Abstract
The authors describe an asymmetric information demonstration that assigns students different probabilities of incurring healthcare expenses. In each round, students choose whether to purchase insurance; then, the instructor randomly determines who gets “sick.” After computing insurer profits, students help determine a new insurance price to maximize future profit. Within three rounds, students recognize that the provider always incurs losses from adverse selection, opening a discussion of market failures pertaining to health insurance and asymmetric information. The experiment features idiosyncratic, but not systematic, risk as such; the same number of students get “sick” every round. Therefore, the instructor can straightforwardly demonstrate the benefits of risk pooling. The experiment is applicable to economic principles as well as intermediate courses in healthcare economics and microeconomic theory.
Acknowledgments
The authors thank participants at the SEA Conference in the fall of 2020 and the CTREE conference in the spring of 2021 for their insightful comments. In particular, they wish to thank their discussants Simon Halliday and Eva Dziadula. Finally, they appreciate the help of the anonymous referees who helped them strengthen the demonstration.
Notes
Notes
1 Idiosyncratic risk is that which is faced by an individual, so the chance that a student gets “sick” is idiosyncratic. Systematic risk is the risk faced by society as whole, so the chance that overall healthcare costs rise (fall) because more (fewer) students get “sick” is systematic.
2 Enrollment periods are limited to prevent people from buying insurance in response to new information about their health, dumping consists of refusing to treat less healthy patients who might use services in excess of their premiums, creaming is seeking to attract more healthy patients who will use services costing less than their premiums, and skimping is providing less than the optimal quantity of services for any given condition in a given time period.
3 This experiment has never been run in classes of more than 50 students, but there is no reason that it would not work. The only difficulties would be handing out the roles, which could take some time if there are hundreds of students, and counting the number of students who buy insurance and get “sick.”
4 Mellor (Citation2005) is similar enough to Eckles and Halek (Citation2007) that we did not include it in our taxonomy.
5 For example, on a roll of 1, all students in the demonstration get sick.
6 This matches real world data based on .
7 A twenty-sided die can be easily purchased in a gaming store or online; however, for instructors without access to a physical die there are plenty of online random number generators: https://rolladie.net/roll-a-d20-die.
8 We recommend that the instructor print and bring appendix C to class so they can easily determine which groups to assign based on the total number of students in class.
9 The existence of adverse selection puts a natural limit on the monopolist price, so it is not necessary to have a competitive market with many insurers.
10 The use of $20,000 as the medical cost along with the twenty-sided die to determine risk probabilities makes calculating expected value extremely simple. This can be done during the demonstration by students without the need to teach the definition of expected value beforehand.
11 For simplicity, students do not have any income in the experiment and are simply making a choice about whether they think insurance is worthwhile given their risk profile and the price. Lack of an explicit income has never created a problem during the demonstration.
12 Again, students do not explicitly have any income to lose, but this has never disrupted the demonstration in the past.
13 If a group of students is in charge of the insurance company, the instructor should encourage them to try both higher and lower premiums. This will help during the post-demonstration discussion.
14 This demonstration was run in 2019. Since then, the COVID pandemic has offered the best example of systematic healthcare risk.