Tuesday, April 25, 2017

The Kaiser misunderstanding

The Kaiser Family Foundation has a useful measure of how close to capacity the ACA marketplace is operating. Well, check that...Kaiser has two measures, and one has generated some confusion for some time.

The more useful measure is a national and state-by-state estimate of  Marketplace Enrollees Receiving Financial Assistance as a Share of the Subsidy-Eligible Population. For this, Kaiser combines analysis of enrollment data provided by CMS with Census data on income and insurance status. Importantly, Kaiser accounts for people whose income would qualify them for subsidies but who are disqualified by an offer of employer insurance, as well as people disqualified by immigration status.

As of  March 31, 2016*,  Kaiser estimated that 64% of those eligible for premium subsidies nationally were enrolled. State scores ranged from 92% in Florida, which has developed a culture of enrollment, with plenty of assistance available and advertised, to 31% in Colorado.** Kaiser estimated last October that 5.3 million of the uninsured were eligible for marketplace subsidies.

The less useful -- and more often cited -- Kaiser measure is Marketplace Enrollment as a Share of the Potential Marketplace Population -- without reference to subsides. The problem here is that the measure does not take into account those who enroll in individual market plans sold outside the ACA marketplace. By now most of those plans are ACA-compliant. People who know they are not eligible for subsidies have no reason to buy their plans through the marketplace rather than directly through an insurer or via a broker or online broker such as eHealth. By most estimates a bit more than a third of ACA-compliant plans are sold off-marketplace.

Hence the "Share of Potential Marketplace Population Enrolled" is confusingly low -- 40% nationally at present. That number doesn't tell us much.

The individual market as a whole is operating at over 60% of capacity -- too low, perhaps, but far from a dysfunctional 40%.  Kaiser estimates total potential "marketplace" enrollment at 27.4 million. That's effectively total potential individual market enrollment -- which at present stands at about 18 million, with about 1.7 million of those in pre-ACA ("grandfathered" or "grandmothered") plans, according to Charles Gaba's estimate. Insurance professionals sometimes cite 75% capacity as a desirable level.

Kaiser's broader "percent of potential marketplace enrolled" sometimes shows up in misleading contexts. A recent example is in the detailed "innovation waiver" proposal put together by a task force in Oklahoma and published under the auspices of state HHS secretary Terry Cline, which proposes a radical reshaping of Oklahoma's marketplace.  The report justifies the reshaping in large part by the poor performance of Oklahoma's marketplace to date, claiming, with a cite to Kaiser's percent of potential marketplace chart:
Oklahoma only had 31% of its eligible population (those with incomes between 100-400% of the FPL) purchasing coverage through the FFM in 2016, relative to an average of 43% among other states similar to Oklahoma.
That's inaccurate. Kaiser's 31% figure includes people in all income brackets, and it does not account for off-marketplace individual market enrollees. Kaiser's estimate for takeup among the subsidy-eligible in Oklahoma is 48% -- low, but not 31%.  

In fact, the Oklahoma  waiver concept paper estimates total individual market enrollment at 198,300. That's 47% of Kaiser's total potential enrollment estimate for the state  of 424,000 -- almost the same as Kaiser's estimate for subsidized potential enrollees. Again, not good, but not 31% either.

The Oklahoma task force error found its way today into a Kaiser Health News article about state attempts to stabilize their marketplaces, which asserted that  "only about 30 percent of eligible Oklahomans are enrolled" in the marketplace -- technically correct, but discounting off-marketplace enrollment. 

"Percent of potential marketplace population enrolled is not a meaningless statistic. While insurers who sell both on- and off-marketplace in a given market must combine their customers in a single risk pool, most states allow insurers to sell only off-marketplace if they choose -- which does fragment the risk pool.  In addition, as mentioned above, 1-2 million current enrollees are in ACA-noncompliant plans, and so also out of the risk pool.  The marketplace will be healthier when all those pre-ACA plans phase out, and it would be healthier if states required insurers in the individual market either to sell on-marketplace only (as D.C. and Vermont do) or to sell on-marketplace if they want to sell off-marketplace, as Maryland and Massachusetts do (or did, as of 2014).

Still, to the extent that Kaiser's estimate of the total potential marketplace is accurate, both the marketplace and the individual market as a whole are closer to reaching their potential than some are led to believe.

*  Based on a CMS Enrollment Snapshot as of that date, which captures the first wave of attrition following the end of open enrollment. Every year, a significant percentage of enrollees never pay their first-month premium. In 2016, enrollment as of March 31 was about 87% of enrollment at the end of open enrollment, Jan. 31.

** The District of Columbia shows just 17% of potential enrollees enrolled, but in DC a) Medicaid is available to people in households with incomes up to 200% of the Federal Poverty Level, removing the income group that generally has the highest takeup rates, and b) the totals are small -- only 1224 subsidy eligible people are enrolled out of a potential population of just 7000. While the percentage is anomalously small, DC's percentage of all potentially eligible marketplace enrollees who are enrolled is anomalously high at 58%. That in turn is because DC requires all plans offered on the individual market to be offered on the exchange. 

The Colorado figure also is anomalous, as there is a wide discrepancy between CMS's reported enrollment for the state as of 3/31/16, 108,311, and the state's own report that enrollment as of May was 146,000 (the state runs its own marketplace). That would boost the percentage of subsidy-eligible people enrolled to a still anemic 42%. As of the end of open enrollment 2017, the state reported 178,000 enrollees (March 2 report - probably does not account for post-OE attrition). Thanks to Louise Norris of healthinsurance.org and Colorado Health Insurance Insider fame for providing the state-reported stats. 


  1. To clarify on the Colorado data... I believe only 61 percent of the 2016 enrollees were eligible for subsidies. So although the 146,000 effectuated enrollment number is higher than what CMS had at the end of open enrollment, a significant portion of those enrollees were not receiving subsidies.

    1. Right, Louise -- I actually figured 90.5k subsidized out of Kaiser's estimate of 215,000 subsidy-eligible, for 42%. Kaiser had 67/108k subsidized as of 3/31/16, so I took the same proportion.