Economics, Organization and Management
Chapters 7 & 11

Milgrom, P. and J. Roberts (1992). Economics, Organization and Management. Englewood Cliffs, NJ: Prentice Hall, chapters 7, 11 + pp. 154-61, 408-13.

Chapter 7: Risk Sharing and Incentive Contracts

·  A response to moral hazard problems are incentive contracts

o  Incentive contracts are where individual incentives are strengthened by holding people at least partially responsible for the results of their actions, even though doing so puts them to risks that could be more easily borne by insurance companies

Incentive Contracts as a Response to Moral Hazard

·  Actual insurance contracts are also incentive contracts

o  Contain provisions that restrict and condition claim payments that provide better incentives than full insurance without removing the essential part of coverage

§  E.g. deductible clause common to homeowners’ fire and theft insurance policies requires them to bear the initial part of any loss but protecting them against large financial losses

§  Automobile insurance is experience rated

§  Health-insurance policies often require copayments

·  Features designed to encourage insureds to take care and to deter their excessive use of the insurance

·  Similar moral hazard issues arise when devising compensation contracts for employees in a firm

·  Desirable to hold employees responsible for their performance but subjects them to risk in current or future incomes

·  Efficient contracts balance the costs of risk bearing against the incentive gains that result

Sources of Randomness

·  Having pay depend on performance would not generate risk-bearing costs if employees’ performance could be determined precisely and they performed as required

·  Perfect measures of behaviour are hardly ever available, exposing employees to risk with PRP

·  Compensation based on results instead of care and effort could be more effective

·  Results are frequently affected by random factors outside of employee control

o  E.g. population growth, poor maintenance, design flaws, etc.

·  Second source of randomness arises when performance itself is measured instead of the result, but performance evaluation measures could be random/subjective

o  E.g. supervisor’s subjective perception of employee’s attitude

·  Third source of randomness from the possibility that outside events beyond the control of the employee may affect their ability to perform as contracted

o  E.g. health problems of the employee, family problems, or weather and traffic conditions

Balancing Risks and Incentives

·  Eliminating risk is possible through making compensation completely unrelated to performance or outcomes

·  But in this case, no rewards for good behaviour or punishments for bad

·  Important to understand how rational people act in risky situations to analyse how rational people respond to incentives in insurance-like contracts:

  1. Describe risks precisely with statistics
  2. Describe how rational people, acting individually, can choose consistently among risky choices
  3. Examine how groups of people can share risks and form insurance pools

Risk Sharing and Insurance

How Insurance Reduces the Cost of Bearing Risk

·  Many kinds of institutions that assist in sharing risks including insurance companies

·  Enables companies to reduce individual risks greatly

o  If risks are independent and number of policyholders quite large, risks are effectively eliminated and insurance works very well

o  E.g. risk that you suffer an automobile accident is very nearly independent of the risk that any other particular person will do so, therefore automobile insurance is a feasible enterprise

·  Insurance companies specialise in evaluating individual risks and they can reduce the cost of risk bearing by pooling the risk-bearing capacity of policyholders

·  But some risks are too large and pervasive that they cannot be made negligible by sharing and cannot be managed by traditional insurance agreements

o  E.g. oil price increase would have such widespread effects, reducing the effective incomes of most people in oil-consuming countries, that no amount of risk sharing could insulate them from the loss

·  General risks are shared through other markets especially in financial markets

o  E.g. investment risks taken by firms, such as those associated with a new technology; risk of failure of the technology is borne by shareholders and this capacity for risk sharing reduces the firm’s cost of financing the investment, helping to promote technical change

Logic of Linear Compensation Formulas

·  Commonly observed in the form of commissions paid to sales agents, contingency fees paid to attorneys, piece rates paid to tree planters or knitters, crop shares paid by sharecropping farmers, etc.

·  When sales targets used they are often set to cover short periods of time so that the periods during which incentives are too low are not extended ones

·  Makes compensation of additional sales efforts more nearly equal over time

·  Linear systems are also simple to understand and administer

The Informativeness Principle

·  In designing compensation formulas, total value is always increased by factoring into the determinant of pay any performance measure (with appropriate weighting) that allows reducing the error with which the agent’s choices are estimated and excluding measures that increase error (random factors outside agent’s control)

·  Comparative performance evaluation according to which the compensation of an employee (typically manager or executive) depends not just on their own performance but on the amount it exceeds or falls short of someone else’s performance

·  When is comparative performance evaluation a good idea?

o  E.g. in automobile insurance, collision coverage is insurance that pays the owner of an automobile when their own auto is damaged in a collision vs. comprehensive damage coverage is insurance that pays for damage to the persons’ automobile when it is stolen or damaged by other means (environmental etc.)

o  Both kinds of coverage usually work by specifying a deductible which is the portion of the loss that the insured person must pay before any payment is due from the insurance company

o  If a car owner takes care of his car but in the case of a collision or theft, the owner has no control over the size of the loss suffered

o  Size of loss provides no information about the care taken by the owner

o  According to informativeness principle, the owner’s contribution toward any loss should not depend on the size of loss but only on the most informative performance indicator

o  In an optimal insurance contract, the owner’s contribution should not depend on the size of the loss but should be a fixed contract

The Incentive-Intensity Principle

·  The optimal intensity of incentives depends on four factors: incremental profits created by additional effort, precision the desired activities are assessed, agent’s risk tolerance and agent’s responsiveness to incentives

·  E.g. counterproductive to use incentives to encourage production workers to work faster when their production already exceeds what the next stage in the production line can use

·  More risk averse agents ought to be provided with less intense incentives

·  Low precision of measurement leads to high variance, thus weak incentives should be used; wage incentives should only be used when good performance is easy to identify

·  Incentives should be most intense when agents are able to respond to them

o  E.g. in Japanese practice, the amount paid by a manufacturing firm for its inputs depends on the actual costs as measured in the supplier company’s accounting records rather than being a contractually fixed price

o  Evidence obtained is consistent with the theory: incentive contracts for Japanese suppliers do appear to depend on the considerations identified by the theory

Monitoring Intensity Principle

·  Comparing two situations, one with B set high and another with it set lower, we find that V (variance) is set lower and more resources are spent on measurement when B is higher: when the plan is to make agent’s pay very sensitive to performance, it will pay to measure performance carefully

Equal Compensation Principle

·  When there are several activities being conducted, the employer will be concerned that employees allocate their time and efforts correctly among the various activities

o  E.g. marketing representatives for a company making speciality steel alloys perform several activities i.e. solicit business from new customers, problem-solving services and advice to customers, gather information, etc.

o  If the firm were to only base marketing representatives on accurately measured current sales figure, it might induce distortion in their behaviour so they switch their efforts towards the immediate high-payoff activity

·  Profit and reputation goals can be in conflict

·  Equal compensation principle: if an employee’s allocation of time/attention between 2 activities cannot be monitored by the employer, then either the marginal rate of return to the employee from time or attention spent in each of the two activities must be equal or the activity with the lower marginal rate of return receives no time or attention

Application: Cost Centres and Profit Centres

·  Important part of designing incentives is to determine the employees’ performance measures

o  E.g. manager of a manufacturing facility that deals with the factory as a cost centre or a profit centre

o  2 activities that the manager must handle are cost reduction and revenue generation

o  Equal compensation principle implies that if the factory manager is to be provided with sales-generation incentives at all, then the incentives need to be of the same strength as those for manufacturing cost control

Application: Incentives for Teachers

·  According to equal compensation, if it is desirable to have teachers devote some efforts to each of many activities, and if it’s possible to distinguish the effort put into each, then all kinds of efforts must be compensated equally

·  Responsibilities and compensation should be determined together

·  One proposal for teachers would be to install a system of specialist teachers who are compensated based on student test scores but not responsible for other aspects of student performance

·  Determining the job design and compensation together can sometimes solve problems that cannot be solved by compensation policy alone

Moral Hazard with Risk-Neutral Agents

·  Agent can be perfectly motivated at zero cost by setting b=1, or making them bear the entire risk

·  In the case of automobile insurance, it amounts to having drivers paying full cash compensation to those who they have damaged

Problems with the risk-neutral agent scenario

·  Solution will fail whenever the agent lacks sufficient funds

·  Solution will also fail when the risk is nonfinancial and is therefore difficult/impossible to transfer

Summary

·  Several principles govern design of optimal incentive contracts

  1. Informativeness principle: cost of providing incentives increases with the variance of the estimator of the employee’s effort
  2. Incentive-intensity principle: strength of incentives should be an increasing function of the marginal returns to the task, the accuracy with which performance is measured, the responsiveness of the agent’s efforts to incentives and the agent’s risk tolerance
  3. Monitoring intensity principle: more resources should be spent on monitoring when it is desirable to give strong incentives
  4. Equal compensation principle: if an employee’s allocation of time and effort between alternative tasks cannot be monitored by the employer, then the marginal returns earned by the employee in any tasks to which they actually devote effort to must be equal
  5. Ratchet effect: the practice of basing performance targets on past performance in the same activity

·  Various factors not included in models that would make the principal-agent problem hard to solve: agent may lack sufficient capital, some losses are nonfinancial or private information might be had by the agent

Chapter 11: Internal Labour Markets, Job Assignments and Promotions

·  Internal labour markets: long-term employment relationships, limited ports of entry for hiring, career paths in the firm and promotions from within

·  May be more than one internal labour market within a single firm

Labour Market Segmentation Patterns

·  ‘Internal labour markets’ defined by Peter Doeringer and Michael Piore in 1970s

o  Distinguished between primary and secondary sectors in the economy

o  Primary sector: skilled blue-collar work, most white-collar positions and technical, managerial, and professional employment

o  Secondary sector: unskilled, manual-labour, blue-collar jobs; low or unskilled service positions (janitors, check-out clerks, waiters); low or unskilled white-collar positions (office mail-room workers, filing clerks); and migrant, part-time, and seasonal workers

o  Internal labour markets are common but not universal in the primary sector and absent in the secondary

·  Not all primary sector jobs are in internal labour markets

o  E.g. medical doctor in private practice is not in any internal labour market but not in secondary sector either

·  Primary sector redefined later; subdivided into differences in level of entrance standards, promotion and turnover rates, criteria for promotions, and forms of control

Fairness and Efficiency

·  Assignment of some positions in the labour market can be approached as an equity or justice issue

·  Fairness not the only issue, where skills and abilities are needed for specific jobs; efficiency matters too

·  Question of whether assignment of people divided among segmented labour markets reflects and promotes efficiency or not

Pay in Internal Labour Markets

·  Internal labour markets typically has some insulation of compensation within the firm from external market forces

·  At minimum, firms must compete with other employers at the ports of entry

Job Classifications and Pay

·  “Wages attach to jobs rather than individuals”

·  Fairly narrow range of pay specified for any job in the internal labour market

·  What any particular employee is paid is then determined by his job assignment rather than individual productivity or opportunity costs

o  E.g. at least for lower-level jobs, many firms have very explicit salary scales, and the only way to get a raise once an employee’s pay hits the top of the scale is to get a promotion

o  E.g. Stanford University’s practices on job classifications and pay where non-union staff positions are classified into 27 levels and pay is attached to a classification level

Caveats and Qualifications

·  Even within context of job classification scheme there are still differences in pay between people assigned to the same job

o  E.g. Stanford’s maximum salary corresponding to a job classification is on the order of 50% more than minimum for that rank; leaves quite a bit of room within any single job classification for merit pay or seniority