Turnover in Zero-Premium Status Among Health Insurance Marketplace Plans Available to Low-Income Enrollees

Key Points Question How often did silver-tier marketplace plans that were free to low-income enrollees in 2021 become nonfree in 2022—putting enrollees at risk of termination if they did not switch plans or initiate premium payments? Findings In this observational cross-sectional study of plan and enrollment data from 33 HealthCare.gov states, turnover of zero-premium plans occurred in 93% of counties (weighted by enrollment) from 2021 to 2022. Meaning In this study, the risk of coverage loss was elevated among a large number of enrollees experiencing free-plan turnover at the start of 2022; outreach or other steps could help mitigate terminations or reenroll affected people.


Turnover in Zero-Premium Status Among Health Insurance
Marketplace Plans Available to Low-Income Enrollees Edward Kong, Mark Shepard, Adrianna McIntyre Medicaid expansion and total enrollment shares Enrollment in our sample was concentrated in states that did not expand their Medicaid programs. This was especially true for enrollment in the 100-150% FPL income group (eTable 1). Unweighted results for Table 2 The results we present in the manuscript are weighted by enrollment in the 100-150% FPL income band.
This weighting ensures that our findings are not biased by giving sparsely populated rural counties equivalent weight as densely populated urban counties representing more of the affected population.
However, we thought it was important to present unweighted estimates for comparison (see eTable 2 below).

Bivariate regression results
We use bivariate regression to assess features associated with counties experiencing turnover in zeropremium silver plans. Below, we consider county shares of self-reported race (white, non-white, black), and ethnicity (Hispanic, non-Hispanic), and the share of enrollees in the 100-150% FPL income group.
We consider two outcome variables. Our first outcome is a binary variable for whether a county has at least one zero-premium plan in 2021 that has a positive premium in 2022. Our second outcome is a binary variable for whether all zero-premium plans in a county have a positive premium in 2022.
We restrict to counties with at least one zero-premium plan in 2021. All regressions are weighted by the number of enrollees in the 100-150% FPL income group and use robust standard errors. We report full regression results below (eTable 3).

A. At least one zero-premium plan in 2021 becomes a positive-premium plan in 2022
(1)

B. All zero-premium plans in 2021 become positive-premium plans in 2022
(1)

Results by quartiles of each characteristic
The relationship between the probability that one or all zero-premium plans become positive-premium plans and each characteristic above may not be linear (this is the case, in particular, for the share of black enrollees in each county). Below, we report results from regressions of the outcome on quartiles of each characteristic (eTable 4). Regressions omit the constant term to allow inference for all four quartiles without an omitted quartile. Independent variables are shown in each column heading.

Additional analysis: changes in carrier and plan counts
Below, we relate the likelihood that a county is affected (i.e., has a zero-premium plan become a positive-premium plan) to changes in either the number of carriers (i.e., the number of distinct insurers) or the number of plans (each insurer may offer multiple plans). For the analyses that follow, we limit to the 2,187 out of 2,449 counties with enrollment data for the 100-150% FPL income group and at least one $0 silver plan in 2021.

Carrier entry
We find that 56% of these N=2,187 counties experience no change in the number of carriers (synonymous with "issuers"). Among the remaining counties, the majority gain 1 or 2 carriers (27% and 11% of all counties respectively, see eFigure 1, Panel A below). Very few counties (N=10) experience a decline in the number of carriers by 1.
We find that counties gaining carriers are almost certain to have at least one zero-premium plan become positive-premium, whereas counties with no change in the number of carriers (or a drop in the number of carriers) are less likely to be affected (see binscatter below, eFigure 1, Panel B).

A. Change in the number of carriers per county B. Turnover is associated with carrier entry
NOTES: Histogram (Panel A) shows the change in number of carriers from 2021 to 2022 for each county. Binned scatterplot (Panel B) relates the change in number of carriers to the probability that a county has at least one zeropremium plan become a positive premium plan from 2021 to 2022. Counties adding 3 or more carriers are combined. Shaded area denotes 95% confidence interval based on robust standard errors.

Plan entry
We repeated the analysis above with plan entry rather than carrier entry. As shown in the histogram below, many counties (21%) experience no change in plans, but the majority of counties experience plan entry between 2021-2022 (70% of counties, eFigure 2, Panel A). Only 9% of counties experience a decline in plans. The binscatter below reproduces Figure 2 in the main text and shows that counties with no change in the number of plans are least likely to be affected, whereas counties with increases or decreases in the number of plans tend to see turnover in the set of zero-premium plans (eFigure 2, Panel B). The ~35% of counties that add 5 or more plans are virtually guaranteed to be affected by turnover in zero-premium plans.

A. Change in the number of plans per county B. Turnover is associated with plan entry and exit
NOTES: Histogram (Panel A) shows the change in number of plans from 2021 to 2022 for each county. Binned scatterplot (Panel B) relates the change in number of plans to the probability that a county has at least one zeropremium plan become a positive premium plan from 2021 to 2022. Points in the binned scatterplot represent within-bin averages for counties adding between -9 and -1 plans (average = -3), 0 plans, 1 plan, or 2-4 plans. A separate point at x = 5 represents counties adding 5 or more plans. Shaded area denotes 95% confidence interval based on robust standard errors.