bsp1005 lecture notes 12 - asymmetric information

1

Lecture Note 12

ASYMMETRIC INFORMATION

Semester II, AY2013-2014NUS Business School

BSP1005 Managerial Economics

By Jo Seung-Gyu

Outline

Introduction Adverse Selection

Signaling and Screening Introduction Case of Auto Insurance Job Market Signaling Guarantees and Warranties Other Signals

Moral Hazard─ Introduction─ Principal-Agent Problem─ Compensation Scheme 1: Straight Wage─ Compensation Scheme 2 Bonus Plan─ Compensation Scheme 3; Revenue Sharing

2

● asymmetric information Situation in which a buyer and a seller possess different information about a transaction.

3

Throughout most of our lectures thus far, we have assumed that all the economic agents – consumers, producers and governments –have perfect information about the economic environments that are relevant for the choices they face. However, in many transactions, one party may have better information than the other – i.e. there may be asymmetric information.

Transactions in markets with asymmetric information can fundamentally alter how markets perform, yielding some type of inefficiencies.

INTRODUCTON

4

Suppose a bum on the street approaches to you offering Rolex watches for sale.

.

How much would you pay?─ The amount you would like to pay… it self-explains

why the watches must be all fakes

Example

Introduction – cont.

5

Two types of information asymmetry

(i) Hidden Information (or Hidden Characteristics)When one side of a transaction knows some characteristic of itself than the other does not.

- A bum at Orchard offering a golf jewelry for sale- Quality of used car- Health condition for life insurance- Driving habits for auto insurance

The potential inefficiency problem caused is called ‘Adverse Selection’.

(ii) Hidden ActionWhen one side of a transaction can take an action that affects the other side but which the other side cannot directly observe.

- Driving less carefully once insured - Auto repairman asks you who pays for the job…- Tenured professors are poor in research

The potential inefficiency problem caused is called ‘Moral Hazard’.

Outline Introduction

Adverse SelectionSignaling and Screening

Introduction Case of Auto Insurance Job Market Signaling Guarantees and Warranties Other Signals


6

Adverse selection refers to a situation where a selection process under asymmetric information results in a pool of individuals with economically undesirable characteristics.

ADVERSE SELECTION

7

Recall the Orchard bum who was eager to sell you the watch at, say, $200.

Ever wary, you voice some doubt about its genuineness.

“The bum responds by throwing in a gold ring and offers the package for $100.

Seeing this, you would be like, “No, thank you.”

You think, “at that price they couldn’t possibly be real gold.”

The fact that the bum was so eager to sell the Rolex at such a low price is a sure indicator that the jewelry was fake.

Introduction: a street bum selling Rolex watches

Introduction-cont.

8

Now, suppose there was a bum who had a windfall of picking a genuine Rolex. Excited, he wants to sell it on the street.

He would let it go at $1,000, a real good bargain. Yet no one would show interest. It is a real Rolex! Yet you won’t even buy it at one hundredth of the price. Knowing this, the bum won’t even try to sell it and instead carries it around.

Result? No genuine Rolex on the street at all!

The key feature is that one side – the bum – had the information on watch while the other side – you – did not. And as a result,

Such phenomenon is called Adverse Selection.

Bad products drive out good products!

Introduction-cont.

9

Adverse selection refers to a situation where a selection process under asymmetric information structure – in this asymmetric information structure, characteristics or types of the informed side are hidden, not actions – results in a pool of individuals with economically undesirable characteristics.

It is sometimes called “a lemon problem”, called after the original paper “The market for lemons” by George Akerlof who first addressed the issue


10

The Market for Lemons (Case of Used Cars)

George Akerlof, “The market for Lemons: Quality Uncertainty and the Market Mechanism”, Quarterly Journal of Economics, 84, pp 488-500.

Suppose two kinds of used cars are available—high-quality cars and low-quality cars. Also suppose that both sellers and buyers can tell which kind of car is which. There will then be two markets.

In reality, the seller of a used car knows much more about its quality than a buyer does. (Buyers discover the quality only after they buy a car and drive it for a while.)

When making a purchase, buyers therefore view all cars as “medium quality,” in the sense that there is an equal chance of getting a high-quality or a low-quality car. However, fewer high-quality cars and more low-quality cars will now be sold.

As consumers begin to realize that most cars sold (about three-fourths of the total) are low quality, their perceived demand shifts. This shifting continues until only low-quality cars are sold.

11

Left: The demand curve for high-quality cars is DH. However, as buyers lower their expectations about the average quality of cars on the market, their perceived demand shifts to DM.

Right: The perceived demand curve for low-quality cars shifts from DL to DM. As a result, the quantity of high-quality cars sold falls from 50,000 to 25,000, and the quantity of low-quality cars sold increases from 50,000 to 75,000.Eventually, only low quality cars are sold.

Introduction-cont.


12

The Market for Credit

How can a credit card company or bank distinguish high-quality borrowers (who pay their debts) from low-quality borrowers (who don’t)? Clearly, borrowers have better information—i.e., they know more about whether they will pay than the lender does.

Again, the lemons problem arises. Low-quality borrowers are more likely than high-quality borrowers to want credit, which forces the interest rate up, which increases the number of low-quality borrowers, which forces the interest rate up further, and so on.

In fact, credit card companies and banks can, to some extent, use computerized credit histories, which they often share with one another, to distinguish low-quality from high-quality borrowers. Many people, however, think that computerized credit histories invade their privacy. Should companies be allowed to keep these credit histories and share them with other lenders?

We can’t answer this question for you, but we can point out that credit Histories perform an important function: They eliminate, or at least greatly reduce, the problem of asymmetric information and adverse selection—a problem that might otherwise prevent credit markets from operating. Without these histories, even the creditworthy would find it extremely costly to borrow money.


13

The Market for Health Insurance

People who buy insurance know much more about their general health than any insurance company. As a result, adverse selection arises, much as it does in the market for used cars.

• Because unhealthy people are more likely to want insurance, the proportion of unhealthy people in the pool of insured people increases. This forces the price of insurance to rise, so that more healthy people, aware of their low risks, elect not to be insured. This further increases the proportion of unhealthy people among the insured, thus forcing the price of insurance up more. The process continues until most people who want to buy insurance are unhealthy.

• At that point, insurance becomes very expensive, or—in the extreme—insurance companies stop selling the insurance.

One solution to the problem of adverse selection is to pool risks.

• For health insurance, the government might take on this role, as it does with the Medicare program. By providing insurance for all people over age 65, the government eliminates the problem of adverse selection.

• Likewise, insurance companies offer group health insurance policies at places of employment. By covering all workers in a firm, whether healthy or sick, the insurance company spreads risks and thereby reduces the likelihood that large numbers of high-risk individuals will purchase insurance.


14

Think from the Reading (‘More Sex is Safer Sex)

Promiscuity vs Self-restraint:

Which one is more sinful for the epidemic of AIDS?

• Two types of patrons in a club:

the high-risky & the low-risky

• Public campaign for self-constraining

would affect the low-risky more than the high-risky

• Thus, public campaign should induce the low-risky to be more active.

Examples?

Outline Introduction Adverse Selection

Signaling and Screening Introduction Case of Auto Insurance Screening Job Market Signaling Guarantees and Warranties Other Signals


15

SIGNALING AND SCREENING

16

The informed side may want to send a signal about his/her own type, and when it is believed to be credible such signal can resolve the adverse selection problem(Signaling). Typical examples are guarantees or warranties.

The uninformed side may also provide an appropriate incentive scheme which should be clever enough so that the informed side reveals true types (Screening). Typical examples are different insurance premium based on something observable, like no-accident discount.

Signaling and screening usually – not necessarily – coexist.

Signaling and Screening – cont.

17

Screening: Case of Auto Insurance

If insurance company can’t distinguish good drivers and lousy drivers, it will have to base premium on the average experience. Then those with low risk will choose not to insure, raising the accident probability of the pool and rates. Lousy drivers drive out good drivers and adverse Selection problem may occur again.

Possible solutions? Compare the following two policies:

Policy A: This policy has a very high initial premium, but if the purchaser does not have accidents the premium will drop substantially in subsequent years. If the policyholder does have an accident, the premium will remain very high.

Policy B: This policy is priced lower than policy A, but regardless of whether or not policyholder has accidents, the premium will not fall.

Good drivers would prefer A while lousy drivers would prefer B. The smart policy design can help the drivers self-select.


Signaling and Screening Introduction Case of Auto Insurance

Job Market Signaling Guarantees and Warranties Other Signals


18

19

Signaling in Job Markets

• Michael A Spence , ‘Job Market Signaling,’ Quarterly Journal of Economics, 87, pp. 355-374.

When you hire/interview employee, you are not sure about their true ability. Potential adverse selection risk exists.

• Dressing well for the job interview might convey some information.─ Not a good signal since it can be easily mimicked. To be effective, a

signal must be easier for high-productivity people to give than for low-productivity people to give, so that high-productivity people are more likely to give it.

Nobel Laureate Michael Spence introduced the idea of using ‘education as signal’ to get around the adverse selection problem in job markets.

• More productive people are more likely to attain high levels of education in order to signal their productivity to firms and thereby obtain better-paying jobs.

Job Market Signaling – cont.

20

A scenario

A Model

21

A recruiter is hiring a manager:

• Two types of job candidates: High quality and Low quality with 50:50 chance• Types are known only to the candidates but not to the recruiter.• If hired, the job continues for 10 years.

If candidates’ types are identifiable,

Recruiter is to offer annual salaries as follows (based on their productivities):

• High quality candidates: $20,000/year• Low quality candidates: $10,000/year

If candidates’ types are not identifiable?

Due to incompleteness of information, all candidates may pretend as if they were of high quality firm’s profits lower than expected firm recruiter would offer lower salary only low quality candidates will apply: ‘adverse selection’ problem arises!



22

Education as a Signal

Now you can offer a high quality candidate salary ($20,000/year) if she studies for more than N* school program years in higher education (after college)

• Disutility (or Cost) per class

High quality people find education easier, therefore less costly to them than to low quality people

– High quality candidates: $20,000 per higher school year

– Low quality candidates: $40,000 per higher school year

For the above mechanism to sort out the candidates’ types, the following conditions needs to hold:

23

(a) Incentive Compatibility (IC) for Low Quality Candidates

Low-quality candidates should not have incentive to send the signal.

That is,

ΠL(study N years to mimic high type) = $200,000 – $40,000N

< ΠL(no higher education) = $100,000

or

(wage gains from mimicking high quality)= $200,000 - $100,000

= $100,000

(Cost of mimicking high quality)= $40,000N

<

Then, N should be: N > 2.5


24

(b) Individual Rationality (IR) Condition for High Quality Candidates:

High-quality candidates should have an incentive to send the signal.

That is,

ΠH(study N years to signal the true type) = $200,000 – $20,000N

> ΠH(no higher education) = $100,000

or

(wage gains revealing High-quality type)

= $200,000 - $100,000= $100,000

(Cost of mimicking high –quality)= $20,000N

>

Then, N should be: N < 5


25

Equilibriums

We will have two kinds of equilibriums, depending on the recruiter’s requirement for the length of N*.

Separating Equilibrium (Candidates’ types are revealed.)

– 2.5 < N* < 5, then only high quality candidates get higher education.

Pooling Equilibrium (Candidate’s types are not revealed.)

– N*> 5 neither type gets higher education

– N*< 2.5 both types gets higher education



Signaling and Screening Introduction Case of Auto Insurance Job Market Signaling

Guarantees and Warranties Other Signals


26

Guarantees and Warranties

27

Consider the markets for such durable goods as televisions, stereos,cameras, and refrigerators. Many firms produce these items, but some brands are more dependable than others. If consumers could not tell which brands tend to be more dependable, the better brands could not be sold for higher prices. Firms that produce a higher-quality, more dependable product must therefore make consumers aware of this difference. But how can they do it in a convincing way?

The answer is guarantees and warranties.

• Guarantees and warranties effectively signal product quality because an extensive warranty is more costly for the producer of a low-quality item than for the producer of a high-quality item. The low-quality item is more likely to require servicing under the warranty, for which the producer will have to pay. In their own self-interest, therefore, producers of low-quality items will not offer extensive warranties.

• Thus consumers can correctly view extensive warranties as signals of high quality and will pay more for products that offer them.

Other Signals?

• You want to sell your 2 year old car through a newspaper classified section.

o What’s the buyers’ concern? o What’s your concern? How would you get out of the dilemma?

“I am moving out of the country” “My wife/mom says this sports car is dangerous”

• Can you tell why your ‘not so successful’ salesman friend is in an Armani?

‘More sex is safer sex’ from the reading

28




29

MORAL HAZARD

30

What is moral hazard problem?

Moral hazard occurs when the insured party whose actions are unobserved can affect the probability or magnitude of a payoffassociated with an event.

If my home is insured, I might be less likely to lock my doors or install a security system

When your car is insured, you would be less cautious in passing the red Mustang on AYE.

Introduction

Introduction-cont.

31

Example: Fire Insurance

You own a warehouse worth $100,000.

Probability of a fire is knows to be

– 0.005 with a fire prevention programcosting $50 to run

– 0.01 without the program

• With the program the actuarially fair premium is: 0.005 x $100,000 = $500

• Once you purchase the insurance, you no longer have an incentive to run the program, therefore the probability of loss becomes .01.

• $500 premium will lead to a loss to the insurance firm because the expected loss is now $1,000 (.01 x $100,000)

Introduction-cont.

32

In a study of 725 hospitals, from 14 major hospital chains,

Example: Costs for For-profit Hospitals vs Nonprofit Hospitals

After adjusting for services performed, the average cost of a patient day in nonprofit hospitals was 8 percent higher than in for-profit hospitals.

Without the competitive forces faced by for-profit hospitals, nonprofit hospitals may be less cost-conscious and therefore less likely to serve appropriately for the society.

33

Challenge: How would you solve this morally-hazardous behaviors?

Insurance─ Deductibles, co-payment etc

Non-profit hospitals─ Introduce private sector schemes

After all, it is all about adding designing the right incentive scheme.

─ It should be in the best interest of the informed side only to take the desirable action..



Moral Hazard─ Introduction

─Principal-Agent Problem─ Compensation Scheme 1: Straight Wage─ Compensation Scheme 2 Bonus Plan─ Compensation Scheme 3; Revenue Sharing

34

PRINCIPAL-AGENT (P-A) PROBLEM

35

A firm often can be understood as a relation between principal and agents

principal Individual who employs one or more agents to achieve an objective.

agent Individual employed by a principal to achieve the principal’s objective

● principal–agent problem Problem arising when agents (e.g., a firm’s managers) pursue their own goals rather than the goals of principals (e.g., the firm’s owners).

Principal-Agent Model in Corporate

36

The traditional corporate governance structure is straightforward. Shareholders own the firm’s assets and assume the risks of doing

business. Shareholders hire managers (agents) to perform the duties of running the

business.

Principals determine rules that assign agents compensation as a function of principal’s observation of the firm performance. But there is asymmetric information here: agents (managers) have more information about the action relative to the principal. This asymmetric information consists of two basic problems:

The agent’s action is not directly observable by the principal The outcome of the action is not completely determined by the agent’s

actions

And then the result may be:

There is a possibility for the agent to pursue their own goals, even at an expense of owners: Moral Hazard Problem (Or Hidden Action Problem)

Principal-Agent Model in Corporate

37

The most common hidden action problem in the corporate world is determining the effort of agents.

Effort has a disutility to the agent, but a value to the principal because it increases the probability of a favorable outcome (higher profit).

Principal however cannot observe agents’ efforts.

A P-A model

38

Revenue is determined jointly by agent’s effort and market states. Agent’s effort – Low (EL) or High (EH) Three possible market states with 1/3 each

Principal cannot directly observe agent’s effort.

Effort Cost to the Agent

The cost of effort = $0 if low effort$4,000 if high effort

.

State 1 (s1) State 2 (s2) State 3 (s3) Expected Revenue EL = low effort 5,000 10,000 15,000 10,000 EH = high effort 10,000 30,000 50,000 30,000

Assume both principals and agents are all risk-neutral.

Then the situation can be depicted through the following game:

39

Offers a Contract (Wage Scheme) EL EH

(L, EIL) (, EIH)

P

A

E = expected profit for the principalEI = expected income for the agent

Wage (W) = $9,000 (straight wage regardless of revenue outcome)

40

Agent’ Expected Income

EIL = 9,000 – 0 = 9,000EIH = 9,000 - 4,000 = 5,000

Principal’s Expected ProfitEL = 10,000 – 9,000 = 1,000EH = 30,000 – 9,000 = 21,000

Compensation Contract 1: A Straight Wage

P-A problem then can be viewed as below:


41

Wi = 9,000 ; i = L, H

EL EH

() ()

P

A

In equilibrium:

• Agent would exert low effort and the principal gets a lower payoff.

Compensation Contract 1: A Straight Wage – cont.

We would have to design an incentive scheme to induce the agent to exert high effort.

If Revenue < 30,000, then W = $8,000.If Revenue ≥30,000 , then W = $8,000 plus a bonus of $6,500.

42

Agent’s Expected Income

EIL = 8,000 – 0 = 8,000. EIH = [8,000 – 4,000](1/3) + [8,000 + 6,500 – 4,000](1/3)

+ [8,000 + 6,500 – 4,000](1/3) = 8,333.33

Principal’s Expected ProfitEL = (5,000-8,000)(1/3) + (10,000-8,000)(1/3) + (15,000-8,000)(1/3)

= 2,000 EH = (10,000-8,000)(1/3) + (30,000 – 8,000 – 6,500)(1/3)

+ (50,000 – 8,000 – 6,500)(1/3) = 17,666.67

Compensation Contract 2: A Bonus Plan



43

In equilibrium:

• Agent would exert high effort.

Compensation Contract 2: A Bonus Plan- cont.

W = 8,000, if Revenue < 30,000 = 8,000 + 6,500, otherwise

EL EH

(2,000 ; 8,000) (17,666.67 ; 8,333.33)

P

A

Moral hazard problem disappears!

If Revenue > 14,500, then W= 0.95(gross revenues – 14,500)Otherwise, then W = 0

44

Agent’s Expected Income

EIL = (0)(1/3) + (0)(1/3) + 0.95[15,000 – 14,500](1/3) = 158.33EIH = [0 – 4,000](1/3) + {0.95*[30,000 – 14,500] - 4,000}(1/3)

+ {0.95*[50,000 – 14,500] – 4,000}(1/3) = 12,150

Principal’s Expected ProfitEL = 5,000 (1/3) + 10,000(1/3) + {15,000 – 0.95[15,000 – 14,500]}(1/3) = 9,841.67EH = 10,000(1/3) + {30,000 – 0.95[30,000 – 14,500]} (1/3)

+ {50,000 – 0.95[50,000 – 14,500]} (1/3) = 13,850

Compensation Contract 3: Revenue Sharing



45

In equilibrium:

• Again, agent would exert high effort.

Compensation Contract 3: Revenue Sharing - cont.

Moral hazard problem disappears, again!

W = 0.95(R – 14,500) if Revenue >14,500

= 0 Otherwise

EL EH (9,841.67 ; 158.33) (13,850 ; 12,150)

P

A

An episode. …

• ‘Where is all food gone?’ from the reading

46

Thank you!

47

Some Information on the Final Exam

• Closed book exam, but one A4-size cheat sheet is allowed.

• Two hour exam and four questions altogether.─ two from Prof Yang (Q1 and Q2)─ two from me (Q3 and Q4)

• No MCQs this time.

• Questions will be about both concepts and quantitative work. When answering, read the instructions carefully. Some questions require explanations and some don’t.

• You can use a calculator during the exam.

48

bsp1005 lecture notes 12 - asymmetric information

Documents