Incorporating Behavioral Economics into Microsimulation Models: A Case Study of the Individual Mandate

To estimate the effects of health insurance reforms, microsimulation models must replicate consumers’ decisions about whether to enroll in coverage, and what type of coverage to choose. Many microsimulation models have relied on neoclassical economic theory to estimate consumer choices, using utility maximization as the basis for predicting enrollment decisions. This approach assumes that consumers make decisions by weighing the costs and benefits of available health insurance options, understand the implications of cost-sharing requirements and network breadth, have complete and accurate information on the health risks that they face, and behave rationally. While these assumptions enable modelers to create a replicable and consistent framework for estimating consumer choices, they are overly simplistic.

In this analysis, we incorporated principles of “bounded rationality” into RAND COMPARE, a microsimulation model that uses utility maximization to estimate how people respond to health insurance reforms. As a test case, we considered how alternative assumptions about bounded rationality affect predicted response to the elimination of the individual mandate penalty. The aspects of bounded rationality that we considered included inertia in decision-making, incomplete information, a taste for compliance with federal law, and the presence of a “welcome mat” effect that may have caused people to enroll in Medicaid even if their incomes were too low to be subject to the individual mandate penalty. In total, we considered 10 scenarios.

The effect of eliminating the individual mandate penalty varied considerably across these scenarios, reflecting the importance of assumptions about the drivers of choice. When we used a rational choice framework, we estimated that eliminating the penalty would reduce enrollment by 6.5 million people in 2020. When we assumed there was inertia in decision-making (i.e. people stuck with a choice they previously made without reevaluating the decision), we estimated that insurance enrollment would fall by only 2.8 million after the elimination of the individual mandate penalty. However, when we incorporated other aspects of bounded rationality—including a taste for compliance with the law and a lack of awareness of penalty exemptions—the elimination of the individual mandate penalty led to a reduction in enrollment of roughly 13 million individuals.

The assumptions also had a significant impact on the estimated cost of eliminating the penalty. In most scenarios, the elimination of the mandate penalty increased the deficit due to the loss of penalty revenue. However, in scenarios in which heavily-subsidized people dropped insurance in response to the change (a somewhat “irrational” response), the deficit fell.

The analysis points both to the importance of assumptions about consumer decision-making in microsimulation modeling, and the difficulty in choosing appropriate assumptions given the multiple and sometimes unexpected factors that may influence behavior. Additional research on the non-economic drivers of insurance choice could help modelers better estimate the likely response to policy changes.