DISCUSSION DRAFT
A Reconsideration of Fiscal Policy in the Era of Low Interest Rates
Jason Furman and Lawrence Summers
1
November 30, 2020
The last generation has witnessed an epochal decline in real interest rates in the United
States and around the world despite large buildups of government debt. As Table 1 illustrates
U.S. ten-year indexed bond yields declined by more than 4 percentage points between 2000 and
early 2020 even as projected debt levels went from levels extremely low by historical standards
to extremely high by historical standards. Similar movements have been observed at all
maturities and throughout the industrial world. Available market data suggests that the COVID
crisis has depressed real interest rates despite raising government debts, likely by increasing
inequality, uncertainty and the use of information technology.
Table 1
Note: Debt-to-GDP forecast is the CBO 10-year ahead forecast (2030 from June 2019 Alternative Fiscal Scenario for 2020). Real
interest rates are based on 10-year Treasury Inflation Protected Securities (TIPS) from January 2000 and February 2020.
Source: Congressional Budget Office (2000, 2019b); U.S. Department of the Treasury; authors' calculations.
This paper argues that while the future is unknowable and the precise reasons for the
decline in real interest rates are not entirely clear, declining real rates reflect structural changes in
the economy that require changes in thinking about fiscal policy and macroeconomic policy
more generally that are as profound as those that occurred in the wake of the inflation of the
1
Jason Furman is a Professor of the Practice of Economic Policy jointly at the Harvard Kennedy School and the
Economic Department at Harvard University as well as a non-resident Senior Fellow at the Peterson Institute for
International Economics. Lawrence Summers is the Charles W. Eliot Professor and President Emeritus at Harvard.
The authors are indebted to extraordinary research assistance by Wilson Powell III and additional research
assistance by Joseph Kupferberg. The authors have benefited from innumerable conversations on these topics with
many people over the years and specific inputs to this paper from Wendy Edelberg, Richard Kogan and Brian Sack.
2000 2020
Debt-to-GDP (10-year ahead forecast) 6% 109%
Real Interest Rates 4.3% -0.1%
U.S. Debt is Much Higher
But Interest Rates Are Much Lower
2
1970s. In terms of the dichotomy posed by Blanchard and Summers (2019) we believe the
evidence points increasingly towards revolution rather than evolution.
Our analysis begins by considering the downward trend in real rates. We note that with
massive increases in budget deficits and government debt, expansions in social insurance, and
sharp reductions in capital tax rates, one would have expected to see increasing real rates if
private sector behavior had remained constant. We suggest that changes in the supply of saving
associated with lengthening life expectancy, rising uncertainty and increased inequality along
with reductions in the demand for capital associated with demographic changes, demassification
of the economy, and perhaps changes in corporate behavior have driven real interest rates down.
It may well be that, as suggested by Stansbury and Summers (2020), declines in the sensitivity of
spending to real rates due to declines in interest-sensitive spending as a share of GDP and other
factors have exacerbated the decline in real rates. We argue that neutral real rates are as likely to
fall below their levels immediately prior to COVID as they are to rise above them. Finally, we
consider the role of monetary policies, changing asset supplies and risk premiums in affecting
long-term real rates concluding that they are likely minor and, in any event, that the precise
reasons for declining real rates do not much affect their implications for fiscal policy.
We discuss three implications for fiscal policy that follow from low interest rates:
First, fiscal policy must play a crucial role in stabilization policy in a world where
monetary policy can counteract financial instability but otherwise is largely “pushing on a string”
when it comes to accelerating economic growth. The roughly 600 basis point reductions in rates
that have been found necessary to counteract recessions will be infeasible for the foreseeable
future. Limitations on how far interest rates can be reduced given the zero lower bound and the
possible inefficacy of lower rates in stimulating demand raise the possibility that full
employment may be infeasible with overly restrictive fiscal policies. Even if full employment is
feasible with a given fiscal policy there is the possibility that the necessary very low interest rate
level will be associated with excessive leverage and put financial stability at risk. These views
represent a departure from the orthodoxy of the last generation. They open up the prospect that
countries may be less constrained by fiscal space because fiscal expansions themselves can
improve fiscal sustainability by raising GDP more than they raise debt and interest payments.
They also imply that policymakers need to do more to both improve automatic recession
insurance and also find more ways to use fiscal policy to expand demand without increasing
3
deficits, for example through balanced budget multipliers, more progressive fiscal policy and
also expanded social insurance.
Second, we reconsider traditional views about the dangers of debt and deficits. We note
that in a world of unused capacity and very low interest rates and costs of capital, concerns about
crowding out of desirable private investment that were warranted a generation ago have much
less force today. We argue that debt-to-GDP ratios are a misleading metric of fiscal sustainability
that do not reflect the fact that both the present value of GDP has risen and debt service costs
have fallen as interest rates have fallen. Instead we propose that it is more appropriate to
compare debt stocks to the present value of GDP or interest rate flows with GDP flows. We note
that at current and prospective interest rate levels nominal and real Federal debt service is likely
to be low not high by historical standards over the next decade, a point that is strengthened when
account is taken of interest recycled to the Treasury by the Federal Reserve and interest receipts
on Federal financial assets. Moreover, current debt levels are at low rather than high levels
relative to calculations of the present value of GDP or prospective tax receipts. The kind of
reasoning employed in formulating the Maastricht criteria a generation ago does not suggest
alarm about current debt levels or those over the next decade in the United States or most
European countries. While current projections do raise concerns over the fiscal situation beyond
2030 we note that that there is enormous uncertainty and that much of the issue would be
addressed if necessary reforms internal to Social Security and Medicare were undertaken.
Third, we consider the issue of borrowing in the context of how the borrowed funds are
used. We highlight that traditional notions of financial responsibility for households and
businesses hold that borrowing in order to invest in assets that have a return well in excess of the
cost of borrowing increases creditworthiness and benefits future stakeholders. Think for example
of a household that accumulates equity by owning rather than renting the home in which it lives
or a business that owns rather than leases its headquarters. Drawing on recent work considering
dynamic scoring effects of various Federal expenditure programs we argue that borrowing to
finance appropriate categories of Federal expenditure pays for itself in Federal budgetary terms
on reasonable assumptions.
We conclude with thoughts on appropriate guidelines for U.S. fiscal policy. We reject
traditional ideas of a cyclically balanced budget on the grounds that it would likely lead to
inadequate growth and excessive financial instability. We set the goal that fiscal policy should
4
advance economic growth and financial stability. Achieving this goal depends on both improving
responses to downturns and expanding and improving public investment. As a new guidepost,
we propose that fiscal policy focus on supporting economic growth while preventing real debt
service from being projected to rise quickly or to rise above 2 percent of GDP over the
forthcoming decade. We also propose three guidelines that would be consistent with achieving
this broader objective within the guidelines we recommend: (i) undertaking substantial
emergency spending that is not paid for in response to economic downturns; (ii) paying for all
long-term commitments with broad exceptions for ones that plausibly pay for themselves in
present value; and (iii) improving the composition of government to make it more supportive of
demand and also more efficient.
The surprising and likely long-lasting decline in interest rates
As striking as any development in the global economy over the last generation is the
large and sustained fall in real global interest rates. As Figure 1 illustrates it has been a feature of
all the G7 economies and more broadly has been universal in the industrial world. A clear
downward trend in longer term real rates antedates the 2008 financial crisis and has continued
since it was substantially resolved. The observations that the trend has been equally pronounced
in long- and short-term real rates, has lasted over 30 years and has coincided with constant or
slightly declining rather than increasing inflation and inflation expectations suggest that it is a
real rather than a monetary phenomenon.
2
Figure 2 shows that while long-term forecasts largely
expected the slowdown in economic growth due to an aging population, they entirely missed the
large decline in real interest rates.
2
Real interest rates were also negative or low in the 1940s and 1950s (Council of Economic Advisers 2015). This
appears to have been primarily to do with a combination of financial repression, falling debt, high tax rates on
capital, and lack of robust social insurance contributing to high levels of saving.
5
Figure 1
Note: Inflation measured by one-year changes in the core consumer price index (core personal consumption expenditures for
United States).
Source: Bank of Canada; Statistics Canada; Eurostat; Japanese Statistics Bureau; U.S. Bureau of Economic Analysis;
Macrobond; authors’ calculations.
Figure 2
Note: Real interest rates are calculated from forecasted/actual nominal interest rates and forecasted/actual Consumer Price Index
inflation.
Source: The Board of Trustees, Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds (1989, 2020);
authors’ calculations.
Estimates of the “neutral real interest rate” provided by the Federal Open Market
Committee (FOMC) and economists have also fallen to low levels. The FOMC projection of the
6
neutral real rate derived as the difference between its long-run federal funds rate and inflation
projection is now at an all-time low of 0.2 percent down from 1.95 percent eight years ago as
shown in Figure 3. Aronovich and Meldrum (2020) provide new estimates and a summary of the
findings of existing macro models suggesting a downward trend and a real rate very close to
zero. These estimates all significantly exceed what is priced into markets. As this is written 10
year forward federal funds are priced at 1.4 percent and ten year forward inflation is priced at
2.2 implying a real federal funds rate of -0.9 a decade from now.
Figure 3
Note: Expected long-run real federal funds rate calculated by subtracting long-run expected PCE inflation, adjusted for the
historical divergence from CPI inflation, from the long-run expected federal funds rate.
Source: Federal Reserve Bank of St. Louis; Bureau of Labor Statistics; Bureau of Economic Analysis; Macrobond; authors’
calculations.
Longer-term real interest rates are expected to remain negative. The five-year forward
five-year real indexed bond rate is currently -0.5 percent and the ten-year forward expectation of
the ten-year real interest rate is -0.1 percent. These estimates are more likely to be over- than
underestimates of prospective real short-term rates since they make no allowance for term or
liquidity premiums normally present in these markets. In a mechanical sense, increases in
inflation expectations towards currently targeted levels would reduce real rates unless the Federal
Reserve took offsetting actions. Convergence with other industrial countries would also reduce
U.S. real rates.
7
The decline in interest rates has happened even as public debt levels have risen. In the
United States the real interest rate on ten-year debt was 4.3 percent as measured by the yield on
10-year Treasury Inflation-Protected Securities (TIPS) in January 2000, a time when it was
widely expected that the entire federal debt would be largely paid off over the following decade
as shown in Table 1 above. In February 2020 the economy was in a similar cyclical position but
the debt was on course to rise to more than 100 percent of GDP over the following decade (the
pre-pandemic projection) and at the same time the 10-year TIPS rate had fallen to -0.1 percent.
Any discussion of the causes of the decline in real interest rates has to begin with
the observation that the most obvious factors bearing on real rates—government fiscal positions,
social insurance expansions and changes in the after-tax profitability of capital—have all
operated to increase rates. They imply that ceteris paribus neutral real rates in the industrial
world would have declined by about 700 basis points (Rachel and Summers 2019). This suggests
substantial changes in the structure of the private economy. Rachel and Summers (2019) suggest
a tentative decomposition of some of these factors as shown in Figure 4.
Figure 4
Source: Rachel and Summers (2019).
8
At this point we find explanations based on the financial crisis, monetary policy choices
or global factors to be implausible accounts of more than a small part of the decline in real rates.
Declines in real rates predate the financial crisis, continued after financial conditions had
normalized, and are equally pronounced in countries like Canada where the financial crisis was
much less severe and financial institutions remained healthy.
In general, economists doubt the ability of monetary policy to affect real rates over long
horizons. Abnormally easy monetary policies would be expected to manifest themselves in
unusually rapid growth in nominal GDP whereas actual growth has consistently fallen short of
expectations. There is no tendency for declines in real rates to be greater at the short end than at
the long end of the yield curve as one might expect if monetary policy was a primary driver.
Rachel and Summers (2019) consider global factors. There have only been very small
fluctuations in the current account of the industrial countries taken as a group. While the United
States was running a substantial current account deficit in 2005 when Ben Bernanke (2005)
famously invoked a “savings glut” to explain declining U.S. real rates, that deficit roughly
halved over the subsequent decade even as real rates continued to decline.
The evidence is most consistent with structural changes in propensities to save and invest
as the dominant reason for declining real rates. As Summers (2014) argued, factors operating to
raise private saving include longer retirement periods, increased inequality, and rising
uncertainty. Factors operating to reduce private investment include slowing labor force growth,
greater efficiency in the use of capital, for example through companies like Uber and Airbnb, the
impact of information technology in reducing the need for large capital investments, as for
example law firms need much less office space per lawyer, and dramatic reductions in the
relative price of capital goods. Increases in corporate market power and increased pressure on
corporations to pay out cash to shareholders may also contributed to reduced investment.
An alternative line of explanation for low and declining real rates focuses on changes in
risk premiums as relative asset supplies change as argued, for example, by Caballero, Farhi and
Gourinchas (2017). Rachel and Summers (2019) question the importance of this line of thought
noting that required returns on assets aside from government debt like risky debt, real estate and
stocks have moved largely in tandem with debt yields. They note other difficulties with the safe
asset theory including the emergence of negative swap spreads and growing supplies of
government debt in recent years. Another view is that low rates reflect a shift in the risk profile
9
of government debt from providing a source of risk to acting as a hedge (Campbell, Sunderam,
and Viceira 2017), although rates have fallen as much at the short end (which has a beta of zero)
as at the long end so this explanations is unlikely to be an important part of the story.
Ultimately, however, the exact reason for low interest rates makes little difference for the
analysis of fiscal policy except to the degree the explanations bear on the persistence of the
trend. Whatever the cause of low interest rates, their relevant consequence is to limit the ability
of monetary policy to stimulate demand and reduce the cost of borrowing.
The decline in interest rates has three important implications: (i) as monetary policy is
limited in its ability to stabilize the economy and financial system, fiscal policy must play a
critical role; (ii) fiscal sustainability cannot be assessed by traditional debt-to-GDP ratios but
should instead be understood with measures like nominal or real interest as a share of GDP; and
(iii) many public investments pay for themselves, or come close to paying for themselves, and
the risk of not undertaking these investments is larger than the risk of doing too little deficit
reduction. The remainder of this paper discusses these three implications in turn.
Implication 1: Active Use of Fiscal Policy is Essential in Order to Maximize Employment
and Maintain Financial Stability in the Current Low Interest Rate World
Traditional thinking for the last half century has held that monetary policy should take
primary responsibility for stabilizing aggregate demand and ensuring low inflation. This is the
essential economic theory behind the widely implemented idea of independent central banks with
targets. Fiscal policy is seen as operating through automatic stabilizers and through discretionary
stimulus packages at moments of major distress. Fiscal policy is seen largely through the prism
of microeconomic efficiency, fairness and equality, and the desire to promote investment by
avoiding crowding out.
This thinking is no longer appropriate if monetary policy cannot be relied on to stabilize
the economy or to ensure that inflation targets are achieved. We believe that this will be the case
for as long as interest rates remain anywhere near current and prospective levels. Moreover, there
is a real risk that contractionary fiscal policies will jeopardize financial stability by forcing
decreases in interest rates that may encourage excessive leverage and asset price bubbles.
10
The challenges associated with low interest rates
The clearest challenge associated with lower equilibrium interest rates is constraints on
monetary policy. In 2000 David Reifschneider and John Williams published a paper predicting
that the federal funds rate would be at the zero lower bound 5 percent of the time. Since the
publication of the paper it has been at the zero lower bound 38 percent of the time. And it has
been at the bound 59 percent of the time since the onset of the financial crisis. As we write,
option markets suggest that five years out there is a 72 percent chance that nominal rates will be
at their current level of effectively zero or even negative. As shown in Figure 5, interest rates in
other major economies, most notably Japan, have also been at an effective lower bound (often
negative) for much of the time in recent decades.
Figure 5
Source: European Central Bank; Bank of Japan; Bank of England; Federal Reserve; Macrobond; authors’ calculations.
In the United States, the average policy interest rate reduction in the nine recessions
before the pandemic was 630 basis points. Even if the long-run nominal federal funds rate
reaches the FOMC’s expectation of 2.5 percent, something the market doubts, that would leave
less than half as much room to respond to future recessions as past recessions with only a little
additional room if rates are allowed to go negative. The European Central Bank and Bank of
Japan may have even less room to respond to future recessions, and in fact neither of them was
able to cut rates in response to the current recession because they were already at the effective
lower bound. While rate reductions can be augmented by measures like quantitative easing and
11
forward guidance or even yield curve control, the only effect of such measures is to reduce term
premiums. Given their current low level this is unlikely to add much to the efficacy of monetary
policy. How much investment would be done at a zero percent ten-year Treasury rate that would
not be done at a one percent ten-year Treasury rate?
The consequence, if not compensated for by more active fiscal policy, would be longer
and more severe recessions. Moreover, it is at least plausible that recessions have hysteresis
effects and reduce subsequent potential output through effects on both labor force scarring and
subsequent productivity (e.g., Adler et al. 2017; Oreopoulos, von Watcher, and Heisz 2012;
Yagan 2019; DeLong and Summers 1988).
The importance of setting fiscal policy on the basis of the need to maintain aggregate
demand is highlighted by a counterfactual calculation. While there was much controversy over
the content of the 2010 National Commission on Fiscal Responsibility and Reform’s (commonly
known as the Bowles-Simpson commission) recommendations for moving towards budget
balance, there was, at the time, little debate over the merits of their objective. The Bowles-
Simpson plan would overtime have represented about a 4 percent of GDP annual shift towards
austerity by the end of the decade. Given that for much of the period unemployment was above
its sustainable non-accelerating inflation rate of unemployment (NAIRU) level, this would have
adversely impacted aggregate demand. For 5 years during this decade the federal funds rate was
at its lower bound and at no point did it exceed 2.5 percent. It is therefore not remotely plausible
that a lower rate path could have offset more than a small fraction of the reduction in aggregate
demand the fiscal contraction would have produced. The result likely would have been even
more economic slack and inflation further below target.
This is of course a hypothetical calculation. Had a major recession ensued, fiscal policy
responses would surely have been implemented. The point is that with our current economic
environment, fiscal policies need to be set with a view to maintaining full employment.
A second, related, challenge is financial stability. Lower interest rates lead to a shift to
riskier assets, higher leverage, and the possibility of bubbles as investors reach for yield, and
reduce the capital of banks, which have been unable to reduce deposit rates as much as lending
rates (e.g. Borio, Gambacorta, and Hoffman 2015; Dell’Ariccia, Laeven and Suarez 2017). This
may be exacerbated by central bank asset purchases which have increasingly become a tool to
provide additional support for the economy at the effective lower bound (Bernanke 2020).
12
Finally, in addition to complicating countercyclical policy in recessions it is possible that
there is also a chronic lack of demand that makes it impossible for the economy to grow
normally even outside of recessions. The experience of the United States in 2018-19 is
instructive in this regard. These were the ninth and tenth years of an economic expansion with a
record number of months of consecutive job growth and a relatively low unemployment rate.
Nevertheless, the stance of macroeconomic policy in these two years was what one would
associate with a moderate to severe recession as shown in Figure 6. The fiscal stimulus was
larger than that of most recessions since the 1960s and interest rates were cut to levels that were
also more accommodative than in most previous recessions. This public support, not private
demand, helped the economy to grow at a 2.6 percent annual rate. Moreover, even with this
extraordinary monetary and fiscal stimulus the inflation rate still remained below the Federal
Reserve’s 2 percent target over this period.
Figure 6
Note: Increase in budget deficit is measured 8 quarters from peak; federal funds rate is minimum from peak through 24 months
after peak.
Source: Congressional Budget Office (2020c); Federal Reserve; Macrobond; authors’ calculations.
Lower equilibrium interest rates make it clear that economies will be in liquidity traps,
where the interest rate needed to equilibrate saving and investment is an unattainable negative
rate, much more often in future recessions. The experience of 2018-19 raises the prospect that
the liquidity trap may not just be confined to downturns but that secular stagnation may be a
13
chronic problem of the demand for loanable funds falling short of the supply of funds at any
nominal interest rate above zero even in normal times.
Central bank policy interest rates are at or near zero in all of the major advanced
economies leaving little if any additional scope for monetary policy to support additional
demand and speed the closing of output gaps.
Low interest rates mean that countries cannot afford not to undertake fiscal expansions
The main concerns about fiscal expansion in economic downturns is that they will lead to
unsustainable debt and may not be affordable in countries that currently have high debt levels.
This concern is misplaced. At a minimum, countries can always come back later to raise
revenues or reduce spending in order to get debt trajectories back on a desired course. More
importantly, this may not even be needed as fiscal support may help fiscal sustainability by
increasing output more than it raises debt, thus reducing the debt-to-GDP ratio.
3
A range of academic research and modelling by organizations such as the IMF and
Organisation for Economic Co-operation and Development (OECD) in recent years has found
that fiscal expansions in depressed economies can reduce the debt-to-GDP ratio. DeLong and
Summers (2012) found that even a small amount of hysteresis would result in a fiscal expansion
increasing GDP by more than it increases debt, resulting in a reduction in the debt-to-GDP ratio.
This view is supported by the empirical evidence in Auerbach and Gorodnichenko (2017), who
use a variety of specifications of fiscal shocks in a panel data set of advanced economies and find
that “a fiscal stimulus in a weak economy may help improve fiscal sustainability along the
metrics we study,” namely debt-to-GDP ratios and credit default swap spreads.
Most of the major macroeconomic models also have the property that fiscal expansions in
depressed economies at the zero lower bound can improve fiscal sustainability. The OECD
(2016) showed that public investment reduced the debt-to-GDP ratio in most of the advanced
economies in both its National Institute’s Global Econometric Model (NiGEM) and Fiscal
Maquette (FM) models. The IMF found similar results in modelling of demand-side policies
(Gaspar, Obstfeld and Sahay 2016). Ball, Delong and Summers (2017) report a run of the
Federal Reserve staff’s FRB-US model which shows that a 1 percentage point increase in fiscal
3
In the next section we will talk about the limitations of debt-to-GDP as a metric. We use that metric here because it
was the basis of the research we are citing.
14
stimulus reduces the debt-to-GDP ratio by 2.2 percentage points after 20 years, as shown in
Figure 7.
Figure 7
Source: Reifschneider and Summers as reported in Ball, DeLong, and Summers (2017)
Auerbach and Gorodnichenko find their result still holds at high debt levels, a finding
that is echoed in the OECD modelling which does not evidence any relationship between the
level of debt and whether fiscal policy pays for itself. In general, the higher a country’s debt the
more the growth rate-interest rate differential matters for its fiscal trajectory and the less that the
cost of stimulus itself matters for debt dynamics, as shown in the standard debt dynamics
equation:
≈
where r is the real interest rate, g is the real growth rate of GDP, and the primary deficit is the
difference between non-interest spending and revenues. To the degree that it can raise growth
rates relative to interest rates, then fiscal stimulus will have even larger favorable effects on debt
dynamics in a high debt economy than a low debt economy.
The case for a fiscal expansion in a depressed economy at the effective lower bound does
not rest on nor require that fiscal expansions reduce the debt-to-GDP ratio. Fiscal expansions
could easily pass a cost-benefit test even without these effects, but to the degree to which the
dynamic cost is smaller than the static cost, or even negative, the net benefit will be even larger.
15
The importance of automatic recession insurance
The increasingly limited effects of monetary policy call not just for discretionary
responses in situations like the current moment but also for improvements in the automatic
stabilizers, especially in countries like the United States that have weak automatic stabilizers.
Automatic stabilizers are spending increases or tax cuts that happen automatically when the
economy weakens. For example, when more people lose jobs labor tax payments go down and
unemployment insurance benefits go up. The magnitude of automatic stabilizers is generally
related to the size of government, as shown in Figure 8. Some European countries with larger
stabilizers have, however, undone them with discretionary fiscal contractions in past periods of
economic weakness.
Figure 8
Note: Dotted line is linear trend.
Source: International Monetary Fund (2015).
One way to make automatic stabilizers larger would be to increase both revenues and
expand benefits. An alternative approach is to build in specific macroeconomic contingencies
that trigger additional assistance when, for example, the unemployment rate rises above a certain
level (Boushey, Nunn, and Shambaugh 2019). This automatic recession insurance could include
national transfers to subnational units that have a harder time borrowing in downturns, increased
unemployment insurance benefits, or other transfers like nutritional assistance or even across-
the-board cash transfers.
16
There are four arguments for establishing more automatic contingent policies. First, they
can lessen downturns and speed economic recoveries by overcoming the political limitations of
discretionary fiscal policy which often gets slowed down by recognition lags, political debates,
and may end prematurely as the political system tires of change. Second, automatic contingent
policies can be more regionally differentiated, for example they can be based on state-level or
other subnational economic measures so as to provide the greatest assistance where it is most
needed, including in response to regional recessions. Third, they often make microeconomic
sense. For example, the optimal level of unemployment insurance depends on the unemployment
rate because when the unemployment rate rises moral hazard concerns about discouraging job
seeking diminish and the importance of consumption smoothing rises (Baily 1978 and Chetty
2008). Finally, they may advance additional priorities; for example, providing assistance to states
and localities can help prevent damaging cuts to education (Fiedler, Furman, and Powell 2019).
Aggregate demand can be further increased in a budget neutral manner
Given that interest rates are at essentially lower-bound levels around the world even in
the presence of substantial deficits, the case that maintaining full employment requires more than
monetary policy is compelling. The obvious concern is that expansionary fiscal policy may not
be sustainable if it leads to excessive debt accumulation. We have already noted that
expansionary fiscal policy may actually reduce levels of debt relative to GDP by stimulating
growth and increasing revenue collections. There is a further crucial point as well—it is possible
for fiscal policy to stimulate demand without increasing the deficit or the level of government
indebtedness.
First, fiscal policy can take advantage of the balanced budget multiplier whereby
spending has (over time) a higher multiplier than taxes (Haavelmo 1945) because increases in
spending increase demand dollar for dollar whereas, particularly in the case of very progressive
tax increases, taxes are paid out of funds that otherwise would have been saved. A reasonable
estimate is that the spending financed by taxes on high-income households is at least half as
potent in stimulating the economy as spending financed by borrowing
Second, fiscal policy can shift in a more progressive direction. One of the causes of lower
interest rates has been the increase in inequality which has resulted in larger incomes for higher-
income households who are the most likely to save it. Offsetting the increase in inequality would
17
reduce net national saving for any given interest rate, something that has often been viewed as a
minus for economic growth but in the current circumstances would be welcome (e.g., Koo and
Song 2016). Note that in general a larger government generally is a more progressive one so the
first and second recommendations are related.
Finally, public support for retirement, health care, college and other large, lumpy and
sometimes uncertain needs reduce the need for lifecycle and precautionary saving, thus boosting
consumption demand. When Keynes visited the United States during World War II, he
highlighted maintaining demand as an important virtue of the then recently adopted Social
Security system. In all cases the expanded public support could be paid for on a pay-as-you-go
basis so that they need not change the short- or long-run deficit.
Implication 2: Lower Interest Rates Necessitate New Measures of a Country’s Fiscal
Situation
The debt-to-GDP ratio is the most common measure of a country’s fiscal situation used
by policy makers and is enshrined in rules that govern fiscal policy. Recently some have been
alarmed as the debt-to-GDP ratio has risen across the world and now stands at more than 100
percent in the majority of G7 economies, as shown in Table 2.
Table 2
Note: In international comparisons, U.S. values are for general government, not federal government.
Source: International Monetary Fund; Macrobond.
Canada 46
France 110
Germany 54
Italy 149
Japan 177
United Kingdom 98
United States 107
Memo: G7 Countries 110
General Government Net Debt as
a Percentage of GDP, 2020
18
Debt levels have been a leading metric for public policy but the decline in interest rates
shows how problematic this measure is. For example, in 1992 the Maastricht Treaty set a limit of
60 percent debt-to-GDP ratio for countries in the euro zone. At the time, ten-year German bonds
had a nominal interest rate of 7.8 percent or a real interest rate of about 5 percent. In 2019 the
nominal interest rate on ten-year German bonds had fallen to -0.2 percent or a real interest rate of
about -2 percent. At interest rates prevailing in 1992, a country with a 60 percent debt-to-GDP
ratio paid about 5 percent of GDP in interest. Today, Japan with a 177 percent debt-to-GDP ratio
is expected to pay 0.2 percent in interest and the United States with a 107 percent debt-to-GDP
ratio for general government is expected to pay 2.0 percent of GDP in interest, with the real
interest after accounting for inflation being negative or close to zero in both countries. If a 60
percent debt-to-GDP ratio made sense as a ceiling in 1992 it definitely no longer does. Instead a
much higher ceiling would be appropriate today. This is a vivid illustration of how the debt-to-
GDP metric is flawed and why it should be replaced with other measures of a country’s fiscal
position. Importantly, these other measures generally show that fiscal positions are better today
than they were a few decades ago as the favorable debt sustainability dynamic associated with
lower interest rates outweighs the increase in the debt itself.
Shifting from the debt-to-GDP metric is part of a broader reappraisal of the ways in
which the current debt and deficit situation is less of a concern in an economy in which growth
rates have often exceeded interest rates in the past and are likely to continue to exceed them for
sometime in the future (see, e.g., Elmendorf and Sheiner 2017, Blanchard 2019, Furman 2016,
DeLong and Summers 2012).
The debt-to-GDP metric has three shortcomings that make it a misleading metric of a
country’s fiscal position:
1. It ignores the fact that debt can be repaid over time. Debt is a stock (an amount estimated
at a point in time that does not need to be repaid immediately) while GDP is a flow
(measured over a discrete period of time). A stock-stock perspective on the debt
compares the debt to the present value of GDP over the indefinite future. With lower
interest rates, the present value of future GDP is higher and debt is correspondingly more
manageable over time.
19
2. It ignores interest rates. Relatedly, at lower real interest rates a given amount of debt is
less costly. The stock-stock perspective is based on a very speculative measure of the
present value of GDP over an infinite horizon, a metric that requires knowing growth and
interest rates for the indefinite future. In contrast a flow-flow perspective looks at how
much that debt costs today, in interest payments, as compared to income today, providing
a sense of how affordable that debt is.
3. It is a backward-looking concept. The public debt is effectively the sum of the unified
deficits that a country has run from its inception (with some adjustments for financial
transactions). It does not reflect scheduled future policies, like pensions, or likely future
policies, like the cost of responding to future emergencies. It also does not incorporate the
ability to respond to evolving debt concerns with future tax increases or spending cuts.
The next three subsections address these considerations in more detail including showing how
more coherent measures of the fiscal situation dramatically changes our understanding of fiscal
sustainability, making it clear that most major advanced economies are in better fiscal shape
today than they were twenty or thirty years ago when their debt levels were much lower.
The debt is less of a concern: a stock-stock perspective
While interest rates have come down dramatically, growth rates have fallen much less.
As a result, the difference between growth rates and interest rates has risen in all of the G7
economies over the last three decades and is a substantial positive number in all of them ranging
from 1.1 in Italy to 3.7 in Germany as shown in Figure 9. Growth rates exceeding interest rates is
the norm not the exception, occurring about two thirds of the time since 1871 in the United
States and for much of the time in other major economies as well (Blanchard 2019).
20
Figure 9
Note: GDP growth is average annual growth rate over prior 5 years. Interest rates are average of 10-year and 3-month
government bond yields.
Source: World Bank; Macrobond; Global Financial Database; authors’ calculations.
When growth rates exceed interest rates the present value of GDP is infinite. This means
that over time the economy will outgrow its debt and associated interest so that the debt will
disappear relative to the economy. From a stock-stock perspective, the debt stock is 0 percent of
the present value of GDP stock. This is a more favorable fiscal situation than most of the G7
economies appeared to enjoy in the early 1990s when interest rates were substantially above
growth rates so that the present value of GDP was finite and debt was a positive fraction of the
present value of GDP.
Interest rates could rise and growth rates could fall in which case the present value of
GDP could be finite. In the United States the only regular and systematic measure of GDP over
an infinite horizon we are aware of is produced by the Social Security Trustees. Any estimate of
this quantity is highly speculative and subject to massive uncertainty, and we have our quibbles
with some of their specific assumptions, but this measure provides a useful benchmark with
which to assess how the U.S. debt appeared at various points in time.
According to the latest Social Security Trustees Report (2020) the present value of GDP
over an infinite future was $3.8 quadrillion on January 1, 2020. As of November 19, 2020, the
U.S. federal debt held by the public was $21.2 trillion. Adjusting these to have the same dates,
debt is 0.5 percent of infinite horizon GDP. In other words, a 0.5 percentage point increase in
revenue as a share of GDP or reduction in spending as a share of GDP would be sufficient to pay
21
off the entire debt. This is smaller than or similar to what policymakers would have thought at
various times in the past as shown in Figure 10 even as the debt has nearly tripled as a share of
current GDP.
Figure 10
Note: 2021 value is based on debt as of November 19, 2020.
Source: The Board of Trustees, Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds; Federal Reserve
Bank of Philadelphia, Survey of Professional Forecasters; Department of the Treasury; Macrobond; authors’ calculations.
This intuition applies more broadly. For example, a decline in r – g from 1.0 to 0.5
doubles the present value of GDP—and so makes twice as much debt sustainable relative to the
future stock of GDP. This ½ percentage point decline in r – g is an order of magnitude smaller
than the actual 5 percentage point reduction in r – g in the median G7 economy since 1990.
The debt is less of a concern: a flow-flow perspective
The stock-stock perspective relies on a highly uncertain extrapolation of fiscal conditions
into the indefinite future. An alternative coherent metric avoids that problem by comparing the
flow of interest on the debt to the flow of annual GDP. The two measures are, of course, related
as lower interest rates raise the present value GDP and make it more possible to pay the debt off
over time.
The more analytically relevant measure is real interest payments as a share of GDP. Real
interest rates adjust for inflation by comparing the real interest rate paid on the debt to the size of
GDP. Equivalently, it can be understood as nominal interest payments as a share of GDP minus
22
the amount that the debt is inflated away each year as a share of GDP. Specifically, the formula
is:
.
In implementing this concept we smooth inflation by averaging it over five years, which comes
closer to the concept of expected inflation which is relevant for real interest rates. As one would
expect, real interest as a share of GDP falls when, all else equal, nominal interest rates fall,
inflation rises, nominal GDP rises, or when the debt rises. This last effect is because the larger
the debt-to-GDP ratio the more an economy benefits from the inflation that partly erodes the
debt.
Real interest payments as a share of GDP are more analytically relevant than the more
commonly used nominal interest payments as a share of GDP. For example, consider two
economies that both have debt-to-GDP ratios of 100 percent and nominal interest rates of 4
percent. Nominal interest costs as a share of GDP are thus 4 percent in each economy. Assume
now that the first economy has no inflation and the second economy has 4 percent inflation. For
simplicity assume there is no real growth in either economy. The no inflation economy has a real
interest to GDP ratio of 4 percent, the amount it needs to raise revenues or reduce non-interest
spending in order to stabilize the debt-to-GDP ratio. In contrast, the second economy has a real
interest to GDP ratio of 0 percent so it need not raise taxes or cut non-interest spending to offset
the cost of its debt. That is because the second economy is able to deflate away the debt by an
amount equal to its nominal interest payments.
Both nominal interest payments and real interest payments have fallen as a share of GDP
across the G7 economies and are generally lower today than they have been in decades as shown
in Figures 11a and 11b.
4
4
These figures show general government net interest payments, which in the United States includes payments by
federal, state, and local governments.
23
Figure 11a Figure 11b
Note: General government, including for United States.
Source: International Monetary Fund, Macrobond; authors’ calculations.
The U.S. experience over the last twenty years provides a vivid example of how different
the analytically meaningful flow-flow perspective is from the misleading stock-flow perspective
conveyed by debt-to-GDP ratios. In 2000 U.S. Federal debt was 34 percent of GDP, not far from
its post-war low. Since then the debt-to-GDP ratio rose almost monotonically, nearly tripling to
over 100 percent of GDP. At the same time both nominal and real interest rates have fallen so
that both nominal and real interest as a share of GDP have fallen nearly monotonically and are
now towards the low end of the range as shown in Figure 12a and 12b.
Figure 12a Figure 12b
Note: 2021 values are projections.
Source: Office of Management and Budget; Congressional Budget Office; Macrobond; authors’ calculations.
24
BOX I – Adjusting debt and interest payments to reflect the Federal government’s
complete balance sheet
The federal U.S. debt and net interest data used in this paper has followed U.S.
scorekeeping conventions which are at variance with the economically relevant concepts. In
addition to liabilities the Federal government also has financial assets, the largest of which is
direct student loans and the second largest of which is cash at the Treasury. The relevant concept
of debt for both fiscal sustainability and assessing macroeconomic effects is the debt held by the
public net of financial assets. As the CBO (2020d) explains, “Debt net of financial assets also
provides a more comprehensive picture of the government’s overall effect on credit markets than
does debt held by the public. When the government borrows to make loans that will be repaid in
the future, the overall supply of credit is essentially unchanged. Therefore, the issuance of that
debt does not crowd out, or take the place of, debt issued in the private sector to the same degree
that debt issued for other purposes does.” For example, when the Federal government shifted
from guaranteeing private student loans to making direct loans itself its financial position and
risks were essentially unchanged but the debt held by the public rose. Figure I.1 shows the
divergence between debt held by the public and debt net of financial assets which has grown
over time and is now about 9 percentage points.
Figure I.1
Source: Office of Management and Budget; Richard Kogan’s calculations.
The Federal budget defines “net interest” largely as the interest paid on Treasury bonds
with adjustments for other interest paid and received by other Federal agencies (for example, the
equity earnings of the National Railroad Retirement Investment trust partly offset net interest).
The data do not count the Federal Reserve as part of the Federal government even though it is
clearly a Federal agency and the Treasury’s and Federal Reserve’s balance sheets should be
thought of on a consolidated basis for thinking about fiscal sustainability and the macroeconomy.
Put another way, fiscal analysis should essentially not count the Treasury debt held by the
Federal Reserve but should add the Federal Reserve’s reserves because these are effectively
interest-bearing short-term debt. In 2019 the Federal Reserve earned interest of $103 billion
largely on its Treasury and mortgage securities while paying $41 billion in interest mostly on
25
reserves. This $62 billion interest spread reflected the higher interest rates it received on its
longer-term assets than it paid on its shorter-term debt and $55 billion of this spread was remitted
to the Treasury. Thus, the Federal government’s consolidated net interest should subtract out
remittances to the Federal Reserve which are currently inaccurately classified as a receipt (or
revenue item) not as net interest. Figure I.2 shows the gap between net interest and net interest
minus Federal Reserve remittances over the recent past. The gap between these two is likely to
grow substantially over the next several years as the Federal Reserve has expanded its balance
sheet but the latest CBO projections expect it to come back down to 0.2 percent of GDP in 2030
(CBO 2020a).
Figure I.2
Source: Office of Management and Budget; authors’ calculations.
END BOX-------------------------------------------------------------------------------------------------------
Looking forward: the debt is not “spiraling” over the next decade and interest payments are
modest
Debt relative to the present value of GDP and real interest payments relative to GDP are
both coherent and meaningful measures and both of them are superior to the largely incoherent
concept of debt relative to GDP. All three of these measures, however, suffer from the same
shortcoming: they do not reflect the future fiscal trajectory which may dwarf the cumulative
historical trajectory.
Looking forward it is plausible that interest rates will rise from their current
extraordinarily low levels, with the CBO forecasting a steeper increase in interest rates than
26
financial markets are expecting (Figure 13). CBO also projects spending on Social Security and
health programs to rise, with revenues as a share of GDP expected to rise as well.
Figure 13
Note: Market-implied rates as of November 27, 2020. CBO based on Q4 forecasts.
Source: Congressional Budget Office (2020b); Bloomberg.
For the United States, the latest estimates show that the debt will rise but then level off at
a higher level (Figure 14a). Real interest payments will be low as a share of GDP but then
increase towards the end of the ten-year budget window (Figure 14b), but will still end up well
below their historical average. The estimates shown are for three scenarios: one based on the
latest CBO baseline which largely assumes current law continues, one that adds in the cost of
making the expiring provisions of the 2017 tax cuts permanent, and one that follows current law
in assuming Social Security is reformed (in the spirit of Blahous 2017 on the proper baseline).
5
5
Under current law Social Security and Medicare cannot pay full benefits after their trust funds are exhausted,
which are projected to be 2034 and 2026 respectively according to the Trustees (The Board of Trustees, Federal
Old-Age and Survivors Insurance and Federal Disability Insurance Trust Fund 2020; The Boards of Trustees,
Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds 2020) and 2031 and 2024
respectively according to CBO (2020f). Current law essentially requires Social Security reform to happen, as has
happened in the past when Social Security ran up against trust fund exhaustion, most recently in 1983. Instead of
abruptly cutting benefits when the trust fund is exhausted our current law baseline smooths the adjustment out over
time starting at 0.5 percent of GDP in 2025 and growing to 1.7 percent of GDP on and after 2035. This is equal to
the present value of the Social Security shortfall as estimated by the Social Security trustees. It is a conservative
reflection of current law because it does not include Medicare and because it is smaller than the adjustment CBO
projects under current law. Doing everything consistently on a CBO baseline would thus show that current law
would result in an even lower debt and interest rate path than we are showing here and throughout this paper.
27
Figures 14c and 14d show the same three scenarios over the next thirty years although, as
discussed in the next section, the uncertainty around forecasts that go out this far is so large they
should not have inordinate weight in policymaking today. See the Appendix for details on these
projections.
Figure 14a Figure 14b
Figure 14c Figure 14d
Social Security reform phased in linearly from 0.5% of GDP to 1.7% of GDP over 10 years beginning in 2025
Source: Office of Management and Budget; Congressional Budget Office; Macrobond; authors’ calculations.
These baseline predictions can serve a useful purpose in assessing the fiscal trajectory but
they also are limited in their ability to answer the question of whether the debt is sustainable by
their assumption of no change in law or policy. This is because they do not answer the question
of whether the fiscal situation could be made sustainable in the future. The United States collects
31 percent of GDP in general revenue, well below the OECD average of 37 percent of GDP or
the OECD maximum of 57 percent of GDP collected in Norway. The United States has often
28
collected revenue that was higher than what it collects today and similar revenue levels have
been proposed by, for example, the National Commission on Fiscal Responsibility (2010). In that
sense the U.S. fiscal situation is by definition sustainable—effectively it has an asset equal to at
least several percent of GDP that it could choose to collect if it needs to. Higher tax countries
have less room in this regard. France, for example, is closer to the top of its Laffer curve so
would have less space to close a fiscal hole with more revenue. Considerations of debt
sustainability need to take into account not just the amount of taxation under the law but the
capacity for taxation.
A bigger issue with forward-looking projections is the tremendous uncertainty they are
subject to, the topic of the next subsection.
The uncertainty in budget forecasts is enormous especially looking forward several decades
Forecasting deficits and debt is extremely difficult and forecasts have very large standard
errors that derive both from unexpected changes in economic variables like growth and interest
rates and also in “technical” factors like the cost of health care and the efficiency of tax
collections. The CBO does a very good job given all of the uncertainties. Its forecasts of deficits
and debt produced since 1984 have been mostly unbiased in the sense that their forecast errors
tend to even out without about half of their forecasts being too optimistic and about half of them
being too pessimistic after adjusting for legislative changes that they were not trying to forecast
(CBO 2019a). In contrast, CBO’s forecasts of net interest as a share of GDP have been
systematically biased toward being too high, in large part because CBO’s forecasts of interest
rates (like those of other forecasters) have consistently been too high, as shown in Figure 15.
29
Figure 15
Source: Congressional Budget Office.
CBO’s forecast errors have, however, been large in both directions and grow over time.
Table 3 shows their own estimates of a two-thirds error band around their forecasts for deficits
and debt along with our estimates, derived from their statistics, of a 90 percent error band.
Table 3
Source: Congressional Budget Office (2019a); authors’ calculations.
CBO does not assess its track record over a longer period, in part because of limitations
in the data that would be needed for such an assessment. Moreover, we have very few very long-
range fiscal forecasts but the ones we have are consistent with the view that the errors grow over
time. One of the earliest long-range forecasts was made by the Government Accountability
Two-thirds
Spread of
Errors
90 Percent
Confidence
Interval
Two-thirds
Spread of
Errors
90 Percent
Confidence
Interval
Year 1 (Current Year) 1.0 1.8 1.5 2.8
Year 2 (Budget Year) 2.1 4.3 3.3 7.8
Year 3 3.0 5.7 6.5 13.4
Year 4 4.0 6.6 8.5 18.2
Year 5 4.1 7.5 12.3 23.4
Year 6 3.8 8.3 17.2 29.2
Deficit
Debt
Measures of Historical Forecast Accuracy as a Percent of GDP
30
Office (GAO) in 1992 and it projected that the debt would reach 177 percent of GDP by the end
of fiscal year 2020, which is 77 percentage points above the actual debt-to-GDP ratio at the end
of that year as shown in Figure 16. Moreover, legislation adopted since 1992 likely was net debt
increasing so GAO’s economic and technical forecasting error was even larger than this estimate.
Figure 16
Note: GAO estimate converted from GNP to GDP and adjusted for revisions to GDP methodology since publication.
Source: Government Accountability Office (1992); Office of Management and Budget; Bureau of Economic Analysis;
Macrobond; authors’ calculations.
As a back-of-the-envelope way of assessing forecast errors we assume conservatively
that the standard deviation of debt forecast errors grows linearly with time, reflecting the
increased uncertainty as deficit forecast errors accumulate over time. As a result, the 30-year
forecast of the debt-to-GDP ratio would have error bands as shown in Figure 17.
6
In other words,
assuming the tax cuts expire and Social Security is reformed we would expect the debt-to-GDP
ratio to be 112 percent of GDP in 2050 with a two-thirds chance that it falls in the 66 to 157
percent range and a 90 percent chance it falls in the 33 to 190 percent range.
6
The 90 percent uncertainty band is the result of a Monte Carlo simulation of 10,000 draws. The simulation draws
from a distribution that assumes that there is no debt forecast error on average and that the standard deviation of the
debt forecast error for each year increases over time based on the linear relationship observed in CBO’s forecast
errors for years one through year six.
31
Figure 17
Note: Social Security reform phased in linearly from 0.5% of GDP to 1.7% of GDP over 10 years beginning in 2025
Source: Office of Management and Budget; Congressional Budget Office; Macrobond; authors’ calculations.
The impact of uncertainty on current policymaking depends on two factors. First, how
costly and irreversible are steps we take today? If it is costly to take steps today, for example
irreversibly reducing the educational opportunities of children, then in the face of uncertainty it
is better to delay action until more of the uncertainty has been resolved (Dixit and Pindyck
1994). Second, how costly is waiting to act? To the degree that one wants to act on the tax side
there is a cost associated with waiting in that it involves a larger tax increase and thus does not
efficiently smooth tax rates (Barro 1979), although in practice that cost is likely to be relatively
small, especially for countries that are far from the peak of their Laffer curves. To the degree one
wants to act on the spending side waiting may be more constraining as it is harder to give notice
that would allow people to adjust their plans, but even here there is little adjustment that people
understood or adjusted much in response to the decades of notice they got about the increase in
the Normal Age of Retirement for Social Security.
Overall, the conclusion we draw from the uncertainty on forecasts is that policymakers
should put relatively little weight on projections for ten years or more in the future and that large
changes should not be made well in advance based on highly uncertain and possibly inaccurate
forecasts, especially when it is very feasible to make adjustments later.
32
BOX II: The fiscal gap as an alternative measure of the fiscal situation-------------------------
The fiscal gap is an alternative measure of the fiscal situation that aims to address the
three shortcomings we have discussed: the stock-flow disconnect, falling interest rates and the
need to look forward. It was originally developed by Laurence Kotlikoff and Alan Auerbach and
is regularly updated by government institutions (e.g., CBO 2019b) and academics (e.g.,
Auerbach, Gale, and Krupkin 2019). The concept is the immediate and permanent change in the
primary balance—either through an immediate increase in taxes or an immediate reduction in
non-interest spending—that would be needed to stabilize the debt as a percentage of GDP at its
current value for a specific period of time. As such, the fiscal gap incorporates information not
just about the past debt but also about future primary deficits, GDP, and interest rates.
The fiscal gap is a more meaningful and useful concept than the debt-to-GDP ratio. It
does, however, suffer from three problems. First, unlike measures like nominal or real interest as
a share of GDP it does not provide an objective measure that can be used across time, across
countries, or even measured at a point of time because it depends on projections about the
uncertain future.
Second, and relatedly, fiscal gaps have extremely large error bands because they depend
on deficit and debt forecasts that have extremely large error bands. To put this in perspective, if
the debt follows the mid-course trajectory shown in Figure 17 then the fiscal gap would be 0.5
but at the upper and lower 90 percent confidence intervals the fiscal gap could be anywhere from
-2.0 to +3.2 percentage points as shown in Table II.1. Projecting further in the future results in
even larger errors, with the large majority of infinite horizon fiscal gaps driven by projections
that are decades or more in the future.
Third, the fiscal gap measures the immediate adjustment that would be needed to stabilize
the debt-to-GDP ratio at its current value. Its current value, however, is arbitrary and
uninformative about what the goal of policymakers should be. The primary deficit adjustment
needed to achieve different fiscal targets varies enormously as shown in Table II.1 which shows
the immediate primary balance adjustment needed to achieve different debt-to-GDP goals in
2050. Does the United States need to make an immediate fiscal adjustment of 2.1 percent of
GDP to get the debt down to 50 percent of GDP or would it be reasonable for the debt to rise to
112 percent of GDP through 2050, in which case no adjustment, beyond social security reform,
would be needed? Note that all of the estimates in Table II.1 assume the equivalent of current
law on Social Security (i.e., Social Security reform happens). If a law is passed to continue
paying full benefits after the trust fund is exhausted that would add about 1.3 percent of GDP to
these fiscal gap measures.
33
Table II.1
Note: No change includes Social Security reform phased in linearly from 0.5% of GDP to 1.7% of GDP over 10
years beginning in 2025. Relative to the CBO baseline which assumes current policy for Social Security the fiscal
gaps would be about 1.3 percentage point larger.
Source: Congressional Budget Office; Macrobond; authors’ calculations.
The forward-looking nature of the fiscal gap makes it more informative about fiscal
sustainability but requires forecasts and subjectivity. The tradeoff between an objective but
uninformative measure and a subjective but informative one is largely unavoidable. The fiscal
gap is one metric that policymakers should use but it is important to put it in a broader context
and to that end we think that debt service ratios projected out over a period of about a decade are
a better way to minimize uncertainty and put context on the best fiscal targets.
END BOX-------------------------------------------------------------------------------------------------------
Implication 3: The Scope and Need for Public Investment Has Greatly Expanded
One of the principal arguments made for fiscal rectitude is a desire to avoid burdening
future generations with debt or to make them poorer as debt crowds out private investment. As
Blanchard (2019) recognizes the low level of real interest rates undercuts the force of the
crowding out argument because it implies the “certainty equivalent” productivity of capital is
low. More broadly the argument itself rests on an ethically ambiguous foundation because future
generations are likely to be richer than the current generation even net of any additional fiscal
obligations inherited by them. The utilitarian logic for redistribution from richer to poorer should
also apply across generations.
Even accepting the premise of intergenerational equity it is not obvious that deficit
reduction will make future generations richer and could even leave them poorer. The narrowest
example of this is deferred maintenance which is a cost that is akin to the debt even if it is not
Debt in 2050
Immediate
Primary Deficit
Reduction
90% Confidence Interval
for Debt with Associated
Deficit Reduction
Primary Deficit Reduction
Associated with 90% Confidence
Interval for Debt Target
50 2.1 -29 to 129 -0.5 to 5.1
97 (Fiscal Gap) 0.5 19 to 176 -2.0 to 3.2
112 (No Change) 0.0 33 to 190 -2.4 to 2.7
150 -1.2 71 to 229 -3.4 to 1.3
Immediate Primary Deficit Reduction Necessary to Achieve Selected Debt Targets in 2050
(Percent of GDP)
34
explicitly included or measured in the Federal balance sheet. Investments in infrastructure and
other areas that reduce deferred maintenance can reduce this (unaccounted for liability) and
replace it with a smaller but accounted for liability. The result is measured debt goes up even if
the meaningful liabilities of the Federal government go down. Put another way, it is better to fill
potholes today than to wait and fill them at a cost that grows faster than the interest rate, which is
currently around zero in real terms.
From a demand-side perspective in certain circumstances fiscal expansions can offset
some, all or even more than all of their cost, as discussed above. This happens if they expand
output more than they increase the debt by increasing the utilization of the economy’s potential.
From a supply-side perspective, public investment can also offset some, all or even more
than all of its cost if it has a sufficiently high rate of return in expanding the economy’s potential
itself. More important for a broader set of policies, public investments that have a rate of return
in excess of the interest rate can repay themselves in present value terms. Dickens, Sawhill, and
Tebbs (2006) provided a long-term analysis of investments in early education and found that
they more than pay for themselves over a 75-year horizon. Recently in an important paper
Nathaniel Hendren and Ben Sprung-Keyser (2020) synthesized high-quality research on 133
policy changes and found that numerous policy changes partly or even more than fully paid for
themselves in present value by raising future wages and reducing future government transfers.
As an example of their analysis consider the Perry preschool program:
“For example, “Perry Pre-School [cost] $17,759 in 2006 USD. However, we
estimate that the long-run reductions in transfer payments and increases in tax
revenue offset roughly 92% of these upfront costs. Heckman et al. (2010) estimate
significant earnings increases from ages 19-40, and an increase in earnings of
26% at age 40. We combine their estimated earnings effects with a forecast to age
65 this into a lifetime earnings impact of $70,535. Using a state and Federal
combined tax rate of 12.9% , this implies an increase in tax revenue of $9,607.
Heckman et al. (2010) also estimate that the policy led to a reduction of payments
on welfare programs of $3,941. In addition, there are also induced costs of college
attendance and vocational training whose incidence falls on the government.
Heckman et al. (2010)'s estimates actually imply a fall in such costs, saving the
35
government $2,805. This suggests $16,353 is repaid to the government, implying
a net cost of $1,406 (95% CI of [-9,235, 12,126]). Roughly 92% of the upfront
spending is repaid to the government.”
Many of the programs they examined more than repaid their initial costs including “four major
health insurance expansions to children over the past 50 years. We calculate an average across
those policies and find that for each $1 of initial expenditure they repaid $1.78 back to the
government in the long run.” They found similar results for numerous children’s education
programs like Head Start (in one of three estimates) and more K-12 spending, some college
programs like grants for tuition, and moving to opportunity housing vouchers. Even where they
did not find programs paid for themselves they found a substantial offset in costs.
Hendren and Sprung-Keyser used a 3 percent real discount rate for their analysis, well in
excess of the current or likely future cost of Federal borrowing. With a more realistic discount
rate many more programs, especially investments in education and children, would repay
themselves over time. Note even if an investment does not repay it may still be worth making.
Similarly, investments in infrastructure that have a rate of return higher than the cost of
government debt are worth making and with a sufficient rate of return they will repay themselves
as well. For example, research from the IMF and OECD finds that increased public investment
leads to increased economic growth, which, particularly during periods of economic slack, can
lower the debt-to-GDP ratio (Abiad, Furceri, and Topalova 2015 and Mourougane, et al. 2016).
Research also finds that there are substantial spillovers from investment in research and
development, particularly basic research, and implies current levels of investment are below their
socially optimal level (Bloom, Schankerman, and Van Reenan 2013; Akcigit, Hanley, and
Serrano-Velarde 2020).
The above points depend heavily on what the additional debt is used for. If it is used to
fund effective public programs with high rates of return, like research, infrastructure, education
and investments and support for children, it is very likely to have benefits far greater than the
costs of any additional debt accumulation. Wasteful and poorly designed spending programs or
tax cuts, however, are not justified by this logic.
Also, in the case of infrastructure, even if the investment pays for itself or offsets much of
its cost it still may be desirable to pay for it if the payfor itself has a policy rationale, like a gas
36
tax or vehicle miles travelled fee that addresses other externalities and helps ensure that existing
infrastructure is used better. Nevertheless, if these first best policies are not possible for political
reasons it is still worth doing the second best of unpaid for infrastructure investments.
Overall, it is impossible to be sure exactly what the right balance is but given the very
low interest rates currently and in the foreseeable future it is more likely to be a mistake to
excessively reduce the debt at the expense of more deferred maintenance and foregone
investments than it is to make the opposite mistake and overinvest.
Going Forward: New Objectives, New Guideposts, and New Guidelines
Currently the primary worry for policy in the United States and several other countries is
doing too little to expand the debt, not doing too much. Low interest rates create more scope and
need for expansionary fiscal policy, a reason to reassess views of debt sustainability, and more
reason to undertake public investments. Overall U.S. debt service obligations are currently
modest and the debt is modest relative to future GDP and the ability to generate taxes from this
GDP. Even the more conventional and misleading measure of the debt-to-GDP ratio is stable
over the next decade and assuming current law is complied with, which requires both the tax cuts
to expire and Social Security reform, it will be essentially stable over the next three decades as
well, although could plausibly be anywhere from among the lowest in postwar history to around
190 percent of GDP. Additional investments of about 1 percent of GDP that initially raised the
debt above this path could potentially pay for themselves and to the degree they do not would
still leave interest as a share of GDP below its historic levels.
Understanding how to respond to our challenges requires setting new objectives for fiscal
policy, adopting new guideposts to assess fiscal sustainability, and using new guidelines to set
fiscal policy.
Any fiscal policy approach should be a combination of optimal, understandable and
achievable. Some approaches, like the German balanced budget requirement, are understandable
but would not be achievable for some countries and are very far from optimal for many
countries, including Germany. Even if we knew the optimal approach—and we do not—it would
be of little use if it was not readily understood by the public and possible for policymakers to
37
follow. What follows is something we think is a reasonable combination that is not too far from
what policymakers could understand and implement in a manner that would foster stronger
economic growth, better responses to recessions and greater financial stability.
The objective of fiscal policy
The objective of fiscal policy should be growth and financial stability including the
avoidance of recessions and stronger long-term growth. This calls for more expansionary fiscal
policy both in the short run to combat the current recessionary conditions and over the medium-
and long-run to support demand and expand supply, including through some measures that
increase short-run deficits and debt and also other measures like balanced budget multipliers,
redistribution and expanded social insurance that can expand demand without increasing short-
run deficits and debt.
A new guidepost for fiscal policy: keeping real interest payments below 2 percent of GDP
The room for fiscal expansion is not, however, unlimited and policymakers need a
guidepost to assess fiscal sustainability, especially when the objective for fiscal policy can be
consistent with many different debt levels and no single prescribed fiscal trajectory.
7
When the growth rate is greater than the interest rate there is substantially more room to
run primary deficits and any given primary deficit will not lead to an unlimited explosion of debt
but instead will lead the debt to asymptote to a finite value. The finite value the debt asymptotes
to could, however, be very large—and large enough that the upward pressure it exerts on interest
rates makes it unlikely that growth rates will remain above interest rates in which case the
resulting dynamic would be an explosion of the debt and interest payments.
We propose the following approach: policymakers need not worry about the fiscal
outlook as long as the debt service-to-GDP ratio, measured in real terms, is expected to stay
comfortably inside of historical experience over the next decade and does not spiral upwards
7
Modern Monetary Theory (MMT) has suggested that high or rising inflation is an indicator that fiscal policy
should tighten (Tymoigne and Wray 2013). This raises the question of why not use monetary policy and interest
rates to curb inflation? And should policymakers really have engaged in fiscal contractions in the periods like the
early 1980s and even the early 2000s when inflation was above target and rising? Moreover, although debt defaults
in countries that borrow in their own currency and control their monetary policy are essentially impossible, what are
the limits of fiscal dominance by the monetary authority and the potential cost of not being able to roll over debt at
an acceptable interest rate?
38
over that period. This could be roughly operationalized as limiting real interest payments to
comfortably below about 2 percent of GDP ideally measured in the economically meaningful
sense of net interest less remittances from the Federal Reserve and interest on Federal financial
assets (see Box I for a discussion). To the degree that there is room to increase revenue a country
can have more latitude and to the degree that there are major adverse signals from bond markets
that are crowding out investment it has less latitude. Were it to become the case that interest
payments were projected to exceed our rough ceiling then it would be appropriate to pursue
policies directed at reducing the debt-to-GDP ratio. It is a feature not a bug that our approach
does not provide a prescription for the appropriate level of debt to GDP itself.
The levels of debt and primary budget balances associated with different real interest rate
goals depend on the real interest rate and the growth rate. When real interest rates are low there
is substantially more room to have elevated debt levels without triggering excessively high real
interest as a share of GDP. The higher the growth rate the larger the primary deficit could be
consistent with this interest and debt path. Different required steady-state primary deficits and
resulting debt levels are shown for alternative real debt service goals and assumed interest rates
and growth rates in Table 4.
8
If real interest rates stay below 1.33 percent—which is currently
well above what is expected—then a debt level of 150 percent of GDP would be comfortably
sustainable according to our criteria. If real interest rates, implausibly, were still 0.5 percent as
the debt-to-GDP ratio rose to 400 percent that would indicate that the extremely low interest
rates were such a powerful force that that degree of debt would be warranted as shown in Table
4. Conversely if real interest rates rose to 2 percent then getting the debt-to-GDP ratio should be
lower as all of the problems associated with low interest rates would be diminished and thus a
debt-to-GDP ratio of 100 percent would be more reasonable.
8
Debt dynamics can be analyzed using the identity discussed earlier:
≈
. In steady-state with a stable debt-to-DGP ratio this becomes:
≈
39
Table 4
Source: Authors’ calculations.
New fiscal policy guidelines for the United States
Operationalizing a fiscal plan that advances the broad objectives we have set out for
fiscal policy while staying withing the sustainability guideposts we propose depends on the
outlook for the fiscal situation. In the United States, overall U.S. debt service obligations are
currently modest and the debt is modest relative to future GDP and the ability to generate taxes
from this GDP. Even the more conventional and misleading measure of the debt-to-GDP ratio is
stable over the next decade and assuming current law is complied with, which requires both the
tax cuts to expire and Social Security reform, it will be essentially stable over the next three
decades as well, although could plausibly be anywhere from among the lowest in postwar history
to around 190 percent of GDP. Additional investments of about 1 percent of GDP that initially
raised the debt above this path could potentially pay for themselves and to the degree they do not
would still leave interest as a share of GDP below its historic levels.
Given the current outlook, a set of three broad guidelines would help move fiscal policy
towards a better stance for achieving our objectives while also offering a limiting principle for
pinning down this policy. Our starting point is current law, which includes the expiration of the
2017 tax cuts and Social Security reform. Debt is largely stable assuming policymakers stick to
the law and concerns about debt stability are largely predicated on the worry that they will pass
laws in the future that result in a higher debt trajectory.
We then propose three broad guidelines for the conduct of fiscal policy in practice which
could potentially be operationalized as more specific rules:
Real
Interest
Debt
Required
Primary
Balance
Debt
Required
Primary
Balance
0.5% 100% -1.5% 25% 0.1%
1.0% 200% -3.0% 50% 0.3%
2.0% 400% -6.0% 100% 0.5%
Hypothetical Debt Sustainability Examples
(Percent of GDP)
Illustrative Scenario 1
Illustrative Scenario 2
g = 2.0 and r = 0.5
g = 1.5 and r = 2.0
40
1. Temporary emergencies should not be paid for, with a broad definition of what
constitutes a temporary emergency and what can be done in that situation. As discussed
above, more fiscal support is needed today—and will likely be needed going forward in
future recessions. In many cases this fiscal support may effectively pay for itself resulting
in a lower debt-to-GDP ratio and potentially lower net real interest as a share of GDP as
well. Dynamic scoring using demand-side models would be appropriate for assessing the
fiscal impact of short-run emergency spending. Given the potentially persistent effects of
the current, and future, downturns, the possibilities of scaring reducing potential GDP,
and the amount of fiscal room, it would make sense for policymakers to err on the side of
a broad definition of unpaid emergency spending that lasts for several years and includes
items that do not spend out immediately, like infrastructure spending.
2. Permanent programs should be paid for, with broad exceptions for programs that
plausibly pay for themselves in net present value, like investments in children. The
principle of paying for permanent changes broadly makes sense and is consistent with a
sustainable real debt service trajectory given the current outlook. Moreover, a limiting
principle is needed for budgeting and as a way to assess whether any given program or
proposal makes sense. In principle, permanent programs should be assessed using
dynamic scoring. In practice such dynamic scoring may be difficult to do in real time and
misses the fact that much of the payoff of investments occurs outside the budget window.
As a result, we propose a crude way to take account of this by excluding a specific set of
programs and investments from the constraints of pay-as-you-go when strong evidence
from academic research implies they would plausibly pay for themselves in present
value. This includes well-designed investments in areas like children, education, and
research. Infrastructure would ideally be paid for with Pigouvian revenue measures that
improve infrastructure utilization, but it too could get an exception to the pay-as-you-go
principle.
3. Improve the composition of government to make it more supportive of demand and more
efficient. This includes many of the steps discussed earlier in this paper: improving
41
automatic stabilizers to better respond to recessions and increasing demand through
expanded balanced budget multipliers, more progressive fiscal transfers and expanded
social insurance. The composition of what the government does matters and to make
room for additional investments there are numerous changes that should be made for their
own sake, like reforming health care delivery systems in a way that would save costs and
potentially improve or not worsen the quality of outcomes and better enforcement of the
tax system to collect what is owed under current law (Sarin and Summers 2019).
Figures 18a, 18b and 18c project the resulting debt, nominal interest and real interest
paths under a program consistent with these guidelines. We believe policymakers should only
focus on the next decade because of the increasingly dramatic uncertainty around budget deficits
after that window but we still present thirty years of estimates to show a current best guess of the
impact of this program over the longer horizon, a subject of interest to some policy analysts.
Specifically, by way of illustration, these estimates assume $2.5 trillion in additional
fiscal support over the next there years and an investment program that starts at a net cost of 1
percent of GDP but eventually starts to result in deficit decreases over a longer period of time.
Overall, this would mean about $5 trillion of deficit-financed investments over the next decade
plus additional investments paid for by added revenue or other spending reductions. The result is
that the debt would stabilize at less than 150 percent of GDP which would be the highest the
United States has ever experienced but nominal interest payments would still be only 3.8 percent
of GDP and real interest payments would be only 1.0 percent of GDP (around the 77th percentile
of historical experience).
42
Figure 18a Figure 18b
Figure 18c
Note: Proposed framework includes Social Security reform phased in linearly from 0.5% of GDP to 1.7% of GDP over 10 years
beginning in 2025. $2.5T additional stimulus over 2021-2023. Investment in early education adds 1.0% of GDP to primary deficit
through 2035 after which deficit impact linearly shrinks until it reduces primary deficit by 0.5% of GDP in 2050.
Source: Congressional Budget Office; Office of Management and Budget; The Board of Trustees, Federal Old-Age and
Survivors Insurance and Disability Insurance Trust Funds; Macrobond; authors’ calculations.
The guidelines we propose are not universal or set in stone. They are what we believe is
understandable and achievable while moving the United States closer to an optimal approach to
achieving its broader economic objectives based on the current fiscal situation and outlook. To
the degree the economic situation, bond markets or fiscal outlook changes then a revised set of
guidelines—including a shift to even more deficit spending or potentially active deficit
reduction—might be warranted.
43
Conclusion
Six years ago one of us reintroduced the term “secular stagnation” to describe this period
of low interest rates, arguing that it was the result of an excess supply of saving meeting a
declining demand for investment and thus putting downward pressure on interest rates,
potentially with the zero lower bound for interest rates preventing the market from clearing. The
term “secular stagnation” has a negative resonance and was intended as a warning and prod for
action. But it should not be understood as an unqualified negative event that happened to the
economy.
Whether low interest rates are good or bad for our economic future depends on our
choices. Since interest rates cannot go well below zero as long as cash is still in existence—and
even low interest rates may lead to financial stability problems—this creates a challenge for the
economy and especially for attempts to manage recessions with countercyclical monetary policy.
Low interest rates also create numerous opportunities. They expand the scope for expansionary
fiscal policy, make the debt more sustainable and increase the scope of public investments that
will pay for themselves over time. Whether the era of low interest rates becomes a time of more
prolonged and severe recessions and greater financial market bubbles or instead becomes an
opportunity for public investment and stronger economic growth depends on macroeconomic
policy decisions. The correct diagnosis of our situation is the starting point for better
macroeconomic policy going forward.
44
Appendix – Methodological Assumptions Underlying Long-run Budget Forecasts
The Congressional Budget Office (CBO) generally produces three ten-year projections
annually and one thirty-year projection annually (e.g. CBO 2020a; 2020e). These projections are
intended to be baselines, that is to reflect unchanged laws or policies going forward. The
meaning of “unchanged” is clear in some cases, but in others it is ambiguous since absent
additional legislation discretionary spending would go to zero and major entitlement programs
would expire. In many cases, history has been repeated enough that there is a relatively clear
concept of what should be assumed in the baseline (e.g., major entitlement programs are
assumed to be reauthorized, as they have been historically. In other cases, it is trickier. For
example, CBO assumes that a set of individual tax cuts expire in 2025 whereas in the past many
but not all tax cuts have been extended instead of allowed to expire. CBO also assumes that
Social Security and Medicare continue to pay full benefits after their trust funds are exhausted
even though historically Congress has taken action to raise revenues or reduce spending to avoid
this situation.
We generally show three sets of estimates which assume:
Appendix Table 1
Expiring tax
cuts
Social Security
Medicare
Rest of
Budget
CBO-based
Baseline
Current law
Current policy
Current policy
Generally
current law
except
discretionary
spending
Tax Cuts
Permanent
Current policy
Current policy
Current policy
Social Security
Reform
Current law
Current law
Current policy
The second issue is how to handle discretionary spending and some miscellaneous other
mandatory spending not counting Social Security and health programs. CBO assumes that these
grow with inflation in the ten-year budget window and with nominal GDP outside the window.
Growing these series with nominal GDP is essentially assuming that Congress is annually
passing what would be described as a spending increase and that the real value of what the
government is buying with its discretionary spending is rising over time. Moreover, as shown in
Appendix Figure 1 this is at odds with the historic pattern of discretionary spending which has
fallen as a share of GDP. As such this is less a baseline than an estimate that shows what would
45
happen if the government passes annual spending increases. For our long-run forecasts we
instead assume that discretionary spending and other mandatory spending grow with inflation
plus population. This is both a more realistic description of what has happened in the past (also
shown in Appendix Figure 1) and also a more realistic estimate of what it would cost the
government to continue its current policies in terms of what is done by discretionary spending. If
anything, it is an overly high estimate of discretionary spending—and thus an overly high
estimate of the debt—because not all government costs grow with population, some are fixed,
and also because adjusting for inflation implicitly assumes zero productivity growth in
government.
Appendix Figure 1
Note: Series adjusted for population and inflation is adjusted to 2019 values using the GDP price index and the civilian
population aged 16 and older.
Source: Congressional Budget Office; authors’ calculations.
Finally, our long-run estimates include dynamic feedback from choices about budget
policies to GDP and interest rates. These are chosen to be very conservative and come closer to
standard estimates so do not reflect the fact that higher debt, for example, enables more effective
monetary policy and thus smaller output gaps and a higher level of long-run output. We assume
that deviations from CBO’s baseline debt path lead to deviations from CBO’s baseline GDP path
of 0.04 percent lower GDP for each percentage point increase in the debt-to-GDP ratio relative to
the CBO-based baseline in the preceding year. This approximates the effects of macroeconomic
feedback due to higher debt from CBO’s 2019 long-term budget outlook (CBO 2019b). In the
CBO-based baseline, we also assume that interest rates increase after 2030 based on the increase
46
in the prior year’s debt-to-GDP ratio compared with its 2030 value in order to approximate the
macroeconomic feedback effects in CBO’s 2019 long-term outlook. For other scenarios, interest
rates increase based on the increase in the prior year’s debt-to-GDP ratio compared with its value
in the CBO-based baseline
47
References
Abiad, Abdul, Davide Furceri, and Petia Topalova. 2015. “The Macroeconomic Effects of Public
Investment: Evidence from Advanced Economies.” IMF Working Paper 15/95.
Adler, Gustavo, Romain Duval, Davide Furceri, Sinem Kiliç Çelik, Ksenia Koloskova, and
Marcos Poplawski-Ribeiro. 2017. “Gone with the Headwinds: Global Productivity.” IMF
Staff Discussion Note 17/04.
Akcigit, Ufuk, Douglas Hanley, Nicolas Serrano-Velarde. 2020. “Back to Basics: Basic Research
Spillovers, Innovation Policy, and Growth.” The Review of Economic Studies.
Aronvich, Alex, and Andrew Meldrum. 2020. “New Financial Market Measures of the Neutral
Real Rate and Inflation Expectations.” FEDS Notes. Washington: Board of Governors of
the Federal Reserve System.
Auerbach, Alan J. William G. Gale, and Aaron Krupkin. 2019. “Revisiting the Federal Budget
Outlook.” Tax Notes Federal 813-829. August 5.
Auerbach, Alan J. and Yuriy Gorodnichenko. 2017. “Fiscal Stimulus and Fiscal Sustainability.”
In Fostering a Dynamic Global Economy, 217-270. Kansas City, MO: Federal Reserve
Bank of Kansas City.
Baily, Martin. 1978. “Some Aspects of Optimal Unemployment Insurance.” Journal of Public
Economics 10(3): 379-402.
Ball, Laurence, J. Bradford DeLong, and Lawrence H. Summers. 2017. “Fiscal Policy and Full
Employment.” In Confronting Policy Challenges of the Great Recession: Lessons for
Macroeconomic Policy, edited by Eskander Alvi, 85-106. Kalamazoo, MI: W.E. Upjohn
Institute for Employment Research.
Barro, Robert J. 1979. “On the Determination of the Public Debt.” Journal of Political Economy
87(5): 940-971.
Bernanke, Ben S. 2005. “The Global Saving Glut and the U.S. Current Account Deficit.”
Remarks at the Sandridge Lecture, Virginia Association of Economists. March 10.
______. 2020. “The New Tools of Monetary Policy.” American Economic Review 110(4): 943-
983.
Blahous, Charles. 2017. “Fix the Federal Spending Baseline First.” Economics21. New York:
Manhattan Institute. September 10.
48
Blanchard, Olivier. 2019. “Public Debt and Low Interest Rates.” American Economic Review
109(4): 1197-1229.
Blanchard, Olivier, and Lawrence H. Summers (eds). 2019. Evolution or Revolution? Rethinking
Macroeconomic Policy after the Great Recession. Cambridge, MA: MIT Press.
Bloom, Nicholas, Mark Schankerman, and John Van Reenen. 2013. “Identifying Technology
Spillovers and Product Market Rivalry.” Econometrica 81(4): 1347-1393.
Borio, Claudio, Leonardo Gambacorta, and Boris Hofmann. 2017. “The Influence of Monetary
Policy on Bank Profitability.” International Finance 20(1): 48-63.
Borio, Claudio, and Haibin Zhu. 2008. “Capital Regulation, Risk-Taking and Monetary Policy:
A Missing Link in the Transmission Mechanism?” BIS Working Paper 268. Bank for
International Settlements.
Boushey, Heather, Ryan Nunn, and Jay Shambaugh (eds). 2019. Recession Ready: Fiscal
Policies to Stabilize the American Economy. Washington: Brookings.
Caballero, Ricardo J., Emmanuel Farhi, and Pierre-Olivier Gourinchas. 2017. “The Safe Assets
Shortage Conundrum.” Journal of Economic Perspectives 31(3): 29-46.
Campbell, John Y., Adi Sunderam, and Luis M. Viceira. 2017. “Inflation Bets or Deflation
Hedges? The Changing Risks of Nominal Bonds.” Critical Finance Review 6: 263–301.
Chetty, Raj. 2008. “Moral Hazard versus Liquidity and Optimal Unemployment Insurance.”
Journal of Political Economy 116(2): 173-234.
Congressional Budget Office (CBO). 2000. The Budget and Economic Outlook: Fiscal Years
2001-2010. Washington.
______. 2019a. An Evaluation of CBO’s Past Deficit and Debt Projections.
______. 2019b. The 2019 Long-Term Budget Outlook. Washington.
______. 2020a. An Update to the Budget Outlook: 2020 to 2030. Washington.
______. 2020b. An Update to the Economic Outlook: 2020 to 2030. Washington.
______. 2020c. Automatic Stabilizers in the Federal Budget: 2020 to 2030. Washington.
______. 2020d. Federal Debt: A Primer. Washington.
______. 2020e. The 2020 Long-Term Budget Outlook. Washington.
______. 2020f. The Outlook for Major Federal Trust Funds 2020 to 2030. Washington.
Council of Economic Advisers (CEA). 2015. “Long-term Interest Rates: A Survey.” Report.
49
Dell'Ariccia, Giovanni, Luc Laeven, and Gustavo A. Suarez. 2017. "Bank Leverage and
Monetary Policy's Risk‐Taking Channel: Evidence from the United States." The Journal
of Finance 72(2): 613-654.
DeLong, J. Bradford, and Lawrence H. Summers. 1988. “How Does Macroeconomic Policy
Affect Output?” Brookings Papers on Economic Activity 19(2): 433-480.
______. 2012. “Fiscal Policy in a Depressed Economy.” Brookings Papers on Economic Activity
43(1): 233–97.
Dickens, William T., Isabel Sawhill, and Jeffrey Tebbs. 2006. “The Effects of Investing in Early
Education on Economic Growth.” Brookings Policy Brief No. 153. Washington:
Brookings.
Dixit, Avinash K. and Robert S. Pindyck. 1994. Investment Under Uncertainty. Princeton, NJ:
Princeton University Press.
Elmendorf, Douglas W., and Louise M. Sheiner. 2017. “Federal Budget Policy with an Aging
Population and Persistently Low Interest Rates.” Journal of Economic Perspectives
31(3): 175-94.
Fiedler, Matt, Jason Furman, and Wilson Powell III. 2019. Increasing Federal Support for State
Medicaid and CHIP Programs in Response to Economic Downturns. Unemployment
Insurance and Macroeconomic Stabilization. In Recession Ready: Fiscal Policies to
Stabilize the American Economy, eds. Heather Boushey, Ryan Nunn, and Jay
Shambaugh, 93-127. Washington: Brookings.
Furman, Jason. 2016. “The New View of Fiscal Policy and Its Application.” Remarks at Global
Implications of Europe’s Redesign Conference. New York. October 5.
Gaspar, Vitor, Maurice Obstfeld, and Ratna Sahay. 2016. “Macroeconomic Management When
Policy Space Is Constrained: A Comprehensive, Consistent, and Coordinated Approach
to Economic Policy.” IMF Staff Discussion Note, September 2016.
Government Accountability Office (GAO). 1992. Prompt Action Necessary to Avert Long-Term
Damage to the Economy. Washington.
Haavelmo, Trygve. 1945. “Multiplier Effects of a Balanced Budget.” Econometrica 13(4): 311-
318.
Hendren, Nathaniel, and Ben Sprung-Keyser. 2020. “A Unified Welfare Analysis of Government
Policies.” The Quarterly Journal of Economics 135(3): 1209-1318.
50
International Monetary Fund (IMF). 2015. Can Fiscal Policy Stabilize Output? In April 2015
Fiscal Monitor: Now Is the Time, 21-48.
Koo, Jaewoon, and Yunxing Song. 2016. “The Relationship Between Income Inequality and
Aggregate Saving: An Empirical Analysis Using Cross-Country Panel Data.” Applied
Economics 48: 892-901.
Mourougane, Annabelle, Jarmila Botev, Jean-Marc Fournier, Nigel Pain and Elena Rusticelli.
2016. “Can an Increase in Public Investment Sustainably Lift Economic Growth?” OECD
Economics Department Working Papers No. 1351.
National Commission on Fiscal Responsibility and Reform. 2010. “The Moment of Truth.”
Washington.
Oreopoulos, Phillip, Till von Wachter, and Andrew Heisz. 2012. “The Short- and Long-Term
Career Effects of Graduating in a Recession.” American Economic Journal: Applied
Economics 4(1): 1–29.
Organisation for Economic Co-operation and Development (OECD). 2016. “Chapter 2: Using
the Fiscal Levers to Escape the Low-Growth Trap.” In OECD Economic Outlook 2016,
no. 2. Paris: OECD Publishing.
Rachel, Lukasz, and Lawrence H. Summers. 2019. “Our Secular Stagnation in the Industrialized
World.” Brookings Papers on Economic Activity 50(1): 1-54.
Reifschneider, David, and John C. Williams. 2000. “Three Lessons for Monetary Policy in a
Low-Inflation Era.” Journal of Money, Credit, and Banking 32(4): 936-966.
Sarin, Natasha and Lawrence H. Summers. 2019. “Shrinking the Tax Gap: Approaches and
Revenue Potential.” Tax Notes. November 18.
Stansbury, Anna, and Lawrence H. Summers. 2020. “The End of the Golden Age of Central
Banking?” Secular Stagnation is About More than the Zero Lower Bound?” Unpublished
manuscript.
Summers, Lawrence H. 2014. “U.S. Economic Prospects: Secular Stagnation, Hysteresis, and the
Zero Lower Bound.” Business Economics 49(2): 65-73.
The Board of Trustees, Federal Old-Age and Survivors Insurance and Federal Disability
Insurance Trust Funds. 1989. The 1989 Annual Report of the Board of Trustees of the
Federal Old-Age and Survivors Insurance Trust Fund and The Federal Disability
51
Insurance Trust Fund, Pursuant to 42 U.S.C. 401(c)(2), 1395i(b)(2), 1395t(b)(2).
Washington.
______. 2020. The 2020 Annual Report of the Board of Trustees of the Federal Old-Age and
Survivors Insurance and Federal Disability Insurance Trust Funds. Washington.
The Boards of Trustees, Federal Hospital Insurance and Federal Supplementary Medical
Insurance Trust Funds. 2020. The 2020 Annual Report of the Boards of Trustees of the
Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds.
Washington.
Tymoigne, Eric, and L. Randall Wray. 2013. "Modern Money Theory 101: A Reply to Critics."
Levy Economics Institute, Working Papers Series.
Yagan, Danny. 2019. “Employment Hysteresis from the Great Recession.” Journal of Political
Economy 127(5): 2505-2558.
