A Simple Yardstick for HIE Enrollment: CBO’s Revised Estimate Looks Like an Outer Bound

Print Friendly, PDF & Email

Richard Evans / Scott Hinds / Ryan Baum


203.901.1631 /.1632 / .1627

https://twitter.com/images/resources/twitter-bird-blue-on-white.png revans@/shinds@/rbaum@ssrllc.com


September 24, 2013

A Simple Yardstick for HIE Enrollment: CBO’s Revised Estimate Looks Like an Outer Bound

  • CBO* estimates that roughly 20 million subsidy-eligible persons will enroll on the individual health insurance exchanges (HIEs) by 2017, and that this level of enrollment will remain roughly stable over the forecast period (2023)
  • We believe the 20 million enrollee level is much more of an outer-bound than an expected value; and, barring changes in the Affordable Care Act (ACA) that enrollment will deteriorate over time, rather than remaining stable
  • Using data from the last 14 years of the employer sponsored insurance (ESI) market, we compare the percentage of cash wages employees pay for their share of health insurance (‘price’) to acceptance rates for employer sponsored insurance (‘quantity’) to establish a simple health insurance demand curve
  • Assuming premiums paid by subsidy-eligible households buying coverage on the HIEs are equal to the first year (2014) percent-of-income limits specified in the ACA; and, that the price/quantity relationships evident in the ESI market translate to the HIE market, we would expect HIE enrollment of roughly 23.5 million subsidy-eligible households – essentially on par with the CBO figure
  • If anything, the ESI-derived demand curve over-estimates HIE enrollment – the uninsured tend to have weaker preferences for health insurance than the insured; ESI enrollment benefits from an ‘enrollment by default’ standard, where HIE participants must actively ‘opt-in’; lower-income (and thus subsidy-eligible) households may place a greater value on the same percentage of marginal income than higher income households; and, lower-income households are on average younger
  • More importantly, the percent-of-income limits ACA places on households’ premium costs apply only in 2014. In subsequent years, if health cost growth exceeds wage growth – as it is likely to do – households’ HIE premium costs would exceed the 2014 maximums. Absent significant further reforms, subsidy-eligible households would be asked to pay larger percentages of incomes for HIE coverage, thus our expectation (under current law) that HIE enrollment erodes over time
  • Even if the 20 million CBO figure is right, the roll-out of the individual HIE’s would grow the total commercial (ESI + individually purchased coverage) market by about 11pct in enrollee terms, and 9pct in beneficiary terms. Stretched out over the 3 to 4 years required for the HIE’s to hit peak enrollment, the annual growth effect is around 3 to 4pct in enrollee terms, and 2 to 3pct in premium terms
  • For the commercial HMOs, the cost of these modest volume gains is a raft of legal, regulatory, and structural changes that are highly adverse – particularly for the larger ‘national account’ oriented commercial HMOs (e.g. CI, AET, WLP, UNH). Chief concerns include shifting of ESI sponsored beneficiaries to (public or private exchanges), where shifted beneficiaries buy less generous (i.e. cheaper) plans from an expanded (including local) list of underwriters, implying both share losses and falling average contract values
  • Medicaid HMOs (e.g. CNC, MGLN, WCG) are far better positioned. The Medicaid expansion (eventually) should be larger than the expansion of coverage to individuals on the HIEs; the Medicaid HMOs are likely to gain share of Medicaid beneficiaries overall; and, average contract values stand to improve as higher cost dually-eligible enrollees are enrolled

*Congressional Budget Office

Price : Quantity relationships in the employer sponsored (ESI) market

To estimate price elasticity of demand for health insurance in the employer sponsored (ESI) market, we define the price of health insurance as the percent of cash wages an employee spends on his or her share of the health insurance premium, and quantity as the percentage of employees that accept an employer’s offer of employer-sponsored coverage

Since 2000, ‘P’ has risen from 2.2pct of cash wages to an estimated 4.8pct in 2013 (Exhibit 1). In other words, an average employee refusing an employer’s offer of employer sponsored insurance would ‘save’ his or her share of the premiums; refusing coverage would have increased the employee’s cash wage by 2.2pct in 2000, and by 4.8pct in 2013

As ‘P’ has risen, ‘Q’ predictably has fallen – the percent of employees offered ESI who accepted the offer fell from 85pct in 2000, to 80pct in 2013 (Exhibit 2)

Exhibit 3a is a simple demand graph, zoomed in to the relevant price range, comparing delta ‘P’ to delta ‘Q’. The dotted line is a best-fit trendline, whose familiar shape is more easily seen in the zoomed-out version in Exhibit 3b


What ESI ‘P/Q’ suggests about HIE enrollment

Having established a relationship between the percent of income paid for insurance (‘P’) and the percent of persons offered insurance who accept (‘Q’), we can use this to at least roughly estimate what demand might be for individually-purchased coverage on the health insurance exchanges (HIEs) that begin enrollment next week

Within the income ranges that are subsidy-eligible (from 138pct to 400pct of the federal poverty level, or FPL), the Affordable Care Act limits first-year premiums to a maximum percentage of household income. The maximums rise with income, and range from 2 to 9.5 percent. Because households purchasing coverage on the HIEs are paying premiums with after-tax dollars, we make a corresponding adjustment to put these premiums on the same tax-affected basis as the ESI market (where premiums are paid with pre-tax dollars). As a final step, we adjust for applicable penalties[1]. Net of tax and penalty effects, the maximum percentage of income paid for coverage by subsidy-eligible households buying coverage on the HIEs will range from 2.2 to 9.9 percent (Exhibit 4)

The P/Q relationships in the ESI market suggest that about 85 percent of households offered coverage that costs them (net of subsidies) 2.2 percent of cash wages will accept the offer, and that about 69 percent of households offered coverage that costs them (net of subsidies) 9.9 percent of cash wages will accept the offer

‘Q’ times ‘eligible households’ …

We know the number of households in each ‘tranche’ of the subsidy-eligible income ranges, and also know roughly the percent of households in each tranche with our without ESI, and with a head of household younger than 65. By applying the income-appropriate, ESI-derived acceptance rates to the ‘un-ESI-insured’ and under-65 households in each income tranche, we can estimate the total number of subsidy-eligible households that might choose to purchase coverage on the HIEs (9 million); and, by multiplying this number by an average 2.6 persons per household could estimate individual HIE enrollment of roughly 23.5 million subsidy-eligible persons (Exhibit 5). This 23.5 million figure is greater than CBO’s most recent (May 2013) estimate of 20 million[2]; however for several reasons we believe the ESI-based estimate of 23.5 million HIE enrollees (and the CBO estimate of 20 million) are best-case scenarios that are more realistically viewed as estimates of maximum potential enrollment, rather than as estimates of likely enrollment

On net, ESI P/Q appears more likely to over- than to under-estimate HIE enrollment

Using ESI P/Q relationships as a benchmark for HIE P/Q outcomes is rational, within limits. Both groups are trading a predictable percent of cash wages to partially fund health insurance that is heavily subsidized by a third party – employers in the case of ESI, and the federal government in the case of HIE’s. Both groups will contain a broad range of ‘earner to beneficiary’ ratios, which heavily influences the cost of insurance as a percent of wages – for example in both markets single earners buying single coverage will pay substantially less as a percent of cash wages than single earners buying household coverage (Exhibit 1, again)

The similarities between ESI and HIE are sufficient to legitimize the use of ESI P/Q relationships as a basis for estimating HIE enrollment, but important differences should be borne in mind. Included among these are the following, several of which would raise HIE enrollment versus the ESI-based prediction, and several of which would result in lower HIE enrollment. On net, factors that should lower HIE enrollment relative to the ESI-based estimate appear to have a greater weight, accordingly we believe the ESI-based estimate of HIE enrollment is effectively an ‘outer-bound’

  1. ‘Right Shift’ influence (more HIE than ESI demand at a given premium : income level)

ESI participation rates are lowered by employed spouses that refuse their employer’s offer to participate in their spouse’s coverage. Thus the true ESI demand curve is at least somewhat right-shifted versus our estimate – i.e. a higher percentage of persons presumably would accept ESI sponsored coverage at given premium levels. The magnitude of this effect is difficult to quantify, but we can at least say the effect has gotten smaller during the time period of this analysis. Since the rate at which employers (especially smaller employers) offer insurance fell steadily during the period of analysis, the odds of a spouse having an alternative source of ESI also fell

  1. ‘Left Shift’ influences (more ESI than HIE demand at a given premium : income level)

The uninsured are more likely to have weak preferences for health insurance than the insured[3]. Notwithstanding the unfortunate truth that many who want and need insurance do not have it, evidence suggests that the average person without insurance places less value on insurance than the average person who is insured

ESI participation rates are inflated relative to HIE participation rates by opt-out v. opt-in dynamics. ESI beneficiaries generally have to opt-out to avoid being covered; HIE beneficiaries plainly will have to opt-in to be covered. It is well established that all else equal, opt-out benefit participation rates tend to be higher than opt-in rates

Marginal income may be more valuable to households with lower absolute incomes. Maslow’s hierarchy of needs applies: a lower-income (and thus subsidy-eligible) household may not have covered all of what it defines as its more basic (relative to health coverage) needs; correspondingly the lower-income household is more likely to use marginal income for some other purpose than health insurance

Lower income households tend to be younger, and may correspondingly have lower perceived need of health coverage

  1. Movement along the curve

HIE coverage will at first be less expensive than the percent of income maximums spelled out in the ACA, but is then likely to become more expensive than these maximums. Because the maximums are based on the second cheapest ‘silver’ plan (70 actuarial value, or ‘AV’) available in a given household’s pricing region, and because many households will buy less generous ‘bronze’ coverage (60 AV), many subsidized households’ actual premium costs as a percent of income may be lower than their applicable maximum – at least in 2014

Over time continued health cost growth in excess of wage growth is likely to continue for a number of structural reasons; in this case ACA’s subsidy provisions would result in subsidy-eligible households’ net premiums rising substantially faster than their incomes. Because the percent of income limits we used to calculate HIE uptake only apply in 2014, subsidy-eligible households’ premium costs as a percent of income are likely to exceed these limits in years beyond 2014, in which case enrollment would be expected to fall accordingly.[4] Absent substantive changes in the law, this is exactly the outcome we expect. I.e., where the CBO estimate calls for HIE uptake to reach the 20 million person level by 2017 and to remain at roughly this level through their forecast period (2023), because premiums almost certainly will inflate as a percent of incomes under current law, we would expect enrollment to quickly erode from whatever peak enrollment was achieved

  1. These have the effect of reducing the marginal cost of HIE-purchased coverage. We define the cost of coverage as the savings a household would achieve if they did not buy coverage – and because the penalty lowers the potential savings associated with not buying coverage, it has the effect of also lowering the true marginal cost of buying coverage
  2. http://www.cbo.gov/sites/default/files/cbofiles/attachments/44190_EffectsAffordableCareActHealthInsuranceCoverage_2.pdf
  3. See for example: “Health Insurance Enrollment Decisions: Understanding the Role of Preferences for Coverage” Alan C. Monheit and J.P. Vistnes, available here: www.rwjf.erui.org/pdf/wp31.pdf
  4. For details please see: “Why Premiums Should Grow Faster than Health Costs under ACA” SSR Health LLC, October 26, 2012
Print Friendly, PDF & Email