978-0393919684 Chapter 8 Lecture Note

Unlock access to all the studying documents.

View Full Document

PART THREE—Some Broad Classes of Games and Strategies

CHAPTER 8

Uncertainty and Information

Teaching Suggestions

Section 1

The crucial idea to convey is that reducing risk by sharing or pooling depends on the risks

not being too positively correlated. A good example is the 2008 financial and economic crisis. Too

many of the risks associated with the assets that were created during the previous boom hinged on

one single factor: house prices. When these stalled and then declined, all those assets declined in

value at the same time, a case of near-perfect positive correlation. Sharing and diversification

availed nothing. Political contests also provide good examples of strategies to manipulate the size

and correlation of risk. Favorites and incumbents take safer stances and ideologically more neutral

positions; underdogs and challengers take more extreme or riskier positions and propose novel

ideas and policies in the hope of catching a wave of support.

Section 2

We have found that most students have an intuitive appreciation of information

asymmetries in strategic interactions. A good way to build the ideas in discussion is to ask them to

produce examples from their own experience and observation, and then to guide the discussion to

fit their examples into the conceptual categories of moral hazard and adverse selection. You can

then move on to discuss the strategies—cheap talk, incentive payments, signaling, screening—that

the players in the game could devise to cope with the situation.

Section 3

The key ideas should be intuitive and evident: unsupported verbal statements are fully

credible if the parties’ interests are perfectly aligned, untrustworthy (and to be ignored altogether)

if the interests are totally opposed. A spectrum runs from one extreme to the other, along which

partially credible communication is possible. A good question for discussion may be the credibility

of political promises made during campaigns. The candidates’ interests are not perfectly aligned

with those of the voters, but they are not totally opposed either. A politician who promises a policy

the voters like and then reneges after election will hurt her chances in future reelections or in

contests for higher offices. Small deviations can be explained away and may be acceptable, big

ones less so. A natural quantification of this idea is the fineness of the Crawford-Sobel partition in

cheap talk equilibria. See Vincent P. Crawford and Joel Sobel, “Strategic Information

Transmission,” Econometrica, vol. 50, no.6 (1982), pp. 1431–1451.

Sections 4–6

For more formal examples of signaling and screening other than those in the book, your

sources will depend on your focus. If your course is economics oriented, you can use the standard

example from Spence (education as a signal of quality), available at a relatively accessible level in

many intermediate microeconomics textbooks. If your course is more politics oriented, there are

many examples in Jeffrey Banks, Signaling Games in Political Science (Reading, U.K.: Harwood

Academic, 1991). For science- or biology-oriented courses, you can develop and use the examples

of signaling in evolution, based on the discussion in The Red Queen that is cited in the text.

Another possibility, now that the text is in its fourth edition, is to make use of an example

that appeared in an earlier edition. The second edition of the text used an attacker-defender

interaction to motivate a standard two-player signaling game. That game, found in Section 5 of

Chapter 9 of the second edition, has a structure similar to the Tudor-Fordor example in the current

edition and can be used as a classroom example to help students understand the more complex

components of the analysis in the current Section 6.

Here are some examples of situations in which moral hazard or adverse selection may

arise. Moral hazard: (1) when a firm’s owner can’t observe the effort (or work quality) of a

manager or worker, (2) when an insurance company can’t observe contributory negligence of a

policyholder, and (3) when an insurance company can’t accurately value a loss and must rely on the

policyholder’s valuation. Adverse selection: (1) when an employer doesn’t know the skill level of a

potential employee, (2) when an insurance company doesn’t know the risk level of an applicant for

insurance, and (3) when players don’t know actions available to other players as in a war game.

The health-care system has all kinds of information asymmetries and is a good topic for

discussion to consolidate the ideas of moral hazard and incentive contracts, and of adverse

selection, signaling, and screening. Here is a summary of such a discussion we conducted in class.

The topic proved far too big for one week’s discussion, and we managed to touch on only a few of

the relevant points. That is hardly surprising; numerous experts have struggled for years without

resolving all the issues. You may want to start the conversation when you cover this chapter and

return to it when you get to Chapter 13 and mechanism design, where the analytics of incentive

contracts are covered more fully.

DISCUSSION TOPIC: INFORMATION AND INCENTIVES IN THE HEALTH-CARE

SYSTEM

Information Limitations and Asymmetries

Users (Potential and Actual Patients)

1. Know more about their own health risks (creates adverse selection) and control aspects of

their behavior that affect the risk (creates moral hazard).

2. Don’t know their own diagnosis, and don’t know the appropriate treatment (are poorly

informed consumers). But they do have some information supplied by the media and the drug

companies (which have their own incentives to misinform).

3. Don’t know the providers’ quality or effort for sure (face adverse selection and moral hazard)

but have various sources of partial information, for example, word of mouth, the media, databases

on malpractice suits, and so on.

Providers

1. The doctor knows much more about the patient’s diagnosis and appropriate treatment. But

there is some inherent uncertainty.

2. Doctors and hospitals know much more about their own quality (create adverse selection).

Insurers

1. Don’t know enough about new customers’ health risks and behavior (face adverse selection

and moral hazard).

2. Know the historical success record of providers and therefore have better idea of quality.

Government

1. Can centralize information and reduce adverse selection and moral hazard.

2. Large organizations face limits on their span of control and cannot monitor quality or actions

of all employees, users, and so on. This especially increases moral hazard.

Systemwide Objectives

1.Universal coverage

2.Quality of care

3.Cost control

4.Availability of choice

5.Public-private mix

Different systems of insurance, charges to users, payment to providers, and so on, affect

the participants’ incentives and therefore the achievement of the system objectives differently.

There are trade-offs: a better performance in one dimension may come at the cost of a worse

performance in some other dimension. We looked at just a few examples of this.

Users’ Incentives

Conventional insurance, with 80% or more coverage, gives the users an incentive to

overdemand services. This includes insistence on coverage of very expensive and experimental

treatments. Insurance companies try to restrict coverage but with limited success when faced with

lawsuits.

The problem would be much more severe for prepaid or HMO-type systems, because the

patient’s marginal contribution to the cost of each use is essentially zero. HMOs control this by

limiting coverage to what is available within the group of providers. But that restricts consumer

choice. In national (government-run) systems, queues are used to restrict access to certain

procedures deemed less urgent or essential. This is an even more extreme restriction on choice.

If consumers had to pay a much greater share of the cost of each use, this incentive

problem would be reduced. But then access would be limited by income or wealth even more than

it is now and coverage would be even less universal.

Providers’ Incentives

Under the fee-for-service system, doctors and hospitals have the incentive to overprovide

services. When the patients have insurance coverage, they offer no opposition. For items such as

diagnostic tests, this is worsened by the threat of malprac tice suits. To counter this tendency and to

control budgetary costs, the government (in its Medicare and Medicaid sys tems) keeps the

reimbursement rates low and uses primary providers as gatekeepers for specialists. But this has

reduced the availability of care to people in these systems. Private insurers have instituted prior

approval mechanisms; this has reduced consumer choice and perhaps (at least the doctors claim)

has reduced the quality of care.

Under a prepaid or HMO-type system, the providers can have the opposite incentive—to

shirk or to provide minimal care. This can be controlled using reputation mechanisms in repeated

relationships. For example, insurance companies might cut off doctors with persistent poor

performance. (The profession’s own procedures for detecting and removing poor quality seem to

work only imperfectly and with very long lags.) More generally, success-based compensation

schemes could be used to self-select the most able doctors into specialties.

HMOs have more incentive to offer better preventive care. So should public (national

health) systems, but these are often subject to cost-cutting squeezes.

Doctors, hospitals, and equipment and drug makers have mixed incentives to innovate.

Technical innovations can bring fame and profit. Cost-cutting ideas may be less attrac tive, and also

limited because many procedures need a certain labor time and their costs go up as productivity

and wages in the rest of the economy go up (this is often called Baumol’s disease).

Insurers’ Incentives

Adverse selection can lead to cream skimming. Each insurance company practices

screening by offering a contract that is attractive only to the healthiest individuals. Then a residual

insurer, like Blue Cross or the government, is left with the worst risks and the highest costs. In

fact, employment-based insurance can act as a screening device of this kind, to the extent that a

person needs a certain minimum of good health to get and hold a full-time job.

Cream skimming and the resulting loss of coverage or increase in cost to residual groups

can be avoided by requiring pooling: each plan has to have certain basic provisions and to accept

all comers. A nationalized system does this automatically, enrolling everyone into the same pool at

birth.