Monday, June 25, 2018

ACA marketplace enrollment, 2018: Answers and questions

In 2018, ACA marketplace enrollment dropped 5% in the 39 states using the HeathCare.gov and 4% nationally. The more detailed data that CMS compiles for the HealthCare.gov states (accounting for three quarters of all enrollees) show that enrollment dropped most sharply at the lowest income levels -- where, thanks to Cost Sharing Reduction (CSR) subsidies, the most comprehensive coverage is available.*

Here's the breakdown of enrollment by income level in 2018 vs. 2017** on HealthCare.gov.

Enrollment by Income Level on HealthCare.gov, 2018 vs. 2017


Total enrollment
100% to 150% FPL
150% to 200% FPL
200% to 250% FPL
250% to 300%  FPL
300%- 400%  FPL
Other FPL*
2018
              8,743,642
       2,979,236
            1,885,778
         1,277,488
       747,165
   867,198
            986,777
2017
              9,201,805
       3,208,242
                        2,050,555
         1,312,520
       752,403
   786,678
     1,091,407

2018 as % 2017
95.0%
92.9%
92.0%
97.3%
99.3%
110.2%
90.4%
 "Other FPL" is comprised mostly of unsubsidized enrollees. About one quarter are likely enrollees with incomes under 100% FPL, most of whom are likely legally present noncitizens time-barred from Medicaid, who are subsidy-eligible.

Why was enrollment down sharply at 100-200% FPL, down modestly from 201-300% FPL, and up at 300-400% FPL? The higher the income, the more definite the answers.

At 300-400% FPL, steep premium hikes rendered many people who were previously ineligible for subsidies subsidy-eligible.  At 300-400% FPL, subsidies become available if the benchmark silver plan premium exceeds 9.56% of the buyer's income.

There's no subsidy cliff to speak of for those for those with incomes under 400% FPL with respect to the benchmark (second cheapest) silver plan that determines subsidy size. For those at the edge of subsidy eligibility, the subsidy for the benchmark could be as little as $5 per month. The subsidy can be more valuable for plans that cost less than the benchmark, however. And this year,   the steep discounts in bronze and gold plans triggered by Trump's cutoff of CSR payments and the resulting "silver-loading" of CSR costs (see explanatory note below) may have made enrollment attractive to those who were previously subsidy-ineligible.

Silver loading also pretty plainly limited enrollment drops at 201-300% FPL, particularly in states where gold plans were cheaper than silver and where free or near-free bronze plans were widely available. In Pennsylvania, for example, 42% of enrollees with incomes in the 200-400% FPL range chose gold plans. In Kansas, gold selection was in the high thirties in that income bracket. Overall, nearly 2 million enrollees in the 200-400% FPL range benefited from bronze and gold discounts generated by silver loading.

The mystery is why enrollment dropped sharply in the 100-200% FPL range. There, CSR-enhanced silver was available on pretty much the same terms as always: benchmark silver would cost between 2% and 6.3% of income for a plan with an actuarial value of 94% (up to 150% FPL) or 87% (at 151-200% FPL). In some places, the cheapest silver plan might offer a CSR discount. This year, bronze plans were available for free to many at this income level, and in some places gold plans cost less than silver, though they also have a lower AV (80%) than CSR-enhanced silver plans (87% or 94%). So why the drop? Some possibilities:
  • Low income voters were more susceptible to the drop in enrollment assistance and advertising, as well as to the shortened enrollment period. 

  • To the extent that low income correlates with low information, more people in this bracket may have been under the impression that the ACA (or the individual mandate) had been repealed.

  • While AV stays the same year by year, out-of-pocket costs keep rising, as a fixed percentage of a rising cost of care. For a person with an income of $23,000, an 87% AV plan may not look so attractive if it carries a deductible of $1,000 or more. Similarly, continued narrowing of networks and tiering of networks may alienate some (e.g., note this apparently sophisticated reaction to network tiering in the Philadelphia area).
I suspect that cuts to in-person enrollment assistance may have had a strong impact among lower income prospective enrollees, as well as the shortened enrollment period.

More simply, the enrollment drop in the 100-200% FPL bracket (7.5%), where offerings were pretty much status quo ante, or degraded somewhat by rising costs and reduced competition, illustrates the extent to which the discounts triggered by Trump's CSR cutoff mitigated losses at higher income levels.
---
* CSR boosts the actuarial value of a silver plan to 94% for enrollees with incomes up to 150% of the Federal Poverty Level and to 87% for this with incomes in the 150-200% FPL band, higher than most employer-sponsored plans).

** These numbers are based on Public Use Files (20182017) published by CMS.

A note on those bronze and gold plan discounts

When Trump cut off federal reimbursement of insurers for the Cost Sharing Reduction subsidies they're legally required to provide to lower income ACA marketplace enrollees who select silver plans (57% of marketplace enrollees in 2017), most states allowed or required insurers to concentrate the cost of CSR in premiums for silver plans only. States in which 70% of individual market enrollees live also allowed insurers to sell cheaper silver plans off-exchange, with no CSR cost attached.

Since ACA premium subsidies are keyed to the price of the benchmark (second cheapest) silver plan in each rating area, subsidies rose to cover inflated silver premiums, generating often dramatic discounts in non-silver plans, i.e. gold and bronze (platinum availability and purchase is negligible). In many states, steep increases in silver plan premiums resulted in zero-premium bronze plans becoming available to many buyers (or nominal $1-3/month premiums), and gold plans that were either cheaper than silver or close in price.

Cheap gold plans were a particular boon to enrollees with incomes between 200% and 400% of the Federal Poverty Level (FPL). These buyers are not eligible for strong CSR, which makes silver plans roughly equivalent to platinum plans for buyers up to the 200% FPL threshold. Normally,  enrollees in the 200-400% FPL range would pay between 6% and 10% of their income (percentage rising with income) for a benchmark silver plan with an actuarial value of 70%, i.e. with an average deductible of around $3600). With CSR priced into silver plans in 2018, gold plans  (80% AV, with an average deductible of around $1100) came came within reach of many in this income range. Gold plan selection more than tripled in Pennsylvania and Maryland, for example.

---
Note: Before this post was published, I  rearranged it to highlight a striking but potentially unreliable spike in Hispanic enrollment, treated more fully in a subsequent post. For the record, I've attached my initial take in this post at bottom.

Update, 6:10 p.m. [revised 6/27] Re the apparent spike in Hispanic enrollment in 2018 noted at top below: while it appears to be an apple-to-apples comparison, it's hard to say whether it's meaningful. CMS reports enrollees' response to separate questions about their race and ethnicity. 2018 shows a large spike -- over 500,000 -- in the "unknown" category for race. In ethnicity, where Hispanic identity is recorded, CMS changed a category in 2018. In 2017, all those who didn't identify as Hispanic were simply classed as "not Hispanic"; in 2018, "unknown" was broken out separately from "not Hispanic." That change theoretically should not affect the difference in Hispanic self-identification, which rose by 77,000 in 2018. But the category shifts make me leery of taking the increase at face value. In any case, here is the scotched prior writeup.
---
Okay, I had to turn around a whole blog post because I've stumbled on a mystery.

I was trying to figure out why ACA marketplace enrollment dropped 7.5% among those with incomes between 100% and 200% of the Federal Poverty Level, while rising among those in the 201-400% FPL bracket (actually, rising sharply at 301-400% FPL and dropping slightly at 201-300% FPL).

The rise among those at 300-400% FPL is fairly easy to explain but the sharp drops at lower incomes is something of a mystery. After running through some likely reasons (see below), I thought that immigrants' wariness of  providing detailed personal information to the government in the Trump era might have something to do with it, so I took a look.

Guess what: Hispanic enrollment [self-identified in the ethnicity question] rose in 2018 --  at least in the 39 states using HealthCare.gov -- which include Texas and Florida. Hispanic enrollment on the federal platform was reported by CMS at 956,516 in 2017 and 1,033,699 in 2018. That's a 7.5% increase. Among all enrollees on HealthCare.gov in the 100-200% FPL income range, enrollment was down 7.5% in 2018.

What about Florida and Texas, which rank 1st and 3rd nationally in ACA marketplace enrollment? Hispanic enrollment in Florida was way up in 2018, from 327,965 to 378,471. That's a 15% increase, and accounts for most of the 39-state increase. Hispanic enrollment was up more modestly in Texas, from 306,741 to 313,098. Meanwhile, overall enrollment was down 2.5% in Florida and 8% in Texas.*

So what gives? A marketing blitz in Florida? That started years ago, when the Miami-Dade area busted out as the area with the highest concentration of ACA marketplace enrollees in the country. As Charles Gaba just pointed out today, the percentage of Floridians enrolled in the individual market is roughly double that of the country as a whole, at about 13%. (That's in large part because the state has refused to expand Medicaid; nearly half of the state's 1.7 million enrollees have incomes that would qualify them for Medicaid if the state accepted the ACA expansion.)

I do not have any insight into this Hispanic spike. Do you?  Meanwhile, here are my prior thoughts about why enrollment dropped at the lower range of subsidy eligibility and rose at the higher end.

No comments:

Post a Comment

Share