Choice Frictions in Health Insurance Markets
Paper Session
Monday, Jan. 4, 2021 10:00 AM - 12:00 PM (EST)
- Chair: Timothy Layton, Harvard University
Endogenous Information and Simplifying Insurance Choice
Abstract
In markets with complicated products, individuals may choose how much time and effort to spend understanding and comparing alternatives. Focusing on insurance choice, we find evidence consistent with individuals acquiring more information when there are larger consequences from making an uninformed choice. Building on the rational inattention literature, we develop and estimate a parsimonious demand model in which individuals choose how much to research difficult-to-observe characteristics. In contrast to standard demand models, counterfactual simulations imply that simplifying choice by reducing the number of plans or shifting cost to premiums, which are easy to compare, can raise welfare. The results imply endogenous information can have important implications for regulating complex products.The Effect of Clarifying Risk Consequences on Insurance Choice
Abstract
Individuals often make insurance choices that cannot be rationalized by expected utility theory and appear sub-optimal. We experimentally test a new decision aid approach that allows us to investigate and address the possibility that poor insurance choices are largely a result of peoples' poor understanding of the risk consequences of insurance options. We use both incentivized and unincentivized choice experiments mirroring choices people typically face in health insurance menus. We compare choice patterns when people get typical feature-based information about plan options to our decision aid that provides information about the distribution of financial consequences of each plan. We also compare our distribution-based approach to providing people with estimates of the expected value of their costs, which is the most common decision-support available in most insurance markets. We find that when people choose plans using standard feature-based information, they violate state-wise and second-order stochastic dominance at high rates. When people receive our consequence distribution information, choice patterns change and rates of dominance violations are reduced substantially. Our distribution-based approach reduces rates of dominance violations and generally changes choice patterns about twice as much as the expected-value information treatments. The results show that people fail to map the premiums and cost-sharing features of insurance plans to their distribution of financial consequences and that it is possible to design a simple decision aid that helps to clarify these risk consequences.The Behavioral Foundations of Default Effects: Theory and Evidence from Medicare Part D
Abstract
Regulated, subsidized health insurance markets are commonly used to provide social health insurance benefits in the U.S. and around the world. These markets depend on consumer choice to drive firms to compete on price and quality and ensure market efficiency. In this paper, we study decision-making in Medicare Part D, a large, regulated social health insurance market. We document a new stylized fact: Only 16% of low-income beneficiaries make an active plan choice when they initially qualify, with the rest instead receiving the default, a randomly-assigned plan. Default assignments are persistent, with only one-third of randomly-assigned beneficiaries actively choosing to switch to a different plan after five years of enrollment. We show that this behavior is explained by default-following behavior, rather than switching costs: When the default quasi-randomly switches from remaining in their incumbent plan to being randomly assigned to a new plan, 92% of beneficiaries switch. The consequences of this passivity are substantial, with default-driven re-assignment resulting in a significant decrease in drug utilization and a small uptick in use of hospital services. Those who are assigned to plans that are particularly poor matches for them (in terms of coverage of their drugs by the plan formulary) experience larger effects. However, these beneficiaries are not more likely to make active decisions, neither ex ante nor ex post. We show that these results are inconsistent with the class of boundedly-rational models of choice commonly used in the literature, implying that optimal default policy should be more paternalistic rather than one that attempts to `shock' consumers into making active choices.Discussant(s)
Jason Abaluck
,
Yale University
Mark Shepard
,
Harvard University
Anna Sinaiko
,
Harvard University
Dmitry Taubinsky
,
University of California-Berkeley
JEL Classifications
- I1 - Health
- D8 - Information, Knowledge, and Uncertainty