Adverse Selection in the Health Insurance Exchange

Monday, June 24, 2019: 10:00 AM
Hoover - Mezzanine Level (Marriott Wardman Park Hotel)

Presenter: Karen Zhang

Co-Authors: Andrea DeVries; Ezra Fishman;

Discussant: Stephen Pickett

Adverse selection has been a key concern in the Health Insurance Exchange (the exchange). Prior to the enactment of the Affordable Care Act, empirical evidence on adverse selection in health insurance concentrated on the elderly population and group insurance market. Quantitative evidence on adverse selection in the non-elderly individual insurance market is lacking due to its historical experience rating and small market size. Under the new community rating regulation, the unknown effects of the population-specific adverse selection make the on-exchange plan design and participation decisions difficult. To the extent that the regulatory changes have contributed to increased adverse selection in the individual market, this can and likely has had an impact on the product offerings available in the marketplace.

The challenge eventually impacts individuals who seek coverage on the exchange, since they have to face fewer and less affordable plan choices.

A full assessment of adverse selection on the exchange promotes marketplace stability and consumer welfare. In this paper, we examine the adverse selection among the on-exchange insurees using administrative and claims data from a large U.S. insurer between 2014 and 2017. We focus on two types of adverse selection: riskier insurees’ enrollment in more generous plans, and their strategic decisions on when to enroll and how long to stay covered.

For adverse selection in plan generosity, we first detect a positive correlation between coverage and expected costs, which attests to the overall adverse selection among on-exchange insurees. We then leverage the similarity in the actuarial levels between on-exchange plans and commercial small-group plans to perform a difference-in-difference (DID) style analysis. In this analysis, we first match plans sold in the two markets based on cost-sharing levels, e.g. identifying “silver-equivalent” and “gold-equivalent” small-group plans. We then regress the expected costs on the full interaction between cost-sharing levels and on-exchange status. Our DID-style estimation quantifies the additional average costs in each non-Bronze tier relative to the Bronze among the on-exchange insurees using the small-group insurees as a comparison. We find clear evidence of adverse selection in all non-Bronze tiers in both on-exchange and small-group plans. The level of selection increases with plan generosity, and is more pronounced among on-exchange insurees than small-group insurees across all tiers.

For the adverse selection in enrollment timing and enrollment duration, we first conduct a positive correlation test using enrollment periods leading to and following a coverage gap. We detect a positive correlation between coverage and expected costs in enrollment following a coverage gap, but not in those leading to a coverage gap. We then conduct a difference-in-difference-in-difference style regression of the expected costs on the three-way interaction between enrollment duration, cost-sharing levels, and on-exchange status. We find no evidence of adverse selection in enrollment duration among small-group insurees. For on-exchange insurees, only those who enroll in Gold or Platinum are making strategic enrollment choices, based on analysis of enrollment duration and costs; In our survival analysis on enrollment duration, we find premium subsidy receipt significantly increases an on-exchange insuree’s probability to stay in coverage.