Asymmetric Information 583for its employees and their spouses. Employees who elect this coverage pay
more than with the standard plan. Among other benefits, the premium plan
will pay for maternity-related health expenses. The firm estimates that 1 in 20
of its employees will have a new baby in a given year. (This estimate comes from
records for the last 10 years.) Accordingly, the company has set the premium
to cover its expected payouts at this 1-in-20 rate. Postscript: In the first two years
of the program, the company has lost an enormous amount of money on the
program. Employees covered by the plan are having babies at the rate of 1 in
10 per year. Is this bad luck or bad planning?
The company’s losses are not due to bad luck. Today’s workforce does
not differ in its composition or behavior from that of the last 10 years.
Instead, the firm’s losses are due to adverse selection.The following table lists
the hypothetical, but plausible, numbers for the first year of the program.
Notice that the overall rate of new babies is 200/4,000 or 1 in 20, exactly the
average rate of the previous 10 years. The rate of having babies has not
changed. However, among policyholders, the rate of having babies is 1 in 10
(100/1,000); among nonpolicyholders, it is 1 in 30. This result should not
surprise us. Couples who are planning to add to their families will tend to
elect the policy; those who are not will forgo the coverage. This behavior usu-
ally is termed self-selection.From the company’s point of view, the result is
called adverse selection. Couples who are most likely to have babies (and
know this) will most likely elect the coverage.Baby No Baby TotalPolicy 100 900 1,000
No policy 100 2,900 3,000
Total 200 3,800 4,000Adverse selection occurs because of asymmetric information. Individuals
have better information about their true risks than the insurance provider
does. As a result, individuals at the greatest risk elect insurance coverage. To
avoid losses on their policies, insurance companies must anticipate this behav-
ior and set their premiums accordingly. In the preceding example, the com-
pany would have to double its premium to break even.A “LEMONS” MARKET The used-car market is a famous example of asym-
metric information.^2 Consider someone trying to sell a car that is six months
old and has been driven only 4,000 miles. Even though almost new, it now
may sell for as little as 75 percent of its original sale price. The steep discount(^2) The classic article on this topic is G. A. Akerlof, “The Market for ‘Lemons’: Quality Uncertainty
and the Market Mechanism,” Quarterly Journal of Economics(1970): 488–500.
c14AsymmetricInformationandOrganizationalDesign.qxd 9/26/11 11:03 AM Page 583