ACA Enrollment and Adverse Selection Pressures – An Update

Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0

Richard Evans / Scott Hinds / Ryan Baum


203.901.1631 /.1632 / .1627 revans@/shinds@/


February 20, 2014

ACA Enrollment and Adverse Selection Pressures – An Update

  • For a podcast of this and other research notes, please see the SSR Health YouTube channel
  • Current health insurance exchange (HIE) enrollees skew older, to the extent that adverse selection pressures appear highly likely. Only 25% of enrollees are aged 18-34, as compared to the 40% enrollment share presumably needed for premiums and claims to balance across the HIEs
  • The argument is being made that enrollees’ age is less relevant than their health status, which is to draw a technically correct but practically meaningless distinction between known age and probable health. For this argument to pan out, the typical HIE enrollee must be dramatically healthier than his or her age and income would imply
  • We put numbers to this. At the currently enrolled age mix, for premiums collected to cover claims paid, and assuming HIE enrollees aged 35-64 have normal health for their age and incomes, the 18-34 year old HIE enrollees must:
    • be 21pct less like to incur an injury or develop an illness requiring urgent care;
    • weigh 17 pounds less;
    • have a 28pct lower rate of obesity;
    • have no physical limitations; and,
    • rate their health 25pct higher on a 5-point Likert self-assessment scale than the average 18-34 y.o. in the HIE eligible population
  • The likelihood of HIE enrollees being on average at all healthier than the broader population of persons eligible to enroll is very nearly zero; the likelihood of younger enrollees being this much healthier than their age and income matched peers is too small to take seriously
  • Adverse selection on the HIEs is far more likely than not, which implies significant policy changes in the relatively near term. Feasible options include: allowing greater premium differences based on age, allowing higher out-of-pocket maximums, reducing the scope of benefits, increasing subsidies, increasing penalties, and/or merging the relatively poor HIE risks into an existing risk pool having better average risks

The age-mix of health insurance exchange (HIE) enrollees (Exhibit 1, green columns) skews older than the age-mix we believe is necessary (Exhibit 1, grey columns) to avoid adverse selection, i.e. for premiums and claims to balance. Our independently derived estimate of ‘balanced’ age-mix – specifically the need for roughly 40pct of enrollees to fall in the 18-34 age range – is consistent with estimates[1] made by others – including CBO – before the enrollment process began

At present only 25% of HIE enrollees fall into the 18-34 age range, which strongly suggests the current enrollment mix will result in adverse selection, forcing policy changes in the near-term

The argument has been made that the age mix of enrollees is less relevant than the health of enrollees[2]. Despite being technically correct this argument is hopelessly evasive. It relies on the exceedingly narrow possibility that actual claims will be sufficiently different from normal age-related patterns (specifically that the enrolled are healthier than those that don’t enroll) for premiums and claims to balance. This begs the question of just how different actual claims would have to be from typical age-related patterns; quantifying that difference is the simple point of this research note

We estimated relative demand for health care by age and income using Medical Expenditure Panel Survey (MEPS) data for 2011 (the most recent year available). To quantify relative demand, we relied on variables that would not be affected by the presence or absence of health insurance, e.g. odds of illness or injury, body mass index (BMI), and self-assessed measures of health

Using age and subsidy-eligibility data from the most recent report on HIE enrollment[3], we then calculated enrollment weighted averages for each MEPS health-status variable, assuming in this calculation that HIE enrollees are a random grab of health risks, i.e. that HIE enrollees are no more or less healthy than their age- and income-matched peers in the general population[4]. The results are in columns (b) (all ages) and (d) (18-34 y.o.) of Exhibit 2. For reference, column (a) provides weighted average values that we would expect if the targeted age mix (e.g. 40pct of enrollees aged 18-34) were enrolled

Because the currently enrolled population is on average older than the targeted age mix, the age- and income-weighted health measures for the enrolled population obviously imply worse health (column (b)) than would be expected under the targeted age mix (column (a))

The question then simply becomes how much healthier than average the enrollees must be in order for these enrollees’ claims to be no greater than the claims that would have been expected under the targeted age mix. Specifically, we calculated how much healthier the enrolled 18-34 y.o.’s would have to be as compared to their age- and income-matched peers in the general population, assuming that 35-64 y.o. enrollees are a random draw from their age- and income-matched counterparts in the general population. We express the result in two forms: as the value the MEPS health status variables would have to reach for the 18-34 y.o.’s to be sufficiently healthy to pull the enrolled population’s health status in line with that of the targeted age mix (column (e)) ; and, as the required percentage change in each MEPS health status variable (column (f))

Our results indicate that in order for the enrolled age-mix to have claims (and thus a relationship between premiums and claims) on par with that of the targeted age-mix, enrolled 18-34 y.o.’s would have to be 21pct less likely to incur an injury or have an illness that required urgent care, have an 11pct lower body mass index (BMI – N.B. this implies being 17 lbs. lighter on average), have a 28pct lower rate of obesity, have no physical limitations, and to rate their health an average of 25pct higher on an 5-point Likert self-assessment scale, all as compared to age- and income-matched peers

It is wholly unrealistic to expect HIE enrollees to be at all healthier than non-enrollees, much less by the degree necessary to overcome the current skew in enrollment toward older beneficiaries. We believe health insurance exchanges are a crucial structural feature of any truly efficient health system; as such if we have any subjective bias, it takes the form of a genuine hope that the HIEs succeed. Despite this hope, we see no reason to ignore the considerable objective evidence pointing to failure of the HIEs in their current form, and believe that otherwise credible sources simply reduce the odds of righting the HIEs by suggesting enrollment is on track

  1. See, among many examples,
  4. If anything, we believe HIE enrollees are less healthy than non-enrollees. However by assuming the HIE enrollees have the same average health as the general population (on an age- and income-adjusted basis), we stack the odds in favor of the analysis finding less potential for adverse selection, thus the assumption is conservative
Print Friendly