ECON6021 (Nov 17&19)
Information Asymmetry
Informational economics
 When a person buys medical insurance, the insuring
company does not know whether the person is healthy.
Nor does it know how well he is going to take care of
himself after buying insurance.
 The former type of asymmetry information is called a
hidden type problem, or adverse selection problem.
 The latter type of asymmetric information is called a
hidden action problem, or moral hazard problem. But
the notion of moral hazard has subsequently expanded.
 Information economics is the study of decision makings
between agents when their information is asymmetric.
Adverse Selection
Why called “Adverse Selection”?
 Adverse selection refers to a situation where a selection
process (here market) results in a pool of
products/individuals with economically undesirable
characteristics.
 With “hidden type”, either (1) bad products drive out
good products or (2) good products subsidize bad
products (both receive the same price).
 Gresham’s law: bad money drives out good. Or, where
two media of exchange come into circulation together
the more valuable will tend to disappear.
Adverse selection: Used
Cars (Lemons) Market
good cars bad cars
buyers' valuation
$30K
$20K
sellers' valuation
$25K
$10K
number of cars
100
200
number of buyers
infinite
Assumption: all of the above is commonly known in the
following exercises.
Scenario I: Full Information
good cars bad cars
buyers' valuation
$30K
$20K
sellers' valuation
$25K
$10K
number of cars
100
200
number of buyers
infinite
Suppose that every buyer and every seller know
the type of the car they are negotiating.
Then both good cars and bad cars will be
traded.
There are simply two products (good and bad
cars).
Scenario II: No Information
good cars bad cars
buyers' valuation
$30K
$20K
sellers' valuation
$25K
$10K
number of cars
100
200
number of buyers
infinite
 Suppose buyers don’t know the type of the cars they are
interested. Also suppose no sellers know the type of the
cars they own. Assume all agents are risk neutral.
 Expected valuation of a car to buyers= 1/3 * $30K +
2/3 * $20K = $23.33K
 Expected valuation of a car to sellers = 1/3 * $25K +
2/3 * $10K = $15K
 Both good cars and bad cars will be traded!
Scenario III: Asymmetric (Unequal)
Information
good cars bad cars
buyers' valuation
sellers' valuation
number of cars
number of buyers
$30K
$25K
100
infinite
$20K
$10K
200
 Sellers know the types of cars they own. But buyers
don’t know the types of cars they are going to buy.
 Is a buyer willing to pay at a price greater than $25K
(say $26K)?
 No, because there is 2/3 of probability that the car is
bad, and the expected valuation to the buyer=1/3*$30K
+ 2/3*$20K= $23.33K < the price
Scenario III: Asymmetric (Unequal)
Information
good cars bad cars
buyers' valuation
$30K
$20K
sellers' valuation
$25K
$10K
number of cars
100
200
number of buyers
infinite
 Is a buyer willing to pay a price of $22K to buy a car?
 No, at such a low price, only bad cars owners will sell
their cars. But bad cars are worth only $20K to the
buyer. $22K is too high a price.
 The market price is even lower, at $20K or somewhat
lower. Only bad cars will be traded. Good cars don’t find
a buyer!!!
 Remark: What matters is not the amount of information.
Scenario III: Asymmetric
Information
good cars bad cars
buyers' valuation
$30K
$20K
sellers' valuation
$25K
$10K
number of cars
100
200
number of buyers
infinite
 Good cars may still find a buyer, if the probability of bad
cars in the pool is low.
 Let p be such prob. A buyer is willing to pay $25K if (1p)x$30K + px$20K>$25K, or p<0.5.
 Good cars of 2-3 years old will easily find a buyer, while
good cars of 10 years old don’t find a buyer
Solving the Problems:
Guarantees & Warranties
Liability Laws
Reputation of a store or the manufacturer
Experts--a disinterested party
Standards & Certifications
signaling by the party who has more information
[education, money burning, etc.]
Moral Hazard
Recap:
When a person buys medical insurance, the insuring company
does not know whether the person is healthy. Nor does it know
how well he is going to take care of himself after buying
insurance.
The latter type of asymmetric information is called a hidden
action problem, or moral hazard problem. That is, when people
are insured by medical insurance, they tend to exert less effort
to take care of themselves than the case they are uninsured.
Why called moral hazard? People buy insurance for their houses,
and then burn them and claim for compensation greater than
the value of the houses.
Some knowledge about
decisions under uncertainty
 Expected value of a gamble=EV=p1v1+p2v2+…+pnvn where pi is the
probability for the prize vi and p1+…+pn=1
 A gamble (game, or lottery) is fair if its expected value is zero.
 St. Petersburg paradox:
a coin is flipped until a head appears. If the head first appears on
the nth flip, the player is paid $2n. How much are you willing to pay
to play the game?
 EV of the game =(1/2)x$2+(1/4)x$4+…(1/2n)x$2n+… =
 But rarely are we willing to pay a few thousand dollars for such a
game!!
 This is because people are generally risk averse! One prefers
$1,000 to 1/2 probability of $2,000 and 1/2 probability of $0.
Contracting between a risk
neutral and risk averse agent
 In the absence of moral hazard, the risk neutral agent should bear
all risk.
 Suppose there are only two states: good and bad (each with 1/2
probability), and a joint venture between a risk neutral agent (N)
and a risk averse agent (R) have a revenue of 150 and 100 under
the two states.
 Consider any contract stipulates that the agent get x at the good
state and y at the bad state. N’s expected income is (150x)/2+(100-y)/2=125-(x+y)/2. R’s expected income is (x+y)/2.
 Consider another contract with a sure payment of (x+y)/2 at each
state. The each party’s expected income remains unchanged. While
N is indifferent to the change, R is strictly better off because risk is
eliminated.
Principal-Agent Problem &
Insurance
 A risk neutral principal hires a risk averse agent to help him on a
task. Suppose the agent’s effort affects the profits nondeterministically (e.g. farming).
 Without moral hazard, the optimal contract is to let the
principal bear all the risk and the agent get a fixed income.
[Optimal risk sharing]
 But in the presence of moral hazard, such a contract is infeasible in
the sense that the agent will shirk. A tradeoff is to have a contract
stipulating a bonus positively related to the profits generated.
 The above analysis can be extended to insurance. In the absence of
moral hazard, the optimal insurance contract is a full insurance.
 In the presence of moral hazard, full insurance is no longer feasible.
Insurance can only be partial [deductible, etc.]
Moral hazard as a
verification problem
Even if actions are observable, as long as they
are not verifiable to a third party, then a contract
made contingent on these actions is not
enforceable.
The following insurance contract is not
enforceable. “If you take good care of yourself,
I’ll pay you $5,000 in case of a loss; otherwise, I
won’t pay you any in case of a loss.”
Hence, moral hazard may well be a result of
incomplete contracts, rather than asymmetric
information.
Moral hazard as a
commitment problem
 The idea of moral hazard has subsequently expanded to
account for related phenomena.
 The most discussed are commitment or time
(in)consistency problem.
 Actions are observable and even verifiable and hence
contractible.
 However, when the undesirable action takes place, both
parties of the contract will choose not to enforce it.
Instead, they renegotiate the contract.
A Contract between
Mother and Son
 A contract between mother and son: “If the son goes to sleep by
10pm, he’ll receive no penalty; otherwise, he will be shot to death.”
 Ex ante, the mom wishes the contract enforceable so that,
foreseeing the death penalty, the son will go to bed on time.
 Suppose HK law allows this contract to be legal.
 Suppose they can arrange videotaping so that the court can verify
when the son goes to bed.
 However, ex post in case the son doesn’t go to bed by 10pm, it
would be in both parties’ interest to renegotiate the contract (no
one likes to see the son die anyway).-->a time inconsistency
problem (or commitment problem)
 The son knows this and hence does not go to bed by 10pm.
Moral hazard &
Government Policy
 Moral hazard as a time consistency problem is closely
related to government policy, where government is one
party of a contract and citizens (or firms) are another
party.
 An interesting quote: “There are two types of
countries in the world; those with deposit
insurance, and those not knowing they have
deposit insurance.”
Moral hazard in a natural
disaster
 Suppose residents near Yellow River are more vulnerable to attack
of typhoons.
 Two states of nature: good state and bad state with probability 0.9
and 0.1.
 In good state: loss =0
 In bad state: loss = $1m
 Residency elsewhere no loss
 moving cost = $50k<expected loss=0.1x$1m=100k
 in case of bad state, the government can step in to reduce the
resident’s loss to $200k at a cost of $500k
Moral hazard in a natural
disaster
Scenario
1st best (not helping at
bad state and
hence,
foreseeing this)
moving out
rd
3 best not moving &
not helping at
bad state
2nd
not moving &
best
helping at bad
state
government's
cost
0
resident's
loss
total
cost
moving cost
= $50k
$50k
0
0.1*$1m
=$100k
$100k
0.1*$500k 0.1*$200k
=$50k
=$20k
$70k
The efficient outcome is to have the resident move out. But the
government cannot commit not to rescue the resident at the bad state (a
benevolent government should rescue). Knowing this, the resident will
not move out!
Remedies for moral hazard as a
commitment problem
Mandatory insurance
Coercion (force the resident to move at the
outset)
divided government makes discretion less likely
In general, don’t postpone! Solve the problem
now!!
Further Applications
 Why unemployment can be an equilibrium outcome
 Why there is equilibrium credit rationing
 Why debt financing and equity financing are different
 Why firms exist and what explains their boundaries
(holdup problem)
 Why currency attack occurs even when the
fundamentals are sound