In order to standardize health insurance plans, the ObamaCare law requires insurers to offer plans that are categorized as Bronze, Silver, Gold and Platinum. These plans, in turn, must cover 60%, 70%, 80% and 90%, respectively, of the expected costs. The “actuarial value” of a claim is the average amount a plan with a given set of benefits is likely to pay given a standard population.
If the Department of Health and Human Services (HHS) overestimates actuarial value, which it appears that it has done, the bloated insurance plans required by ObamaCare can be offered with higher deductibles, copayments, and other out-of-pocket costs than would otherwise be the case. All else equal, higher deductibles and out-of-pocket costs also enable lower premiums.
Lower premiums improve the political palatability of the exchange plans. Because higher deductibles and out-of-pocket costs are relatively more attractive to the healthy than to the chronically ill, they are attractive to insurers seeking to improve the risk profile of the group of people signing up for their plans.
I saw the light
There is no standard for defining a standard population or for determining its health expenditures. Insurers have historically combined experience and actuarial judgment to arrive at their spending estimates. In 2011, The Kaiser Foundation conducted an experiment in which it asked three highly respected actuarial firms to estimate ObamaCare’s allowable Silver plan deductibles under a common set of assumptions. The standard populations differed, and the estimates of allowable deductibles ranged from $1,850 to $4,200.
People who purchase health insurance with their own money balance the extra cost of having an insurer cover smaller costs like routine screening against the higher premiums it will charge to do so. In 2009, almost half of all individual policies purchased for a single person had deductibles that were higher than $2,500. Some of the more popular pre-ObamaCare individual policies featured deductibles of $5,000 with no additional cost sharing after the deductible. People who chose those policies knew their annual financial risk in advance.
ObamaCare makes those policies illegal. It generally limits deductibles to $2,000 per individual and $4,000 per family in the unsubsidized individual and small group markets, even though total cost sharing can be as high as $6,250.
In its estimates, HHS defines actuarial value as
Proprietary data from Milliman suggest that the average annual per individual claims cost for a commercially insured population in 2010 was $4,000. Assuming that the deductibles were the entire amount of cost sharing, and a market average deductible of $2,500, the commercial insurance market’s HHS actuarial value would have been about 0.38. In order to qualify under ObamaCare, deductibles would have had to fall to $1,600, likely triggering an increase in premiums.
When the actuarial value calculations were carried out, the amount of cost sharing that ObamaCare allowed depended upon the government’s estimate of the overall plan costs for plans that have not yet taken effect. In a methodology report, HHS explained how it estimated plan costs for its standard population. With one exception, the choices made would likely overstate the cost of essential benefits coverage.
The HHS estimates are based on claims by 12.6 million people. They were culled from 2010 claims data for 39 million people provided by Blue Health Intelligence, a spin-off of the BlueCross and BlueShield Association. HHS discarded claims from individual policies, claims from groups that were not covered by PPO/POS plans, claims from groups without maternity claims, claims from groups of fewer than 50 people, and claims from groups that didn’t have at least one claim over $5,000. The report does not provide basic sample characterizations and is silent on a number of issues that are known to affect claims, things like age, geographic distribution, and the prevalence of union plans.
Plan design matters because plan design affects the claims that people make and the claims that plans pay. For example, the RAND Corporation has shown that the design of consumer directed policies tends to reduce health expenditures. If people with individual policies tend to have less expensive total claims than people with employer group policies, dropping them would inflate the estimates of actuarial value for the standard population.
Eliminating plans without claims of at least $5,000 is also likely to overstate the cost of covered benefits. In any given year, relatively few people have high health care claims costs. Willey et al. (Health Affairs, 2008) reported on average spending by people drawn from a managed care database of 26.8 million people in 2004/2005. In their sample, median spending was $989. Mean annual health spending for the plan population in normal health was $3,075 for both out-of-pocket and plan costs. People with chronic disease spent more, with mean spending of $8,225 and median spending of $3,321. For the severely ill, mean spending was $29,273 and median spending was $9,300.
After selecting its sample, the Centers for Medicare and Medicaid Services (CMS) adjusted the data in a variety of other ways that could produce inaccurate estimates of average annual health care payments.
Utilization of “habilitative services” was assumed to be cost-free, raising the question of why coverage for services that have no cost is required by law. Kids’ dental care and eyeglasses payments were estimated, annualized and multiplied by expected participation rates, without explanation.
Only in-network services were used to calculate maximum out-of-pocket limits. HHS defends this practice with the statement that the data suggest that out-of-network services were seldom used, a situation likely to change given the narrow networks characteristic of exchange plans. Without any information on how claims for individual policies behaved, CMS simply assumed that “small group participation is similar to individual market participation in terms of age and sex distribution.”
All cost data were trended forward to 2014, and “[s]everal adjustments were made to this data to more closely represent the expected population of enrollees in 2014.” The report does not say how this was done. It also fails to explain how CMS arrived at the seemingly arbitrary spending brackets it used to determine cost shares. It imputed benefit characteristics to the brackets by age and sex, and used nonlinear least squares to estimate the actuarial value based on the imputed cost shares in each bracket.
The HHS methodology generates further cause for concern when it notes that the overall means of the groupings used to model the Bronze and Silver plans were “distorted due to random observations of extreme spending above the 99th percentile.” But Milliman reports that just 0.2 percent of a typical commercially insured population will incur more than $100,000 in allowed medical claim costs in a year, an occurrence rate that though not random would be small enough to fit the HHS description of extreme spending above the 99th percentile. HHS says that it dealt with those “random observations” using an unspecified “ad hoc” adjustment. Whether the ad hoc adjustment under or over-estimates plan spending for the actuarial value calculation is impossible to determine.
In the future, HHS plans to allow states to determine their own actuarial values. It would be simpler to end the individual mandate and replace ObamaCare’s regulatory rat’s nest of subsidies based on estimates piled on estimates with a tax credit for the purchase of health insurance.