Decomposing the Effect of Health Plan Characteristics on Premiums: A Comparison of the Individual and Small Group Markets in 2016

As individual health insurance markets continue to undergo some turmoil, movements to merge the individual and small group markets may return to the forefront at least in some states. Any such policy discussions will need to be guided by greater understanding than we currently have on differences and similarities between the two markets. In this paper, we estimate hedonic premium models to measure the effect of health plan characteristics on premiums in the individual and small group health care markets. Our primary data source is the Robert Wood Johnson Foundation’s 2016 HIX Compare dataset, which includes detailed information on premiums and cost-sharing provisions (deductibles, coinsurance and maximum out-of-pocket limits) for all plans offered in the 50 states and District of Columbia. We then go on to conduct an Oaxaca-Blinder decomposition to estimate how much of the average difference in premiums in the two markets is due to differences in the distribution of plan characteristics and how much is due to differences in how the plan characteristics are priced in the two markets.

In 2016, we find that, on average, premiums in the individual market were approximately $1,650 lower than they were in the small group market. Roughly half of that difference is due to plan characteristics, with the largest contributors being higher deductibles, higher out-of-pocket maximums, and fewer more expensive plan types (PPO, POS) in the individual market. The other half is due to the net effects of differential pricing in the two markets. The intercept for the small group market is approximately $1,000 larger than that for the individual market which likely represents unmeasured quality. On the other hand, several plan characteristics, in particular out-of-pocket maximums, are more costly in the individual market as compared to the small group market. This may be due to less healthy risk pools in the former. Our results suggest that policies to merge the two markets will need to be attuned to quite substantial differences between them, differences that we observe both in the distributions of plan characteristics and the pricing of those characteristics.