Doug Elmendorf
September 13, 2018
IMF Conference on Sovereign Debt

I am delighted to be joining you for this interesting and important conference, and to be discussing this excellent paper on sustainability of public debt.

I have two qualifications for my assignment today: First, I have made some modest contributions over the past 20 years to the research literature on fiscal policy. Second, I spent six years as director of the U.S. Congressional Budget Office advising the Congress about fiscal policy. During that time, U.S. federal debt more than doubled, jumping by 35 percent of GDP—so I know a bit about unsustainable paths for debt. I will draw on both of these qualifications in my remarks today.

I like the paper a lot. It is a wide-ranging and thoughtful summary of a large amount of both pure research and applied analysis of debt sustainability over the past few decades. I learned things I did not know or had forgotten, and I agree with its main points. I think the paper will be a terrific resource for scholars and practitioners. I also appreciate that the draft I received already included a footnote thanking Elena and me for our insightful comments, which was a very optimistic and gracious approach.

I will use my time to extend the authors’ discussion in three directions—the first about uncertainty and debt sustainability, the second about the consequences of persistent changes in interest rates, and the third about communicating with a nontechnical audience.

Uncertainty and Debt Sustainability

Uncertainty is the central challenge in assessing debt sustainability, which is why I want to highlight it. There are at least two important aspects of this challenge.

First, forecasts of budget deficits conditional on current policies are often poor, because of the difficulty of predicting both economic growth and tax revenues and government spending given economic growth. I have spent a noticeable portion of my career working with very smart people trying to predict economic outcomes and to predict budget outcomes given those economic outcomes, and these are two hard challenges. Part of the problem is that annual budget deficits are the difference between two much larger numbers, so small errors in predicting those larger numbers cause big errors in predicting deficits. Another part of the problem is that deficits compound to debt in an exponential fashion, so small errors in predicting annual deficits cause big errors in predicting debt.

The result is that assessments of sustainability generally involve tremendous uncertainty.

As just one example, CBO’s latest long-term budget outlook for the United States (CBO, 2018) shows a range of possible debt-to-GDP ratios 30 years ahead, all assuming that current law is generally unchanged but allowing for different possible outcomes for labor force participation, total factor productivity growth, interest rates, and health care spending growth. The range is from 85 percent to 247 percent, or 3-to-1, and CBO cautions that actual outcomes could fall outside this range. The bottom edge of this fan chart is probably sustainable, while the top is almost certainly not.

The second challenge regarding uncertainty is that forecasts of changes in budget deficits in response to changes in policies are complicated by the feedback between budget policies and economic growth—which varies depending on budget and economic circumstances.

For example, some deficit-reducing policy changes in some circumstances will depress short-term growth, and different changes in different circumstances will boost short-term growth. Alberto Alesina and others have written a number of papers about reductions in public spending that boost household or business confidence or reduce interest rates sufficiently to raise private spending by more than the reduction in public spending, and Dave Reifschneider and I did some modeling of this possibility for the United States about 15 years ago (Elmendorf and Reifschneider, 2002). On the other hand, CBO has always concluded that proposed or actual changes in U.S. budget policy that would reduce budget deficits would depress U.S. economic growth in the short term.

The result is that assessments of the impact of policy changes on sustainability generally involve significant uncertainty.

Consequences of Persistent Changes in Interest Rates

The second topic I want to cover is the consequences of persistent changes in interest rates, which can make a big difference in assessing debt sustainability. I will focus on persistently low interest rates, which may seem an odd scenario when one is examining potentially unsustainable public debt, but a scenario that is relevant today in a number of countries. I will focus on the example of the United States.

In the United States, the interest rate on federal debt has generally been below the growth rate of the economy. Let r equal the average interest rate on public debt, defined as government interest payments divided by the par value of outstanding debt. Let g equal the growth rate of nominal GDP. Larry Ball and Greg Mankiw and I showed that r  was less than g, on average, for the 1871-to-1992, 1920-to-1992, and 1946-to-1992 periods for the U.S. federal government (Ball, Elmendorf, and Mankiw, 1998). CBO (2018) reports the recent history of r and g, as well as the agency’s projections of r and g for the coming three decades (see the table on the following page). The interest rate was higher than the economic growth rate for the 1988-to-1997 decade, but below it for the 1998-to-2007 and 2008-to-2017 decades. Indeed, r has been much lower relative to g during the past five years than earlier. Going forward, CBO projects that the interest rate will be less than the growth rate for the next two decades but edge ahead of it thereafter.

How can r be less than g if the U.S. economy is dynamically efficient? Larry and Greg and I surmised that the historical reason is uncertainty: The return on government debt is below the average return to private capital because of the debt’s lower risk, and it can be below the growth rate of the economy most of the time as long as there is some risk of a sufficiently adverse economic outcome.

CBO’s historical data and 2018 projections
Average interest rate + Growth rate 1988-97 1998-2007 2008-17 2018-27 2028-37 2038-47
r 7.3 5.2 2.5 3.1 3.6 4.1
g 5.9 5.4 3.0 4.1 4.0 4.0

Moreover, as Louise Sheiner and I argued in the Journal of Economic Perspectives last year: “Both market readings and detailed analyses by a number of researchers suggest that Treasury interest rates are likely to remain well below their historical norms for years to come.” That is, the downward trend we have seen in Treasury rates during the past few decades will probably not be fully reversed in the coming years. The reasons why Treasury interest rates will probably be lower in the future than the past include much slower growth of the labor force, a larger share of total income going to higher-income households, and an increase in investors’ preference for safety.

Low interest rates matter for assessments of debt sustainability in two ways: First, lower interest rates make debt dynamics more favorable: For given primary deficits, debt compounds more slowly. Second, as Louise and I showed: “Many—though not all—of the factors that may be contributing to the historically low level of interest rates imply that both federal debt and federal investment should be substantially larger than they would be otherwise.” That is, the amount of debt the United States should be trying to sustain is larger than it would have been before.

The broad lesson is that assessments of debt sustainability need to walk a fine line: They should neither assume permanence of changes in interest rates nor ignore the implications of persistent changes.

Communicating Debt Sustainability to a Nontechnical Audience

My third topic is the communication of assessments of debt sustainability. Communicating debt sustainability to policymakers and the public is both difficult and crucial.

One perspective on sustainability is to assess whether a country will be able to meet its debt obligations using evidence on historical adjustments to government spending and taxes. A related but different perspective on sustainability is to advise countries about the adjustments to government spending and taxes they should make in the future. I find the latter perspective more useful, in part because I have served as an adviser on this issue, and in part because historical political reaction functions may not provide much information about future political behavior. But if the focus is changing future behavior, then communication is key.

Let me offer three suggestions regarding communication of debt sustainability. First, analysts should focus on estimated changes in policy needed to achieve specific debt outcomes rather than other measures of sustainability. Analyses of likely outcomes based on historical adjustments to policies can end up shifting some responsibility away from current policymakers and citizens, while showing the adjustments needed to reach specific goals places the responsibility squarely where it belongs.

Moreover, announcing critical thresholds is not constructive because such sharp lines cannot be reliably estimated (they vary across countries and over time within a country) and because policymakers should make adjustments well before those extreme values are reached (in part because of risks and in part because political constraints on tax rates are usually reached well before the top of the Laffer curve is crossed). The evidence cited in the paper about how sovereign debt ratings and credit spreads depend on many variables besides public debt and deficits reinforces this point.

As a positive example, CBO (2018) estimates that holding U.S. federal debt at its current percentage of GDP (78 percent) over the next 30 years would require a reduction in the primary deficit of about 2 percent of GDP or roughly $400 billion per year today, while returning debt to its historical percentage of GDP (41 percent over the past 50 years) would require a reduction in the primary deficit of about 3 percent of GDP or roughly $600 billion per year today. Those amounts can be readily compared to current spending on various government programs, to provide a reference point for the scale of the changes needed.

Second, analysts should focus on straightforward calculations that can be easily explained. Present values of infinite future streams of revenues or spending—as near and dear as they may be to our hearts—do not resonate as well with most policymakers as the simpler annual figures for revenues and spending that they see and deal with on a regular basis. That is why CBO talks about how many hundreds of billions of dollars of annual adjustments are needed rather than describing the unfunded open-group liability of Social Security as so many trillions of dollars.

Third, analysts should be open about uncertainty and discuss the value of insurance against negative budget shocks—but focus their attention on point estimates. When I was CBO director, my colleagues and I spent a great deal of time formulating confidence regions for certain estimates and developing ways to present and explain those confidence regions. I am proud of that work intellectually and pleased that CBO is continuing it. But I think that our analysis of uncertainty contributed less, on balance, to Congressional decision-making than we had hoped. The temptation for each policymaker and citizen to choose the side of a confidence region that provides the greatest support for their preconceived position is nearly irresistible, and letting people essentially pick their preferred estimates can make the policy debate less honest and less comprehensible.

With that, I will stop and simply thank the authors again for their fine work.


Ball, Laurence, Douglas Elmendorf, and Gregory Mankiw, “The Deficit Gamble,” Journal of Money, Credit, and Banking, November 1998.

Congressional Budget Office, 2018 Long-Term Budget Outlook, June 2018.

Elmendorf, Douglas and David Reifschneider, “Short-Run Effects of Fiscal Policy with Forward-Looking Financial Markets,” National Tax Journal, September 2002.

Elmendorf, Douglas and Louise Sheiner, “Federal Budget Policy with an Aging Population and Persistently Low Interest Rates,” Journal of Economic Perspectives, Summer 2017.