CHAPTER 7
DEMAND FOR INSURANCE
Why buy insurance?
Demand for insurance driven by the fear of the
unknown
Hedge against risk -- the possibility of bad outcomes
Purchasing insurance means forfeiting income in
good times to get money in bad times
If bad times avoided, then money lost
Ex: The individual who buys health insurance but
never visits the hospital might have been better off
spending that income elsewhere.
Bhattacharya, Hyde and Tu – Health Economics
Risk aversion
Hence, risk aversion drives demand for
insurance
We can model risk aversion through utility from
income U(I)
Utility increases with income: U(I) > 0
Marginal utility for income is declining: U(I) < 0
Bhattacharya, Hyde and Tu – Health Economics
Income and utility
Graphically,
Utility increasing with income U’(I) > 0
Marginal utility of income decreasing U’’(I) < 0
Bhattacharya, Hyde and Tu – Health Economics
Adding uncertainty to the model
An individual does not know whether she will
become sick, but she knows the probability of
sickness is p between 0 and 1
Probability of sickness is p
Probability of staying healthy is 1 - p
If she gets sick, medical bills and missed work will
reduce her income
IS = income if she does get sick
IH > IS = income if she remains healthy
Bhattacharya, Hyde and Tu – Health Economics
Expected value
The expected value (mean) of a random variable X,
E[X], is the sum of all the possible outcomes of X
weighted by each outcome’s probability
If the outcomes are X=x1, x2, . . . , xn, and the probabilities
for each outcome are p1, p2, . . . , pn respectively, then:
E[X] = p1 x1 + p2 x2 + · · · + pn xn
In our individual’s case, the formula for expected
value of income E[I]:
E[I]p = p IS + (1- p) IH
Bhattacharya, Hyde and Tu – Health Economics
Example: expected value
Suppose we offer a starving graduate student a
choice between two possible options, a lottery and a
certain payout:
A: a lottery that awards $500 with probability 0.5 and $0
with probability 0.5.
B: a check for $250 with probability 1.
The expected value of both the lottery and the
certain payout is $250:
E[I]p = p IS + (1- p) IH
E[I A] = .5(500) + .5(0) = $250
E[I B] = 1(250) = $250
Bhattacharya, Hyde and Tu – Health Economics
People prefer certain outcomes
Studies find that most people prefer certain
payouts (like B) over uncertain scenarios (like A)
If a student says he prefers uncertain option,
what does that imply about his utility function?
To answer this question, we need to define
expected utility for a lottery or uncertain
outcome.
Bhattacharya, Hyde and Tu – Health Economics
Expected Utility
The expected utility from a random payout X
E[U(X)] is the sum of the utility from each of the
possible outcomes, weighted by each outcome’s
probability.
If the outcomes are X=x1, x2, . . . , xn, and the
probabilities for each outcome are p1, p2, . . . , pn
respectively, then:
E[U(X)] = p1 U(x1) + p2 U(x2) + · · · + pn U(xn)
Bhattacharya, Hyde and Tu – Health Economics
Example
The student’s preference for option B over option A
implies that his expected utility from B, is greater
than his expected utility from A:
E[U(B)] ≥ E[U(A)]
U($250) ≥ 0.5 U($500) + 0.5 U($0)
In this case, even though the expected values of
both options are equal, the student prefers the
certain payout over the less certain one.
This student is acting in a risk-averse manner over the
choices available.
Bhattacharya, Hyde and Tu – Health Economics
Expected utility without insurance
Lottery scenario similar to case of insurance
customer
She gains a high income IH if healthy, and low
income IS if sick where her income I is a random
variable
Uncertainty about which outcome will happen,
though she knows the probability of becoming
sick is p
Expected utility E[U(I)]p is:
E[U(I)]p = p U(IS) + (1- p) U(IH)
Bhattacharya, Hyde and Tu – Health Economics
E[U(I)] and probability of sickness
Figure shows how expected utility changes as the probability of sickness changes.
Consider a case where the person is sick with certainty (p = 1):
E[U(I)] = U(IS) equals the utility from certain income I S (Point S)
Consider case where person has no chance of becoming sick (p = 0):
E[U(I)] = U(IH) equals utility from certain income I H (Point H)
Bhattacharya, Hyde and Tu – Health Economics
What if p lies between 0 and 1?
For p between 0 and 1, expected utility falls on a
line segment between S and H
Bhattacharya, Hyde and Tu – Health Economics
Ex: p = 0.25
For p = 0.25, person’s expected income is:
E[I] = 0.25·IS + (1 - .25)·IH
Utility at that expected income is E[U(I)]0.25 (Point A)
Bhattacharya, Hyde and Tu – Health Economics
Expected utility and expected income
Crucial distinction between
Expected utility E[U(I)]
Utility from expected income U(E[I])
This distinction in the case when p=0.5:
The individual's expected income from the sickness
lottery:
E[I]0.5 = 0.5·IS + (1 - 0.5)·IH
And expected utility from the sickness lottery:
E[U(I)]0.5= 0.5·U(IS) + (1 - 0.5)·U(IH)
Bhattacharya, Hyde and Tu – Health Economics
For risk-averse people, U(E[I]) > E[U(I)]
Bhattacharya, Hyde and Tu – Health Economics
Risk-averse individuals in the utility-
income model
Synonymous definitions of risk-aversion:
Prefer certain outcomes to uncertain ones with the
same expected income.
Prefers the utility from expected income to the
expected utility from uncertain income
U(E[I]) > E[U(I)]
Concave utility function
U’(I) > 0
U’’(I) < 0
Bhattacharya, Hyde and Tu – Health Economics
A basic health insurance contract
The individual approaches a health insurance company
that offers a policy with the following features:
The individual pays an upfront fee r regardless of
whether she stays healthy or becomes ill.
Payment r is known as the insurance premium
If ill, customer receives q -- the insurance payout
If healthy, customer receives nothing from the insurance
company
Either way, customer loses the upfront fee r
Customer’s final income is:
Sick: IS + q – r
Healthy: IH + 0 – r
Bhattacharya, Hyde and Tu – Health Economics
Income with insurance
Let IH’ and IS’ be income in the healthy and sick
states with the insurance contract.
Sick: I S’ = I S + q – r
Healthy: IH’ = IH + 0 – r
Remember that risk-averse consumers want to
avoid uncertainty in buying insurance
For them, optimally
E[I]p = IH’ = IS’
An insurance contract that fulfills this equation is said to
be actuarially fair, full insurance. Bhattacharya, Hyde and Tu – Health Economics
Full insurance
Full insurance means no income uncertainty
IS’ = IH’
Final income is state-independent
Regardless of healthy or sick, final income is
the same
Risk-averse individuals prefer full insurance
to partial insurance (given the same price)
Bhattacharya, Hyde and Tu – Health Economics
Full insurance payout
State independence implies
IH’ = IS’
So
IH + 0 – r = IS + q – r
IH = IS + q
q = IH – IS
The payout from a full insurance contract is
difference between incomes without insurance
Bhattacharya, Hyde and Tu – Health Economics
Actuarially fair insurance
Actuarially fair means that insurance is a fair bet
i.e. the premium equals the expected payout
r=pq
Insurer makes zero profit/loss from actuarially
fair insurance in expectation
Bhattacharya, Hyde and Tu – Health Economics
Actuarially fair, full insurance
The above algebra shows that with contract, the individual’s
income is E[I]p regardless of whether she turns out to be sick or
healthy. Notice that consumers with actuarially fair, full insurance
achieve their expected income with certainty!
Bhattacharya, Hyde and Tu – Health Economics
Bhattacharya, Hyde and Tu – Health Economics
Insurance and risk aversion
As we have seen, simply by reducing uncertainty,
insurance can make this risk-averse individual
better off.
The nature of the insurance contract is that the
individual loses income in the healthy state (IH >
IH) and gains income in the sick state (IS < IS)
relative to the state of no insurance
In other words, the risk-averse individual willingly
sacrifices some good times in the healthy state to
ease the bad times in the sick state.
Bhattacharya, Hyde and Tu – Health Economics
Insurer profits
Now consider the same insurance contract from
the point of view of the insurer
Premium r
Payout q
Probability of sickness p
E[] = Expected profits that the insurer makes
Bhattacharya, Hyde and Tu – Health Economics
Fair and unfair insurance
In a perfectly competitive insurance market, profits
will equal zero
Same definition as actuarially fair!
An insurance contract which yields positive profits is
called actuarially unfair insurance:
An insurer would never offer a contract with
negative profits
Bhattacharya, Hyde and Tu – Health Economics
When insurance is fair, in a sense, it is also free.
The customer’s expected income does not
change from buying the contract, so she
effectively pays nothing for it.
Despite the fact that the Premium r is positive in
an actuarially fair contract, the price is actually
zero.
Thus, the Premium associated with an insurance
contract is not its price.
Bhattacharya, Hyde and Tu – Health Economics
Full vs. partial insurance
Partial insurance does not achieve state-
independence
Size of the payout q determines the fullness of the
contract
Closer q is to IH – IS , the fuller the contract
Bhattacharya, Hyde and Tu – Health Economics
Comparing insurance contracts
AF -- Actuarially fair & full
AP -- Actuarially fair & partial
A – Uninsurance
U(AF) > U(AP) > U(A)
Bhattacharya, Hyde and Tu – Health Economics
The income uncertainty the individual faces is
largest in the case of no insurance: She receives
either IH or IS, which may be significantly
different.
Partial insurance lowers her uncertainty but does
not eliminate it altogether since ISP is still less
that IHP.
Only with the highest quantity of insurance – full
insurance – does the individual reach state
independence and fully eliminate income
uncertainty. Bhattacharya, Hyde and Tu – Health Economics
The ideal insurance contract
For anyone risk-averse, actuarially fair & full
insurance contract offers the most utility
Hence, it is called the ideal insurance contract
Ideal and non-ideal insurance contracts:
Bhattacharya, Hyde and Tu – Health Economics
Comparing non-ideal contracts
AF – Full but actuarially unfair contract
AP – Partial but actuarially fair contract
Bhattacharya, Hyde and Tu – Health Economics
Comparing non-ideal contracts
In this case, U(AF) > U(AP)
Even though AF is actuarially unfair, its relative
fullness (i.e. higher payout) makes it more desirable
But notice if contract AF became more unfair, then
expected income E[I] falls
If too unfair, AF may generate less utility than A P
Similarly, AP may become more full by increasing its
payout
Uncertainty falls, so point AP moves
At some point, this consumer will be indifferent between
the two contracts
Bhattacharya, Hyde and Tu – Health Economics
Conclusion
Demand for insurance driven by risk aversion
Desire to reduce uncertainty
Diminishing marginal utility from income
U(I) is concave, so U’’(I) < 0
U(E[I]) > E[U(I)]
Risk aversion can explain not only demand for
insurance but can also help explain
Large family sizes
Portfolio diversification
Farmers scattering their crops and land holdings
Bhattacharya, Hyde and Tu – Health Economics
