On Gruber: I Don't Want Apologies. I Want Independent Analysis.
Between the auto show and the Prop 8 trial and associated travel, last week was tremendously exhausting for me and it will take me several days to actually report on those two events. But it seems one thing hasn’t moved on very much since last Sunday–the reporting surrounding Jonathan Gruber’s role in pitching the Administration’s health care. Gruber’s defenders are still falsely claiming I accused Gruber of tainting his analysis for pay (I said, “I don’t doubt he believes all this stuff”) and suggesting that I’m ignoring Gruber’s qualification for the HHS contract (I wrote an entire post affirming that the sole source on it made sense). Now, the debate has ratcheted up as some very able commentators call for apologies.
Unfortunately, that debate–like Gruber’s failure to reveal his conflicts in the first place–has supplanted what is really long overdue in this policy debate: real analysis of the assumptions behind the $850 billion plan about to be enacted by Congress, the assumptions that Gruber had a key role in formulating.
Gruber’s public claims delayed real analysis of the claim that the excise tax would raise workers’ wages
To explain why this is important let me make a suggestion that I can’t prove, but which is the reason I started looking at this in the first place: because someone as credible as Gruber made certain claims about the excise tax, others in his field did not examine his claims in timely fashion.
Gruber, in conjunction with the Joint Committee on Taxation, has long been claiming that the excise tax would raise workers’ wages. I first started challenging that claim in October, in response to an Ezra Klein post that relied on Gruber’s faith-based claim that the excise tax would lead to higher wages. On November 5, Gruber quantified the benefit as $74 billion in 2019. And by December, I was in full panic mode, given that no economist could point me to a study proving the point, even in the face of benefit consultants’ surveys refuting it. Economists kept pointing me to Gruber’s papers and telling me not to worry my sweet little non-economist head about such matters.
Perhaps because of the work of the Economic Policy Institute, people finally started looking at this key claim in the last two weeks. No lesser economist than Gruber’s chief defender, Paul Krugman, judged that those making the claim (Krugman implied, but did not say explicitly, that this criticism was directed at Gruber) were exaggerating. And Gruber, who backed off the claim slightly after having had his conflicts exposed, has since admitted privately that he “over-reached” in his earlier statements.
So to review what happened: for a number of months, unions and benefits professionals and dirty fucking hippies like me challenged this assumption, but no one in Gruber’s field appears to have done any independent analysis of his claims. As a result, the excise tax was passed by the Senate based on at least one erroneous assumption. But now, either because economists have weighed in or because Gruber’s conflicts have been exposed, a key part of those assumptions has been challenged (and, in a perhaps not unrelated development, unions have been able to negotiate a palatable deal on this issue).
This kind of analysis should have happened last fall, but it did not, at least partly (I would argue) because someone of Gruber’s prominence had strongly made the claim. His colleagues didn’t do what scholars normally do, regardless how prominent the scholar, which is to check his work. And again, none of this is meant to say Gruber “over-reached” with this claim intentionally. Rather, because the normal peer review of Gruber’s claims didn’t happen, he (and with him, the Administration and the Senate) made a mistake, one that has already had real policy implications.
The entire basis for the excise tax remains unexamined
Now, this matters to me not because a bunch of prominent people are accusing me of being a scandal-monger, but because I believe some more key assumptions about the health care reform have not been adequately examined. In fact, there are two key claims about the excise tax that have, at the least, gotten far too little scrutiny.
Start with the revenue model. Here’s how Gruber very helpfully explains the revenue model the Joint Committee on Taxation used to come up with revenue estimates that will be raised through the excise tax.
This analysis relies on three documents issued by the JCT. The first is their October 13, 2009 memo which provided the score of the revised High-cost insurance tax as in the Senate Finance Committee mark. This memo shows the year-by-year revenues raised by the High-cost insurance tax. Importantly, the memo highlights the two different ways the High-cost insurance tax raises revenues. The first is through actual excise tax receipts paid by those high cost plans that remain above the High-cost insurance threshold. The second is through the fact that firms will spend less on health insurance – and this reduced spending will be shifted to workers in the form of higher wages. This conclusion of wage shifting is supported by both economic theory and evidence, and is assumed in modeling by both the JCT and the CBO. This division is very informative: the JCT estimates that about 80% of the revenues raised by the High-cost insurance tax will come from revenue from higher wages, not from the excise tax itself. [my emphasis]
To lay out what Gruber says the assumptions he (along with CBO and JCT) make:
- The excise tax will lead employers to switch from “gold” to “silver” insurance plans
- This will mean “reduced spending” as “firms will spend less on health insurance”
- Those companies will pass those savings onto employees in the form of higher wages
- Employees will pay more taxes on those higher wages which will lead to higher tax revenues
80% of the revenues raised under this tax, Gruber and JCT estimate, will come from higher wages.
Now, we’ve already seen Lawrence Mishel and Krugman and even Gruber challenge or rethink assumption 3, the claim that employers will pass on savings to employees. So there is already a clear problem in the revenue model JCT used to assess this plan.
But (as some economists reassured me when I was having my December panic) if the other assumptions remain the same, it wouldn’t necessarily be a problem, because employers would simply keep the increased profits they got from those savings in health care, and pay higher taxes on them.
Which brings us to assumption 2, that “firms will spend less on health insurance.” Note what this claim implies. It argues that companies will not just aim to keep their health benefit spending constant over time (that is, move from a gold plan, which had been $700/mo but had been raised to $750/mo by the insurance company and therefore fell into the excise tax, and switch to a $700/mo silver plan to stay under the excise tax level). Rather, Gruber describes the JCT as assuming that companies will “spend less” on health insurance, implying a switch from a $700/mo plan to a $600/mo plan as the plan began to hit up against the excise tax.
And that must be what the JCT assumes. If there weren’t either increased profits or wages to come out of this, then profit and or wages would remain the same, and tax receipts would not go up at all (except through companies actually paying the excise tax itself, which JCT says will only account for 20% of the revenue). Yet the only way there would be actual savings (as opposed to flat spending over several years) is if companies dramatically cut back health care (from a Platinum to a Silver plan, or a Gold to a Bronze plan) or if they stopped offering health care altogether (which, given that there will be no employer mandate, they may well do).
Now, I’m fairly sure that Krugman, at least, believes this to be an erroneous assumption. He says,
Maybe it will help the plausibility of this case to notice that we’re not actually asking whether a fall in premiums would be passed on to workers. Even with the excise tax, premiums are likely to rise over time — just more slowly than they would have otherwise. So what we’re really asking is whether slowing the growth of premiums would reduce the squeeze rising health costs would otherwise have placed on wages. Surely the answer is yes. [my emphasis]
But, as Gruber’s language clearly suggests, he and JCT do appear to have assumed that there would be “a fall in premiums,” they do appear to have ignored that “premiums are likely to rise over time.”
And as one example of why this is probably another erroneous assumption, take this story on Safeway, which has been celebrated for the kind of innovative cost-cutting that policy makers would like to see come out of the health care reform. As the WaPo reports, after Safeway made some fairly aggressive insurance plan changes, their costs did drop; but costs have already climbed back up to where they were after the reform.
But a review of Safeway documents and interviews with company officials show that the company did not keep health-care costs flat for four years. Those costs did drop in 2006 — by 12.5 percent. That was when the company overhauled its benefits, according to Safeway Senior Vice President Ken Shachmut.
The decline did not have anything to do with tying employees’ premiums to test results. That element of Safeway’s benefits plan was not implemented until 2009, Shachmut said.
After the 2006 drop, costs resumed their climb, he said.
Even as Burd claimed last year to have held costs flat, Safeway was forecasting that per capita expenses for its employees would rise by 8.5 percent in 2009. According to a survey of 1,700 health plans by the benefits consultant Hewitt Associates, the average increase nationally was 6.1 percent.
Today costs are slightly higher than in 2005, Shachmut said.
So when Safeway said it had flatlined costs since 2005, “we defined that, you might say, loosely,” he said. “Perhaps a more precise way to say it is that our costs today on a per capita basis are essentially the same as they were in 2005.
So even in the case of a fairly radical change, a company only got savings (and therefore only paid higher taxes) for a few years, at which point their costs returned to the same levels. More moderate changes, it seems, may well result in no savings and therefore no revenue gains. And if companies don’t actually spend less–if they instead simply spend the same but get less health care in return, as seems likely–then 80% of the presumed revenue tied to the excise tax would fail to materialize.
In other words, it appears that economists, if they looked at Gruber’s and JCT’s claims closely, would challenge the entire revenue model to the excise tax, which would surely change the debate about the tax. But that hasn’t happened.
And will the excise tax even “bend the cost curve”?
But then there’s the really troubling unexamined assumption: that it will “bend the cost curve”–that is, make costs go up more slowly over time.
Now, curiously, all the justifications for this claim speak abstractly of the benefit of forcing employees to pay for more of their health care costs themselves, which will result in them choosing which care they need, using less care, and therefore saving money. I’ve yet to see any analysis of what happens in real life when this happens.
The thing is, though, it has happened in real life.
It has been happening for decades.
And yet no economists that I’ve seen have argued that such switches have led to a slow in the rise of health care costs. If, as companies have tried to trim health care costs over the last several decades by making precisely the kind of changes that excise tax boosters envision incenting in the future, but those changes haven’t already slowed health care costs (not to mention resulted in increased tax revenues), then there is zero reason to expect it to work in the future. Zero.
Let me go back.
In 2003, Jonathan Gruber actually studied the degree to which employers were shifting some of their health care costs onto employees. He and his study co-author, Robin McKnight, found that “in 1982, 44% of those who were covered by their employer-provided health insurance had their costs fully financed by their employer, but by 1998 this had fallen to 28%.” Gruber and his co-author attributed the shift to the following factors: 32% to the rise of HMOs, 9% in employees becoming newly eligible for Medicaid, 14% due to changes in taxation, and 47% to the rise in medical spending. Significantly, part of this cost-shifting was already a response to taxation changes, but even the shift because of rise in medical spending would look–in practice–precisely like what excise tax boosters want to see happen in response to the excise tax. The cost to employers goes up, and as a result employers make employees pay more of their own health care out-of-pocket.
In other words, over the last thirty years, employers have already been shifting costs onto employees. As long ago as 1998, fewer than 30% of employers were shielding their employees from the rising costs of insurance. And surely those numbers are far, far lower now. So if the excise tax were going to work, then the “cost curve” would already be bending; health care cost increases would already be slowing in response.
Now, frankly, I’m less sure this assumption is totally bogus. For example, employer supplied health care costs increased more slowly last year (though analysts have suggested that’s just because fewer people have it). But still–since we’ve seen 30 years of employers shifting costs onto employees, shouldn’t there be data–and therefore a testable argument–that shows that all that cost-shifting has resulted in decreased costs?
Because if it hasn’t worked over the last 30 years, there is absolutely no reason to think it will happen in the next ten. And this, of course, is the big “cost savings” that bill proponents point to to excuse the abandonment of a public option or drug reimportation.
More analysis, please, not more defensiveness
Now, we don’t really know what role Gruber played; we don’t know why the White House sold his pieces as “objective” analysis justifying their own policy decisions. The Gruber story may or may not end up being a scandal. We don’t know whether or why yet.
But what we do know is that the guy we all thought was conducting independent analysis of these policies was in fact just affirming policies he conducted the analysis for the Administration on, the same analysis the Administration used in selecting this policy. We do know that on at least one count, Gruber (by his own admission, as well as Krugman’s) overreached in his defense of the Administration’s favored policies. And we know that all those other policies Gruber was pitching remain largely unexamined–at least, unexamined by Gruber’s colleagues, the ones with the ability to identify other problems in his assumptions. And the risks are real. The entire revenue model may be (and I suspect is) based on a faulty assumption. And the claim that the excise tax will slow cost increases may (I’m less sure about this one) also prove to be based on a measurably wrong assumption.
So rather than defending Gruber for what his colleagues judge to be a mistake, can they (I’m looking at you, Krugman) start checking some of his work? As I’ve said, this country is about to make an $850 billion bet on Gruber’s analysis and assumptions. And yet rather than doing what all scholars do–check each others’ work–Gruber’s colleagues would prefer to spend their time focusing on me, rather than on Gruber’s analysis.