1、IntroRecall our example:a man walks into the office of a life insurance company.He wants to buy a$1 million life insurance policy for a term of one day.Your company will have to pay$1 million to his heirs if and only if he dies tomorrow.You know nothing else about this man.How much do you charge?Ch
2、10|Adverse selection in real marketsPREDICTIONS OF ASYMMETRIC INFORMATION MODELSAsymmetric information models make three predictions about these markets1)Positive correlation between risk and coverage2)Bulk markups3)Adverse selection death spiral1)Positive correlation between risk and coverageRecall
3、 Rothschild-Stiglitz:Separating equilibrium high-risk types have full insurance(1),low-risk types have incomplete insurance(2)2)Bulk markupsBulk discounts a lower per-unit price for a large purchase of a commodityBulk markups a higher per-unit price for large purchases of a commodityInsurance compan
4、ies use bulk markups to protect themselves from risk customers who want a lot of insuranceThis is exactly what the Rothschild-Stiglitz model predicts3)Adverse selection death spiralBoth frail and robust individuals are pooled togetherPremium is average cost of people in the poolFrail types are indir
5、ectly subsidized by robustRobust types exit the pool,leaving only unhealthy individualsCycle repeatsReadjusted premiumHealthier types leaveCh 10|Adverse Selection in Real MarketsADVERSE SELECTION IN HEALTH INSURANCEEmpirical evidence for adverse selection in health insurance marketsRAND HIEIndividua
6、ls are able to predict their health care costs for the year to a good degree of accuracySpecifically,they are able to predict health care costs more accurately than insurance companiesFamilies with high predicted costs were more likely to want supplemental insuranceEmpirical evidence for adverse sel
7、ection in health insurance marketsSeveral studies find a positive risk-coverage correlation in various marketsBrown and Finkelstein 2009(elderly US Medicare beneficiaries)Van de Ven and van Vliet 1995(Dutch supplemental private insurance)Cutler and Zeckhauser 1998(Harvard professors)Spenkuch 2012(lo
8、w-income Mexican families)Cardon and Hendel 2001(young graduates joining US workforce)Empirical evidence for adverse selection in health insurance marketsEmpirical evidence against adverse selection in health insurance marketsUS workers with employer-sponsored insurance(2001)Positive correlation dis
9、appears when adjust for age,race,and genderMedicare beneficiaries(2008)Negative correlationMaybe“advantageous selection”Seniors with greater cognitive ability are healthier and more likely to purchase supplemental insuranceCh 10|Adverse selection in real marketsADVERSE SELECTION IN OTHER MARKETSAdve
10、rse selection in other types of insurance marketsAutomobile insurancePositive risk-coverage correlation in Israel,but not FranceLife insuranceBulk discounting(not markups)in life insurance market for teachersNegative risk-coverage correlation:people with life insurance live longer!Viatical settlemen
11、tsWhat is a viatical settlement?A firm purchases a life insurance contract from a sick person,then collects the payout when he dies.Adverse selection theory suggests that relatively healthy HIV patients should sell their insurance contracts more readily,but this was not the case in the early 1990s v
12、iatical settlement market in the US.Ch 10|Adverse selection in real marketsWHAT PREVENTS ADVERSE SELECTION?Why would adverse selection not occur?Four main reasons:Customers misperceive their own riskCustomers do not act on their private informationInsurers can accurately observe customer risks“Advan
13、tageous selection”Why would adverse selection not occur?Customers misperceive their own riskUS and Swedish novice drivers demand less insurance88%of US students believe they are safer than median driverWhat is the chance that a man aged 85-89 will live to 100?US survey respondents aged 85-89 reporte
14、d an average subjective probability of 31%Mortality tables indicate the answer is about 3.4%for USWhat is the chance that a woman aged 70 to 74 will live to 85?US survey respondents aged 70 to 74 reported a subjective probability of 51%Mortality tables indicate the answer is about 57%for USWhy would
15、 adverse selection not occur?Customers do not act on their private informationReal customers have more important things to think about than small bargains on insuranceCustomers may fail to realize their advantageCar owners can predict how many miles they will drive in the coming year with good accur
16、acyYet,they do not seem to take that into account when deciding how much insurance to purchaseWhy would adverse selection not occur?Insurers can accurately observe customer risksLong-term care insurers,with access to historical data and a team of analysts,are better at predicting whether middle-aged
17、 customers will eventually need nursing home care than the customers themselvesIn the pre-HAART days,viatical settlement firms could accurately predict HIV life expectancy from a simple blood testRemember that all of our models rest on an assumption of information asymmetryWhy would adverse selectio
18、n not occur?“Advantageous selection”Sometimes it is healthier,less risky people who are more likely to buy insurance Why?more risk aversehigher incomebetter understanding of insurance benefitsWhy?Fang(2008)show that cognitive ability is positively correlated with Medigap insurance purchaseNegatively
19、 correlated with health expendituresHeterogeneous risk preferencesAssuming homogenous risk types and heterogeneous risk preferences:Negative correlation:between risk and coverageBulk discounts:will exist to encourage people to buy more insuranceNo death spiral with insurance pooling:people who are m
20、ore risk averse will stay in the pool when premiums rise but will still be the same risk type to the insurerConclusionAkerlof and RS models make specific predictions about how insurance markets will look.Evidence confirms these predictions in some markets,not in others.Adverse selection is crucial to understanding the challenges of health policy.
