Professional Documents
Culture Documents
By the end of this chapter, and having completed the Essential readings and activities, you should
be able to:
describe the inefficiencies that may arise in the presence of asymmetric information
discuss and analyse simple cases of adverse selection, with and without the presence of
possible signals
discuss and analyse the problem of adverse selection in insurance markets
explain the conditions under which signalling can overcome the adverse selection
problem
define the moral hazard problem and give examples of situations in which it may arise
discuss and analyse the principal–agent problem and suggest appropriate solutions.
Essential Reading
1. General Overview
We say that there is asymmetric information if we are in a situation in which one party
has more information than another when both parties are involved in the same
transaction.
For example, if we consider an insurance problem; the buyer of motor insurance has
better information about his driving ability and the ‘care’ he takes to prevent accidents
than the insurance company.
There are two types of information that an economic decision-maker (the less informed
party) might lack but desire.
In the first case, one party knows the characteristic (e.g. driving ability) of himself that
the other does not – a situation of hidden characteristics. This problem is referred to
as the Adverse Selection problem. In general, the adverse selection problem occurs
before the transaction takes place.
In the second case, one party can take hidden actions that the other party cannot
observe. This problem is referred to as the Moral Hazard problem. In general, the moral
hazard problem occurs after the transaction takes place.
We shall focus on a few economic terms in asymmetric information and a few models
that will explain these terms.
2. Adverse Selection
Adverse selection refers to the situation that the informed party knows about his own
hidden characteristics that the uninformed party do not know.
This will cause inefficiencies. To illustrate this, let us consider the Akerloff’s Lemon and
Plum model.
The sellers are informed about their quality of the cars that they are selling. However, the
buyers do not know the quality of the cars that they are buying.
Suppose that the good quality car sellers are willing to sell their cars for a minimum of
$20,000. Their supply function is shown below.
Price of Cars
20,000
N Quantity of Cars
Suppose that the bad quality car sellers are willing to sell their cars for a minimum of
$10,000. Their supply function is shown below.
Price of Cars
10,000
N Quantity of Cars
Buyers will be willing to buy a bad quality car for $12,500 and a good quality car for
$25,000.
Their demand functions for both bad and good quality cars are shown below.
Price of Cars
25,000
Demand for good cars
12,500
Demand for bad cars
2N Quantity of Cars
As a result, there will be two market equilibriums, one for good cars, and the other for
bad cars.
Price of Cars
25,000
Demand for
good cars
20,000
Supply for
good cars
N 2N Quantity of Cars
Hence the trading price for good cars will be $25,000 with N good cars traded.
Price of Cars
12,500
Demand for
bad cars
10,000
Supply for
bad cars
N 2N Quantity of Cars
Hence the trading price for bad cars will be $12,500 with N bad cars traded.
This is a separating equilibrium where the good and bad cars are separated into two
different markets.
This is the issue of adverse selection as the buyers do not know if the cars are good or
bad (hidden characteristics).
The price they are willing to pay for the good cars is 25,000 and the price they are willing
to pay for the bad cars is 12,500. As they cannot differentiate between the good and
bad cars, their expected price to pay for any car will be 0.5(25,000) + 0.5(12,500) =
$17,500.
To calculate expected price, the valuations are multiplied by 0.5 as there are N good car
sellers and N bad car sellers. That is the proportion of good car sellers is 0.5 and the
proportion of bad car sellers is also 0.5.
20,000
17,500
Demand for
cars
10,000
Supply for bad cars
N 2N Quantity of Cars
The supply for good cars does not intersect the demand for cars. As a result, no good
cars are traded in equilibrium.
Over time, the buyers realise that no good cars are traded in equilibrium. As a result,
they know that only bad cars are traded in equilibrium. Their willingness to buy becomes
12,500 where only the bad cars are traded in equilibrium.
The diagram illustrated below shows the final equilibrium for the car market.
This is inefficient as the good cars are not traded and only the bad cars are traded in
equilibrium.
The bad car sellers receives producer surplus while the good car sellers receive none.
Price of Cars
12,500
Demand for
bad cars
PS
10,000
Supply for
bad cars
N 2N Quantity of Cars
Activity 12.1
2013 ZA Question 8
If lenders cannot observe the quality of projects of borrowers, the usual competitive market
supply logic of lending more at higher interest rates does not always hold. Is this true or false?
Explain your answer.
Consider two individuals, Andy being a high-risk driver, and Ben being a low-risk driver.
Initial wealth for both drivers is denoted by W.
Andy faces an accident (loss of D) with probability PH and Ben faces an accident (loss of
D) with probability PL. (PH > PL)
The insurance company cannot differentiate between a high-risk driver and a low-risk
driver. This is the issue of adverse selection.
Both Andy and Ben wants to buy actuarially fair insurance. We shall analyse this in the
state contingent commodities model.
No Accident
Accident
Since Andy is a high-risk driver, the gradient of his fair odds line will be which is
steep.
Under actuarially fair insurance, the insurance company will be able to identify Andy as a
high-risk driver. Andy will fully insure in this case. His indifference curve will be tangent to
his fair-odds line while intersecting the 45° line.
At the tangent point, we can see Andy will be fully insured as he will get the same amount
regardless if the accident occurs or not.
Similarly for Ben (low-risk driver), without asymmetric information and with actuarially fair
insurance, the insurance company will be able to identify Ben as a low-risk driver. Ben will
fully insure in this case. (gradient of fair odds line is .
No Accident 45°
Accident
Under Asymmetric Information
Under asymmetric information, the insurance company cannot differentiate between a
high-risk driver and a low-risk driver.
Hence, the expected probability of the accident happening is the average between
and . Therefore, the insurance company sets its insurance policy (the constraint for
the contingent commodities diagram) to be an average between Ben’s fair odds line and
Andy’s fair odds line. The new policy line is not as flat as Ben’s fair odds line. It is also
not as steep as Andy’s fair odds line
No Accident
45°
UB
UA
Accident
Assuming that both individuals are fully insured, both agents’ indifference curve are on the
45° line as well as the new policy line (Point C).
Notice that as the indifference curves intersect the 45 degree line, Andy and Ben’s gradient
of the indifference curve must be equal to the gradient of their fair odds line.
No Accident
45°
UB1
C
Andy’s UB2
Fair
Odds
line
UA2
UA1
Accident
Andy (high-risk driver) will be better off (UA2 > UA1) Ben (low-risk driver) will be worse off
(UB2 < UB1) assuming that they fully insure.
This is inefficient as Ben will not be willing to enter the market. Eventually, only Andy will
buy the insurance. At the C, the premium will be too low if all the customers are high-
risk drivers. Consequently, the insurance company will make losses and the market will
fail. Thus, the pooling equilibrium at Point C cease to exist.
Suppose the assumption of full insurance breaks down, then Andy (the high-risk driver)
no longer chooses Point C. He will choose a point such that his indifference curve
tangent to the new policy line. Consequently, he will over-insured as he will be paying a
higher premium. This is shown below.
No Accident
No insurance
45°
Premium
with Full
insurance
Premium with
Over‐insurance
Andy’s New Policy
Fair
odds UA1 UA2
line
Accident
Suppose the assumption of full insurance breaks down, then Ben (the low-risk driver) will
under insure. This is shown below:
No Accident
45°
Premium:
Under‐insured UB1
Ben’s
Fair
Odds
Premium: Full line
insurance
UB2
New
Policy
Accident
The high-risk drivers will select a policy that they prefer (and that the low-risk drivers do not
prefer) and the low-risk drivers will select a policy that they prefer (and that the high-risk
drivers do not prefer).
This will lead to a separating equilibrium such that the insurance company will be able to
differentiate between the high-risk drivers and the low-risk drivers since both the high-risk
and low-risk drivers choose different policies.
No Accident 45°
B
Andy’s UB3
Fair
Odds
line
UA1
Accident
Andy (the high-risk driver) will prefer Policy A to Policy B. This is because if Andy were to
take Policy B, he will be on a lower indifference curve.
Ben (the low-risk driver) will prefer Policy B to Policy A. This is because if Ben were to take
Policy A, he will be on a lower indifference curve.
Hence with these 2 policies, we will have a separating equilibrium such that the high-risk
drivers will choose A while the low-risk drivers will choose B.
Notice that the high-risk drivers fully insure at A, while the low-risk drivers under insured at
B as B is above 45° line.
However, this is a better outcome than the pooling equilibrium at Point C. We shall compare
the separating equilibrium (A and B) with the pooling equilibrium (C) and show that the
separating equilibrium is better than the pooling equilibrium.
Recall that under the pooling equilibrium, Andy is better off (UA2 > UA1) while Ben is worse off
(UB2 < UB1) and that market will fail over time.
However, with Points A and B introduced as the self-selection device, Ben will choose policy
B while Andy will choose Policy A. The insurance company will charge high premium on Andy
and low premium for Ben. The market will not fail. This is an improvement compared to the
pooling equilibrium.
4. Signalling
As we have seen in the above sections, in a market with hidden characteristics, adverse
selection problems may arise such that good buyers or good sellers end up getting poor
deals or dropping out of the market.
Signalling means that the high quality party may be able to signal their type in a credible
way such that the information between parties will be symmetric. The device will only
work provided the signal is sufficiently costly.
We shall study education as a costly signal that differentiates high-ability workers and
low-ability workers.
Ability is measured by with 5.5 being high ability and 2 being low
ability.
There are the same amount of high ability workers and low ability workers. Hence a
randomly picked worker will have high ability with probability 0.5 and low ability with
probability 0.5.
Personal cost in terms of effort spent to acquire the chosen education is given by .
Notice that higher ability implies lower cost of personal effort for education. Thus for
high-ability workers, education costs 8/5.5 = 1.45. For low-ability workers, education
costs 8/2 = 4.
Notice that education costs nothing if the worker chooses zero education, i.e.,
0 as no personal effort is required.
No Asymmetric information
As firms are making zero profit in perfect competition, firms will pay low ability workers
2 and high-ability workers 5.5. High-ability workers have utility = 5.5
while low-ability workers have utility = 2.
Firms will not be able to differentiate between low ability workers and high ability workers.
Hence, they will pay the expected wage = expected ability = 0.5(2) + 0.5(5.5) = 3.75
Assuming that initially there is no education, the utility of the high-ability worker will be
the same as the utility of the low-ability worker which is w = 3.75.
The high ability worker could have had utility = 5 under no asymmetric information.
However, due to asymmetric information, their utility falls to 3.75. This is inefficient.
Recall that the cost of education for high ability workers is 1.45 and the cost of education
for low ability workers is 4.
High ability workers will choose education if their utility with education is more than 0.
This satisfies his Participation Constraint. i.e.,
8 22.25
Utility with education (high ability) 5.5 1 0.
5.5 5.5
High ability workers will choose education if their utility with education (wage of 5.5 with
signal) is more than their utility without education (wage of 2 with no signal). This satisfies
his Incentive Compatibility Constraint. i.e.,
8 22.25
Utility with education (high ability) = 5.5 1 2 (Utility with no education = 2).
5.5 5.5
Low-ability workers will choose no education if their utility with no education is more than
0. This satisfies their Participation Constraint. i.e.,
Low-ability workers will choose no education if their utility with no education is more than
their utility with education. This is such that they have no incentive to mimic the high
ability workers. This satisfies their Incentive Compatibility Constraint. i.e.,
Hence this is a separating equilibrium such that the high ability will choose education
while the low ability chooses no education. Firms will pay workers with education as a
signal a high wage (w = 5.5) and workers with no education a low wage (w = 2).
Activity 12.2
2011 ZA Question 1
Consider a market in which goods of different qualities are offered for sale and the quality of a
good is the seller’s private information. To solve the adverse selection problem, the government
considers a policy that would force all sellers to offer a warranty against breakdown. Explain
carefully whether this policy would be effective.
5. Moral Hazard
The moral hazard problem arises when a party to a contract engages in post-contractual
opportunistic behaviour because of an information imbalance between parties
After buying insurance, the car owner can engage in post-contractual opportunistic
behaviour of driving carelessly, knowing that he has already been covered by insurance
Similarly, after the worker is hired by the manager, the worker can engage in post-
contractual opportunistic behaviour by not working to his full potential, knowing that he
has already been hired.
One important example of the Moral Hazard problem is also known as the Principal-
Agent problem, which we will analyse in a model below.
His utility is U(w) – e where e is the effort he puts into his work.
The principal does not observe effort (hidden action, moral hazard).
Payoff to the principal is S with probability P1 if agent works and F with probability 1 – P1
if agent works.
Payoff to the principal is S with probability P0 if agent shirks and F with probability 1 – P0
if agent shirks.
P1 > P0.
Success Fail
e=1 P1 1 - P1
e=0 P0 1 - P0
If all agents work and not shirk, the principal can pay a common wage to all agents.
However, knowing that the principal pays a common wage to all agents, and given that
the other agents all work hard, there is a temptation to free-ride on the hard work of the
others and shirk while still receiving a common wage. This is the action of moral hazard.
The principal will pay the wage based on performance depending on if the outcome is S
or F.
The principal will pay wage WS if the outcome is S and WF if the outcome is F.
The agent will work hard only if utility (working hard) > 0 (reservation utility). This satisfies
participation constraint.
The agent will work hard only if utility (working hard) > utility (shirking). This satisfies
incentive compatibility constraint.
However, the principal will induce agents to work hard only if E(profit for working hard) >
E(profit for shirking).
E(Profit for working hard) = P1(S – WS) + (1 – P1)(S – WF) > P0(S – WS) + (1 – P0)(S – WF).
If all these inequalities are satisfied, then all workers will work hard and the moral hazard
problem is solved.
This is because with a common wage, U(shirking) will always be more than U(working
hard).
Activity 12.3
2010 ZA Question 3
Making automobile insurance mandatory for all drivers might increase the number of car
accidents. Is this true or false? Explain your answer.
Activity 12.1
The Key idea here is adverse selection. There are good and bad projects.
If the bad projects fail and get a zero return, then lender will get a zero return as well. As rates
increases, the borrowers with good projects (less risky projects) will drop out of the market due
to their low return as they are unable to make a profit due to the high rates.
However, for the bad projects, they do not care if the rates increase. This is because, even if
the rates do increase, the worst case scenario is such that they will get a zero return and thus
default on their borrowing. Thus the increase in the rates do not affect the bad projects.
Hence as rates increases, the lender’s profit will fall as good projects drop out and more bad
projects will default. Hence, the lender might want to lend less at higher rates.
Activity 12.2
A warranty serves as a signal to buyers and can therefore create a separating equilibrium.
It should be incentive compatible such that only high quality sellers have the incentive to provide
a warranty and low quality sellers have no incentive to provide a warranty. As such, having a
warranty signals to the buyer that the quality is good.
Activity 12.3
If insurance is not properly designed, in this case, mandatory, then it can create a moral hazard
problem. If full insurance is offered, drivers might change their behaviour and take less care
when driving, resulting in a higher number of car accidents.