Sunday, June 05, 2011

Grading PolitiFact: Barbara Boxer and Medicare overhead

To assess the truth for a numbers claim, the biggest factor is the underlying message.
--PolitiFact editor Bill Adair

The issue:

(clipped from PolitiFact.com)


The fact checkers:

Louis Jacobson:  writer, researcher,
Bill Adair:  editor


Analysis:

I favor an examination of context no less for liberals than for conservatives (transcript mine; source video below):
BOXER:
And this is something people should never forget:  You know, Newt Gingrich said, 15 years ago, let Medicare wither on the vine.  Bob Dole, when--I went back into the 60s when Medicare was put into place--said something like it's socialism and it's horrible and it's the worst thing in the world.  And the fact of the matter is, it works.  There's a 1.5 to 2 percent overhead in Medicare.  The insurance companies have a 20 to 30 percent overhead.  What are these boys doing on the other side.

The video version:




I would identify four propositions in Boxer's statement.

1)  Medicare carries an overhead of 1.5 to 2 percent.
2)  Private insurance companies carry an overhead of 20 to 30 percent.
3)  Medicare is at least 10 times (20 divided by 2) more efficient than private insurance in terms of overhead.
4)  Medicare operates more efficiently than private insurance (the underlying message).

This time, PolitiFact is not blind to the underlying message:
Boxer’s comment cuts to the core of whether a government-run program like Medicare has advantages over one in which private insurers take a primary role.
The term "advantages" might easily turn out a weasel-word.  Boxer asserts better efficiency, using as her representative example the claim that overhead is at least 10 times greater for private insurance.  If correct, Boxer's assertion would constitute an important evidence of Medicare's overall greater efficiency.  Individual advantages one way or another carry modest importance except where one model has an overall advantage over the other.

Weighing the evidence?

PolitiFact:
To measure the administrative costs for Medicare, we first turned to the 2011 Annual Report of the Boards of Trustees of the Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds -- the document prepared by Medicare’s fiscal overseers.

The trustees’ summary listed total Medicare expenditures of $522.8 billion for 2010, of which $7 billion was characterized as "administrative expenses." That works out to 1.3 percent -- not far off from what Boxer stated.
Perhaps the gap in efficiency is even greater than Boxer asserts. The figure is less than Boxer's lower boundary.   That's worth noting before we move on:
For the private insurance market, we turned to a 2008 study by the Congressional Budget Office, the nonpartisan number-crunching arm of Congress. CBO cited data, compiled by the McKinsey Global Institute, that estimated administrative costs for private insurers at 12 percent. That’s quite a bit lower than the 20 percent to 30 percent Boxer cited.
Boxer's lower-boundary number represents a 66 percent inflation of the CBO number (which assists the growing PolitiMath database), though if we use the 1.3 percent figure cited earlier then the figure for private insurance would come close to Boxer's lowest estimate for the supposed greater efficiency of Medicare over private insurance.

PolitiFact:
However, the administrative burden for private plans get more complicated the deeper you dig. There are large variations between different types of insurance plans. The data cited by CBO found that administrative costs were about 7 percent for employers with at least 1,000 employees, but 26 percent for firms with 25 or fewer employees. Meanwhile, in the individual insurance market -- that is, plans secured by individuals on their own, rather than through an employer -- the rate was nearly 30 percent, CBO said.

A big reason for this variation, CBO and others have concluded, is that bigger plans can spread costs over a larger number of people.
When one is spreading administrative costs over a larger number of people, I'm pretty sure we're talking about economies of scale.  The PolitiFact explanation here is less than clear.

The story delves into a poor justification of the numbers by a Boxer spokesperson before asking the key question:  Is Medicare overhead really as low as the trustee reports make it seem?

PolitiFact:
A lively academic debate has broken out over whether Medicare’s administrative costs are really as low as 1 percent or 2 percent.

The difference stems from whether Medicare essentially freeloads off other parts of the federal government for services that private insurers have to pay for on their own. Adjusted estimates for Medicare’s administrative costs cited by the Urban Institute, a think tank that does research on issues such as poverty and economics, range from 3.6 percent to 5 percent, rather than the 1.3 percent using the data in the trustees’ report.
Using the bottom threshold from the Urban Institute study would make the private insurance share of overhead to about 3.3 times Medicare overhead.  In PolitiMath terms that would put the rate of Boxer's exaggeration at 203 percent.

That doesn't sound so good.  Can't PolitiFact do something to help Boxer out?  How about citing an opinion even further to the left than the Urban Institute?

PolitiFact:
But Edwin Park, a health policy specialist at the liberal Center on Budget and Policy Priorities, said that the differences are overblown, since Medicare’s administrative cost total already includes payments to other agencies for such services.
As usual with PolitiFact's use of expert sources, I'd like to see Park's entire set of comments.  Does he question the numbers from the center-left Urban Institute?  Or is he talking about something else?  What services make up "such services" for which Medicare makes payments?  PolitiFact leaves us with no context, no proof and no helpful citation in support of Park's claim.

PolitiFact does, however, leave out any and all commentary from two sources that appear to contradict Park.

William G. Schiffbauer, from "The Level Playing Field Myth: Comparing Administrative Costs for Public, Private Health Insurance" (yellow highlights added):
The expenses of ‘‘general administration’’ include all of the costs of operating the health insurance plan other than the costs of ‘‘claims administration.’’

Private Health Plans. These expenses include: sales and marketing (other than commissions); contract and legal staff work; enrollment; eligibility determinations; issuance of identification cards; underwriting; communications materials (such as Summary Plan Descriptions and other booklets); billing activities; customer service; accounting and data reports and analysis; wellness programs; quality improvement; utilization review; research and advice concerning laws and regulations; plan design advice and plan revision implementation; problem resolution and general accounting; salaries and wages; health and retirement benefits; payroll taxes; supplies; and amortization of buildings and facilities. Importantly, these expenses would also include legal and administrative expenses in connection with compliance with federal and state laws and regulations, and efforts to detect and recover fraud and abuse, and enforcement of contract provisions including coordination
of benefits. 

Medicare. None of these ‘‘general administrative’’ expenses is counted in the 3 percent ‘‘overhead’’ cost cited for Medicare. These ‘‘general administrative’’ functions are clearly performed for the Medicare program; however, they are expensed elsewhere in the federal budget.
"Medicare Administrative Costs Are Higher, Not Lower, Than for Private Insurance" (yellow highlights added):
Advocates of a public plan assert that Medicare has administrative costs of 3 percent (or 6 to 8 percent if support from other government agencies is included), compared to 14 to 22 percent for private employer-sponsored health insurance (depending on which study is cited), or even more for individually purchased insurance.
Apparently swayed by Park's persuasive assertion, PolitiFact eventually allows that Medicare's administrative costs may be a few points higher than Pelosi's figures but since her numbers were extrapolated from the trustees' report they are "defensible."

Again, welcome to the wonderful world of PolitiFact fact checking.

PolitiFact:
We won’t settle this question, but we will point out evidence that even when you control for the differences, Medicare is still considerably more cost-efficient. In one study, CBO found that privately run Medicare plans had 11 percent overhead, compared to 2 percent for traditional Medicare.
Serial sophistry serves as a poor substitute for adequate fact checking.  The list of sources includes one originating with the CBO.  That report apparently does not contain the cited information.  The information appears to come from a CBO report from 2006 titled "Designing a Premium Support System for Medicare."  The language employed by PolitiFact suggests that comparing administrative costs was a significant goal of the study.   That notion appears overpoweringly unlikely.  From the report (Page 12):
Administrative costs and return on investment account for about 11 percent of private plans’ costs of delivering Medicare benefits, whereas the administrative costs of the fee-for-service Medicare program (as reported by CMS) account for less than 2 percent of its expenditures.
Citation 17 follows the above.  In the corresponding footnote we find this:
CBO computed the average administrative cost of Medicare Advantage plans from 2005 ACR data. Information on Medicare’s administrative costs is available from the 2006 Annual Report of the Boards of Trustees of the Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds (May 1, 2006), p. 5.
There was no CBO study designed to "control for the differences."  It was the same kind of back-of-the-envelope guesstimation PolitiFact used earlier in the story by uncritically assuming that the numbers reported by the Trustees provide a complete picture of overhead costs under Medicare.  This is a sloppy and extraordinarily misleading way to check facts.

In weighing the evidence, PolitiFact apparently ignored evidence from its cited sources that would work against Boxer while creating the misleading illusion that a figure near 2 percent represents a reasonable estimate of Medicare's overhead percentage.  Not even the unabashedly liberal Ezra Klein (also cited for this story) buys that.

An appropriate metric?

Next PolitiFact takes up the issue of whether measuring overhead spending serves as a useful measure of plan efficiency.  Again, the flawed analysis misleads the reader.

PolitiFact correctly allows that the right kind of administrative costs lead to advantages for patients.  But the story never delves into the differing ways of determining overhead costs.  The story throughout carries the assumption that figuring overhead costs as a percentage of total expenses serves as the most appropriate method.

Again we find PolitiFact ignoring information from a cited source, the Heritage Foundation.  The author of that paper, Robert A. Book, Ph.D., argued that overhead was best determined on a per person basis.  For example, suppose insurance company A insures one client only.  The client incurs $9.9 million in expenses while the company spends $100,000 on overhead.  Insurance company B also insures but a single client.  That client incurs only $50 in expenses while the company spends $5,000 on overhead.  Using overhead as PolitiFact uses it, the company that spent over $9 million on patient care was the more efficient of the two.  Book's method suggests the opposite.

Probably neither metric alone tells a sufficiently complete story to enable a reliable judgment.  The problem here stems from the fact that PolitiFact never addresses the controversy.  Rather, Jacobson and Adair encourage the reader to uncritically accept a poor measure--coincidentally the one that best supports Senator Boxer.

Henry Aaron of the Brookings Institution dropped a tidbit in this section of the story that received far too little emphasis:  Greater spending on administration can lead to decreased expenses as a result of fraud.

Medicare fraud cost an estimated $60 billion in 2009.  Sixty billion dollars represents over 11 percent of total Medicare expenditures for the calendar year 2010.  How's that for efficiency?

Conclusions

PolitiFact dispenses the conclusion in three parts.

Part one:
There is some disagreement over how much Medicare pays in overhead. It could be a few percentage points higher than the 1 to 2 percent that Boxer cites. But Boxer’s numbers are defensible since they come straight from the Medicare trustees’ report.
The trustees' report does not give the 1 to 2 percent figure.  Rather, the figure is obtained by extrapolating information from the report while ignoring the apples-to-oranges nature of the comparison.  Klein opted for a figure of 5 to 6 percent.  Klein's lower figure gives Boxer an error of 60 percent with her top figure of 2 percent.

Part two:
Meanwhile, Boxer’s 20 percent-to-30 percent figure for the private sector is more squishy. Some plans have overhead rates that high, but only a fraction do, and the industry-wide average is quite a bit lower -- 11 to 12 percent.
Again using most favorable number Boxer offered, as well as the kinder end of the presumably accurate range of 11 to 12 percent, Boxer ends with an error of 67 percent.

Part three, the underlying message:
We’re convinced that Boxer’s underlying point -- that private plans have higher overhead than government plans -- is correct, if for no other reason than that profits matter only for private insurers. But for most plans and patients, the difference between Medicare overhead and private-sector overhead is not as great as she suggests. So we rate her statement Half True.
It must be nice to have judges focus on such a puny underlying point and accept such thin evidence in favor of it.  Did PolitiFact give Boxer a pass on its "burden of proof" criterion or simply provide us an example of partisan credulity?

Update (6-5-2011):  I should have noted the impact of this final set of numbers on the the third of the four claims I attributed to Boxer.  Using the end of the range most favorable to Boxer from the realistic set of numbers, Medicare--solely in terms of the overhead percentage of total costs--is 2.2 times more efficient than private insurance rather than 10 times more efficient as Boxer implicitly claimed.  Boxer's exaggeration of Medicare's efficiency comes to 355 percent.  The exaggeration figure does not take into account the questionable nature of Boxer's attempted comparison.


The grades:

Louis Jacobson:  F
Bill Adair:  F

This fact check gives the appearance that the PolitiFact team arbitrarily preferred sources that permitted the best possible rating for the subject.  That makes for the very worst sort of journalism and earns this PolitiFact story the special tag "Journalists Reporting Badly."



Correction 6/5/2011:  Several hours after publishing, corrected a miscalculation of Pelosi's error as to the percentage of overhead costs.  Originally I reported a 60 percent error.  The correct figure was 67 percent.

No comments:

Post a Comment

Please remain on topic and keep coarse language to an absolute minimum. Comments in a language other than English will be assumed off topic.