Tuesday, September 01, 2015

Five factors driving (or inhibiting) CSR takeup in ACA private plan markets

I want to lay out some notes here for a regression analysis of what drives takeup (or the lack of takeup) of Cost Sharing Reduction (CSR) subsidies among buyers of private health plans on ACA exchanges. CSR reduces out-of-pocket costs for buyers with incomes under 251% of the Federal Poverty Level (FPL).  The benefit is quite strong up to 200% FPL, but almost negligible in the 201-250% FPL range.

CSR is best understood not as some obscure secondary benefit but as the ACA's best defense against underinsurance -- that is, against leaving plan holders on the hook for more medical expenses than they can afford. Thanks mainly to CSR, about 60% of buyers on ACA exchanges buy insurance with an actuarial value of 80% or higher - - coverage comparable to or more comprehensive than that offered by most employers. Without CSR, only 10% of exchange customers would access that AV level. CSR provides insurance with AV 87% or 94% to about half of ACA exchange customers.

CSR is a leaky vessel, however, Only about three quarters of those who are eligible access the benefit, including probably a bit over 80% of those eligible for "strong" CSR (AV 87% or 94%). Silver plan premiums can be a hard swallow for low income buyers. In 2015, somewhere between 15% and 20% of buyers under 201% FPL probably opted for cheaper bronze plans with their sky-high deductibles (usually over $5,000 per individual).

CSR takeup among all eligible buyers varies quite a bit from state to state, most commonly between 70% and 80%, more broadly between about 68% and 85% -- discounting a few states that have layered their own benefit structures on the national ACA template (e.g., Vermont and Massachusetts).  In various posts, I've spotlighted factors that have an impact (or may have an impact) on CSR takeup levels, though none form a basis for consistent predictions. Below, I've listed those factors in what I would guess to be descending order of likely impact.

1. Income distribution of the state's subsidized private plan buyers. Two factors are at work here. First, most obviously, is state median household income. Second, perhaps even more importantly, is whether the state accepted the Medicaid expansion initially mandated by the ACA and later made optional by the Supreme Court. In nonexpansion states, eligibility for private plan subsidies begins at 100% FPL versus 138% FPL in expansion states (where applicants below that threshold are eligible for Medicaid). Silver plans are most affordable for the should-have-been-Medicaid-eligible group -- the premium for the benchmark silver plan is capped at 2% of income up to 138% FPL. For a buyer at 200% FPL, the benchmark costs more than 6%  of income ($118 for an individual in 2015).

Most but not all of the states that refused the Medicaid expansion are low income, and all but one of the refusenik states also refused to establish their own exchanges and so rely on healthcare.gov.  In the 21 nonexpansion states using healthcare.gov, an astounding 50% of private plan buyers for whom income data is available had incomes below 151% FPL; in expansion states using the platform, just 24% did.

For the most part, low income nonexpansion states have high CSR takeup. In Mississippi (median household income $40.9k), 85.5% of CSR-eligibles bought silver; in Alabama (median HH income $41.4k), 83.2% did. Louisiana breaks the pattern, however: despite a median income of $39.6k, lowest in the nation, just 70.6% of CSR-eligibles selected silver plans and accessed the benefit. What gives, Louisiana? I do not know.

2. Website design. Some state-based exchanges provide clear explanations in the right context of how CSR works. Others, not so much. Several -- e.g., Connecticut, New York, Rhode Island and Maryland (which adapted CT's interface in 2015) actively steer CSR-eligible buyers toward silver, putting silver plans at the top of the menu when it comes time to select a plan. Connecticut, New York and Rhode Island all reported CSR takeup in the 86-89% range for buyers under 201% FPL. Healthcare.gov, serving 37 states, does a better job highlighting CSR than some states and a worse job than others: it provides a pop-up warning to CSR eligibles if they're about to select another metal level, but then jumps them right back to a menu that ranks plans by premium (rather than foregrounding silver plans) if they continue shopping. As noted above, there is wide variation in CSR takeup among states using hc.gov, even when their income distributions seem similar.

3. State health measures. Sick people who know they need a lot of medical attention may not know what a deductible is when they shop on an ACA exchange, but I suspect they're likely to figure it out: when a plan summary shows a large "$6,600" over the word "deductible, that's a pretty strong inducement to click through and read the definition. In this post, I laid together state measures for life expectancy, diabetes and obesity with median household income for the 10 states with the highest and lowest rates of bronze plan selection. There's a relationship between state health measures and CSR takeup, but those measures tend to correlate pretty closely with state household income, and the relationship of all these factors with metal level selection is by no means linear.

4. Enrollment assistance.  I have not studied the relationship between the quantity and type of enrollment assistance provided in a state and CSR takeup in that state, in part because there is no obvious relationship. Some states that worked really hard on ACA outreach have low silver selection rates, and many of the poorest states have really high silver selection (e.g., Mississippi, where outreach and plan choice were both quite limited). In some states where the government did not lift a finger, the nonprofit community may have been effective. California, where state officials went all-out on outreach, learned to rely heavily on for-profit brokers. In some parts of Florida's Miami-Dade County, where enrollment in 2015 went through the roof, storefront brokerage seems to have taken off. There's just a lot of variation on this front. Broadly, about half of ACA enrollees seem to have gotten some third-party help.  Poorer buyers are likelier to have been outreach targets - - and so, perhaps, to have been steered into CSR.

5. Price structure. Excepting a handful of blue states that added their own benefits  -- via a Basic Health Plan for those under 200% FPL  (MN and, in 2016, NY) or added state subsidy (MA) -- the ACA ensures that every subsidy-eligible individual will pay the same amount for the benchmark silver plan in her region as everyone else in the country with the same income. For example, everyone earning $23k in 2015 paid $118 (give or take a buck) for the benchmark second-cheapest silver. Higher base prices in many cases simply mean higher subsidies. What varies, though, is the spread between the benchmark plan and other options. The best permutation for the CSR-eligible is a cheapest silver plan that is much cheaper than the benchmark second-cheapest silver, allowing discounted access to CSR. This happened in the Philadelphia area (and to a lesser extent in the Pittsburgh area) in 2014, and outreach groups signed up large quantities of applicants with incomes under 138% FPL for a silver plan that cost many buyers a pittance per month.(That spread disappeared in 2015 -- but many who benefited from it became eligible for Medicaid as the state belatedly accepted the expansion.)

In some areas, too, bronze plans may have lower-than-average deductibles (e.g., $2,500 in my own zip code, 07079, in a suburb outside of Newark, NJ) or significant benefits before the deductible kicks in,,such as relatively low co-pays for doctor visits or $5 generic drugs. Cheap bronze plans may be particularly tempting to older buyers, whose larger subsidies (to cover higher base prices for older buyers) can cover all or most of a bronze premium.

Those are the five factors I know of. I suppose that variations in the age and gender composition of different state markets may play a role. Readers, do you know of others? If you have the knowledge, means and inclination to do a regression analysis, I'd like to work with you and learn from you.

(originally published 8/31/15)

No comments:

Post a Comment