Click here for a guide to following the health care reform story online.
The Government Accountability Office concluded that Scully had broken the law, but the inspector general at the department of health and human services concluded Scully had not (though it acknowledged he might have violated the department's own standards of ethical conduct, a point mooted by the fact that Scully had already left government service at the time of its finding). Foster's estimate turned out to be 37 percent higher than the current cost estimate for the same period. *
Update, Nov. 18: Foster sent me an e-mail taking issue with various points made here. I asked him whether I could share it with readers, and also asked whether, while he had Slate readers' attention, he might also explain why his 2003 ten-year estimate of Medicare drug costs turned out to be 37 percent higher than the current cost estimate. He said yes to both.
First, then, Foster's rebuttal to this piece:
A couple of comments on your article today in Slate.
First, it's true enough that the Office of the Actuary doesn't normally do health reform estimates, as my longtime friend, colleague, and former boss Nancy-Ann DeParle stated. However, this statement is true primarily because there normally aren't any health reform proposals. Prior to this year, the last major proposal was the Clinton Administration's "Health Security Act" in 1993. The Office of the Actuary provided the Administration's cost estimates for that proposal.
Second, I believe that the productivity adjustments to Medicare payment updates, based on economy-wide productivity, could pose a problem with or without the baby boom. The issue could be exacerbated as Medicare becomes an increasing share of hospitals' overall caseloads, but Medicare already pays about 28 percent of total hospital costs in the U.S.—enough to cause financial problems if hospitals can't improve efficiency enough to match the adjustment. The idea of adjusting Medicare payments to providers for productivity gains, incidentally, makes a lot of sense, and I've raised it with policy makers on various occasions over the years. The potential problem arises from using economy-wide productivity, rather than a level more consistent with what most providers are able to achieve.
Third, I was puzzled by your thought that CMS would "lose funding under health care reform" and that our memorandum for H.R. 3962 might be aimed toward preventing such an outcome by discussing possible access issues for Medicare beneficiaries. I know Slate uses humor in its reporting, so at first I thought this was a tongue-in-cheek comment. As I read it more carefully, however, I concluded that you actually mean "it would be remarkable if his pronouncements were completely unbiased by any consideration of his agency's self-interest."
In practice, I expect that CMS will get an increase in funding, not a decrease, along with new responsibilities for implementing the Medicare and Medicaid provisions and possibly other aspects of health reform. Moreover, lower Medicare payments to health providers or MA plans under the bill wouldn't affect CMS's administrative funding or its ability to perform its work.
In reality, funding for CMS never crossed my mind and certainly had no effect on our estimates. Believe it or not, we're dedicated to a fairly simple goal: Providing objective, nonpartisan, technical information to policy makers to help ensure that the policies they develop will operate as intended and without surprises. I've devoted a 37-year actuarial career to this principle. Moreover, my own staff would be the first to throw me out the door if they thought I tried to bias an estimate in any direction, for any purpose. No kidding—we take our mission seriously.
My final comment is that I did, in fact, mean to use the word "substantive," as in "considerable in amount or numbers: substantial." I hesitate, however, to debate word meanings with a professional wordsmith.
And here's Foster's follow-up explaining why his 2003 estimate overshot the mark:
Our original cost estimates for Part D of Medicare were made in 2003, using survey data on beneficiaries' prescription drug use from 1998. In 2003, prescription drug expenditures had been increasing annually at double-digit growth rates for more than a decade. We assumed that this growth trend would continue but gradually abate over time. As it turned out, drug cost growth plummeted, starting in 2004, and in 2008 was only a little over 3 percent—the lowest level since the early 1960s. This sharp deceleration in growth rates was unexpected by us, CBO, and most industry experts. It occurred largely as a result of (i) effective efforts by health insurance plans to induce people to use generic equivalents and (ii) a slowdown in the number of new "blockbuster" drugs coming into the market.
Another (smaller) factor affecting actual costs versus our original estimates was the degree to which Part D plans could reduce costs through negotiating retail price discounts and drug manufacturer rebates and through utilization management. We thought that it would take Part D plans a few years to produce results, on average, that equaled the best savings that pharmacy benefit managers achieved in 2002. In practice, the plans equaled these results in the first year (2006) and actually improved upon the rebate levels significantly.
The last (and smallest) reason for the difference was that fewer Medicare beneficiaries enrolled in Part D than we had anticipated. In addition, many beneficiaries waited well into 2006 before signing up, whereas we thought most would enroll by January 1, 2006.
These differences between actual and estimated Part D costs highlight the substantial uncertainty that is inherent in projecting future health care expenditures, particularly with new programs where there isn't any historical experience available (such as Part D or the health care reform proposals currently under consideration). Projections almost always turn out to be "wrong." If you're doing a good job, then you're too high half the time and too low the other half. These estimates are useful, however, despite their uncertainty, because they give policy makers a reasonable idea of what to expect for costs, savings, enrollment numbers, etc. And projections often turn up unexpected consequences of draft legislation, helping to ensure that the proposals can be modified to operate as intended. Thus, I believe that our estimates (and CBO's) provide important information—and that they should be used carefully, with full regard for their inherent uncertainty and limitations.
Correction, Nov. 17, 2009: An earlier version of this update stated incorrectly that Foster's 2003 estimate was too low.
E-mail Timothy Noah at email@example.com.