• How the ACA will affect employer-sponsored insurance: ranges and point estimates

    A new NBER paper titled The Importance of the Meaning and Measurement of “Affordable” in the Affordable Care Act, by Richard Burkhauser, Sean Lyons, and Kosali Simon, is an exploration of the range of possible implications of the ACA for employer-sponsored insurance (ESI). The paper also includes a review of work by others in this area.

    First, from the abstract, are some findings of the NBER paper:

    This working paper highlights the practical importance of two critical but under-explored assumptions […]. The first assumption is whether ACA’s affordable coverage rule will be interpreted to mean that employers must provide affordable single coverage or that they must provide affordable family coverage policies to workers with families to avoid paying a fine. The second assumption is how much employers and employees will cooperatively agree in the future to designing new compensation contracts to take advantage of the way “affordability” is determined. We show that depending on these assumptions, the ACA could lead to far more lower to moderate income families gaining access to affordable coverage through exchanges or, conversely, to far fewer of these families being covered by ESI, even if no employers drop their health insurance plans as a result of the new law. Using our stylized models, we find at one extreme that the share of private sector workers covered by ESI would fall by as much as 12.7 percentage points, relative to a case of full compliance with the law, if the ACA affordability coverage rule is interpreted to apply to family coverage and employees directly pay 100 percent of the cost of the ESI in premiums, with compensating higher wages making them no worse off. At the other extreme, we find no changes in the share of private sector workers covered by ESI along this margin if employee contribution shares do not change in the future and affordability is interpreted to refer to single coverage.

    The key thing to keep in mind about this work is that the results are, as stated, driven in large part by assumptions. There’s nothing wrong with that. It is a perfectly fine way to explore the possible range of consequences. That is, in fact, one of the points of the paper. The authors are not making a prediction about what will happen, so much as showing the extremes of the range. If you have a lot of confidence that one extreme will turn out to be true, you are probably putting a lot of faith in something outside the scope of the paper, as the authors note:

    Our paper is only able to display a full range of possible estimates rather than provide a point estimate. The extent to which employers act in ways that are closer to one extreme or another depend on factors that are beyond the scope of our analysis.

    The paper includes a brief literature review, which I’ve further condensed:

    Dubay [and] Cook […] (2009) simulate the effects of health reform on the uninsured, modeling a Medicaid expansion to 133 percent of the poverty line and an individual mandate. They find that these two reform components would bring some financial assistance to three-quarters of the uninsured, with about half being eligible for subsidies and half being eligible for Medicaid. Holahan and Garrett (2011) discuss the impact of both health insurance and health care provisions in the ACA on employment and conclude that on net, there would be very small consequences.

    Holtz-Eakin and Smith (2010) were the first to point out that the federal subsidies in the ACA intended only for those not eligible for ESI are so large that they could induce strategic behavior on the part of some employees, and tempt some employers to change their labor contracts to take advantage of these subsidies. Using hypothetical workers and a fine of $2,000 per worker (the case where large firms do not offer health insurance to their workers) they demonstrate when it will make sense for an employer to drop health insurance coverage, and the types of workers who will benefit from a switch from ESI to subsidized exchange coverage. […]

    There is also a growing literature assessing the impact of health reform in Massachusetts. In July 2007, Massachusetts was the first state to implement a package of reforms that included an expansion of public health insurance, subsidized exchange coverage, and employer and individual fines. While the specific issue of the definition of affordability and employee premium shares is not present in those reforms, it is noteworthy that there is no evidence that employers have dropped health insurance coverage (Long and Masi, 2008, and Gruber, 2011) as critics contended they may.

    I’ve also commented on Holtz-Eakin and Smith (2010) and Gruber (2011) in prior posts. Also in prior posts you’ll find more on employer responses to the Massachusetts reforms.

    Finally, Steve Pizer, Lisa Iezzoni, and I have made our own predictions about the future of public and private insurance due to the ACA. (See also our working paper.) Though some assumptions are involved, which cannot be avoided, our predictions are based on analysis of historical data that characterizes actual firm and individual behavior.


    Dubay, L. and A. Cook (2009). How Will the Uninsured be Affected by Reform?

    Gruber, J. (2011). “The Impacts of the Affordable Care Act: How Reasonable Are the Projections?” NBER Working Paper 17168.

    Holahan, J. and B. Garrett, 2011. How Will the Affordable Care Act Affect Jobs?

    Holtz-Eakin, Douglas and Cameron Smith. 2010 Labor Markets and Health Care Reform: New Results, May 2010. American Action Forum.

    Long, S. and P. Masi. 2008. “How have Employers Responded to Health Reform in Massachusetts? Employees’ Views at the End of One Year,” Health Affairs, 27 no.6 w576-583 2008.

    • Austin are you familiar with the Lewin Group’s microsimulation model?

      I’ve been working with it quite a bit, and it also does not predict a huge amount of employer dumping. (And by the way met Carol Simon, a former colleague of yours who works for them now). I think there are plausible scenarios under which it will happen, but it depends a lot on how savvy the employers are in the small group market, and in the large group market due to the competitive pressure I think it would take one or two big employers taking the leap which could lead to a domino effect.

    • Thanks for your thoughts on this paper. You echo my initial thoughts. I expect that it will be cited, using the most negative assumptions, by those who oppose the ACA.


    • I think you are (deliberately) missing the point.

      One Extreme: ESI penalties are based on individual plan affordability so employers have a greater incentive to cease offering health care coverage. In which case ACA is a lot more like a government takeover of the healthcare system than you are comfortable admitting on this site… 12.7% more if you believe the authors of the paper.

      Other Extreme: ESI “affordability” refers to family plan affordability in which case ACA costs $150 to $500 billion (in gross costs) more than the CBO score said that it would when Congress voted on ACA and completely blows a hole in America’s fiscal stability.

      Unlike any esoteric discussion of ranges, this issue is an either-or binary decision. Do the penalties kick in based on the cost of family coverage or the cost of individual employee coverage? Both answers cannot be right…

      • I quoted the authors. I believe they understand the nature of their own work better than anyone.

        Also, there are assumptions in your statements of the extremes and in your claim that this is binary (not extremes of continuums) that are outside the scope of the paper and not consistent with other point estimates, including in the paper by me and colleagues to which I linked.

        • Cute… the old “Bait-and-Switch”

          I am sure that the authors do know the nature of their work. However, what the authors of this paper are giving as a “range” is the range of response that employers (to some extent with the cooperation of their employees) will do in response to the adoption of ACA.

          You on the other hand misrepresent their “possible range of consequences” so as to minimize the danger that people will view the ACA as a budget-busting monstrosity that was poorly thought out before it was passed into law.

          But while the authors are (properly) clear that they cannot give a point answer as to how much employer responses to ACA will differ from the CBO estimate of the cost of the act… it is nonetheless clear that they believe it will cost more than the CBO estimated in 2010:

          1) They explicitly state (in the quotes that you chose to cite) that the range is zero to $500 billion… there is no possibility in their model that the CBO was wrong on the high-side.

          2) The title of their paper puts “Affordability” in quotes and its whole purpose is to explain how much more the ACA will cost depending on whether or not the “Meaning and Measurement of “Affordability” ” turns out to be based on family coverage or individual coverage.

          And let’s not pretend that you are unaware of the nature of this debate… in your response to Steve you make clear that you know how this paper is being received in other forums.

          You may think that by slicing and dicing the quotes of the authors that you can fool your readers… but does that have any real value?

          • I want you to know that I take research and how it is characterized and used very seriously. Therefore, in many cases I correspond with the authors whose work I write about, just to be sure I don’t accidentally misrepresent their work. In those cases, when an author has comments or suggestions, I change the post accordingly. This happens to be one of those times where I did have correspondence with the authors. So, I can say with confidence, those of us on this side of the debate that you seem eager to engage in are satisfied with the post.

            In addition, one study does not a conclusion make. I pointed to others. If you follow the links for the work I participated in, you’ll find that I am also not certain CBO considered everything in their analysis. In time, as things clarify and as new research comes to light, I expect they will update their forecasts. Perhaps the ACA will cost more than they expected. What’s to be done about that? Here’s where we enter the land of opinion. I’m not going to fight over that land. You’re entitled to your piece of it, but you may not insist I live on it too, as you seem to want me to do.

      • If only a couple of hundred billion (you write: “$150 to $500 billion in gross costs… completely blows a hole in America’s fiscal stability) is all it would take to fix the hole in America’s fisacl stability.

        • Ha! Good Point!

          I guess that I should have stated that price tag as the latest in additional costs associated with ACA that blows a hole in America’s Fiscal stability. Other CBO “miscalculations” have already turned this from being Deficit Friendly to a budget-buster. Another $150 to $500 billion expense over ten years is just icing on Disaster Cake!

          Of course that assumes that the “range” is as extreme as Austin and I have been discussing. And that is based on a study that makes some questionable assumptions of its own. Lots of ranges and assumptions there.

          Still, the only relevant question facing policymakers is with Obamacare currently garnering the support of only 40% of Americans, will that support drop faster if people discover:

          A) That it costs (another) half-trillion over what the CBO promised that it would (based on “Affordability” being keyed on family coverage) in order for American workers to keep family coverage.

          B) Or to keep the costs in line (“Affordability” being keyed on individual coverage) that Americans may very well wake up to find that their employers have the right to drop them into the Exchanges… where employees get covered but their families do not.


          That is the binary decision facing policymakers… no matter what the “range” of costs are estimated to be.

    • You — or a source — write(s):

      “…it is noteworthy that there is no evidence that employers have dropped health insurance coverage (Long and Masi, 2008, and Gruber, 2011) as critics contended they may.”

      Unfortunately there is a lot of evidence that employees have stopped subscribing to insurance offered by employers in Massachusetts — to the tune of 125,000 fewer subscribers to ESI just since 2007 and a 10% drop 2004 to 2010 in the percent of those eligible for ESI actually taking it (see page 2 of May 2011 Premium Trends report and slide 10 of February 2011 Quarterly report respectively, both from Massachusetts Department of Healthcare Finance and Policy).

      Also, like much of the data underlying these discussions, there is data somewhere that explains the drop in “Eligible employees” of employers that offer ESI vs. all employess of employers that offer ESI but it is not typically available except to academics. Some of you academics ought to dig it out and see what its implications are for the Patient Protectiona nd Affordable Care Act.

      • Do you know if there is any work that examines employer drops in MA relative to trend? I seem to recall seeing something about that somewhere. It’s relevant because one needs to know if the dropping is due to the law or just a continuation of historical patterns.

        • If by “examine” you mean look at in a rigorous academic fashion, I am not aware of any research. The only thing I see is survey data the state puts out to the public typically a year or more old (most recently the results of the April 2010 Employer Survey released in July 2011). [The problem in my opinion with all the data being used to justify/modify RomneyCare (and apparently to plan the Patient Protection and Affordable Care Act–PPACA) is that the methodologies and taxonomies are changed constantly. I have never even seen a discussion of what should be considered the base year — I would say 2005 or earlier — and what should be considered the “first year” of RomneyCare — I would say CY2008 or FY2008 for fiscal issues.]

          You may be thinking of Slide 21 of the just released July 2011 report referenced above which shows the percent of “all employees of employers offering health insurance that take health insurance” trending down pretty consistently from 2001 to 2010. This likely (but I have never seen it discussed in the RomneyCare/PPACA context) is due to employers not offering health insurance to some category (part time help) and purposely hiring that category to avoid the higher personnel costs of hiring another category (full time help). That trend is almost surely not due to RomneyCare but it is something that should have been considered for both RomneyCare and PPACA (the McDonald’s et. al. PPACA waiver brouhaha is tangentially related probably). I am not saying it wasn’t considered; it’s just that I have never seen it mentioned.

          As for the percent of “eligible employees of employers offering health insurance that take health insurance,” that was pretty consistently around 80% pre-RomneyCare and has dropped to 75% (but see comment vis a vis what is pre-RomneyCare and what is post). The state seems to think this is the important percentage. Actual numbers — not even including the 2010 actuals that crosstab to the 75% — are on page 2 of the Premium Trends report released in May 2011 and show a 125,000-person decrease in people getting ESI. How much — if any — of this decrease is caused by unemployment rates vs. RomneyCare is not known. Interestingly the state is not arguing this drop is caused by unemployment so I would guess it thinks it’s a RomneyCare trend.

      • What about ? See page 6. Offers went up in MA. Or am I missing something?

        • Yes, I think you are. I am apparently having a hard time expressing the distinction I am drawing.

          1. Your original blog post (I think maybe quoting someone else?) and the source you cite in your latest reply to me both refer to employers offering insurance.
          2. The data I am referring to (from the same base document in slide 21 but also in many other state of Massachusetts reports on this subject, some of which include the actual numbers rather than squishy percentages) relates to employees taking insurance, which I think is the much more important metric.

          Simplistically, more employers are offering insurance to avoid the fines but fewer employees are subscribing to it. The question is why and how it relates to health care insurance reform (in other words, should we just blame it on RomneyCare?). I would like to but I think
          — Some percentage of employees not subscribing is because they are not eligible for the insurance; this trend is most likely from before RomneyCare but seems to have accelerated post RomneyCare.
          — Some percentage is because of increased unemployment; but that has not been as big an issue in Massachusetts as in the rest of the country and even RomneyCare proponents don’t claim that is the reason
          — Some percentage must relate to health care reform.

          I have seen no academic research on this subject and was simply suggesting it as a subject for you academics.

          • No, I get you. Your points and questions are fine.

            I’m now looking at , specifically page 21. Take up went up from 2005 to 2007 and only then went down. The big drop is from 2009 to 2010. I think we can only speculate. It strikes me as entirely possible this could be recession-related, mixed with trends that had been on-going prior to recession (see 2001-2005 trend). The pattern is very consistent with the prediction made in my paper with colleagues and illustrated here.

    • PS: What is wrong with the formatting of this site’s comments? I am viewing this on Firefox and the site seems to cut off the comments on the right margin. Is that true in IE or Chrome as well?

      • I am not asking you to change a post. I am asking you to acknowledge that your original description of this study was designed to minimize its impact and thus was misleading to your blogs readers.

        I suppose this gets to a question about this blog entirely… what is its purpose? Is it to be a clearinghouse for academic healthcare studies or does it have a point of view? Despite your claiming above that you are not fighting over any “land” in the healthcare world of opinion… I think a casual reader would realize that isn’t true. Nothing wrong with that… just don’t pretend to be a disinterested observer of the debate.