In October 2017, the Trump administration announced that it would stop reimbursing Affordable Care Act (ACA) Marketplace health insurers for cost-sharing reductions (CSRs), the reduced cost sharing they are required to provide to certain lower-income consumers. President Donald Trump, along with most market participants and analysts (including myself), predicted that stopping CSR payments would cause chaos in the ACA Marketplaces. At the same time, analysts also predicted that—following the initial disruption—ending CSR payments would ultimately lead to increases in total financial assistance for Marketplace consumers, lower net premiums for many moderate-income people, and higher Marketplace enrollment.
Almost two years later, the ACA Marketplaces have transitioned to the loss of CSR payments more smoothly than most observers expected. Meanwhile, the change has indeed resulted in lower premiums for millions of people. Analyzing enrollment data across states and income groups, I find that these premium reductions have increased enrollment in ACA plans, compared to a scenario in which the administration continued to make CSR payments.
A year ago on the Health Affairs Blog, Andrew Sprung and David Anderson similarly examined enrollment changes from 2017 to 2018 by income group, finding higher enrollment growth or smaller declines for income groups benefiting more from silver loading. This post updates and builds on their analysis by also taking advantage of state variation in the impact of silver loading on premiums for subsidized consumers, examining the total ACA-compliant market as well as Marketplace enrollment for 2018, and quantifying the overall impact.
I estimate a roughly 500,000 (5 percent) increase in 2019 plan selections by consumers eligible for premium tax credits (PTCs), with no evidence that this increase was offset by large reductions in enrollment in ACA plans among consumers not eligible for PTCs. Moreover, the policy change had a larger effect on enrollment in 2019 than 2018, and its impact will likely continue to grow.
Background
The ACA requires insurers to provide cost-sharing reductions, known as CSRs, to consumers with incomes below 250 percent of the federal poverty level who enroll in silver-tier Marketplace plans. CSRs increase the actuarial value of silver plans from 70 percent to 94 percent for eligible consumers below 150 percent of the poverty level, to 87 percent for those between 150 percent and 200 percent of the poverty level, and to 73 percent for those between 200 percent and 250 percent poverty level. This substantially lowers deductibles and other cost sharing.
The ACA envisioned that the federal government would reimburse insurers for the cost of CSRs, but these reimbursements ended in October 2017, when the administration dropped its appeal in litigation challenging the payments.
While the federal government is no longer providing reimbursement, Marketplace insurers must still provide CSRs to eligible consumers who enroll in silver-tier plans. Insurers have therefore built the cost of CSRs into premiums—in most states, into silver-tier premiums only.
Known as “silver loading,” this practice not only ensures that insurers are still compensated for CSR costs, it also lowers premiums for many consumers. That’s because PTCs are determined based on silver-tier premiums—specifically on the premium of the second lowest-cost available silver plan, known as the benchmark plan—but can also be used to purchase plans in other metal tiers. Since silver loading increases silver premiums but not bronze, gold, or platinum premiums, the increase in PTCs reduces net premiums for these non-silver plans, often substantially. In a typical state, silver loading reduced annual premiums for non-silver plans by about $1,100 in 2018 for a 45-year-old, PTC-eligible consumer. For PTC-eligible consumers purchasing silver plans, silver loading leaves net premiums roughly unchanged, since premiums and PTCs go up by about the same amount.
Analytic Approach
I use variation in the impact of silver loading across states and income groups to estimate how silver loading has affected Marketplace enrollment, compared to a scenario in which the administration continued to make CSR payments.
I measure the impact of silver loading in a state by calculating the 2017–19 change in the premium wedge between the benchmark silver plan and the lowest-cost bronze plan (the change in the silver-bronze wedge). This measure captures the differential impact of silver loading on premiums for silver plans compared to other metal levels, which in turn determines the extent to which silver loading increases PTCs and reduces net premiums for PTC-eligible consumers.
Across states using the HealthCare.gov eligibility platform, the average benchmark premium was 41 percent higher than the average lowest-cost bronze premium in 2019, compared to 21 percent higher in 2017. This 20-percentage-point change in the average silver-bronze wedge is consistent with ex ante projections of the impact of ending CSR payments from researchers at the Henry J. Kaiser Family Foundation and the Congressional Budget Office. But this measure varies widely across states: Not all state regulators have endorsed silver loading; states differ in the share of subsidized enrollees eligible for the most generous CSRs; and insurers differ in their actuarial assumptions, for example, their projections of how many non-CSR-eligible people would continue to purchase silver plans despite silver loading (issues David Anderson explored recently on this blog).
Important to my analysis, the impact of silver loading also differs significantly across income groups.
- Silver loading is least beneficial to people with incomes between 100 percent and 200 percent of the poverty level, for whom silver plans are generally still the best choice due to the substantial cost-sharing assistance they offer. Consistent with this, exhibit 1 shows little shifting out of silver plans by people in this income group since 2017.
- Silver loading has some benefit for people with incomes between 200 percent and 250 percent of the poverty level, but these consumers still give up some cost-sharing assistance by enrolling in non-silver plans.
- It is most beneficial to people with incomes between 250 percent and 400 percent of the poverty level, who are eligible for PTCs but not cost-sharing assistance. Consistent with this, exhibit 1 shows that this group has made the greatest shift out of silver plans since 2017.
- Consumers with incomes above 400 percent of the poverty level (those ineligible for PTCs) should be largely unaffected by silver loading, compared to if CSR payments had continued. While silver loading increases premiums for silver plans, unsubsidized consumers can avoid these premium increases by purchasing plans in other metal tiers. In most states, they can also avoid premium increases for silver plans by purchasing them outside the Marketplaces, as about 70 percent of unsubsidized consumers purchasing ACA-compliant plans do.
Exhibit 1: Share of HealthCare.gov consumers purchasing silver plans
Source: Author’s calculations from the Centers for Medicare and Medicaid Services public use files. FPL is federal poverty level. Note: * Includes those above 400 percent of poverty or not reporting income as well as those below 100 percent of poverty (potentially eligible for subsidies).
My approach is to regress changes in state enrollment on states changes in the silver-bronze wedge from 2017 to 2018 and 2017 to 2019. Of course, the drivers of the change in the silver-bronze wedge could be correlated with other drivers of state enrollment changes. For this reason, I run the regression separately for five income groups. If the regressions are in fact capturing the causal impact of silver loading, rather than other state-level drivers of enrollment, estimates should be near-zero for people with incomes below 200 percent of the poverty level (since they derive little benefit from silver loading, as discussed above) and should be largest for those with incomes between 250 percent and 400 percent of the poverty level.
I obtained data on enrollment and average premiums paid for 2017–19 for HealthCare.gov states from the Centers for Medicare and Medicaid Services (CMS) public use files, data on total ACA-compliant market enrollment by state for 2017–18 from the CMS risk adjustment reports, and data on premiums by state from the Henry J. Kaiser Family Foundation.
Premium Reductions, Enrollment Increases From Silver Loading
Exhibit 2 confirms that silver loading has reduced premiums for PTC-eligible consumers. It shows a strong correlation between the change in the silver-bronze wedge 2017–19 and the change in average net-of-PTC premiums for subsidized consumers (correlation = 0.70). The relationship is similar for 2017–18 (correlation = 0.66).
Exhibit 2: Relationship across states between silver loading and net premiums for PTC-eligible consumers
Source: Author’s calculations from the Centers for Medicare and Medicaid Services public use files and the Henry J. Kaiser Family Foundation data. Note: PTC is premium tax credits.
Exhibit 3 presents the key findings of the analysis. It shows the coefficients and standard errors from regressions of the percent change in a state’s enrollment in the specified income group on changes in the silver-bronze wedge in that state, for 2017–18 and 2017–19.
Exhibit 3: Regression results
Source: Author’s calculations. Notes: Exhibit reports coefficients and standard errors from a regression of changes in enrollment on changes in the silver-bronze wedge. + , ^ , and * indicate statistical significance at the 90 percent, 95 percent, and 99 percent confidence level, respectively. FPL is federal poverty level.
As shown in exhibit 3 and exhibit 4, by 2019, there is a strong positive relationship between the change in the silver-bronze wedge and changes in enrollment among the income groups benefiting most from silver loading. For example, for people with incomes between 300 percent and 400 percent of the poverty level, each additional percentage-point increase in the silver-bronze wedge results in a 0.75 percent increase in enrollment, an estimate that is statistically significant at the 99 percent confidence level. It is also substantively significant, implying that the 20-percentage-point increase in the silver-bronze wedge across HealthCare.gov states more than accounts for the 11 percent increase in enrollment for this income group between 2017 and 2019.
Exhibit 4: Relationship between the impacts of silver loading and enrollment for PTC-eligible, non-CSR-eligible consumers
Source: Author’s calculations from the Centers for Medicare and Medicaid Services public use files and the Henry J. Kaiser Family Foundation data. Notes: PTC is premium tax credits. CSRs is cost-sharing reductions.
Supporting the interpretation that these estimates reflect the causal impact of silver loading on enrollment, the coefficients are smaller for income groups that benefit less from silver loading. For people with incomes between 100 percent and 200 percent of the poverty level, estimates are much smaller and in no case statistically significant at the 95 percent confidence level. For those between 200 percent and 250 percent of the poverty level, estimates are smaller than for groups ineligible for cost-sharing assistance.
Exhibit 3 also shows that the regression estimates for 2017–18 follow a similar pattern by income group but are uniformly smaller than for 2017–19, suggesting that the impact of silver loading is growing over time. One would expect this to occur for several reasons.
First, about a fifth of HealthCare.gov plan selections are automatic re-enrollments, in which current consumers who do not take action to select a plan are automatically re-enrolled into the same plan or another plan in the same metal tier. About 1.3 million people who selected a silver plan before silver loading were automatically enrolled into silver plans for 2018. Supposing 40 percent attrition of automatically re-enrolled consumers throughout the year, that would mean about 750,000 such people were automatically re-enrolled for 2019, and about 450,000 will be for 2020.
Second, lower premiums resulting from silver loading likely increase retention over the course of the year, boosting enrollment for the next year. That would have no effect on enrollment in 2018 and a growing effect over time.
Third, consumers may simply become more responsive to silver loading over time as they become more familiar with it or as states or insurers undertake additional outreach or education.
Exhibit 4 does not show estimates for consumers with incomes above 400 percent of the poverty level because such estimates are difficult to interpret. While enrollment changes for this group are negatively correlated with changes in the silver-bronze wedge, this could be either because these consumers are exiting the ACA-compliant market altogether or because they are switching from Marketplace plans to off-Marketplace ACA-compliant plans (which in most states are not affected by silver loading).
Consistent with the latter interpretation, the last two rows of exhibit 3 show results from a regression of 2017–18 changes in total ACA-compliant enrollment on changes in the silver-bronze wedge. There is a strong positive relationship, which is only slightly reduced if I exclude states that responded to the loss of CSR reimbursements by requiring insurers to increase premiums for all metal tiers. (Excluding these states maintains the counterfactual of continued CSR reimbursements.)
Comparing the coefficients and taking into account that consumers with incomes between 200 percent and 400 percent of the poverty level comprise only about a fifth of the full ACA-compliant market, the results if anything suggest a larger increase in total compliant market enrollment than in PTC-eligible enrollment as a result of silver loading. This could in part result from silver loading increasing retention, as discussed above. But it is difficult to reconcile these results with large negative effects of silver loading on enrollment among unsubsidized enrollees.
Summing Up
My analysis implies that the administration’s decision to end CSR payments has increased HealthCare.gov enrollment among PTC-eligible consumers by about 400,000 people. Extrapolating these results to state-based Marketplaces (for which data on enrollment by income group are available for 2019 but not 2017 and 2018) implies a total increase of about 500,000, or about 5 percent.
This estimate is lower than the RAND Corporation’s recent model-based estimate of an 800,000, or 7 percent, rise in subsidized Marketplace enrollment. That’s not surprising, since its model-based estimates approximate the long-run impact of the policy. As discussed above, the enrollment impact of silver loading appears to have grown significantly between 2018 and 2019 and will likely continue to grow.
Already, however, the data show that silver loading has increased Marketplace enrollment, likely contributing to lower uninsured rates among moderate-income people and a healthier Marketplace risk pool. Restoring CSR payments would reverse these effects, unless it were paired with increases in PTCs of similar magnitude to those that have resulted from silver loading.
Along with other evidence on the impact of increased financial assistance, the enrollment gains from silver loading also suggest that PTC increases could significantly increase Marketplace enrollment and reduce uninsured rates, particularly among the already PTC-eligible consumers who make up the majority of the Marketplace-eligible uninsured.
