The New York Review of Books
Volume 52, Number 4  March 10, 2005

America's Senior Moment
By Paul Krugman

Review of:  The Coming Generational Storm: What You Need to Know About America's Economic Future by Laurence J. Kotlikoff and Scott Burns, MIT Press, 274 pp., $27.95; $16.95 (paper)

 Link to actual article:  The New York Review of Books:  "America's Senior Moment" by Paul Krugman

1.  Two Problems, Not One

America in 2030 will be "a country whose collective population is older than that in Florida today." It will be in "desperate trouble" because the expense of caring for all those old people will cause a fiscal crisis. The nation will be plagued by "political instability, unemployment, labor strikes, high and rising crime rates." That's the picture painted in The Coming Generational Storm by Laurence Kotlikoff and Scott Burns, a book that has helped to feed a rising tide of demographic alarm.

But is that picture right? Yes and no. America does have an aging population, and a responsible government would take preparatory action while the baby boomers are still in the labor force. America also has very serious long-run fiscal problems. But these issues aren't nearly as closely linked as much of the discussion would lead you to believe. The view of demography as destiny is only a half-truth, and in some ways it's as damaging as a lie.

In this essay I'll try to set the record straight. Unfortunately, I can't do that by following Kotlikoff and Burns closely. Kotlikoff is a fine economist, one of the world's leading experts on long-run fiscal issues. His book with Burns is full of valuable information and sharp insights. Yet in their effort to grab the lay reader's attention, Kotlikoff and Burns do little to alert readers to the distinction between two quite different issues--an aging population and rising spending on health care. And their failure to make that distinction grossly distorts their discussion.

The demographic problem is, of course, real. It is, however, of manageable size--exaggerating the problem by confounding it with the problem of medical costs just gets in the way of dealing with it. The problem posed by rising medical costs, on the other hand, would be there even if the population weren't aging--and misrepresenting the problem as one of demography gets in the way of confronting it.

I'll start here by looking at the demographic problem--the aging population--which mainly concerns Social Security, then at proposals for Social Security "reform"--the scare quotes are there because the scheme currently under discussion would undermine our social insurance system, not save it. At the end I'll talk briefly about the much bigger, more intractable issue of paying for the expanding quality and quantity of health care, and the current state of political debate.

2.  Social Security and the Demographic Challenge

Chapter 1 of Kotlikoff and Burns's book is called "From Strollers to Walkers"--a catchy way to describe the aging of the US population. It's followed with a chapter called "Truth Is Worse Than Fiction," centered on a chart familiar to everyone who has looked at this issue: long-run projections from the Congressional Budget Office showing the combined expense of Social Security, Medicare, and Medicaid rising from less than 8 percent of GDP now to more than 20 percent by 2075. It seems natural to assume that the grim cost projections follow directly from the aging of the population, and the book doesn't tell you that this assumption is wrong.

One way to describe the truth is to say that there is no program called Socialsecuritymedicareandmedicaid: these are separate programs with separate problems. Look at the accompanying chart which shows the same CBO projection that Kotlikoff and Burns present, but breaks it down by program. Yes, the total rises drastically--but Social Security, although it is the biggest of the programs now and the only one of the three programs whose costs are driven mainly by demography, accounts for only a small part of that rise. That tells us that demography is not the main driver of these long-run projections.

How big is the demographic challenge? Pundits who want to sound serious love to contrast Social Security as it was in 1950, when sixteen workers were paying in for every retiree drawing benefits, with Social Security as it will be once the baby boomers have retired, with only two workers per retiree. But most of the transition from sixteen to two happened a long time ago. Since the mid-1970s there have been about three workers per retiree --and Social Security has been running a surplus. The real issue is what happens when three goes to two. How big a problem is that?

The answer is, medium-sized. As you can see in the chart, the aging of the population will cause Social Security spending to rise from its current level of 4.2 percent of GDP to a little over 6 percent by 2030, at which point it will stabilize. If demography were the only factor driving rising Medicare spending, it would rise in roughly the same proportion, from 2.7 to around 4 percent of GDP. So if demography were the whole story, we'd be looking at an eventual demography-driven rise in spending of between 3 and 3.5 percent of GDP by 2030, and no further increase after that. That's not a trivial increase, but it's also not overwhelming; a tax increase big enough to cover that rise in spending would still leave overall taxation in the United States well below the average for other advanced countries.

Still, a responsible government would prepare for the aging of America. Textbook fiscal economics says that when a government knows that its expenses will rise in the future, it should start running a surplus now. At first, this surplus should be used to pay off debt, which reduces the government's future interest costs. If the government runs out of debt to pay off, it can start to invest in assets such as stocks and bonds, which will yield future income. That's exactly the path the Social Security system, though not the government as a whole, has been following.


Social Security has its own budget, with its own dedicated revenue base. In 1983, following the recommendations of a commission headed by Alan Greenspan, Congress tried to prepare the program to deal with the baby boomers: it raised the payroll tax, so that Social Security would run a surplus, with the express intention of building up a trust fund to help pay benefits once the baby boomers had retired. At first, it seemed that this action, together with some changes in benefits, had done the job: "For the next 75 years, the OASDI program is estimated to be in close actuarial balance," declared the Social Security trustees in their 1985 report.[1] Later, the trustees lowered their estimates; the public's impression of a looming Social Security crisis largely dates from the mid-1990s, when they were predicting exhaustion of the trust fund by 2029. But the trustees have lately become more optimistic again: they now say the trust fund will last until 2042. The Congressional Budget Office says 2052, and many economists now think that the original optimism was right after all: if the economy grows as fast over the next fifty years as it did over the past fifty years, Social Security will be sound for the foreseeable future. And if the economy doesn't grow that fast, by the way, the high rate of return on stocks needed to make privatization work can't possibly materialize, either.

At this point a loud chorus on the right insists that such estimates are irrelevant, because the Social Security trust fund is just a meaningless piece of bookkeeping: it's a claim by one part of the government on another part of the government. The real crisis will come much earlier than 2042, that chorus says, because payroll tax receipts will no longer cover the full cost of providing Social Security benefits as early as 2018.


Let's take this argument a step at a time. There are two ways to look at Social Security: you can view it as a stand-alone program with its own funding, or you can view it as just part of the federal budget. These aren't mutually exclusive views. On one side, Social Security has always been run as an independent program, and the independence of its budget has considerable legal and political force. On the other side, Social Security is, of course, part of the federal government, and its benefits must ultimately be paid out of the government's revenue. Depending on the question, it's sometimes useful to focus either on Social Security's specific finances or on its role in overall budgeting. What one can't do, however, is switch views in mid-argument. If you want to discuss the budget of the Social Security system, the trust fund and the interest paid on that fund must be part of the picture. If you want to discuss Social Security's role in the overall federal budget, well, you have to talk about the federal budget as a whole; the fact that one particular tax brings in less revenue than one particular category of spending has no significance.

What the crisis-mongers do, however, is switch between views to suit their convenience. For example, in his magisterial survey of Social Security issues in The New York Times Magazine of January 16, Roger Lowenstein caught Michael Tanner of the Cato Institute red-handed. Mr. Tanner's estimate of a $26 trillion deficit for Social Security turned out to be the result of a calculation based on the principle of heads I win, tails you lose: when Social Security runs a surplus, Mr. Tanner doesn't count it, because the system is just part of the government, but when Social Security runs a deficit, he treats Social Security as an independent entity.

If all this seems metaphysical, let's put it this way: What will actually happen when payroll tax receipts no longer cover 100 percent of benefits? The answer, quite clearly, is nothing.

There are only two ways Social Security could be unable to pay full benefits in 2018. One would be if Congress voted specifically to repudiate the Social Security trust fund, that is, not to pay interest or principal on the trust fund's bonds, which would in effect be a decision not to honor debts to retirees. In 2018 the payments on the trust fund's bonds would be sufficient to cover Social Security benefits. Repudiation of those payments is pretty much inconceivable as a political matter; writing in the periodical The Economists' Voice, David Kaiser of the National War College suggests that such a repudiation might even violate the Constitution. In that sense, the trust fund is as real an obligation of the US government as bonds held by Japanese pension funds. The other way would be if the United States found itself in a general fiscal crisis, unable to honor any of its debt. Given the size of the current deficit and the prospect that the deficit will get much bigger over time, that could happen. But it won't happen because of Social Security, which is a much smaller factor in projected deficits than either tax cuts or rising Medicare spending.

The grain of truth in questions about the meaning of the trust fund is that the rest of the federal budget has not been run responsibly. The Social Security surplus should have been kept in a "lockbox." Although this term has come in for a lot of derision, it was a useful shorthand way of saying that the federal government as a whole should in an average year run budget surpluses at least equal to the surplus of the Social Security system. And this in turn was a shorthand way of saying that the federal government as a whole should do the responsible thing and try to prepay some of the costs of an aging population.

In the 2000 campaign both candidates pledged to honor the lockbox. President Bush clearly never had any intention of honoring that pledge; his first tax cut would have broken the lockbox all by itself, and his insistence on pushing through another major tax cut after launching the Iraq war made it clear that this wasn't a fluke. But that's not a Social Security problem. Viewed on its own terms, Social Security has been run responsibly and is a sustainable system.

And the policy implication of that observation is also clear: the problem isn't with Social Security, it's with the rest of the budget. Social Security has already taken the steps needed to cope with an aging population; at most, it needs some minor tinkering. The main thing we need to do to cope with the demographic challenge is for the rest of the federal government to do its part, by dealing with the huge deficit we already have in the general fund.

3.  What About Privatization?

Let's now turn to the sort-of plan ("sort-of" because the administration still hasn't provided key details) to partially privatize Social Security, diverting part of payroll taxes from their current uses, paying benefits and building up a trust fund, and placing them in private accounts instead.

The administration's rationale for privatization is that it is needed because Social Security is in crisis. As we've seen, that's a huge exaggeration, and many of the things President Bush says--such as his assertions that the system will be "flat broke, bust" when the trust fund runs out--are just plain false. Also, the administration pretends that the core of our failure to prepare for an aging population resides in the finances of Social Security; again, as we've seen, Social Security has actually done a lot to prepare for the baby boomers. Mr. Bush's own actions-- above all, his insistence on cutting taxes while waging war--are largely responsible for the real problem, the huge deficit in the general fund.

But even if a drastic change in how Social Security operates isn't necessary, there's still the question of whether such a change is a good idea.

When they aren't warning that only privatization can save us from doom, privatizers often make their case with the argument that people can do better investing their own money than the deal they get from Social Security. Here's a classic example of the genre: during the 2000 campaign, then-candidate Bush urged his listeners to "consider this simple fact: even if a worker chose only the safest investment in the world, an inflation-adjusted US government bond, he or she would receive twice the rate of return of Social Security." Vice President Cheney made a similar comparison, although he spoke about investing in stocks rather than bonds, just a few weeks ago.

As I pointed out at the time Mr. Bush made his remarks:

That's an amazing fact; it's even more amazing when you realize that the Social Security system invests all its money in, you guessed it, US government bonds. But the explanation--which Mr. Bush's advisers understand very well, even if [Bush himself] does not--is that today's workers are not only paying for their own retirement, but are also supporting today's retirees.

Or to put it a different way, you could equally well say that my family would have more cash on hand if we took all my mother-in-law's money and let her starve. Somebody must pay the cost of caring for retirees and older workers, whose own payroll taxes went to support a previous generation. If the payroll taxes of younger workers are no longer available for that purpose because they are being placed in private accounts, some other source of money must be found. This problem is often summarized with the deceptively innocuous term "transition costs," but it's an enormous one.

 Kotlikoff and Burns offer a privatization plan that doesn't try to fudge the issue of transition costs. They call for a 12 percent national sales tax to pay benefits to current retirees and older workers. This tax would gradually be reduced as the beneficiaries of the current system died off, but it would remain high for a long time. That should give you an idea of what a responsible privatization scheme would entail.

I'd argue that even if we had some way to pay the transition costs, it would be a mistake to privatize Social Security: it was always intended to be an insurance program, not a 401(k), and we need that insurance more than ever in the face of growing economic insecurity. In any case, however, Mr. Bush isn't about to propose a tax increase on that scale or any other.

Instead, he proposes covering the costs of paying benefits to older Americans by borrowing the money. Private accounts would be created using payroll taxes that are currently used to pay for benefits; the government would therefore have to borrow to make up for lost revenue. The government would offset this loss of revenue in the long run by gradually reducing benefits relative to those under current law. These future benefit cuts supposedly wouldn't hurt workers, however, because they would be more than offset by the growth in their personal accounts.

Such schemes come wrapped in fine phrases about the "ownership society," but stripped down to their essence they are equivalent to an investment adviser telling you that you won't have enough money when you retire, but that you should make up for this shortfall not by saving more but by borrowing a lot of money, investing it, and trusting in capital gains.

Even if this strategy were successful, the payoff would be a long time coming. A Congressional Budget Office analysis of "plan 2" from Mr. Bush's social security commission, which is widely believed to be what Mr. Bush will eventually propose, found that it would increase the budget deficit every year until 2050. A similar analysis in last year's Economic Report of the President concluded that the debt incurred to establish private accounts, which would peak at almost 24 percent of GDP, wouldn't be paid off until 2060.

It's likely that financial markets would be made very nervous by borrowing on that scale, with the prospect of repayment so far in the future. Bear in mind that the debt incurred during the four decades of increased deficits would be a real, legally binding promise to repay, while the claim that privatization would save money in the long run depends on the assumption that whoever is running America half a century from now will follow through on benefit cuts, even if private accounts have performed poorly and left many retirees in poverty. In the real world, the bond market would consider the solid fact of soaring debt a lot more significant than projections of savings through politically determined benefit cuts many decades in the future. In practice, privatization would significantly increase the risk that international investors will stop lending to the United States, provoking a fiscal crisis, sometime in the not too distant future.

Even if we ignore the danger of provoking a fiscal crisis, the claim that borrowing to create private accounts will somehow benefit everyone is a remarkable exercise in free-lunch thinking. If nobody suffers any pain, where does the gain come from? If private accounts were invested in government bonds, as Mr. Bush suggested back in 2000, there would be no possible gain; the interest earned by private accounts would be completely offset by the interest paid on the government borrowing to fund these accounts. So the claim that there will be gains from privatization always comes down to this: part of the private accounts will be invested in stocks, and privatizers insist that stocks are more or less guaranteed to yield a much higher rate of return than the government bonds issued to pay for the creation of those accounts.

As Michael Kinsley of the Los Angeles Times has pointed out, there's something very peculiar about that assertion: if stocks are a clearly better investment than government bonds, why would anyone out there be willing to sell all the stocks that would end up in private accounts, and buy all the bonds the government would have to issue along the way? Are politicians pushing for privatization asserting that they know more about future rates of return than investors making decisions about where to put their own money?


In response to such questions, privatizers duck the conceptual issue, and take refuge in history: stocks have, in fact, been a much better investment than bonds in recent decades. But as the mutual fund ads say, "Past performance is no guarantee of future results." Stocks are much more expensive relative to underlying profits than they were in the past, which means that they can be expected to yield a lower return. The best bet, suggested both by a look at the numbers and by basic economic theory, is that prospective returns in the form of dividends and capital gains on stocks are somewhat higher than those on bonds, but not much higher--and that the higher expected return on stocks is offset by higher risk. That's why prudent investors hold portfolios containing both stocks and bonds, and why borrowing to buy stocks--which is, to repeat, what Bush-style privatization boils down to--is a very bad idea.

Taking away the assumption that stocks will yield very high rates of return fatally undermines the arithmetic of privatization. Again, consider the analogy of borrowing and using the money to buy stocks: if those stocks end up yielding a lower rate of return than the interest rate on the loan, you've made yourself worse off. Even if your best guess is that the return on stocks will somewhat exceed the interest rate, you can't be sure of that, and you'll be in a lot of trouble if your guess proves wrong. Most privatizers assume, when selling their schemes, that stocks will yield about 7 percent a year on average after inflation, while the interest rate after inflation will be only 3 percent. If the equity premium --the spread between the average return on stocks and the average return on bonds--really were that large, borrowing to buy stocks wouldn't be a sure thing, but the odds would be strongly in favor of coming out ahead. But if the expected rate of return on stocks is only 5 percent or less, which many economists think is more reasonable, the chances that borrowing to buy stock will end up being a los-ing proposition are quite high--especially if one takes mutual fund fees into account.

Privatizers hate it when you talk about fees--about the fact, for example, that the much-touted Chilean system has administrative costs about twenty times those of Social Security, or that according to Britain's Pensions Commission, "providers' charges" in that country's privatized system reduce the size of retirement nest eggs by between 20 and 30 percent. But when we're talking about the narrow equity premium produced by realistic expectations of future yields, fees become a central issue.

The plan of Kotlikoff and Burns for personal accounts is useful as an example of what would be necessary to keep fees minimal: it calls for a system in which workers have no control at all over how their personal accounts are invested. Instead, all accounts would be placed in a global index fund administered by the government: "a single computer, situated in the Social Security Administration, would be programmed to buy and sell securities." In essence, the government, not individuals, would be doing the investing, and the personal accounts would simply be an accounting device. The administrative costs of running this system would be very low.


But it's very unlikely, if Social Security is privatized, that the system will look like that. For one thing, the advertising for privatization stresses "choice." In fact, in 2002 the Cato Institute quietly renamed its Project on Social Security Privatization the Project on Social Security Choice (focus groups said that "privatization" had negative connotations). It's hard to see how to reconcile that advertising with a system in which a computer programmed by bureaucrats does all the choosing. Also, as a matter of political reality, the investment management industry isn't going to accept the idea that a huge pool of money and potential profits is off-limits. Investment companies gave lavishly to the inaugural celebrations, and are major contributors to the lobbying organizations that have been set up to push privatization. They aren't spending that money simply because they think privatization is in the public interest.

Suppose that we end up with a system like that of Britain or Chile, in which mutual funds compete to attract private accounts. In that case, there's every reason to believe that fees will take a large bite. In 2003, the average "expense ratio" on US stock funds-- the ratio of all the various fees charged by management to the amount invested --was 1.5 percent. In Britain, providers' charges used to take more than 2 percent off the return of the average retirement account; new regulations have reduced that, but only to about 1.1 percent. Put fees of that magnitude plus a realistic rate of return on stocks into a typical numerical model of privatization, like the one in the CBO report on plan 2, and privatization quickly turns into a sure-fire losing proposition: the government borrows to establish private accounts that if anything yield an expected rate of return lower than the rate the government pays on its bonds; yet those accounts introduce a major new element of risk.

If Bush-style privatization actually goes through, the end game is fairly predictable: it's what is happening in Britain now. A couple of decades from now, it will be obvious to everyone that the returns on private accounts have fallen far short of expectations, and that America is about to experience a resurgence of poverty among the elderly. There will be irresistible demands for the government to call off cuts in benefit levels. (Remember, the over-sixty-five population will be an even larger share of the electorate than it is now.) And the result will be to make the fiscal outlook much worse than it would have been without privatization: the government will have borrowed trillions of dollars with the promise of future budget savings, but those savings will never materialize.

4.  Medicare, Medicaid, and the Health Care Challenge

If demography is only a medium-sized problem, why do long-run federal budget projections look so scary? The answer is that they assume that the long-term historical tendency of health care spending to rise faster than gross domestic product will continue. That trend has not reflected runaway government spending: private spending on health care has risen almost as fast as government spending. (In 1980, private health spending was 5 percent of GDP, and government health spending was 3.8 percent. By 2003 the numbers were 8.3 and 7.0, respectively.) Nor is it a case of runaway inflation: rising medical costs have not historically been driven by rising prices for existing medical procedures. There is plenty of gouging and waste in the US health care system, but there always has been, so that's not a big factor in the trend. The main reason health care is continuing to absorb a larger share of the economy is innovation: that the range of things that medicine can do keeps increasing.

A good example of what drives rising health care spending is the recent decision by Medicare to pay for implanted cardiac devices in many patients with heart trouble, now that research has shown them to be highly effective. Should this be considered a cost increase? Only if we're careful about what we mean by "cost." It doesn't increase the cost of providing the same care as before; Medicare is spending more to take advantage of a new opportunity to save lives.

Because rising health care spending is, for the most part, driven by increased opportunities, it's not clear that a rising share of health care spending in the economy should be considered a bad thing. Here's what the Congressional Budget Office, the source of those frightening long-term projections, had to say:

Although the rise in health care costs is a serious concern for many policymakers, it largely reflects private choices.... As income rises, consumers may prefer to allocate a larger share of their resources to health care and a smaller share to other goods and services.[2]

Still, there is a problem--but it is social and moral as much as economic: How much inequality in the human condition are we prepared to accept? In Charles Dickens's Britain there were huge class differences in health and longevity, because only the well-off had access to adequate nutrition and, if living in urban areas, to a more or less sanitary environment. Today those differences still exist but are much narrower, in part because of economic growth (which means that more people can afford an adequate diet), but also in large part because of public spending on sanitation, disease control, and health insurance systems that try, however, imperfectly, to provide essential care to everyone.


But what do we do as medical advances make it possible to extend lives or greatly improve their quality, but only at a very high cost? Today we expect the public sector to pay for essential care when individuals cannot pay, and we do so for good reason. Imagine the inequalities that would already exist in America if Medicare wasn't there: high-income Americans would receive hip replacements and bypass surgery in their old age, while low-income Americans would find themselves crippled or dead. Yet the cost of preventing fundamental inequalities in medical care will grow over time.

This isn't just, or even primarily, a question of whether we are prepared to raise federal taxes to pay for rising Medicare and Medicaid spending. The clear and present dangers, health econ- omists tell me, are the inability of state governments to pay their share of Medicaid, and the threat to private health insurance, which is gradually unraveling in the face of rising costs. Between 2001 and 2004, according to the Kaiser Family Foundation, the percentage of American workers receiving health insurance through their employers fell from 65 to 61, and this decline will continue unless the government starts helping out. (John Kerry's plan to have the government pay catastrophic health costs was an example of the sort of thing that may be required, but even that would have provided only limited relief.)

The problem of rising medical costs is much harder to resolve than that of an aging population. In the long run, in fact, it may be impossible to resolve. But there are things we could do to postpone the day of reckoning. One would be to prepay some of those future medical costs; at the very least, we ought to be building up a Medicare trust fund to deal with the demographic component of rising costs, i.e., the increase resulting from the rising proportion of people over sixty-five.

Another would be to find ways to make the US health care system more efficient. For the most part, that's a subject for another essay, but it seems worth making one point: when it comes to health care, the free-market ideology that currently dominates American political discourse seems utterly wrong. Systems that provide universal coverage, like those of France or Canada, are much cheaper to run than our market-based system, yet they yield better results with respect to life expectancy and infant mortality. Or if you don't trust foreign examples, consider the remarkable renaissance of the Veterans' Administration hospital system, described in an important article by Phillip Longman in the February Washington Monthly: he shows that the VA system's centralization of information and control over resources allow it to provide better care at lower costs than any private system.[3]

In other words, whatever the current administration and congressional majority propose to deal with the health care crisis--you can be sure they'll declare a crisis as soon as they're done with Social Security-- will actually move our system in the wrong direction.

5.  Back to the Future

Unless something very unexpected happens, Kotlikoff and Burns's vision of an America that in 2030 has an older population than Florida today will come to pass. It's also quite possible that the state of the nation will be as bad as they suggest in their opening account. But one won't be the result of the other, and in a perverse way exaggerating the demographic challenge makes that grim future more likely.

Here's how the debate is really playing out, in four easy steps:

1. Talking heads and other opinion leaders perceive the issue of an aging population not as it is--a middle-sized problem that can be dealt with through ordinary changes in taxing and spending--but as an immense problem that requires changing everything. This perception is, alas, fueled by books like The Coming Generational Storm, which blur the distinction between the costs imposed by an aging population and the expense of paying for medical advances.

2. Because the demographic problem is perceived as being much bigger than it really is, the spotlight is off the gross irresponsibility of current fiscal policy. As you may have noticed, right now everyone is talking about Social Security, and nobody is talking about the stunning shift from budget surplus to budget deficit since Bush took office.

3. The focus on Social Security-- the one part of the federal budget that is actually being run responsibly--is, in practice, offering the architects of our budget deficit an opportunity to do even more damage.

4. Finally, we're not having a serious national discussion about the bigger problem of paying for health care, and we probably can't in today's ideological climate.

Four years ago, I and many other economists urged policymakers to think about the future cost of Social Security benefits, not because we thought there was anything wrong with Social Security itself, but because we regarded the future costs as a compelling reason not to cut taxes even if the overall budget was in surplus. Today, with the overall budget deep in deficit, and the administration considering "tax reform" that will amount to even more tax cuts for the well-to-do, it all seems a moot point. The first priority is to do something about the fiscal crisis we have right now, not worry about the fiscal crisis we might face a generation from now.

--February 10, 2005


[1] See trustyears.html, pp. 2-3.

[2] See 4916&sequence=2.

[3] Phillip Longman, "The Best Care Ever," Washington Monthly, February 2005.