Time to end the reign of terror of scary upward-sloping graphs

Once a year, the Congressional Budget Office (CBO) publishes long-run debt projections under their assumptions about budget policy under future Congresses, known as the alternative fiscal scenario (AFS). It is used extensively by many—including House Budget Committee Chairman Paul Ryan (R-Wis.)—to argue that we face a catastrophe that can only be solved by effectively dismantling the social safety net and retirement systems that we have in place. But it’s also misleading.

Michael Linden at the Center for American Progress recently released a great analysis showing that this scary long-run debt projection is only scary because CBO assumes that future policymakers will make policy decisions that will make the deficit much worse. If you remove those assumptions to arrive at a more honest baseline, then the problem of an unsustainable rising debt mostly disappears.


But let’s back up a bit and marvel at the absurdity of long-term debt projections. Remember, these projected deficits are largely the product of CBO’s economic and demographic projections, coupled with assumptions about decisions made by future policymakers and long-term health costs. Moreover, economic, demographic, and other budgetary projections are most reliable in the near-term, and their margin of error compounds with time.

Imagine trying to figure out how the 138th Congress that will govern in 2063—likely with many members that aren’t even born yet—will feel about marginal tax rates, or the cost of an MRI. Will we still even use MRIs? On the subject of healthcare, consider this: a device was recently developed that, once implanted into a patient’s body, can provide ongoing monitoring of various blood chemicals and report the findings to a smartphone. This could have a hugely transformative impact on our currently reactive health care system. But would it reduce or increase the share of the economy we spend on health care? Would health costs fall because visits to the doctor would be less frequent and diseases and emergencies would be detected early? Or would costs rise because every patient would start demanding what may be an expensive implant that ends up telling doctors nothing they don’t already know about 99 percent of their patients?

Slate columnist Matthew Yglesias proposes an apt thought experiment regarding 75-year projections: imagine trying to make a projection, in 1940, of what health costs will be in 2015. It’s hard enough to figure out how digital records might affect health costs, but in 1940 you’d first have to anticipate their invention because back then they didn’t even exist. Neither did antibiotics, for that matter.

These are questions probably best described by “Keynesian uncertainty,” or likelihoods so indeterminate that Keynes equated them to assessing the probability of another European war. “About these matters,” Keynes wrote, “there is no scientific basis on which to form any calculable probability whatever. We simply do not know!” In other words, the most accurate long-term debt projection might not be this:


…but rather this:


But the CBO barrels ahead, making policy and health care assumptions based on the only thing it has: the recent past. This may be reasonable for policies with recent political precedent—such as the history of regular adjustments for the Medicare Sustainable Growth Rate (i.e., the “doc fix”)—but policies like the health care reform excise tax have no such history.

Its health care costs projection methodology is luckily more transparent: CBO takes the average of health care cost growth over the last 25 years (weighted such that the most recent year counts twice as much as the first year) and simply applies that to each future year in (near) perpetuity. (In the later years of the projection they assume a slightly slow-down in cost growth because they assume consumers will shift away from health care to keep their real disposable income from falling.) But a model’s output is only as good as its inputs, and it’s far from clear that the last 25 years of health care cost growth rates are a strong predictor of the next 75 years.

In fact, there are strong signs that the unsustainable health care cost growth rates that we’ve experienced over the past few decades may be slowing down. From 1961 to 2007, health spending grew by more than 5 percent annually, and in many years hit double digits. Yet in the last three years—2009, 2010, and 2011—health spending has grown at 3.9 percent each year, the lowest rate in over 50 years. More importantly, rising national health expenditure over the last few years has been stable as a share of the economy.

This slowdown has caused CBO to revise its health care cost projection downward, albeit slightly—after all, CBO’s model takes into account the last 25 years, so a few good years won’t drastically impact CBO’s projections. In fact, it may take upwards of a decade of slow health care costs before the trajectory of CBO’s health care projections is significantly altered—in CBO’s model, the latest 10 years are roughly equal in weight to the first 15 years, so it would take roughly a decade of low growth before the projected rate of excess cost growth was cut in half. This lag means that CBO’s current model will continue to project unsustainable health care costs long after the problem has actually been solved.

So we come to a point where the entire political system is clamoring for a solution to a problem that (a) may not be a problem, and (b) may have already been solved.

What are the implications for policymaking? Simply put, policymakers should focus solely on solutions to long-run debt and budgetary issues that make good policy sense even in the absence of a long-run debt problem. Reforming how Medicare reimburses providers or adding a public option to health care exchanges has the potential to make the provision of health care cheaper not just for the government, but for states and families as well. In contrast, Paul Ryan’s proposal to strip seniors of their guaranteed coverage by converting Medicare to a voucher program would actually increase national health care expenditure because private health coverage costs are higher than public health coverage costs. This is a policy that would never even get a public hearing if not for the invocation of the “looming debt crisis,” which belies the policy’s inherent weakness. On the other hand, defunding public investment would raise unemployment, lower middle-class living standards, and decrease potential economic output regardless of whether the long-run fiscal outlook incrementally proves sustainable or unsustainable.

It is easy to look at CBO’s projection and believe we have a long-term debt problem. But those numbers are likely to be no more accurate than predicting in 1957 that in ten years most households would be able to purchase a flying car. If the political system is determined to use these long-term debt projections, it should at least take them with a heaping load of salt. The best projections for guiding policy are the near-term projections, and the story there is clear: The United States faces a huge demand shortfall, big cyclical budget deficits, high unemployment, and anemic growth—all of which will be exacerbated by more austerity being touted by those who hope to scare us with upward-sloping graphs.