Social Insurance: The

New Function of
Government
Emmanuel Skoufias
LKY School
2023-24:S2

1 of 38
 Social Insurance: The New Function of Government

1. What Is Insurance and Why Do Individuals Value It?

2. Why Have Social Insurance? Asymmetric Information and
Adverse Selection
3. Other Reasons for Government Intervention in Insurance
Markets
4. Social Insurance versus Self-Insurance: How Much Consumption
Smoothing?
5. The Problem with Insurance: Moral Hazard
6. Putting It All Together: Optimal Social Insurance
7 Conclusion

2 of 38
Social Insurance: The New Function of
Government

• Preamble to the United States Constitution:

Establish justice, insure domestic tranquility, provide
for the common defense, promote the general
welfare, and secure the blessings of liberty to
ourselves and our posterity.
• For most of the country’s history, emphasis on
“common defense.”
• Since 1950 or so, shift toward “the general welfare.”

3 of 38
Government Spending by Function,1953 and
2014

1953 2014
Defense 69.4% 17.2%
Disability and 5.0 14.7
Unemployment
Social Security 3.6 24.3
Health 0.4 26.4
Other 21.6 17.5

Government today devotes a much larger portion of its

budget to social insurance than it did 50 years ago.

4 of 38
Social Insurance: The New Function of
Government

• Government spending now focuses on social

insurance programs.
o Social insurance programs: Government
interventions in the provision of insurance against
adverse events.
• For most programs, eligibility is not means-tested.
o Means-tested: Refers to programs in which
eligibility depends on the level of one’s current
income or assets.

5 of 38
What Is Insurance?

• Insurance is a promise to make some payment in case

of a particular event, in exchange for a payment,
called a premium.
• Insurance premiums: Money that is paid to an
insurer so that an individual will be insured
against adverse events.
• Insurance products in the United States include health
insurance, auto insurance, life insurance, and casualty
and property insurance.
• Annual private premiums for these products totals
more than $1.6 trillion.

6 of 38
Why Do Individuals Value Insurance?

Insurance is valuable because it helps individuals’

insurance consumption across states of the world.
• Consumption smoothing: The translation of
consumption from periods when consumption is
high, and thus has low marginal utility, to periods
when consumption is low, and thus has high marginal
utility.
• States of the world: The set of outcomes that are
possible in an uncertain future.

7 of 38
Why Do Individuals Value Insurance?
Diminishing Marginal Utility

• Diminishing marginal utility means that the fourth

slice of pizza is less important than the first.
• Always having two slices is better than sometimes
having four and sometimes having zero.
• Always a moderate amount of consumption for sure
is better than a 50–50 chance of having a lot or
nothing.
• Individuals will demand full insurance in order to fully
smooth their consumption across states of the world.

8 of 38
Formalizing This Intuition: Expected Utility
Model

We formalize these ideas in the expected utility model.

• Expected utility model: The weighted sum of utilities
across states of the world, where the weights are the
probabilities of each state occurring.
• Suppose an adverse event occurs with probability .
Expected utility is

9 of 38
The Expected Utility Model: Health insurance

• 1% chance that Sam gets hit by a car, resulting in

$30,000 in medical expenses.
• Insurance costs b for each dollar of coverage.
o If Sam buys $m of coverage, his premium is $mb.
• To analyze Sam’s choice, assume , and premiums are
actuarially fair.
o Actuarially fair premium: An insurance premium
that is set equal to the insurer’s expected payout.

10 of 38
 Full Insurance Is Optimal

Purchase Hit? C Expected Utility

No Yes 30,000
insurance
No 0 0
Full Yes 38,700
insurance
($300) No 38,700
Partial Yes 38,850
insurance
($150) No 14,850

11 of 38
 Full Insurance Is Optimal

No Insurance Full insurance Partial Insurance

Premium 0 300 150
C if not hit 30,000 38,700 38,850
C if hit 0 38,700 14,850
U if not hit 173.2 172.34 172.77
U if hit 0 172.34 121.86
Expected 0.99 × 173.2 0.99 × 172.34 0.99 × 172.77
utility + 0.01 × 0 + 0.01 × 121.86
= 171.5 + 0.01 × = 172.26
172.34
= 172.34

12 of 38
The Role of Risk Aversion

• Risk aversion: The extent to which individuals are

willing to bear risk.
• Risk-averse people may still want to buy some
insurance even if it is not actuarially fair.
• People may differ in their risk aversion, and if
insurance premiums are extremely unfair, then
only the most risk averse will want it.

13 of 38
Why Have Social Insurance? Asymmetric
Information and Adverse Selection

Why should the government provide insurance?

• Information asymmetry can lead to a key market
failure called adverse selection.
o Information asymmetry: The difference in
information that is available to sellers and to
purchasers in a market.
o Adverse selection: The fact that the insured
individuals knows more about their risk level than
does the insurer might cause those most likely to
have the adverse outcome to select insurance,
leading insurers to lose money if they offer
insurance.

14 of 38
Adverse Selection Example

• Two kinds of people:

o Careless people have a 5% chance of being in a
car accident (half the population).
o Careful people have a 0.5% chance (half the
population).
• If the insurance companies knows each person’s type,
it can charge them separate prices.
• If the insurance company doesn’t know their type, it
could try charging a price that is fair on average, or
try charging separate prices.

15 of 38
Insurer Breaks Even with Full Information
Pricing

• What happens if the insurance company could charge

each type their actuarially fair price?
o Charge careless people $1,500.
o Charge careful people $150.
o Earn $150,000 per 100 careless people, pay out
$150,000.
o Earn $15,000 per 100 careful people, pay out
$15,000.

16 of 38
Asymmetric Information Pricing: Separate

• What if the insurance tries to charge different prices

but cannot tell who is careless?
o Careless people pretend to be careful, pay $150.
o Careful people pay $150.
o Earn $15,000 per 100 careless people, pay out
$150,000. Lose $135,000.
o Earn $15,000 per 100 careful people, pay out
$15,000.

17 of 38
Asymmetric Information Pricing: Separate

• What if the insurance company tries to charge

average price?
o Average price: $825.
o Insurance is a great deal for careless people, so
they buy it, pay $825.
o Careful people decline it.
o Earn $82,500 per 100 careless people, pay out
$150,000. Lose $67,500.
o Earn nothing from careful people.

18 of 38
The Problem of Adverse Selection

• Adverse selection: The fact that insured individuals

know more about their risk level than does the
insurer might cause those most likely to have the
adverse outcome to select insurance, leading insurers
to lose money if they offer insurance.
• Selling to both requires that low-risk people subsidize
high-risk people.
• Low-risk people may not want to do this.
• Sometimes, only high-risk people end up with
insurance.

19 of 38
Does Asymmetric Information Necessarily Lead
to
Market Failure?

If low-risk people have a high enough risk premium,

they will subsidize high-risk people in a pooling
equilibrium.
• Risk premium: The amount that risk-averse
individuals will pay for insurance above and beyond
the actuarially fair price.
• Pooling equilibrium: A market equilibrium in which all
types of people buy full insurance even though it is
not fairly priced to all individuals.
• Separating equilibrium: A market equilibrium in which
different types of people buy different kinds of
insurance products designed to reveal their true
types. 20 of 38
APPLICATION: Adverse Selection and Health
Insurance “Death Spirals”

• In 1995, Harvard stopped subsidizing its most

generous plans, which were experience-rated.
• Experience rating: Charging a price for insurance that
is a function of realized outcomes.
• Before 1995, there was a pooling equilibrium.
o Healthy employees chose the cheap, generous
plan.
• After 1995, there was a separating equilibrium.
o Healthy employees dropped the now expensive
generous plan.

21 of 38
APPLICATION: Adverse Selection and Health
Insurance “Death Spirals”

• Because the less-healthy employees used much more

medical care, the experience-rated premiums of the
more generous plans increased substantially.
• By 1998, the most generous plan had gotten so
expensive that it was no longer offered.
o Adverse selection had led to a “death spiral” for
this plan.
o It kept getting more expensive, and healthy
people kept leaving, driving its price ever higher.

22 of 38
How Does the Government Address Adverse
Selection?

Adverse selection leads to market failure since healthy

people may not be able to buy insurance.
• The government can address adverse selection, and
improve market efficiency, in a number of ways…
• …but they involve redistribution from the healthy to
the sick, which may be quite unpopular.

23 of 38
Other Reasons for Government Intervention in
Insurance Markets

• Externalities: Vaccines have positive spillovers; car

crashes negative ones.
• Administrative costs: Government-run Medicare has
much lower administrative costs than private
insurance.
• Redistribution: Governments may want to
redistribute from healthy to sick.
• Paternalism: Governments may feel that people
would choose to buy too little insurance for
themselves.

24 of 38
APPLICATION: Flood Insurance and the
Samaritan’s Dilemma

The Samaritan’s Dilemma is another rationale for

intervention.
• Compassionate governments want to bail out hard-
hit citizens.
• But, knowing this, citizens may not buy insurance,
making bailouts expensive.
• This is especially important for floods.
• Congress established National Flood Insurance
Program (NFIP) in 1968 to address this.

25 of 38
APPLICATION: Flood Insurance and the
Samaritan’s Dilemma

• NFIP has paid out $51.1 billion since 1969 and has
lead to improved building standards.
• But nearly half of the victims of Hurricane Katrina in
2005 did not have flood insurance, and the claims of
people with insurance bankrupted the system.
• NFIP failed in part because people avoid buying flood
insurance if they are assured the government is going
to continue to help individuals in disaster areas.
• Acts were passed in 2012 and 2014 to more
accurately price flood insurance and control the rate
adjustments for flood insurance premiums in
vulnerable areas.

26 of 38
Social Insurance versus Self-Insurance: How
Much Consumption Smoothing?

• Even if private insurance markets do not function

well, people may still be able to insure with self-
insurance.
• Self-insurance: The private means of smoothing
consumption over adverse events, such as
through one’s own savings, the labor supply of
family members, or borrowing from friends.

27 of 38
Example: Unemployment Insurance

People can insure against unemployment in many ways:

• They can draw on their own savings.
• They can borrow, either in collateralized forms (such
as borrowing against the equity they have in their
homes) or in uncollateralized forms (such as on their
credit card).
• Other family members can increase their labor
earnings.
• They can receive transfers from their extended
family, friends, or local organizations.

28 of 38
Example: Unemployment Insurance

• Unemployment insurance provides benefits through

the replacement rate.
• UI replacement rate: The ratio of unemployment
insurance benefits to pre-unemployment
earnings.
• A higher replacement rate corresponds to more
generous insurance.
• But private insurance reduces the consumption-
smoothing value of this insurance.

29 of 38
Example: Unemployment Insurance

30 of 38
Lessons for Consumption-Smoothing Role of
Social Insurance

The importance of social insurance for consumption

smoothing will depend on two factors:
• Predictability of the event: It is easier for people to
self-insure against a predictable event, such as
increasing their savings. More predictable risks
reduce the benefits of providing social insurance.
• Cost of the event: It is more difficult to self-insure
against high-cost events, such as becoming injured
and unable to work. Costly risks increase the
benefits of providing social insurance.

31 of 38
The Problem with Insurance: Moral Hazard

The cost of insurance is moral hazard.

• Moral hazard: Adverse actions taken by individuals or
producers in response to insurance against adverse
outcomes.
o “Nothing emboldens sin so much as mercy.”
• The existence of moral hazard means that it may not
be optimal for the government to provide the full
insurance that is demanded by risk-averse
consumers.

32 of 38
APPLICATION: The Problems with Assessing
Workers’ Compensation Injuries

• Prison guard Ricci DeGaetano

o Supposedly injured by an inmate. Collected
$82,500 in claims over three years while
operating a karate school.
• Detective Rocky Sherwood
o Injured in traffic accidents. While claiming
workers’ compensation, coached little league
team to California World Series victory.
• Waitress Christina Gamble
o Too injured to “stand” and “change positions.”
Received $360/week in insurance payments
while working as a stripper.
33 of 38
The Problem with Insurance: Moral Hazard

What determines moral hazard?

• How easy it is to observe whether the adverse event
has happened.
• How easy it is to change behavior in order to
establish the adverse event.

34 of 38
The Problem with Insurance: Moral Hazard

In examining the effects of social insurance, four types

of moral hazard play a particularly important role.
• Reduced precaution against entering the adverse
state.
• Increased odds of entering the adverse state.
• Increased expenditures when in the adverse state.
• Supplier responses to insurance against the adverse
state.

35 of 38
The Consequences of Moral Hazard

Moral hazard is costly for two reasons.

• The adverse behavior encouraged by insurance
lowers social efficiency because it reduces the
provisions of socially efficient labor supply.
• When social insurance encourages adverse events,
which raise the cost of the social insurance
program, it increases taxes and lowers social
efficiency further.

36 of 38
Putting It All Together: Optimal Social Insurance

• Optimal social insurance systems should partially, but

not completely, insure individuals against adverse
events.
• The benefit of social insurance is the amount of
consumption smoothing provided by social insurance
programs.
• The cost of social insurance is the moral hazard
caused by insuring against adverse events.

37 of 38
Conclusion

• Asymmetric information in insurance markets has

two important implications:
o It can cause adverse selection.
o It can cause moral hazard.
• The ironic feature of asymmetric information is
therefore that it simultaneously motivates and
undercuts the rationale for government intervention
through social insurance.

38 of 38
39 of 38
 Chetty and Looney
• Studies of risk in developing economies have focused on
consumption fluctuations as a measure of the value of
insurance. A common view in the literature is that the welfare
costs of risk and benefits of social insurance are small if
income shocks do not cause large consumption fluctuations.
• They present a simple model showing that this conclusion is
incorrect if the consumption path is smooth because
individuals are highly risk averse. Hence, social safety nets
could be valuable in low-income economies even when
consumption is not very sensitive to shocks.
• Small consumption fluctuations need not imply that existing
insurance is “adequate” in developing economies. In fact, the
converse may be true: consumption may be smooth precisely
because the welfare costs of consumption fluctuations are
very high.
40 of 38
 Chetty and Looney
• To evaluate the welfare consequences of insurance policies,
one must determine why and how households smooth
consumption—because of high risk aversion (high γ) or
through good insurance arrangements (low θb)? This question
is of practical relevance because many households in low-
income countries could have high γ, e.g. due to subsistence
constraints. Evidence that household resort to costly
consumption-smoothing mechanisms (e.g. Frankenberg et al.,
1999; Dercon, 2002; Miguel, 2005; Chetty and Looney, in
press) also suggests that γ could potentially be high for many
households. Distinguishing between the two explanations of
consumption smoothness would be a useful direction for
future research on risk and insurance
41 of 38
 Expected Utility Model
• The model is described by the following parameters:
– You are hit by the car with some probability p.
– Your income is W, regardless of whether you get hit or not.
– However, if you get hit, you incur medical costs d.
– You can buy insurance, with premium m per dollar of insurance.
• That insurance will pay you $b if you are hit by the car. In this case, we can write
your expected utility (EU) as
EU = (1 -p)*U(W - mb) + p*U(W – d – mb + b)
• The problem with this expression is that we have one equation, with two
unknowns (m and b). To solve this equation, we need to add one more condition:
that insurance is priced in an actuarially fair manner, so that insurance companies
make zero expected profits (we assume, for now, zero administrative costs). In that
case, the zero expected profit (E) condition for the insurer is
E mb - pb = 0
Or premiums received (mb) minus expected benefits paid out (pb), equals zero.

42 of 38
 Expected Utility Model
This, in turn, implies that the premium equals

That is, if the risk is 10%, then m =10¢ per dollar of insurance. We can now go back
and maximize expected utility by plugging in b from this equation. As in the example in
the text, we assume that utility is of the form , So maximize

Maximizing this equation with respect to b, we obtain

Setting this equal to zero and solving for the optimal level of insurance benefits (b*),
we get b*= d. That is, individuals should buy enough insurance so that if they have the
adverse outcome, their benefits exactly offset their costs: individuals should buy full
insurance to smooth their consumption across states.

43 of 38
 Expected Utility Model
Another way to see this is to plug the optimal benefit level (b*=d) back into the utility
function:

That is, we obtain the result that consumption is equalized ( ) in both states of the
world.

Thus, facing actuarially fair insurance markets, individuals will want to insure
themselves fully against risk.

44 of 38
 Adverse Selection
To understand more formally the implications of adverse selection, we now consider
two groups, the careful and the careless, where the probability of accident for the
careful is pc, and the probability of accident for the careless is pa > pc.
With full information, then the insurance company charges prices
• ma=b*pa for the careless, and
• mc=b*pc for the careful.
Thus, the premium for the careless is higher, since pa > pc; those who are more likely
to have an accident have to pay more for insurance.

45 of 38
 Adverse Selection
If there isn’t full information, so that insurance companies know only the proportions
of types in the population, then there are two possible pricing strategies.
One is to assume that individuals are honest and charge them according to their
reported types. However, as discussed, this strategy will lead all individuals to claim
that they are careful. In this world, the profits earned on the careful are:

that is, the insurance company breaks even on the share of the population that is
careful. However, the profits earned on the careless are

since pa > pc profits are negative overall and thus insurance is not offered.

46 of 38
 Adverse Selection
The other strategy is to offer insurance at an average price, mv, that is based on the
average of the accident probabilities pa > pv > pc. At this price, insurance is a good
deal for the careless but a bad deal for the careful, and may be bought only by the
careless. In that case, the expected profits of the insurer are again negative:

since pa >pv

So even with this alternative strategy (offering insurance at an average price based on
the accident probabilities) expected profits or insurer are negative and insurance may
not be offered Market failure

47 of 38
 Adverse Selection
NOTE: It is possible, however, that the careful still would buy full insurance (the
pooling equilibrium). For example, they would buy insurance if expected utility with
insurance (at the unfair price) is still higher than expected utility without insurance;
that is, if
>
where

Whether > for the careful, depends on two things:

– the extent of risk aversion (or curvature of the utility function) of the careful individuals and
– the relationship between pc and pa.
If the careful individuals are more risk averse, they will be more willing to buy
insurance (even at an unfair premium) to guard against the odds of being left with low
consumption. And the closer the average risk is to the risk faced by the careful, the
closer the premium is to being actuarially fair, and the more likely it is that the careful
individuals will buy the insurance.

48 of 38