Insurance Pricing and Narrow Networks on the Federally Facilitated Marketplace

Tuesday, June 14, 2016: 10:15 AM
B21 (Stiteler Hall)

Author(s): Jeffrey S. McCullough; Jean Abraham; Coleman Drake; Kosali Simon

Discussant: Aditi Sen

The Affordable Care Act of 2010 established newly created Exchanges for individual health insurance. While many aspects of Exchange-based plan benefit designs are heavily regulated, insurers have wide latitude in the construction of their provider networks. There is widespread concern that insurers’ networks are “too narrow” and that beneficiaries may face serious access and choice problems. Narrow networks could, however, be used by insurers to negotiate lower prices; thus lowering the cost of insurance and medical care. We explore the relationship between provider network size and insurance premiums.

We employ three datasets in our analyses. First, the Qualified Health Plan (QHP) Landscape file from 2014 includes detailed information on insurer premia and benefit designs for firms participating in the 34 Federally Facilitated Marketplace (FFM) states. These data describe market structure, insurance premiums, and benefit generosity. The 2012-2013 Area Health Resource File (AHRF) provided demographic and market characteristics that may influence both the cost and demand for medical care. Third, we constructed a database of participating providers for each insurance product offered on the FFM for five states; Georgia, Maine, Utah, and Virginia. These data describe the specialty and location of physicians and related ancillary providers as well as the identity of participating hospitals. We measure the count and share of providers within a market by major specialty categories.

We estimate a hedonic pricing model that regresses insurance premiums on network size. These models condition on product characteristics such as insurance plan type (e.g., PPO, HMO, etc.) and benefit generosity as well as a market fixed effect. Network size parameters are thus identified by variation across insurers within markets. Preliminary results suggest a relatively weak correlation between network size and insurance premiums. This suggests that, at least in the short run, narrow networks do little to control the cost of health insurance. However, we observe almost no cross-product within-insurer network variation despite the fact that the average participating insurer offers approximately 12 products per market. This suggests that constructing a provider network is costly. Anecdotal evidence suggests that many insurers used networks negotiated for other products not offered on the FFM and we might expect the relationship between networks and premiums to change in the coming years.