Healthcare Reform Simulation Summary
Thursday, March 19, 2009
A new NFIB-sponsored study of health insurance markets yields striking results and gives researchers a valuable new analytical tool. In one of the first-ever applications of experimental economics to healthcare policy, Stephen Rassenti and Carl Johnston, with NFIB’s help, tested reform proposals in a laboratory. Rassenti and Johnson are protégés of Nobel laureate Vernon Smith, who calls the study “path-breaking.”
The researchers built an insurance market in a laboratory and tested various scenarios on paid participants. The study compared wages, profits, and bankruptcies under scenarios involving individual mandates, employer mandates, premium restrictions, minimum employer contributions, number of policies offered, and the degree to which employees understand health insurance. The text below lists the study’s results, provides an overview of experimental economics, and explains the methodology developed for this study by Rassenti and Johnston.
What were the findings?
· No panaceas: Employee and individual mandates, separately or combined, don’t improve outcomes for all stakeholders. Some reform scenarios actually come close to making everyone worse off.
· Small is vulnerable: Small employers and their employees are especially vulnerable to policy changes and mandates. Small companies lack the advantage of size when optimizing their health care spending and pay disproportionately higher costs for providing benefits.
· Low-margin resembles small: Large companies with low profit margins tend to exhibit preferences similar to small companies, even though they have many employees and otherwise act ‘large.’
· Choosing for others: Individuals seem better able to pick insurance plans for themselves than for other people. One reason is that employers have better information about their own needs than about their employees’ needs.
· Minimum contributions: Requiring employers to pay for half of individual insurance costs reduces employer earnings, but increases employee incomes – at least in the short run. Small and low-margin companies are especially hurt by minimum contribution requirements. Without a mandatory employer contribution, an individual mandate increases employer profits but reduces employee incomes.
· Restricted rating: Restricted rating increases the earnings of individual employees by shifting the higher costs of premiums from the employees to the employers. Faced with restricted rating, employers seem to shield employees from the costs and, in doing so, cut their own profits.
· Bankruptcy: The relative risk of employer bankruptcy varies widely among scenarios. Chances of bankruptcy went down in some scenarios, including individual mandates alone and individual mandates + employer mandates with no minimum employer contribution. However, the individual mandate can actually increase the risk of bankruptcy when combined with other factors, such as restricted rating or poor estimation of health risks and costs by employees. The greatest likelihood of bankruptcy occurs in the two scenarios where employers face the mandatory burden of providing insurance and paying at least 50% of the cost of premiums.
· Employee cognition: In real-world policy discussion, one rationale for employer involvement in health insurance is the belief that employers are better than employees at choosing health insurance. However, when the virtual employees were programmed to be more error-prone in choosing insurance policies, the results varied considerably. In some scenarios, mandates hurt some stakeholders and helped none.
· More choices: The addition of three extra insurance choices depressed earnings for employees and most types of firms. This effect may have been due to the fact that one of the choices was a policy with a low premium and poor benefits. Employees with lower expected healthcare costs often bought this policy to comply with the individual mandate at the lowest cost.
What is experimental economics?
Public policy innovations expose the public to possible expense and risk. Economic experiments let officials compare ideas in laboratories before launching new policies. Previous experimental economic projects/investigations have helped design markets for stocks, radio frequencies, electricity, airport takeoff and landing slots, and space station resource allocation.
Experimental economists, like experimental psychologists, create controlled environments in which to observe test subjects’ behavior. As in real life, subjects affect the fortunes of those around them, and none knows all the actions and motives of others. Learning-by-doing occurs as participants begin to “feel” the market. Experiments reveal surprises and unintended consequences.
Experimental subjects earn cash, based on how well they perform in these artificial markets. Payoffs are small (from $7.00 to $50.08 in this case), but induce participants to act as they would in real-world markets. So, they are self-interested, though not entirely self-aware. In contrast, traditional survey research gives interviewees less motive for honesty or introspection. With scaled-down rules, laymen perform about as well as experts; in a laboratory stock market, for example, undergraduates perform about as well as stockbrokers, even though they may be less able to explain their own performance.
Results provide evidence, not proof. Like surveys and other research methodologies, experimental markets omit many real-world variables. This experiment omitted wage flexibility and taxes. Also, employers here were motivated only by profit, thus ignoring real entrepreneurs’ love of their work, concern for employees, and desire for independence.
How was the study run?
Subjects, mostly undergraduates, acted as employers for 360 “months.” Each produced two goods, using virtual employees they could hire and lay off. Employers paid virtual salaries to workers. To attract employees, firms chose which insurance policies to offer and how much to contribute toward the premiums. In two scenarios, employers had to offer insurance and cover at least 50% of the cost. Goods, income, employees, salaries, and premiums were virtual – existing only on subjects’ computers.
Insured workers were healthier and missed less work time and salary. Employers sold goods produced. As production increased, goods prices declined, reducing the incentive to produce. Employers could finish with profits or losses. Some employers were large (more than 12 employees in this experiment), and others small. Employers could grow or shrink depending on their success. Some employers had thick profit margins; others had thin margins. Employees had different skill sets, commanding different salaries.
Each subject participated in one of nine market scenarios: (1) The status quo: no employer mandate (EM) or individual mandate (IM); (2) IM, but no EM; (3) EM, but no IM; (4) IM + EM; (5) IM + restricted rating; (6) IM + a larger number of insurance choices; (7) IM + employees with poor understanding of health insurance; (8) EM (with minimum 50% contribution rate) + IM; and (9) EM with minimum 50% contribution rate, but no IM. The scenarios were generic – not replicas of actual reform plans under consideration.
------
This study produced a number of important findings, but, like all studies, omitted many important variables. NFIB hopes other researchers will use this methodology to test a broader array of policy variables.