Price Searching in the Health Insurance Exchanges

This paper empirically explores the premium-setting behavior by 11 Texas insurers who participated in the individual insurance exchange during the three open enrollment periods in 2013-2015.  Premium data from www.healthcare.govfor all Bronze, Silver and Gold plans offered by each health insurer in each of the 254 Texas counties are used together with county level data from the Area Resource File, the AHA Annual Survey of Hospitals, the Bureau of the Census, and the RWJF County Health Ranks. A series of carrier-level fixed-effects regression models are used to examine the premiums charged for a 27-year-old non-smoker for each Bronze, Silver, and Gold plan offered in each county in Texas for each plan year.  Additional, county-level analyses examined premiums for the second-least costly Silver plan, the benchmark for establishing Affordable Care Act subsidies.  Four findings emerge from the analysis. First, there was considerable movement in premiums from year 1 to year 2 of the exchanges across all plan types with most carriers reducing their premiums relative to Blue Cross Blue Shield (BCBS) in year 2.  This process continued for Silver and Gold plans in 2013 but not for Bronze.  Second, narrow provider networks, characterized as HMOs and EPOs [exclusive provider networks], were associated with substantially lower premiums. Third, greater numbers of potential subsidized demanders (e.g., larger counties and those with a higher proportion of uninsured residents) had substantially lower premiums. Fourth, there was little evidence that carriers used publically availably county level data on health status to help set premiums.  The first year of the exchanges appears to have been wracked by insurer uncertainty. All other carriers set their premiums above those of BCBS.  However, in the second year most other carriers had reduced their premiums relative to BCBS and two new carriers had entered the state exchanges.  This refinement of premium offerings continued in the third year, but only for Silver and Gold plans. While BCBS offered coverage in all Texas counties, the regional and national carriers were more selective in their choice of counties. Nonetheless, other carriers appear to have been able to challenge BCBS by their use of local provider networks. Notably, HMOs and EPOs had substantial price advantages over PPO plans.  The findings provide insight into the events that transpired subsequent to the second year of open enrollment in Texas. First, a new second year entrant, Assurant, left the market.  The findings suggest that their offerings were overpriced relative to all other carriers in the State. Second, BCBS announced that it had lost $400 million in the Texas non-group market and it was re-structuring its exchange plan offerings to only include HMOs in year 3.  More generally, the results suggest that health insurance markets are inherently local and largely defined by the ability to establish attractive networks with local providers.