Provider Network Size and Marketplace Plan Premiums

Limiting provider network size has long been a strategy employed by insurance companies to gain bargaining power with providers and keep prices low for consumers.  This strategy gained renewed attention with the Health Insurance Marketplace, where the organized and regulated environment for shopping for insurance primarily for low-income price-sensitive consumers, provided few opportunities for insurers to keep premiums low. Mandated to provide minimum essential benefits at predetermined levels of actuarial value, narrowing network breadth remained a strategy to help hold down costs by allowing insurers to negotiate more favorable rates and select providers that may provide lower cost care. Yet little is known about the extent to which premiums are reduced with reductions in network breadth.  In this paper, we explore this relationship among the silver plans offered on the ACA marketplace. Network size information comes from a unique dataset of physician networks for the silver plans in the 2014 marketplace. To construct this dataset, we identified the 395 unique provider networks offered
by the 268 different issuers from the 2014 list of all unique silver plans sold in the marketplaces for all 50 states plus DC. We used the publicly available provider directories on the issuer websites to gather all providers in specified networks including data on provider characteristics such as specialty, name, gender, and geographic location. Our analysis dataset consisted of 450,232 physicians participating in plans issued by 267 carriers across 355 networks where we were successful in gathering information on all physicians and all physicians that are not in any network but included in the SK&A provider list of all physicians. Network size is estimated from these data by network within rating area as the fraction of participating physicians within a rating area relative to the count of physicians in that rating area. Overall, our data sample included 90% of all silver plan networks in the 2014 exchanges. We supplement these data by the HIX Silver Plan Comparison dataset, which provides the premium and cost-sharing features of all silver plans offered for each rating area where each plan is sold within each state.  We estimate hedonic pricing models with market (rating area) and insurer fixed effects for plan premiums where the unit of observation is the plan-rating area. Our regressand is the log of plan premium for a 27-year-old single policyholder and the regressor of interest is network size. We include control variables to account for non-network related variation in different plans such as plan deductible, copay/coinsurance, and maximum out of pocket information and standard errors are adjusted for ratings area-carrier network level clusters. We weight observations to reflect the relevant population in rating areas with access to each plan. Our preliminary findings suggest that one percentage point increase in network size is associated with a 0.6% increase in premiums. As enrollment information becomes available, consumer sensitivity to the tradeoff between lower premiums and narrow networks can be determined and the value of network breadth can be explored in more depth.