurants day-to-day operations for her while she focuses on running the business.
Now the business has the following structure:
└── Terri (Owner of Terri's Amazing Restaurant)
├── Jackson (Manages Location 1)
└── Jason (Manages Location 2)
Jackson and Jason now control the restaurants that Terri owns. She has hired them to run the restaurants while she focuses on other things.
How does she want them to make choices?
She wants them to make the same choices she would make if she were choosing for them.
Let’s say that each manager has responsibility for:
└── Terri (Owner of Terri's Amazing Restaurant)
├── Jackson (Manages Location 1)
│ ├── Food
│ ├── Menu
│ └── Staff
└── Jason (Manages Location 2)
├── Food
├── Menu
└── Staff
How might Jason’s and Jackson’s priorities differ from Terri’s? What if Jason is interested in moving from Terri’s restaurant, located in Sai Kung, to Lung King Heen. A restaurant in central with three Michelin stars. What if Jackson is nearing retirement age?
These situations may lead Jason and Jackson to make different choices than Terri would, had she retained control of the restaurants. When owners and managers have different interests, priorities, or incentives we refer to these as conflicts of interest. Conflicts of interest can occur at any level of an organization so we need slightly broader language to refer to what is going on here.
We refer to people who are hired to act on behalf of others as agents and those on whose behalf they act as principals. We refer to the conflicts of interest that arise in these cases as “principal-agent problems”
To illustrate lets consider the case that arises when the restaurants begin to operate separate services for lunch and dinner.
└── Terri (Owner of Terri's Amazing Restaurant)
├── Jackson (Manages Location 1)
│ ├── Annalee (Dinner)
│ │ ├── Food
│ │ ├── Menu
│ │ └── Staff
│ └── Caleb (Lunch)
│ ├── Food
│ ├── Menu
│ └── Staff
└── Jason (Manages Location 2)
├── Robin (Dinner)
│ ├── Food
│ ├── Menu
│ └── Staff
└── Seth (Lunch)
├── Food
├── Menu
└── Staff
Now we can have conflicts of interest at multiple levels of the company, not just between owners and managers.
This plays out in a similar way in large companies where shareholders hire a CEO to run the company on their behalf. That CEO in turn hires a divisional manager who then in turn hires various managers.
└── Shareholders
└── CEO
├── African Division Manager
│ ├── Distribution Manager
│ ├── Plant Manager
│ └── Sales Manager
├── Asia-Pacific Division Manager
│ ├── Distribution Manager
│ ├── Plant Manager
│ └── Sales Manager
├── European Division Manager
│ ├── Distribution Manager
│ ├── Plant Manager
│ └── Sales Manager
├── North American Division Manager
│ ├── Distribution Manager
│ ├── Plant Manager
│ └── Sales Manager
└── South American Division Manager
├── Distribution Manager
├── Plant Manager
└── Sales Man
Each link involves a principal and an agent, with the former delegating substantial authority for making decisions to the latter. At every level conflicts of interest arise.
For example, the owner might want the manager to work hard, and avoid unnecessary expenses, so that the owner’s surplus or profit may be high.
But the manager may not want to work hard to increase the owner’s surplus. He may choose to use the company’s money to pursue pet projects and enjoy other perquisites (such as expensive office furniture and first-class travel).
Such a conflict of interest can often be reduced with an incentive scheme, which aligns the interests of principal and agent.
The owner, for instance, can base a CEO’s compensation on company profit, or she can offer him shares of the company. The CEO can then choose an incentive scheme for every divisional manager, based on company profits or performance evaluation of the division (which may in turn be based on output, costs, divisional profits, ROI, etc.).
Consider the manager of a factory. An owner’s return on investment in the factory depends on many factors. For example, the market price of raw materials and finished goods, and the number of equipment breakdowns will affect the factory’s profit and the owner’s return. These factors will cause the performance of the firm to be uncertain. The owner’s return will also depend to some extent on how hard-working and/or talented the manager is. If the manager works hard (i.e., if he can keep the machinery in good condition, react quickly to breakdowns, treat workers well and maintain discipline, etc.), his effort will be productive in the sense that it will increase the overall probability of good performance. Outstanding effort on the manager’s part, however, will not necessarily result in good performance, because factors beyond the manager’s control (such as external price fluctuations or deficient raw material supply) may make bad performance inevitable.
In this situation, how should the manager be paid by the owner? One principle, to which many firms have traditionally adhered, is that the manager should be responsible only for those dimensions of performance that he can control. Since the firm’s good performance cannot be completely guaranteed by the manager, he should not be held responsible for, nor should his pay depend upon, factors he cannot control. We refer to this as the controllability principle.
Well, then, says I what’s the use you learning to do right when it’s troublesome to do right and ain’t no trouble to do wrong, and the wages is just the same? I was stuck. I couldn’t answer that. So I reckoned I wouldn’t bother no more about it, but after this always do whichever come handiest at the time.
The Adventures of Huckleberry Finn Mark Twain, 1884
Beyond a certain point, Huck does not like to work. If every week Huck put in just forty hours, did not work too hard during any one of those hours, and took a reasonable paycheck home every Friday, Huck would be a happy guy. Sadly, Huck must do some things he finds unpleasant and hard in order to maintain his standard of living. Huck will not do these things unless he thinks the rewards outweigh his distaste for those activities.
Huck makes widgets in Miss Watson’s factory. The number of widgets produced depends partly on Huck’s effort and partly on random events. Huck can work hard, experience bad luck and widget output will be low. Or, Huck could shirk his duties, but with good luck, widget output is high. Ideally, Huck always works diligently. On average, if Huck is diligent, widget output is satisfactory.
Assume:
Question: How should Miss Watson compensate Huck?
Consider:
What does it mean if the bonus per widget is equal to the selling price of widgets and the salary is negative?
Can you give a realistic example that corresponds closely to Case 1B and the “salary” is negative?
pay constant, Huck dislikes pay plans that have higher variances. But, Huck may prefer a high variance/high mean pay plan to a low variance/low mean pay plan, depending on just how risk averse he is.
Consider how output and associated risks are shared between Miss Watson and Huck in each of these examples. Do pay plans exist that make Miss Watson and Huck both better off than others? What is the source of such improvements in Miss Watson’s and Huck’s welfare?
In and of itself, the accounting report has no value to either Miss Watson or Huck. It cannot be sold for cash as widgets are, but it permits better contracts to be written between Huck and Miss Watson. Would Miss Watson pay some money to acquire such a report? Would Huck?
General result: Whenever signals are informative of Huck’s actions, the joint surplus available to Huck and Miss Watson can be increased by modifying the contract to incorporate these components.
Observation 1: If the firm’s performance does not depend at all on the manager’s action, he should be paid a fixed salary.
Observation 2: If the firm’s performance depends upon the manager’s actions, and if these actions can be monitored perfectly by the principal, the incentive problem can be solved perfectly.
Observation 3: If the firm’s performance depends upon manager’s action, and this action is not directly observable by the principal, then the manager must be held responsible for observable measures of performance. This is because the performance variable permits the principal to partly infer the manager’s action. Making compensation depend on performance influences the manager’s action in a mutually desirable way.
Observation 4 : If the firm’s performance depends upon the manager’s action, and this action is not directly observable by the principal, the manager may end up with a low rather than a high bonus, even if he chooses the action that the principal wants. That is, he may be held responsible for factors outside his control.
Observation 5 : Relative performance evaluation —Consider several firms operating in the same industry. In such cases, we often find a significant correlation between the performances of the different firms that is due to the commonality of the non-controllable factors (like market fluctuations in input and output prices). At times, the correlation with a given competitor might be particularly high; e.g., the other firm might purchase its raw material from the same source. In this case, the principal will often find it advantageous to base the salary of the manager not only on the performance of her own firm, but also on the performance of the other firms (which is completely beyond the manager’s control).
Incentive schemes for managers base pay on the performance of the firm (or division or plant) to which they are assigned. Since performance depends on a large number of variables beyond managers’ control, these managers may end up bearing a lot of risk.
In other situations—where the firm’s performance is totally independent of their actions, or where their actions can be directly observed by the principal—there is no need to impose so much responsibility on managers. Because the principal can usually absorb risk more easily than managers, it is more efficient to offer managers a fixed salary (and fire the manager if he deviates at all from the course of action expected by the principal).
Basing pay on the firm’s performance is sensible only if this performance is influenced by the manager’s actions, and these actions are not directly observed by the principal. Then the manager must be induced indirectly to choose desirable actions: this is achieved by basing the manager’s pay on whatever variables permit some inference about the manager’s actions. While this may impose risk on the managers, some risk/responsibility is necessary to provide managers with appropriate incentives to work hard. In these situations, the manager may be evaluated on the basis of variables beyond his control.
To implement incentive schemes, organizational performance must be decomposed into components that reflect, as nearly as possible, individual contributions. Responsibility accounting is a system of accounting that recognizes various responsibility centers throughout the organization. Such an accounting system reflects the plans and actions of each of these centers by allocating particular revenues and costs to those having the pertinent responsibilities (also called profitability accounting and activity accounting).
Informal interpretation of risk aversion: Holding the expected amount of ↩