Bianchi, Melosi, Inflation as a Fiscal Limit (Jackson Hole Economic Policy Symposium 2022, 2022-08)
Summary
Paper by Francesco Bianchi and Leonardo Melosi dated August 19, 2022, with a header stating it is embargoed until the presentation time on August 27, 2022. The abstract says low and stable inflation requires an appropriate fiscal framework aimed at stabilizing government debt, and that trend inflation is critically influenced by actual or perceived changes to this framework. It states that the recent fiscal interventions in response to the COVID pandemic altered the private sector's beliefs about the fiscal framework, accelerating the recovery but also increasing fiscal inflation. The authors write that this increase could not have been averted by simply tightening monetary policy.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Bianchi, Melosi, Inflation as a Fiscal Limit (Jackson Hole Economic Policy Symposium 2022, 2022-08).”
Original at www.kansascityfed.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/bianchi-melosi-inflation-as-a-fiscal-limit-jackson-hole-economic-polic-3508782ce3/
, SHA-256 8667f15426b5550dfeca60deab13b9551e00bd76a17b059af35840c13b61f219.Full text
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Inflation as a Fiscal Limit∗
Francesco Bianchi Leonardo Melosi
Johns Hopkins University Federal Reserve Bank of Chicago
CEPR and NBER CEPR
August 19, 2022
Abstract
Low and stable inflation requires an appropriate fiscal framework aimed at stabilizing
government debt. Historically, trend inflation is critically influenced by actual or per-
ceived changes to this framework, while cost-push shocks only account for short-lasting
movements in inflation. Before the pandemic, a moderate level of fiscal inflation has coun-
teracted deflationary pressures, helping the central bank to avoid deflation. The recent
fiscal interventions in response to the COVID pandemic have altered the private sector’s
beliefs about the fiscal framework, accelerating the recovery, but also determining an in-
crease in fiscal inflation. This increase in inflation could not have been averted by simply
tightening monetary policy. The conquest of post-pandemic inflation requires mutually
consistent monetary and fiscal policies to avoid fiscal stagflation.
Keywords: Fiscal limits, monetary/fiscal policy mix, inflation, government debt, fiscal
stagflation.
JEL Classification: E50, E62, E30.
∗
Francesco Bianchi: francesco.bianchi@jhu.edu. Leonardo Melosi: leonardo.melosi@chi.frb.org. Russell
Miles provided excellent research assistance. The views in this paper are solely those of the authors and should
not be interpreted as reflecting the views of the Federal Reserve Bank of Chicago or any person associated
with the Federal Reserve System.
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
1 Introduction
Inflation is back. After two decades of low inflation, policymakers all over the world are
newly confronted with a high bout of fast-growing prices. Much is yet to be learned about
this inflation episode. An aspect of particular relevance for policymakers is the persistence of
the ongoing inflationary pressures. Will inflation fall as rapidly as it rose, following a similar
pattern observed after the Second World War and the Korean War in the United States? Or
will inflation drift up as observed in the late 1960s and 1970s in many advanced economies?
In this paper, we argue that the answer to these important questions hinges predominantly
on the fiscal authority’s credibility in stabilizing a large fiscal imbalance. The central bank’s
anti-inflation reputation, albeit important, is not decisive.
Trend inflation is fully controlled by the monetary authority only when public debt can
be successfully stabilized by credible future fiscal plans. When the fiscal authority is not
perceived as fully responsible for covering the existing fiscal imbalances, the private sector
expects that inflation will rise to ensure sustainability of national debt. As a result, a large
fiscal imbalance combined with a weakening fiscal credibility may lead trend inflation to drift
away from the long-run target chosen by the monetary authority.
This reasoning configures a natural and interesting limit on fiscal policy. This limit takes
the form of incompatibility between lax fiscal policy and a monetary framework aimed at
achieving a low and stable inflation environment. When fiscal imbalances are large and fiscal
credibility wanes, it may become increasingly harder for the monetary authority to stabilize
inflation around its desired target. If the monetary authority increases rates in response to
high inflation, the economy enters a recession, which increases the debt-to-GDP ratio. If the
monetary tightening is not supported by the expectation of appropriate fiscal adjustments,
the deterioration of fiscal imbalances leads to even higher inflationary pressure. As a result,
a vicious circle of rising nominal interest rates, rising inflation, economic stagnation, and
increasing debt would arise.
In this pathological situation, monetary tightening would actually spur higher inflation
and would spark a pernicious fiscal stagflation, with the inflation rate drifting away from
the monetary authority’s target and with GDP growth slowing down considerably. While
in the short run, monetary tightening might succeed in partially reducing the business cycle
component of inflation, the trend component of inflation would move in the opposite direction
as a result of the higher fiscal burden. Fiscal stagflation does not stem from a perceived or
actual loss of anti-inflation reputation by the central bank. Rather, it is caused by the
progressive deterioration of the fiscal authority’s credibility to stabilize its large debt and the
realization that the reputation of the monetary authority is incompatible with the expected
behavior of the fiscal authority.
1
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
To substantiate these arguments, we build and estimate a model that allows for changes
in the Monetary/Fiscal policy mix. In doing so, we build on previous work such as Bianchi
and Ilut (2017) and Bianchi and Melosi (2017), with an important difference. We allow
for the possibility that the recent fiscal interventions implemented in response to the COVID
pandemic might have weakened agents’ beliefs that the large fiscal imbalances will be stabilized
by taking the necessary fiscal adjustments. Technically, this feature is modelled as a change in
the probability of moving to a Fiscally-led policy mix in which the fiscal authority is not fully
committed to stabilizing debt and the central bank accommodates the resulting movements
in inflation.
When we bring the model to the data, we find that movements in trend inflation can
be accounted for by fiscal shocks and by changes in the policy mix. The steady increase
in trend inflation observed in the 1960s and 1970s can be explained by the fast-growing
government spending, which was needed to support long-lasting welfare programs associated
with President Lyndon Johnson’ Great Society initiatives, as well as the long and expensive
war in Vietnam. These shocks, combined with a Fiscally-led policy mix in which the central
bank is not credibly committed to stabilize inflation, generate persistent movements in trend
inflation. On the contrary, cost-push shocks only account for the spikes in inflation around the
oils shocks of those years. In this respect, our model conforms to the notion that “persistent
high inflation is always and everywhere a fiscal phenomenon” (Sargent, 2013).
Trend inflation declined in the early 1980s when policymakers adopted a Monetary-led
policy mix, in which the fiscal authority is credibly committed to stabilize debt and the central
bank responds strongly to deviations of inflation from its desired target. This switch in the
monetary and fiscal policy mix followed President Carter’s appointment of Paul Volcker as
Chairman of the Federal Reserve in August 1979 and coincided with the election of President
Ronald Reagan in 1981. This interpretation of the Volcker disinflation is consistent with
Samuelson (2008) who argues that the Reagan administration provided the much needed
political backing to the Federal Reserve Chairman Paul Volcker’s anti-inflation policies.
Fed officials were arguably aware of the change in the fiscal authority’s attitude. At the
July 1981 meeting of the FOMC, the then President of the Federal Reserve of Kansas City,
Roger Guffey, argued: ”Historically, the Federal Reserve has always come up to the hitching
post and then backed off simply because the administration and the Congress have thrown
bricks at us or have not been supportive of a policy of restraint. Through the course of recent
history at least, we’ve backed off and we’ve made a mistake each time. I think we have an
opportunity this time to carry forward what we should have done before because for the first
time ever we do have, for whatever length of time, the support of the administration at least.
So, we ought to take advantage of that opportunity.” (FOMC meeting Transcripts, July 1981,
cf. Weise (2012).)
2
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Fiscal inflation has not completely vanished after the 1980s disinflation, as agents take
into account that a return to the Fiscally-led policy mix is still possible and fiscal spending
remains elevated. However, the amount of fiscal inflation has remained modest for a long
time because of the prevailing Monetary-led policy mix. This moderate level of fiscal infla-
tion has counteracted exogenous deflationary pressure and it has helped the central bank to
avoid deflation. In fact, absent fiscal inflation, the U.S. economy would have experienced
a larger deflationary bias and larger output losses during the years spent at the zero lower
bound. Thus, historical circumstances might have provided a false sense of irrelevance of fiscal
sustainability considerations.
The COVID pandemic, with a second spell of zero lower bound after only a few years of
slightly positive interest rates, has arguably changed this perception. In the model, we allow
for the possibility of changes in beliefs about the future policy mix while at the zero lower
bound. We show that following the ARPA fiscal stimulus, the probability assigned to moving
to the Fiscally-led policy mix experienced a large increase. This change in beliefs has resulted
in a large jump in fiscal inflation, accounting for approximately half (3.5%) of the recent
increase in inflation, with cost-push shocks contributing by a similar amount. However, as
mentioned above, cost-push shocks and fiscal imbalances affect inflation in very different ways.
Cost-push shocks have only transitory effects on inflation, independently from the policy mix
in place. Instead, shocks to long-term spending propagate very differently across monetary
and fiscal policy regimes and remain a potential threat to inflation stability for a long time,
given that they affect the fiscal burden for many years.
The fact that approximately half of the recent increase in inflation has fiscal roots poses
some specific challenges for policymakers today. Not only fiscal inflation tends to be highly
persistent but it also requires a different policy response. When inflation has a fiscal nature,
monetary policy alone may not provide an effective response. To show this, we ask whether
tightening monetary policy earlier on could have prevented the post-pandemic increase in
U.S. inflation. To answer this question, we build a counterfactual simulation in which we
keep agents’ beliefs about the future policy mix unchanged, but we allow the central bank to
increase interest rates in response to inflation. The increase in rates would have resulted in
only a modest reduction in inflation, at the cost of a large reduction in output. This large
sacrifice ratio arises because when inflation has a fiscal nature, the central bank is not uniquely
responsible for its reduction.
To go back to the initial question: will the ongoing inflationary pressures persist as in
the 1960s and and 1970s? Our study underscores the risk that a similar persistent pattern
of inflation might characterize the years to come. Nevertheless, we emphasize that the fac-
tors behind the heightened inflation risk today are quite different from those that caused the
Great Inflation. As described in several narrative accounts (Meltzer, 2009; Taylor, 2011), the
3
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
heightened inflation rate of the 1960s and 1970s was the result of the lack of the Federal Re-
serve’s independence from the fiscal authority. This lack of independence critically tarnished
the credibility of the monetary framework, paving the way for the surge of trend inflation
in the 1960s and 1970s. In contrast, the risk of persistent high inflation the U.S. economy
is experiencing today seems to be explained more by the worrying combination of the large
public debt and the weakening credibility of the fiscal framework.
Thus, the recipe used to defeat the Great Inflation in the early 1980s might not be effective
today. In the early 1980s, the resolute anti-inflation stance taken by the Federal Reserve and
backed by the new administration was the winning move. An important factor behind this
success was the historically low government debt that provided strong credibility to the fiscal
backing. Today the problem of controlling inflation is compounded by the highly uncertain
fiscal situation, with the Congressional Budget Office (CBO) projecting federal debt to keep
rising after the year 2023 to reach its highest level ever recorded in 2032 (Congressional
Budget Office, 2022).1 Therefore, even though monetary policy independence is a much more
widely respected and better understood value today, high inflation can still be a threat if the
fiscal situation is left unresolved. In fact, we show that if the private sector loses confidence
in the fiscal authority’s willingness to fix this quickly-deteriorating fiscal backdrop, hawkish
monetary policies can mire the U.S. economy in a prolonged period of stagflation.
All told, our results suggest that conquering the post-pandemic inflation necessitates an
overhaul of the fiscal framework aimed at financing the large stock of government debt as well
as the increase in public expenditure needed to cover rising costs associated with population
aging and climate change.
Our analysis is subject to a number of caveats. First, getting robust predictions from
models at the end of the sample period is known to be a challenging task. Lacking information
about the development of the macroeconomic outlook over the next years might affect the
accuracy of some of our results. Specifically, our model’s prediction regarding how much the
ARPA fiscal stimulus has altered the perceived credibility of the fiscal framework may vary as
more observations become available. Second, the evolution of the private sector’s perception
about the strength of the central bank’s anti-inflation commitment and about the credibility
of the fiscal framework cannot be observed directly. Rather, we infer them indirectly from
observing the joint dynamics of the debt-to-GDP ratio, the real interest rate, and inflation.
To mitigate this last caveat, we back the evolution of these beliefs with the available narrative
evidence regarding the monetary and fiscal interactions.
The findings of this paper are consistent with Bianchi, Faccini, and Melosi (2022), where
we develop a fiscal theory of trend inflation and find that fiscal trend inflation has increased
1
The CBO bases its projections on the assumption that in 2023 inflationary pressures will subside. Fur-
thermore, the CBO warns that, if lawmakers amended current laws to maintain certain policies now in place,
even larger increases in debt would ensue.
4
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
during the pandemic. In that paper, instead of modeling regime changes in the policy mix, we
build a model in which a Monetary-led policy mix and a Fiscally-led policy mix coexist at the
same time. The model features unfunded fiscal shocks. These are shocks to transfers that are
not backed by future fiscal adjustments. With respect to these shocks, policymakers follow
a Fiscally-led policy mix in which the central bank accommodates the increase in inflation
necessary to stabilize the unfunded share of debt. As a result, these shocks trigger persistent
movements in inflation and real interest rates, leading to a fiscal theory of trend inflation.
However, policymakers follow a Monetary-led policy mix with respect to all other shocks.
Consequently, the propagation of familiar business-cycle shocks, such as TFP and monetary
policy shocks, is consistent with standard general-equilibrium models and with decades of
empirical work based on VAR models.
Our work is connected to the vast theoretical literature on monetary-fiscal policy inter-
action (Sargent and Wallace, 1981; Leeper, 1991; Sims, 1994; Woodford, 1994, 1995, 2001;
Cochrane, 1998, 2001; Schmitt-Grohe and Uribe, 2000; Benhabib, Schmitt-Grohe, and Uribe,
2002; Bassetto, 2002; Reis, 2016; Bassetto and Cui, 2018; Bassetto and Sargent, 2020). Barro
(1974) shows that an alternative way to generate non-Ricardian effects is if agents erro-
neously regard bonds as net-wealth. Aiyagari and Gertler (1985) study the implications of
fiscal backing of government bonds for the propagation of shocks. They find that for debt to
be irrelevant, the model needs to feature a considerable degree of accommodation with respect
to the monetary authority. Leeper and Zhou (2021) find that inflation plays an important
role in the optimal marginal financing of fiscal needs.
In recent years, a series of papers have provided empirical evidence for the importance of
the ideas put forward in the theoretical literature on monetary/fiscal policy interaction. Davig
and Leeper (2006) study the implications of regime changes in the policy mix in a calibrated
New Keynesian model. Bianchi and Ilut (2017) estimate a model with regime changes in the
monetary/fiscal policy mix and link the high inflation of the 1960s-1970s to shocks to long-
term spending combined with a Fiscally-led regime. Bianchi and Melosi (2017) argue that
the possibility of a return to such regime can explain the lack of deflation in the aftermath of
the Great Recession. Leeper, Traum, and Walker (2017) find that the size of fiscal multipliers
depends on the policy mix in place. Eusepi and Preston (2012) show that imperfect knowledge
breaks Ricardian equivalence, making the scale and composition of the public debt relevant
for inflation dynamics. Hall and Sargent (2011) argue that, historically, most of U.S. debt
stabilization has been achieved through a combination of growth, revaluation effects, and low
real interest rates, as opposed to fiscal adjustments.
Bianchi and Melosi (2019) discuss the risks of the lack of coordination between the mone-
tary and fiscal authority, focusing on fiscal stagflation. In earlier work, Sims (2011) discusses
a similar “stepping on a rake” mechanism in the context of a Fiscally-led policy mix to explain
5
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
why a monetary policy shock might be able to initially lower inflation but then generate even
higher inflation. Finally, Woodford (2001) and Loyo (1999) use a perfect foresight endowment
economy to show that if the central bank follows the Taylor principle while the fiscal authority
does not stabilize debt, an explosive path for the price level arises.
Our analysis also connects with the policy interventions that Chris Sims advocated at the
2016 Jackson Hole Symposium to replace ineffective monetary policy at the zero lower bound
(Sims, 2016). Sims called for central banks “to explain that fiscal, as well as monetary policy
should be aimed at meeting inflation targets. This means, specifically, stating that inflation
will intentionally be at least part of the means for financing current debt and deficits.” We find
that the bold fiscal interventions implemented in response to the COVID recession succeeded
in increasing inflation and sustaining the recovery, lifting the economy from the zero lower
bound. In this respect, the pandemic interventions might have made the zero lower bound
fiscally unsustainable as in Woodford (2003) and Benhabib, Schmitt-Grohe, and Uribe (2002).
However, going forward, setting clear limits on fiscal policy could be necessary to prevent
inflation from remaining persistently above the desired inflation target. We discuss these
kinds of limits on fiscal policy in Bianchi and Melosi (2019) and provide evidence for their
empirical relevance in Bianchi, Faccini, and Melosi (2022). In Bianchi, Fisher, and Melosi
(2021), we show that the macroeconomic effects of the ARPA fiscal stimulus are critically
affected by agents’ expectations about the post-ZLB policy mix. In this paper, we formally
estimate this post-ARPA change in beliefs.
Our explanation of the Great Inflation is consistent with Clarida, Galı́, and Gertler (2000)
and Lubik and Schorfheide (2004), who also find that monetary policy was not responding
strongly enough to inflation in the 1970s. We differ with respect to the mechanism that
accounts for the rise of inflation. In their work, fiscal policy is assumed to remain committed
to stabilizing government debt. As a result, the U.S. economy was subject to indeterminacy
in the 1970s and the Great Inflation occurred because of a higher inflation target and the
possibility of shocks to expectations (sunspot shocks). In our model, agents expected inflation
to rise in order to finance the large and persistent increase in government spending which
started in the 1960s with the launch of the Great Society initiatives and the Vietnam war.
2 A model of monetary/fiscal policy interaction
In this section, we introduce the model that we will fit to U.S. data. The model is based on
Bianchi and Melosi (2017) and it is obtained by augmenting a prototypical New Keynesian
model with a fiscal block and a discrete shock that can push the economy to the zero lower
bound (ZLB). With respect to the original specification in Bianchi and Melosi (2017), we
make an important extension: We allow for the possibility of a change in agents’ beliefs about
6
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
the exit strategy from the ZLB. This idea is modelled by allowing for two ZLB regimes that
are characterized by the same policy behavior, but differ in terms of the probability of moving
to a Fiscally-led policy mix.
2.1 A New Keynesian model
Households. The representative household maximizes expected utility:
"∞ #
X
β t exp ζtd log Ct − ΦCt−1
A
E0 − ht (1)
s=0
subject to the budget constraint:
Pt Ct + Ptm Btm + Pts Bts = Pt Wt ht + Bt−1
s
+ (1 + ρPtm ) Bt−1
m
+ Pt Dt − Tt + T Rt
where Dt stands for real dividends paid by the firms, Ct is the level of consumption obtained
aggregating a series of differentiated goods with a Dixit-Stiglitz aggregator, ht is hours, Wt
is the real wage, Tt is taxes, T Rt stands for transfers, CtA corresponds to the average level
of consumption in the economy, and the parameter Φ captures the degree of external habit.
As in Woodford (2001), we allow for a simplified maturity structure of government debt.
Specifically, there are two types of government bonds: one-period government debt, Bts , in
zero net supply with price Pts , and a more general portfolio of government debt, Btm , in non-
zero net supply with price Ptm . The former debt instrument satisfies Pts = Rt−1 . The latter
debt instrument has payment structure ρT −(t+1) for T > t and 0 < ρ < 1. This second asset
can be interpreted as a portfolio of infinitely many bonds with average maturity controlled
by the parameter ρ. The value of such an instrument issued in period t in any future period
m−j
t + j is Pt+j = ρj Pt+j
m .
The preference shock ζtd is the sum of a continuous and discrete component: ζtd = dt + dξd .
t
The continuous component dt follows an AR(1) process: dt = ρd dt−1 + σd εd,t . The discrete
component dξd can assume two values: high or low (dh or dl ). The variable ξtd controls which of
t
these two values is in place and evolves according a Markov-switching process with transition
matrix H d : " #
d phh 1 − pll
H = ,
1 − phh pll
d = j|ξtd = i . The values of H d , dh , and dl are such that the unconditional
where pji = P ξt+1
mean of the discrete shock dξd is zero. This specification allows us to model recurrent shocks
t
that are large enough to trigger the ZLB. Agents take into account that these episodes can
lead to unusual responses by policymakers, as discussed below.
Firms. The economy is populated by a continuum of firms producing differentiated goods
7
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
that enter the household consumption basket. The representative firm j faces a downward-
sloping demand curve, Yt (j) = (Pt (j)/Pt )−1/υt Yt , where the parameter 1/υt is the elasticity
of substitution between two differentiated goods. Firms take as given the general price level,
Pt , and the level of real activity, Yt . Whenever a firm changes its price, it faces a quadratic
adjustment cost:
ACt (j) = .5φ (Pt (j)/Pt−1 (j) − Π)2 Yt (j)Pt (j)/Pt (2)
where Πt = Pt /Pt−1 is gross inflation at time t and Π is the corresponding deterministic steady
state. Shocks to the elasticity of substitution imply shocks to the markup ℵt = 1/ (1 − υt ) .
We assume that the rescaled markup µt = κ log (ℵt /ℵ) follows an autoregressive process,
1−υ
µt = ρµ µt−1 + σµ ϵµ,t , where κ ≡ υφΠ 2 is the slope of the Phillips curve. The firm chooses the
price Pt (j) to maximize the present value of future profits:
"∞ #
X
Et Qs ([Ps (j)/Ps ] Ys (j) − Ws hs (j) − ACs (j))
s=0
where Qs is the marginal value of a unit of consumption good. Labor is the only input in
the firm production function, Yt (j) = At h1−α
t (j), where total factor productivity At evolves
according to an exogenous process: ln (At /At−1 ) = γ + at , at = ρa at−1 + σa εa,t , ϵa,t ∼ N (0, 1).
Government. Imposing the restriction that one-period debt is in zero net supply, the
flow budget constraint of the federal government is given by:
Ptm Btm = Bt−1
m
(1 + ρPtm ) − Tt + Et + T Pt
where Ptm Btm captures to the market value of debt and Tt and Et correspond to federal tax
revenues and federal expenditures, respectively. Government expenditure is the sum of federal
transfers and goods purchases: Et = Pt Gt +T Rt . The term T Pt is a shock that captures a series
of features that are not explicitly modeled here, such as changes in the maturity structure
and the term premium. This shock is necessary because we treat all the components of the
government budget constraint as observables. We rewrite the federal government budget
constraint in terms of debt-to-GDP ratio bm m m
t = (Pt Bt ) / (Pt Yt ):
bm m m
t = bt−1 Rt−1,t / (Πt Yt /Yt−1 ) − τt + et + tpt
m
where Rt−1,t = (1 + ρPtm ) /Pt−1
m is the realized return of the maturity bond, all the variables
are expressed as a fraction of GDP, and we assume tpt = ρtp tpt−1 + σtp εtp,t , ϵtp,t ∼ N (0, 1).
The linearized transfers as a fraction of GDP, tr
e t , behave according to the following pro-
8
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
cess:
e ∗t e ∗t + (1 − ρtr ) ϕy (b
e t − tr
tr e t−1 − tr
= ρtr tr yt − ybt∗ ) + σtr ϵtr,t
e ∗t = ρtr∗ tr
tr e ∗t−1 + σtr∗ ϵtr∗ ,t , ϵtr∗ ,t ∼ N (0, 1) , ϵtr,t ∼ N (0, 1)
∗
where tr
e t corresponds to long-term transfers. This component captures persistent changes
in spending that arise as the result of large programs.2 The total amount of transfers moves
around this trend component as a result of exogenous shocks and the response to business cycle
yt − ybt∗ ), where ybt∗ is potential output.
fluctuations captured by the log-linearized output gap (b
The government also buys a fraction Gt /Yt of total output. We define gt = 1/(1 − Gt /Yt )
and we assume that get = ln(gt /g) follows an autoregressive process: get = ρg get−1 + σg ϵg,t ,
ϵg,t ∼ N (0, 1) .
Policy Rules. When the high value for the preference shock is realized (ξtd = h), the
economy is out of the ZLB and monetary and fiscal policies are not constrained. In this case,
the central bank follows a standard Taylor rule and the evolution of the policy mix can be
described by a two-regime Markov switching process ξtp , whose properties will be described
in the next section. When the low value for the preference shock is realized (ξtd = l), the
central bank lowers the interest rate until the ZLB is reached and the fiscal authority focuses
on stabilizing the economy as opposed to trying to stabilize the stock of debt.
Specifically, the monetary policy rule reads as follows:
ρR,ξp
h ψπ,ξp
i 1−ρ p
∗ ψ p R,ξt
Rt /R = 1 − Zξd (Rt−1 /R) t (Πt /Π) t (Yt /Y ) y,ξt
t eσR ϵR,t
t
+Zξd (Rt−1 /R)ρR,Z (1/R)(1−ρR,Z )ψZ eσZ ϵR,t (3)
t
where ϵR,t ∼ N (0, 1), R is the steady-state gross nominal interest rate, Yt∗ is the output
target, and Π is the deterministic steady-state level for gross inflation. The parameters ψπ,ξtp
and ψy,ξtp capture the central bank’s response to inflation and the output gap, which depends
on the policy mix ξtp in place at time t. The dummy variable Zξd is zero when the value of the
t
preference shock is high (ξtd = h) and one when it is low (ξtd = l). To match the behavior of the
interest rate in the data during the zero-lower-bound period, we allow for small disturbances
and a gradual decline toward a value close, but not exactly equal, to zero. The size of the
monetary policy shocks at the ZLB, σZ , is assumed to be a tenth of the standard deviation
of the monetary policy shocks when out of the ZLB: σZ = σR /10. The persistence of changes
in the interest rate at the ZLB is controlled by ρR,Z and fixed to .2. Finally, the parameter
2
In what follows, x
bt denotes the percentage deviation of a detrended variable from its own steady state. For
all the variables normalized with respect to GDP (debt, expenditure, and taxes) x et denotes a linear deviation,
while for all the other variables xet denotes a percentage deviation.
9
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
0 < ψZ ≤ 1 controls the average level of the interest rate when at the ZLB. It can be thought
of as the fraction of the steady-state net interest rate. If we set σZ = 0, ρR,Z = 0, and ψZ = 1,
we would obtain Rt = 1 at the ZLB and the net nominal interest rate would be exactly zero.3
The fiscal authority moves taxes in response to debt, spending, and the state of the
economy:
h h ii
∗ ∗
τet = 1 − Zξ d ρτ,ξtp τet−1 + 1 − ρτ,ξtp δb,ξtp bem
t−1 + δ ∗
e te
e + δ y (b
y t − y
b t )
t
+Zξd [ρτ,Z τet−1 + (1 − ρτ,Z ) [δe∗ ee∗t + δy (b
yt − ybt∗ )]] + στ ϵτ,t (4)
t
e ∗t + g −1 get , ϵτ,t ∼ N (0, 1) , and τet is the level of tax revenues with respect to
where ee∗t ≡ tr
GDP in deviations from the steady state. When the economy is in the high state of demand
(Zξd = 0), tax revenues respond to the state of the economy, captured by the output gap, to
t
the sum of the long-run level of transfers and government purchases, and to the level of debt.
The parameter δb,ξtp captures the strength with which the fiscal authority engages in debt
stabilization. This, in turn, depends on the type of policy mix ξtp in place at time t. When
the large negative preference shock hits (Zξd = 1), the fiscal authority temporarily disregards
t
the level of debt to focus on stabilizing the economy. However, we still allow for the fiscal
authority to respond to the level of spending.
2.2 Policy changes and beliefs
To characterize changes in policymakers’ behavior out of the ZLB, we build on the partition
of the parameter space introduced by Leeper (1991). For the sake of exposition, we will
assume that the Taylor rule reacts only to inflation, whereas the fiscal rule reacts only to
debt. In this simplified version of the model, we can distinguish four regions based on the
properties of the model under fixed coefficients (i.e., without regime changes). When the values
of model parameters are fixed, the two policy rules are key in determining the existence and
uniqueness of a solution. There are two determinacy regions with very distinct characteristics.
In the first region, Active Monetary/Passive Fiscal (AM/PF), the Taylor principle is satisfied
and the fiscal authority moves taxes to keep debt on a stable path: ψπAM > 1 and δbP F >
β −1 −1. Fiscal policy is dubbed passive because it passively accommodates the behavior of the
monetary authority by implementing the necessary fiscal adjustments. We refer to this policy
combination as the Monetary-led policy mix. Under the Monetary-led policy mix, absent
regime changes and distortionary taxation, the macroeconomy is insulated with respect to
fiscal imbalances.
3
Our approach to modeling the ZLB differs from the conventional one, which implies Rt = max (0, Rt∗ ) ,
where Rt∗ is the interest rate implied by the Taylor rule. Our approach is computationally more tractable and
allows us to focus on the consequences of changes in the policy mix and in beliefs about the future policy mix.
10
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
The second determinacy region, Passive Monetary/Active Fiscal (PM/AF), corresponds
to the case in which the fiscal authority is not committed to stabilizing the process for debt:
δbAF < β −1 − 1. Now it is the monetary authority that passively accommodates the behavior
of the fiscal authority, disregarding the Taylor principle and allowing inflation to move in
order to stabilize the process for debt: ψπP M < 1. Under this regime, even in the absence
of distortionary taxation, shocks to net taxes can have an impact on the macroeconomy as
agents understand that they will not be followed by future offsetting changes in the fiscal
variables. Therefore, the macroeconomcy is not insulated with respect to fiscal imbalances.
We label this policy combination the Fiscally-led regime.
In the third region, both authorities follow passive rules. Absent regime changes, this area
of the parameter space leads to multiple solutions. Finally, when both authorities are active,
no stationary equilibrium exists if the policy mix is in place forever. This is because this
policy mix leads to explosive dynamics for debt and inflation. However, when introducing
regime changes, a solution might still exist as long as the lack of coordination between the two
authorities does not persist for too long. We will use this policy combination to understand
whether the Federal Reserve could have prevented the recent increase in inflation by simply
increasing rates.
In the benchmark model, when the preference shock is high (ξtd = h), the economy is out
of the ZLB (Zξd = 0) and the evolution of policymakers’ behavior is captured by a two-regime
t
Markov chain that evolves according to the transition matrix H p :
" #
p pM M 1 − pF F
H = ,
1 − pM M pF F
p
= j|ξtp = i . This transition matrix is supposed to capture the stochastic
where pji = P ξt+1
outcome of a game between the monetary and fiscal authorities that is not explicitly modeled
in this paper. Regime M is the Monetary-led regime: ψπ,M = ψπAM > 1 and δb,M = δbP F >
β −1 − 1. Regime F is the Fiscally-led regime: ψπ,F = ψπP M < 1 and δb,F = δbAM < β −1 − 1.
When the low value for the preference shock occurs (ξtd = l), the ZLB becomes bind-
ing (Zξd = 1), and policymakers’ behavior is now constrained. In this third regime, the
t
central bank lowers the interest rate until it hits the ZLB and the fiscal authority temporar-
ily disregards the level of debt in order to focus on stabilizing the economy. Notice that the
zero-lower-bound policy mix can be considered as an extreme version of the Fiscally-led policy
mix. However, while out of the ZLB, switches to the Fiscally-led regime capture the deliberate
choices of policymakers, the zero-lower-bound regime is triggered by an exogenous negative
preference shock that prompts the fiscal authority to forgo fiscal adjustments to counter the
effects of a deep recession. Once the preference shock is back to its high value, policymakers’
behavior is not constrained anymore.
11
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Thus, the occurrence of a large contractionary shock imposes a constraint on policymakers’
behavior. However, agents’ beliefs are not constrained. Therefore, beliefs about the future
policy behavior after the ZLB are key to understanding the behavior of the macroeconomy at
the ZLB. To model agents’ beliefs at the ZLB, we introduce a set of parameters controlling
the expected exit strategy from the ZLB. The parameter pZi M represents the probability that
once the discrete preference shock is reabsorbed, the economy will move to the the Monetary-
led regime, under beliefs i. We allow for two possible beliefs, resulting into two distinct
parameters, pZ1 M and pZ2 M . Thus, we allow for two possible ZLB regimes that only differ
with respect to the probability assigned to moving to the Monetary-led policy mix once out
of the ZLB.
We are going to estimate these two parameters. To facilitate the exposition, and without
loss of generality, we are going to label them pZF M and pZM M , (instead of pZ1 M and pZ2 M ).
Our estimates will show that pZM M > pZF M , implying that under regime ZM , a return to the
Monetary-led policy mix is perceived as more likely than under regime ZF . Note that this is a
result and not an assumption, because we are going to have the same priors on the two ZLB
regimes.
To summarize, the joint evolution of policymakers’ behavior and the discrete preference
shock is controlled by the combined chain ξt ≡ ξtp , ξtd = {[M, h] , [F, h] , [ZM , l] , [ZF , l]}. The
corresponding transition matrix H is obtained by combining the transition matrices H d and
H p with the parameters pZM M and pZF M :
" #
pZM M pZF M
phh H p (1 − pll )
1 −"pZM M #1 − pZF M
H= " # .
.5 .5 1 0
(1 − phh ) pll
.5 .5 0 1
2.3 Solving the MS-DSGE model
The model presents trend growth. The model variables are then rescaled with respect to the
unit root technology process At and the model equations are linearized around the determin-
istic steady state. The model can be solved with any of the solution methods developed for
Markov–switching DSGE models. We use the solution method of Farmer, Waggoner, and Zha
(2009). This solution algorithm takes into account that agents form expectations while aware
of the possibility of entering the ZLB and of changes in policymakers’ behavior. Furthermore,
agents understand that entering the ZLB is an event induced by an exogenous shock that
can modify policymakers’ behavior even once the constraint stops being binding. Thus, our
framework allows us to model recurrent zero-lower-bound episodes and to capture the impact
of different beliefs about policymakers’ exit strategy from the ZLB.
12
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Subsample Policy regime Volatility regime
1955:Q4-1957:Q1 Monetary-led (M ) Pre-pandemic volatility
1957:Q2-1981:Q3 Fiscally-led (F ) Pre-pandemic volatility
1981:Q4-2008:Q3 Monetary-led (M ) Pre-pandemic volatility
2008:Q4-2015:Q4 ZLB low prob of F (ZM ) Pre-pandemic volatility
2016:Q1-2020:Q1 Monetary-led (M ) Pre-pandemic volatility
2020:Q2-2020:Q4 ZLB low prob of F (ZM ) Pandemic volatility
2021:Q1-2022:Q1 ZLB high prob of F (ZF ) Pandemic volatility
Table 1 – Regime sequence. The table reports the regime sequences for the policy mix and stochastic
volatilities.
The solution assumes the form of a MS-VAR with cross equation restrictions in the DSGE
state vector St :
St = C (ξtp , θ, H) + T (ξtp , θ, H) St−1 + R (ξtp , θ, H) Q (ξtv , θv ) εt (5)
where θ, θv , and St are vectors that contain the structural parameters, the stochastic volatil-
ities, and all the variables of the model, respectively. We allow for a break in the volatility of
all innovations to accommodate the large shocks observed during the pandemic. Appendix A
provides more details about the linearization and the solution algorithm.
In the solution above, the law of motion of the model depends on the structural param-
eters (θ), the policy regime in place (ξtp ), and the probability of moving across regimes (H).
This notation highlights that the properties of one regime depend not only on the structural
parameters describing that particular regime, but also on what agents expect is going to hap-
pen under alternative regimes and on how likely it is that a regime change will occur in the
future. In other words, the law of motion governing the evolution of the economy depends on
agents’ beliefs about future regime changes.
3 Fiscal Inflation
The model solution is combined with a system of observation equations. We estimate the
model with Bayesian methods. We include seven observables spanning the sample period
1954:Q4-2022:Q1: real GDP growth, annualized GDP deflator inflation, annualized federal
funds rate (FFR), annualized debt-to-GDP ratio, federal tax revenues to GDP ratio, federal
expenditure to GDP ratio, and a transformation of the government purchases to GDP ratio.
All variables are expressed in percentage points when reporting the results.
We fix the regime sequence for the out-of-the-zero-lower-bound regimes based on the VAR
evidence presented in Bianchi and Melosi (2017). Specifically, we assume that the Fiscally-led
regime was in place from 1957:Q2 to 1981:Q3, while the Monetary-led regime was in place
during the remaining part of the sample except for the ZLB periods. Based on the behavior of
13
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
the FFR, we assume that the economy was at the ZLB a first time from 2008:Q4 to 2015:Q4
and then a second time in correspondence of the COVID Pandemic from 2020:Q2 to the end of
the sample. In our baseline estimation, we allow for a change in beliefs about the future policy
mix in correspondence of the ARPA fiscal shock in 2021:Q1. This implies that, formally, a
regime change occurred at that date, but this regime change did not alter current policy rules;
the only change was in the expected future behavior of policymakers.4 Our discussion will
then focus on the change in the probability of moving to the Fiscally-led policy mix. Finally,
we allow for a break in the volatility of all innovations starting from 2020:Q1 to accommodate
the large shocks observed during the pandemic. The sequences for the policy regimes and the
stochastic volatility regimes are summarized in Table 1.
Before moving to present the results, it is useful to mention the importance of estimating
the model over the whole sample, as opposed to focusing exclusively on the two zero-lower-
bound episodes. The properties of the Fiscally-led regime are mostly identified by the study
of the effects of fiscal imbalances during the 1960s and 1970s.5 Similarly, the properties of the
Monetary-led regime are mostly pinned down by the behavior of the economy between the
post-Volcker disinflation period and the first occurrence of the ZLB. For given properties of
the Monetary-led and Fiscally-led regimes, the parameters pZM M and pZF M are informed by
the effects of fiscal imbalances on the macroeconomy at the ZLB. Thus, these key parameters
are identified by the joint dynamics of fiscal variables, inflation, and real activity during the
ZLB period, taking into account how the economy would respond if agents were to expect
one or the other regime to occur after the ZLB.
3.1 Parameter estimates
We report priors and posteriors for the parameters in Table 2 and Table 3. The priors for the
parameters that do not move across regimes are diffuse and in line with previous contributions
in the literature. As for the parameters of the Taylor rule, the prior for the autoregressive
component is symmetric across regimes, whereas the priors for the responses to inflation
and the output gap differ across regimes to be consistent with the properties of the parameter
space discussed above: Under the Monetary-led regime (M) monetary policy is active, whereas
under the Fiscally-led regime (F), monetary policy is passive. In a similar way, the priors for
the response of taxes to government debt are asymmetric across the two regimes: Under the
Fiscally-led regime and the ZLB regime, this parameter is restricted to zero (δb,F = 0) and
4
We also considered a version of the model in which this change in beliefs at the ZLB occurs as soon as the
economy encounters the ZLB for the second time, i.e. on 2020:Q1. However, this second version of the model
returns a worse fit.
5
The discussion here is simplified to the extent that the properties of one regime depend on the properties of
the other regimes because agents take into account the possibility of regime changes. Furthermore, the model
features feedback effects from the macroeconomy to the fiscal variables.
14
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Mode Mean Median 5 percent 95 percent Type Mean SD
ψπ,M 2.1441 2.1996 2.2003 1.6966 2.6862 N 2.5000 0.3000
ψy,M 0.5539 0.5631 0.5465 0.3471 0.8454 G 0.4000 0.2000
ρR,M 0.8826 0.8829 0.8847 0.8396 0.9195 B 0.5000 0.2000
ρτ,M 0.9780 0.9766 0.9782 0.9573 0.9900 B 0.5000 0.2000
ψπ,F 0.7009 0.7727 0.7559 0.5906 1.0151 G 0.8000 0.3000
ψy,F 0.2806 0.3074 0.3047 0.2340 0.3893 G 0.1500 0.1000
ρR,F 0.6474 0.6688 0.6694 0.5906 0.7431 B 0.5000 0.2000
ρτ,F 0.6070 0.6392 0.6427 0.5090 0.7581 B 0.5000 0.2000
δb,M 0.0493 0.0489 0.0471 0.0261 0.0781 G 0.0700 0.0200
δb,F 0.0000 - - - - F 0.0000 -
dl -0.1756 -0.1733 -0.1762 -0.2206 -0.1141 N -0.3000 0.1000
ψZ 0.9720 0.9715 0.9717 0.9648 0.9778 B 0.9500 0.0200
phh 0.9955 0.9932 0.9953 0.9822 0.9978 D 0.9615 0.0264
pll 0.8640 0.8072 0.8255 0.6343 0.9173 D 0.8333 0.1034
pM M 0.9992 0.9918 0.9973 0.9660 0.9995 D 0.9615 0.0264
pF F 0.9978 0.9907 0.9962 0.9619 0.9991 D 0.9615 0.0264
pZM M 0.9834 0.8599 0.9439 0.3889 0.9904 D 0.5000 0.2236
pZF M 0.7107 0.5363 0.5589 0.1660 0.8469 D 0.5000 0.2236
Table 2 – Prior and posterior moments of policy parameters. The table reports the prior
specification and the mode, mean, median and 5 and 95 percentiles of the posterior distribution of the
model parameters. Subscripts M and F denote B the parameter under the Monetary-led regime and
the Fiscally-led regime, respectively. B stands for Beta distribution; G stands for Gamma distribution;
N stands for Normal distribution; IG stands for Inverse gamma; D stands for Dirichlet distribution;
and F denotes a parameter that is fixed before estimation at the value shown in the column Mean.
not reported in the table, whereas under the Monetary-led regime (δb,M = δb ) it is expected
to be large enough to guarantee stability of debt. We restrict the persistence of the long-
term component of transfers (ρeL = .99) and choose a relatively tight prior on the standard
deviation of its innovations. We fix the discount factor β to .9985, a value broadly consistent
with an annualized 2% real interest rate, and the average maturity to 5 years (this is controlled
by the parameter ρ). The capital share, α is equal to 0.33. Furthermore, we assume that the
persistence of the fiscal rule at the ZLB coincides with its value in the Fiscally-led regime:
ρτ,Z = ρτ,F .
Since we fixed the regime sequence, the estimates of the transition matrix are determined
by the model dynamics across the different regimes and not by the frequency of regime changes.
We choose priors for the persistence of the Monetary-led and Fiscally-led regimes that reflect
the prolonged periods spent under these two regimes. We use a loose and symmetric prior
for the parameters pZM M and pZF M , which capture the two possible beliefs about the policy
mix that will be carried out after the liftoff of the interest rate from the ZLB. The fact that
15
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Mode Mean Median 5 percent 95 percent Type Param 1 Param 2
κ 0.0064 0.0060 0.0059 0.0033 0.0086 G 0.3000 0.1500
ρtr 0.2171 0.2193 0.2178 0.1474 0.2990 B 0.2000 0.0500
δy 0.3560 0.3946 0.3896 0.3067 0.5020 N 0.4000 0.2000
Φ 0.8531 0.8596 0.8599 0.8295 0.8880 B 0.5000 0.2000
δe∗ 0.3492 0.4319 0.4196 0.2561 0.6504 B 0.5000 0.2000
ρg 0.9799 0.9854 0.9840 0.9766 0.9960 B 0.5000 0.2000
ρa 0.5251 0.4377 0.4454 0.1936 0.6484 B 0.5000 0.2000
ρd 0.4275 0.4056 0.4067 0.2938 0.5111 B 0.5000 0.2000
ρtp 0.3488 0.3432 0.3428 0.2491 0.4401 B 0.5000 0.2000
ρµ 0.6208 0.6751 0.6784 0.6044 0.7332 B 0.5000 0.2000
π∗ 0.5396 0.5309 0.5302 0.4542 0.6093 G 0.5000 0.0500
γ 0.3986 0.3948 0.3946 0.3337 0.4593 G 0.4000 0.0500
b∗ 1.0580 1.0493 1.0507 0.9095 1.1843 N 1 0.1000
g∗ 1.0872 1.0868 1.0871 1.0717 1.1010 N 1.0600 0.0400
τ∗ 0.1705 0.1707 0.1707 0.1654 0.1765 N 0.1800 0.0100
ϕy -0.2149 -0.2194 -0.2189 -0.2780 -0.1610 N -0.4000 0.2000
σR,l 0.1779 0.1794 0.1792 0.1657 0.1945 IG 2.9454e-05 2.5891
σg,l 0.2752 0.2743 0.2739 0.2547 0.2956 IG 1.1782e-04 2.5891
σa,l 0.6196 0.7571 0.7337 0.4810 1.1161 IG 1.1782e-04 2.5891
στ,l 0.4038 0.4073 0.4065 0.3778 0.4396 IG 4.7126e-04 2.5891
σd,l 7.0012 7.3907 7.3067 6.2160 8.8046 IG 0.1320 14.6969
σtr,l 0.4229 0.4243 0.4239 0.3855 0.4645 IG 4.7126e-04 2.5891
σtp,l 3.0335 3.0372 3.0291 2.8203 3.2778 IG 1.1782e-04 2.5891
σµ,l 0.1152 0.1042 0.1033 0.0862 0.1252 IG 2.9454e-05 2.5891
σtr∗ ,l 0.2548 0.2642 0.2626 0.2227 0.3107 IG 2.7189e-06 4.1751
σR,h 0.2717 0.3073 0.2943 0.2108 0.4505 IG 2.9454e-05 2.5891
σg,h 0.6591 0.7543 0.7222 0.5231 1.0960 IG 1.1782e-04 2.5891
σa,h 0.5663 0.9544 0.7331 0.4022 1.9290 IG 1.1782e-04 2.5891
στ,h 0.8961 1.0566 0.9989 0.6962 1.6071 IG 4.7126e-04 2.5891
σd,h 25.9036 28.6652 28.0400 21.4327 38.1862 IG 0.5282 14.6969
σtr,h 8.2725 9.1263 8.8623 6.6632 12.3372 IG 0.0019 2.5891
σtp,h 8.7961 10.2272 9.8000 6.8195 15.0638 IG 1.1782e-04 2.5891
σµ,h 0.1819 0.2122 0.2029 0.1434 0.3135 IG 2.9454e-05 2.5891
σtr∗ ,h 0.0725 0.1010 0.0881 0.0537 0.1904 IG 2.7189e-06 4.1751
Table 3 – Prior and posterior moments of the other parameters. The table reports the prior
specification and the mode, mean, median and 5 and 95 percentiles of the posterior distribution of the
model parameters. The subscript l and h denote the pre-pandemic and pandemic volatility regimes,
respectively. B stands for Beta distribution; G stands for Gamma distribution; N stands for Normal
distribution; IG stands for Inverse gamma; and D stands for Dirichlet distribution. For the Inverse
gamma prior distribution Param 1 and Param2 denote the shape and the scale parameters, respectively.
For all the other prior distributions, Param 1 and Param 2 denote the mean and the standard deviation,
respectively.
16
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
the priors for both parameters are centered on .5 implies that we maintain an agnostic view
with respect to which exit strategy agents regard as most likely during the first and second
ZLB episodes.
Regarding the parameters of the Taylor rule, under the Monetary-led regime the interest
rate reacts strongly to both inflation and the output gap. The opposite occurs under the
Fiscally-led regime. Under the Fiscally-led and ZLB regimes the response of taxes to debt is
restricted to zero, while under the Monetary-led regime it turns out to be significantly larger
than the threshold value described in Subsection 2.2 (β −1 − 1 = .0015).
Both the Monetary-led regime and the Fiscally-led regime are quite persistent, implying
that when one of the two regimes is in place, agents expect to spend a significant amount of
time under that regime. The persistence of the high state for the discrete preference shock
is also high. This implies that when out of the ZLB, agents attach a small weight to the
possibility of a large contraction in real activity deriving from the negative preference shock.
This result is consistent with the fact that before the 2008 financial crisis, the U.S. economy
had always been able to avoid the ZLB.
When at the ZLB, under the first set of beliefs, agents regard it as quite likely that once the
negative preference shock is reabsorbed, policymakers will move to the Monetary-led regime
(pZM M = 98% at the posterior mode). However, this probability is smaller than the estimated
persistence of the Monetary-led regime (pM M is around 99%). Furthermore, the persistence
of the low state of demand is also lower than the persistence of the high state of demand
(86% vs 99%). Therefore, when the economy entered the first ZLB episode, the probability
attached to switching to the Fiscally-led regime experienced a sizable increase. However,
the probability of moving back to a Monetary-led regime declined significantly following the
ARPA shock (pZF M = 71% at the posterior mode). As we shall see, this large change in
beliefs had important consequences for inflation and real activity during the recovery from
the COVID pandemic.
3.2 Fiscal policy and historical inflation dynamics
As a first step, we want to understand the relative importance of fiscal shocks in accounting
for the historical evolution of inflation. Figure 1 presents the contribution of shocks to long-
term transfers and cost-push shocks to inflation dynamics. In the figure, the orange dashed
line corresponds to inflation, while the blue solid line corresponds to what inflation would
have been if only the shock of interest had occurred, keeping the policy and beliefs regime
sequences as given. To facilitate the interpretation of the results, the gray areas denote the
different policy regimes in place. Note that toward of the end of the sample, during the second
ZLB occurrence, we distinguish between the ZLB regime with low probability of moving to a
Fiscally-led regime (ZM ) and the ZLB regime with high probability of moving to a Fiscally-led
17
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Long-term tranfers Cost-push shocks
12 12
Monetary-led
10 10 Fiscally-led
ZLB: Low prob. of F
8 8 ZLB: High prob. of F
Inflation
6 6
Shock contribution
4 4
2 2
0 0
-2 -2
-4 -4
1960 1970 1980 1990 2000 2010 2020 1960 1970 1980 1990 2000 2010 2020
Figure 1 – Fiscal inflation and cost-push shocks. The figure reports the contribution of shocks to the
long-term component of transfers and cost-push shocks to inflation dynamics over the sample 1954:Q4-2022:Q1.
The orange dashed line corresponds to inflation, while the blue solid line corresponds to what inflation would
have been if only the shock of interest had occurred, keeping the policy and beliefs regime sequences as given.
The gray areas denote the different policy regimes in place.
regime (ZF ). These two regimes only differ with respect to the probabilities assigned to the
two exit strategies.
A first important result is that cost-push shocks account for the high-frequency movements
of inflation, while long-term transfer shocks account for the low-frequency movements of
inflation. The low-frequency component of inflation started increasing in the mid-1960s when
the Great Society initiatives, a series of large, long-term welfare programs, were introduced.
These shocks to the long-term component of spending, combined with a Fiscally-led policy
mix, led to an increase in trend inflation. The change in the policy mix in the early 1980s
corresponded with a quick decline in inflation. As we shall see below, this decline is the
result of changes in both monetary and fiscal policy, not just monetary policy. After this
quick decline in inflation, the fiscal component of inflation remains fairly stable until the
zero-lower-bound period because of the prevailing policy mix.
Cost-push shocks account for the spikes in inflation in the late 1960s, 1974, and at the end
of the 1970s, but not for the break in trend inflation. In fact, the high frequency component of
inflation has not changed dramatically between the pre- and post-Volcker period, while trend
inflation has a clear break (Stock and Watson, 2007). As we will explain in more detail below,
this is due to the different propagation of fiscal shocks under a Fiscally-led and Monetary-led
policy mix.
When moving to the post-2008 period, we see that cost-push shocks tend to lower inflation
until the onset of the COVID pandemic, when we observe a sharp increase in the level of
inflation attributed to these shocks. Over the same years, the increase in the probability of
moving to the Fiscally-led regime estimated for the ZLB periods contributes positively to
18
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Inflation Debt-to-GDP Change under Counterfactual
6
Actual Output
10 80
Counterfactual Real interest rate
70 4
8
6 60
2
4 50
40 0
2
0 30
-2
Actual
20
-2 Counterfactual
-4
1960 1980 2000 2020 1960 1980 2000 2020 1960 1980 2000 2020
Figure 2 – Counterfactual simulation: Always Monetary-led policy mix. This figure compares the
realized (Actual) historical behavior of the economy with a counterfactual scenario in which policymakers
always follow the Monetary-led policy mix, except when at the zero lower bound. Agents’ beliefs are adjusted
accordingly: Agents believe that when out of the ZLB, only the Monetary-led policy mix is possible. In the first
two panels, the orange dotted line and blue solid line correspond to the Actual and Counterfactual scenarios,
respectively. The third and last panel of the figure reports the change in output and real interest rates under
the counterfactual scenario. This is computed as the difference between the counterfactual and actual variables.
The sample is 1954:Q4-2022:Q1. The scale is percentage points.
inflation. During the first ZLB period, this effect is relatively modest and arguably helps the
central bank to remain closer to the desired 2% target, counteracting the deflationary pressure
of the cost-push shocks. However, following the ARPA shock, the jump in the probability of
moving to a Fiscally-led policy mix determines a large increase in the component of inflation
due to long-term transfers. Note that what drives inflation up is not so much the fiscal shock
itself, but the change in beliefs about the future policy mix. In light of the unprecedented
fiscal interventions implemented in response to the pandemic, it is conceivable that agents
revised their beliefs about the future monetary/fiscal policy mix.
To further illustrate the role of the monetary/fiscal policy mix and agents’ beliefs in
accounting for macroeconomic dynamics, we consider a counterfactual simulation in which
policymakers always follow the Monetary-led policy mix, except when encountering the zero
lower bound. Specifically, we modify the baseline model to have the same probability of high
and low discrete demand shocks, but we assume that when in the high state of demand, policy-
makers always follow the Monetary-led policy mix and agents form expectations accordingly.
When the low-demand discrete shock hits, the economy moves to the ZLB, but agents know
that once the economy moves back to the high state of demand, policymakers will return to
the Monetary-led policy mix. This counterfactual is designed to capture a situation in which
there is no doubt about the resolve of the fiscal authority to stabilize debt.
We then simulate both the actual model and the counterfactual model using the estimated
sequence of shocks and we compare the simulated series of inflation and debt-to-GDP implied
19
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
by the two models in Figure 2.6 The first two panels of Figure 2 report the evolution of
inflation and debt in the data (Actual) and in the counterfactual scenario. The last panel
of the figure reports the change in output and real interest rates under the counterfactual
scenario. This change is computed as the difference between the counterfactual and actual
variables. We define the real interest rate as the difference between the FFR and inflation, so
this rate is precisely pinned by the observables used in estimation.
In line with the shock decomposition presented above, if policymakers had always followed
the Monetary-led policy mix, inflation would have been remarkably lower in the 1960s and
1970s. We still observe the spikes in inflation in correspondence of the oil shocks, but trend
inflation remains low. Consistent with this finding, real interest rates would have been higher,
while output would have been lower. Higher real interest rates and lower output would have
made the debt-to-GDP ratio higher during the middle part of the sample, despite the fact
that under the counterfactual scenario the fiscal authority is increasing taxes to lower debt.
This last result highlights that the low debt and the high inflation of the 1970s are the
two sides of the same coin. A large increase in fiscal spending generated inflationary pres-
sure that was accommodated by the central bank, causing low real interest rates and higher
output, contributing to keeping the debt-to-GDP ratio low. Debt started increasing in the
1980s, when the change in the policy mix determined an increase in real interest rates and
a drop in inflation and output. In the aftermath of the Volcker disinflation, we see that the
counterfactual scenario presents slightly lower inflation and higher output. This is because
in the counterfactual scenario, agents exclude the possibility of a change to the Fiscally-led
regime, while in the baseline model they do not. Thus, in the baseline model, the possibility
of a return to the Fiscally-led policy mix generates some modest inflationary pressure, that
the central bank counteracts. This generates a small output loss under the baseline scenario.
Moving to the more recent period, the discrepancy between the actual and counterfactual
scenarios becomes large again. First, inflation would have been visibly lower during the
two zero lower bound episodes. The difference in the inflation paths is primarily due to
the increase in the probability of moving away from the Monetary-led policy mix and the
large accumulation of debt during the last two recessions. This small level of fiscal inflation
partially offsets the deflationary bias of the 2010s, when average inflation failed to hone in on
the central bank’s two-percent target (Amano, Carter, and Leduc, 2019; Hills, Nakata, and
Schmidt, 2019; Bianchi, Melosi, and Rottner, 2021)
This difference widens significantly once agents’ beliefs react to the large change in spend-
ing following the ARPA shock. During this period, inflation would have been approximately
4% lower under the counterfactual scenario. However, the output loss would also have been
large: 3.7% at the end of the sample. Thus, the model attributes a large part of the post-
6
The simulated series from the actual model are identical to the data used to estimate the model.
20
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
pandemic recovery and increase in inflation to the change in the perceived probability of
moving away from the Monetary-led policy mix. Fiscal inflation contributes to the rebound
in the real economy. A similar pattern was already at work during the first ZLB episode, but
the scale was different, because the probability attributed to moving to a Fiscally-led policy
mix was lower.
Summarizing, the results presented in this section suggest that in a country such as the
United States that issues debt in its own currency, fiscal policy has a natural limit. A stable
and low level of inflation can be achieved only under an appropriate fiscal arrangement. When
the monetary authority’s objective of price stability is not backed by the fiscal authority, large
increases in government spending lead to large and persistent increases in inflation. In the
1960s and the 1970s, a sequence of shocks to the long-term component of spending, combined
with a waning fiscal backing, explains the Great Inflation. The Great Inflation ended when
the whole monetary/fiscal policy mix changed. Under the new monetary and fiscal framework,
the fiscal authority was credibly committed to stabilize the federal debt and the monetary
authority gained its independence in pursuing price stability. Fiscal inflation provided a useful
buffer to avert deflation after the Great Recession and fueled the post-pandemic recovery, but
this came at the cost of a large increase in inflation in the post-pandemic period.
It is worth emphasizing that the findings presented in this historical analysis of U.S.
inflation do not depend on the particular New Keynesian framework used in the empirical
analysis or on the way that we model changes in the monetary/fiscal policy interaction.
In Bianchi, Faccini, and Melosi (2022), we take a different modeling approach by allowing
for unfunded fiscal shocks in an otherwise standard medium-size New Keynesian model to
provide a fiscal theory of trend inflation. Unlike funded fiscal shocks, unfunded shocks are
not expected to be stabilized with future fiscal adjustments and produce movements in trend
inflation. We also consider distortionary taxation and hand-to-mouth consumers to break
Ricardian equivalence even when the fiscal authority is fully committed to stabilize debt. This
richer fiscal structure allows us to take into account that a fiscal stimulus can temporarily
push output above its potential level, generating inflationary pressure, even when agents are
certain to remain in a Monetary-led policy mix.
Despite the different modeling strategy, we reach similar conclusions with respect to the
role of fiscal shocks in accounting for persistent movements in inflation. We find that move-
ments in trend inflation are driven by the amount of unfunded debt, as opposed to funded
fiscal shocks, even in presence of hand-to-mouth consumers. In this respect, our results echo
but differ from the arguments of Summers (2021) and Blanchard (2021) that stimulative fiscal
policy may end up overheating the economy. In line with what suggested by these scholars,
our analysis attributes a pivotal role to the ARPA fiscal stimulus in explaining the high infla-
tion readings of the past months. Nevertheless, we focus on a very different mechanism that
21
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
rests on the increase in trend inflation needed to stabilize a large and growing fiscal imbalance.
In the next section, we provide a brief excursus on the historical evolution of the U.S.
monetary and fiscal policy mix to substantiate the evolution of regimes estimated in the
literature and used in our analysis.
3.3 A brief narrative account
Bianchi and Ilut (2017) and Bianchi, Faccini, and Melosi (2020, 2022) provide detailed nar-
rative evidence to interpret the historical evolution of the U.S. monetary and fiscal policy
mix. To keep this paper self-contained, we briefly summarize the key steps of these historical
accounts. This exercise is useful to understand how to interpret the econometric results in
light of political and economic events.
The increase in spending which triggers the Great Inflation in the model started in the
mid-1960’s with the Great Society initiatives (cf. the left panel of Figure 1). These programs –
launched by President Johnson on May 7, 1964 – boosted current and long-term government
spending in education, medical care, welfare, social housing, environment, and transporta-
tion. The Great Society platform kept expanding throughout the 1970s under Republican
administrations.
At the same time, Federal Reserve Chairman Arthur Burns was pressured repeatedly by
the Johnson and Nixon administrations to keep interest rates low. During a conversation that
occurred on October 23, 1969, just after Burns’s nomination to the Fed had been announced,
President Nixon invited Burns “to see [him] privately anytime” and suggested communicating
through an intermediary in order to preserve “the myth of the autonomous Fed.” (Abrams,
2006). Levin and Taylor (2013) argue that this political interference by the fiscal authority
played a critical role in causing inflation to drift up in the 1960s and 1970s. These scholars also
document that Burns (1979) himself openly acknowledged these political pressures: “...the
central bank’s practical capacity for curbing an inflation that is driven by political forces is
very limited.” These accounts are consistent with the predominance of the Fiscally-led policy
mix regime in those years.
President Carter appointed Paul Volcker as Fed Chairman in August 1979. Chairman
Volcker immediately tried to quash inflation by tightening monetary policy, but this aggressive
action did not succeed initially. In fact, Goodfriend and King (2005) argue that “the start of
a deliberate disinflation dates to late 1980” and that this initial increase in the federal funds
rate did not represent a significant discontinuity from how the Federal Reserve had tried to
reduce inflation in the 1970s. This first attempt by Chairman Volcker was doomed by the
decision to accommodate the Carter Administration’s request to introduce credit controls on
March 14, 1980. This choice proved to be detrimental for two reasons. First, it forced the
Federal Reserve to lower the federal funds rate before seeing considerable pullback in price
22
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
dynamics. Second, it was yet another blow to the independence of the monetary authority.
As a result, the Federal Reserve had to start the effort all over again.
In the meantime, President Carter lost the presidential race against Reagan, who ran on a
strong anti-inflation platform. President Reagan provided the necessary support for fighting
inflation. The credit controls were removed in July 1980, and this time, Volcker kept interest
rates high for a prolonged period of time with no interference by the Reagan administration.
Later on, Paul Volcker said that “[u]nlike some of his predecessors, [President Reagan]
had a strong visceral aversion to inflation.” Despite the fact that the change in monetary
policy caused a large contraction in real activity, President Reagan refused to criticize the
Fed chairman publicly (Samuelson, 2008). Thanks to this support, Volcker was able to keep
interest rates high until inflation finally fell. This fiscal backing did not have to result in an
immediate increase in primary surpluses. This is mostly for two reasons. First, thanks to
the low rates of the 1960s and 1970s, the debt-to-GDP ratio was at an historical minimum.
Second, the recession induced by the change in the monetary/fiscal policy mix naturally
implied a drop in tax revenues and an increase in spending. What is required to reduce fiscal
inflation is that agents expect that primary surpluses will increase over time.
As debt started increasing due to the rise in real interest rates and the large spending
programs, the necessary fiscal adjustments were put into place. While the tax cut contained
in the Economic Recovery Tax Act of August 1981 led to an immediate fiscal deficit, it was
quickly followed by partially compensating deficit-reducing measures. These were aimed at
increasing tax revenues, either through higher tax rates or through expanding the tax base.
These changes included the Tax Equity and Fiscal Responsibility Act of 1982, the Social
Security Amendments of 1983, and the Deficit Reduction Act of 1984.
Romer and Romer (2009) also note that President Reagan was a strong advocate of spend-
ing reductions and that he viewed tax cuts as the most effective way to reduce the size of
the government, following the “starve-the-beast” hypothesis. Thus, the reforms also signaled
the intention of moving toward a small government, focused on some key functions, such as
national defense. Thus, President Reagan was strongly in favor of a reduction in the size of
the government, and, more importantly, not willing to allow inflation to be used as a tool for
fiscal adjustment.
All these accounts support the prediction of our model that the end of the Great Inflation
was the result of a change in both monetary and fiscal policy.
As shown above, fiscal inflation has not disappeared after the Volcker disinflation and
its presence is not always an issue. During the first zero lower bound episode, fiscal inflation
counteracted deflationary forces and prevented a more severe output loss. However, the recent
fiscal interventions might have caused a significant change in the perceived future policy mix.
In the next subsection, we will focus on this last event in more detail.
23
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
4 Conquering the post-pandemic inflation
In the previous section, we have shown that trend inflation is in large part a fiscal phenomenon
that can be understood in light of changes in the monetary/fiscal policy mix. While other
shocks can affect inflation in the short-run, in the long-run a low and stable inflation requires
a credible fiscal framework. In view of these considerations, it is interesting to ask under
which conditions policymakers can regain control of inflation in the post-pandemic era.
4.1 Could monetary policy alone have averted the post-pandemic inflation?
In the previous subsection, we have shown a decomposition of inflation between cost-push
shocks and long-term spending shocks. Both shocks played a role in the post-pandemic surge
in inflation. However, the counterfactual simulation presented above has highlighted that
while the cost push shock component of inflation appears to behave similarly even when
changing the perceived future policy mix, the effects of shocks to long-term fiscal spending
are heavily affected by this perception. Given that half of the recent surge in inflation has a
fiscal origin, a natural question to ask is whether monetary policy could have neutralized it
by stepping on the interest-rate brake more quickly and aggressively and, if so, at what cost.
In order to answer this question, we consider a counterfactual simulation in which we
modify the behavior of the monetary authority, while keeping agents’ beliefs about the future
policy mix unchanged.7 Specifically, we assume that right after the realization of the ARPA
stimulus shock and the associated change in beliefs about the future policy mix, the central
bank leaves the zero lower bound and increases the policy rate as it would do under the
Monetary-led policy mix. We assume that this behavior comes as a surprise for agents in the
economy. This allows us to have the same starting point in 2021:Q1 for the baseline model
and for this counterfactual model.
Figure 3 reports the results. The dotted orange line corresponds to the data (Actual sce-
nario), while the blue solid line is the counterfactual scenario in which the monetary authority
increases the interest rate as it would do under the Monetary-led policy mix, but without the
necessary fiscal backing and without a change in agents’ beliefs about the probability of mov-
ing to the Fiscally-led policy mix. We also report how the economy would have behaved under
the counterfactual policy behavior if no cost-push shocks had occurred starting from 2021:Q1
(yellow dashed line).
From this exercise, it is clear that even if the monetary authority had anticipated the
liftoff and tightened the interest rate aggressively, it would not have been able to suppress the
7
In Section 4.3, we will consider the case in which tightening monetary policy works as a signaling device
coordinating the beliefs of the private sector away from the Fiscally-led policy mix. In this exercise we abstract
from this signaling channel, to focus on the effects of a monetary tightening unbacked by a strong fiscal
commitment to stabilize government debt.
24
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Inflation FFR Output gap
2
Actual
3.5 Counterfactual
6 Counterf. & no cost-push shocks 0
3
-2
2.5
4
2 -4
2 1.5 -6
1 -8
0
0.5 -10
2017 2018 2019 2020 2021 2022 2017 2018 2019 2020 2021 2022 2017 2018 2019 2020 2021 2022
Figure 3 – Counterfactual simulation: Active monetary policy following the ARPA shock, but
no change in beliefs. The figure reports a counterfactual simulation in which, in response to the ARPA
shock in 2021:Q1, the central bank adopts the reaction function of the Monetary-led policy mix, but agents’
beliefs about the future policy mix are unchanged. This policy change comes as a surprise for agents in
the economy. The orange dashed line corresponds to the actual data, the blue solid line corresponds to the
counterfactual simulation, and the dashed yellow line corresponds to the counterfactual simulation without
cost-push shocks starting from 2021:Q1. Until 2021:Q1 the actual and counterfactual series coincide. The
sample is 2018:Q1-2022:Q1.
large increase in inflation. This is because half of the increase in inflation observed in 2021
has a fiscal nature (left plot of Figure 2) and it is driven by a shift in beliefs about how the
large stock of debt and the ARPA stimulus will be financed. The more hawkish monetary
policy would have lowered inflation by only 1 percentage point at the cost of reducing output
by around 3.4 percentage points. This is a quite large sacrifice ratio. This result is not driven
by the presence of cost-push shocks. When we completely remove cost-push shocks (dashed
yellow line), output barely changes while inflation remains elevated at around 4.6%.
This result is driven by agents’ beliefs. In this counterfactual scenario, we keep the fiscal
shocks and agents’ beliefs unchanged. The central bank is increasing the policy rate, but
agents still expect that the ARPA stimulus has increased the probability of moving to the
Fiscally-led policy mix. By increasing interest rates, the Federal Reserve is able to cool down
the economy because of nominal rigidities. However, the assumed monetary tightening alone
is not enough to significantly cut inflation because price dynamics are mostly driven by a
perceived lack of fiscal sustainability.
To sum up, we showed that anticipating the liftoff and swiftly raising the policy rate in
isolation would not have averted the post-pandemic surge in prices. In the next section, we
will show that the reasons behind this dismal outcome have to be found in the fiscal nature
of the recent rise in inflation.
25
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
4.2 Fiscal stagflation
We showed that keeping inflation low and stable may be quite challenging for a central bank
if the fiscal framework is not perceived to be consistent with debt stabilization. When this is
the case, fiscal shocks have a large and persistent effect on inflation. On the other hand, cost-
push shocks have relatively short-lasting effects on inflation. Our counterfactual simulations
suggest that these effects do not vary significantly in response to the current or expected
monetary and fiscal policy mix.
To formalize this point, we study a simplified version of the model in which the fiscal
framework may be perceived to be inconsistent with debt stabilization. We achieve this by
introducing an additional regime, which we call the Inconsistent fiscal framework regime. To
simplify the model, we assume that (i ) the Fiscally-led regime is an absorbing state, (ii )
discrete demand shocks are shut down (dξd ), and (iii ) the ZLB constraint never binds. In
t
the baseline case, the regime with the Inconsistent fiscal framework is assumed to be followed
by the Fiscally-led policy mix. The new regime is introduced to isolate the consequences of
adopting a resolute anti-inflation monetary policy stance without the necessary fiscal backing.
Hence, in that regime, we assume that the monetary authority responds to inflation as in the
Monetary-led regime (see Table 2) but does not respond to output and the last period’s
interest rate. The fiscal authority does not adjust taxes in response to changes in the debt-
to-GDP ratio in the Inconsistent fiscal framework regime. The monetary and fiscal policy
rules under the Monetary-led and Fiscally-Led policy regimes are identical to the posterior
mode – shown in Table 2. Analogously, the model parameters that are not affected by
changes in the monetary and fiscal policy mix are also calibrated to the posterior mode. A
more thorough description of how the simplified model with the Inconsistent fiscal framework
regime is calibrated is in Appendix C.
Figure 4 shows the impulse responses to a shock to long-term transfers (left panel) and to
a cost-push shock (right panel) under three policy regimes: Monetary-led, Fiscally-led, and
Inconsistent fiscal framework. The first result that emerges is that a long-term transfers shock
causes a large and persistent increase in inflation under the Fiscally-led policy mix. Agents
understand that the fiscal authority will not implement the necessary fiscal adjustments.
This creates inflationary pressures that are accommodated by the central bank. Under the
Monetary-led regime the same shock generates a much smaller increase in inflation. The
increase is not zero, because agents consider the possibility of a future regime change. Shocks
to long-term transfers do not dissipate quickly. Thus, even if policymakers are currently
committed to fiscal adjustments, agents know that the fiscal burden might result in high
inflation at some point in the future.
Under the third scenario – when the monetary authority responds to inflation as in the
Monetary-led regime but without fiscal backing – not only does inflation increase persistently,
26
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Long-term transfers Cost-push shocks
1 1 Monetary-led policy mix
Fiscally-led policy mix
Inflation
Inconstent fiscal framework
0.5 0.5
0 0
5 10 15 20 5 10 15 20
Figure 4 – Inflation persistence due to fiscal and cost-push shocks. The figures report the impulse
response function of inflation to a one-standard deviation long-term transfers shock (left panel) and to a one-
standard deviation cost-push shock (right panel) under the Monetary-led policy mix, the Fiscally-led policy
mix, and the absence of a consistent fiscal framework to achieve price stability (inconsistent fiscal framework).
Agents know the true transition probabilities of these three policy regimes and form their beliefs accordingly.
but it takes an explosive path. The shock to spending determines an increase in the fiscal
burden, causing inflationary pressures. The central bank reacts by increasing rates, causing a
recession. The recession and a higher cost of financing debt further increase the fiscal burden
that in turn generates additional inflationary pressure. In this case, not only is the central
bank unable to lower inflation, but by increasing rates it is causing even more inflation and
economic stagnation. We call this outcome fiscal stagflation.
In the data, the scenario is still not so grim. Agents still regard a change to a Monetary-
led policy mix as more likely. However, the large increase in the probability assigned to a
Fiscally-led policy mix explains why, in the counterfactual above, the increase in rates had
only a modest effect on inflation.
Finally, it is interesting to contrast these results with the impulse responses to a cost-push
shock. In this case, the policy mix in place is almost irrelevant. This is because, in the data,
cost-push shocks only imply transitory movements in inflation. Thus, it is entirely possible
that we will see a partial reduction in inflation in the months ahead as the effects of cost-push
shocks fade away. However, based on our analysis, the fiscal component of inflation is likely
to persist, unless the necessary fiscal backing is reinstated. In the next section, we study how
policymakers can deal with this fiscal component of inflation.
4.3 Reining in fiscal inflation and avoiding fiscal stagflation
We now tackle the issue of what policymakers can do to combat a rise in fiscal inflation in
our model. We first show that a more hawkish monetary policy unsupported by a credible
fiscal arrangement cannot resolve the problem of fiscal inflation and, in fact, only accelerates
inflationary dynamics and slows down economic activity even further. We then show that
27
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Inflation Output gap FFR
1.4
-0.5
1.3 4
1.2 -1
3.5
1.1
-1.5
1 3
0.9 -2
2.5
0.8
-2.5 Baseline
2
0.7 More hawksih monetary policy
Very hawkish monetary policy
-3
1.5
5 10 15 20 5 10 15 20 5 10 15 20
Figure 5 – Fiscal stagflation as the monetary authority becomes more hawkish. Impulse responses
of inflation, the output gap, and the federal funds rate (FFR) to a shock to long-term transfers as the central
bank becomes progressively more hawkish. Agents expect that eventually the economy will move to a Fiscally-
led policy mix and the fiscal authority is not currently implementing fiscal adjustments. The different lines
correspond to different values for the parameter controlling the response of the monetary authority to inflation,
ψπ . The blue solid line corresponds to the posterior mode estimate for ψπ,M in Table 2, the orange dashed
line increases the value of this parameter by 25%, while the yellow line increases the value of this parameter
by 50%.
a change in agents’ beliefs about the future policy mix is quite effective in quashing fiscal
inflation.
Figure 5 shows the response of inflation, the output gap, and the interest rate to a shock
to long-term transfers as the central bank becomes progressively more hawkish. However, we
assume that agents remain convinced that eventually the economy will move to a Fiscally-led
policy mix to stabilize the increase in debt resulting from the larger spending. The different
lines correspond to different values for the parameter controlling the response of the monetary
authority to inflation, ψπ . The blue solid line corresponds to the posterior mode estimate for
this parameter under the Monetary-led policy mix (ψπ,M ). Under the orange dashed line the
value of this parameter increases by 25%, while under the yellow line it increases by 50%.
As explained above, the increase in spending determines inflationary pressure because
agents expect that eventually the additional fiscal burden will be stabilized with inflation.
As the central bank increases the response to inflation, the initial jump in inflation becomes
smaller at the cost of generating a larger contraction in real activity. In addition, the success
on inflation is ephemeral. Eventually the paths of inflation cross and the more hawkish
response leads to higher inflation and larger output losses. The reason is as follows: The more
hawkish the monetary policy response, the larger the increase in the fiscal burden, the larger
the acceleration in inflation. Across all scenarios, fiscal stagflation persists as agents expect
that the increase in the fiscal burden will also contribute to generating future inflation.
This result is driven by the fact that while the response of the monetary authority to
28
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Inflation Output gap FFR
1 -0.2
2
-0.4
0.8
-0.6 1.5
0.6
-0.8
1
0.4 Sure Fiscally-led policy mix
-1 High prob. Fiscally-led policy mix
Low prob. Fiscal-led policy mix
5 10 15 20 5 10 15 20 5 10 15 20
Figure 6 – Fiscal stagflation as agents’ beliefs about the future policy mix vary. Impulse responses
of inflation, the output gap, and the federal funds rate (FFR) to a shock to long-term transfers as beliefs about
the future policy mix change. The central bank responds to inflation as implied by the Monetary-led policy
mix while the fiscal authority does not adjust taxes. The different lines correspond to different probabilities
that the next policy mix will be Fiscally-led; – the blue line: Fiscally-led policy mix is expected to follow
with certainty; orange dashed line: Fiscally-led policy mix is expected to follow with probability 0.67; yellow
dotted-dashed line: Fiscally-led policy mix is expected to follow with probability 0.33.
fiscal inflation becomes progressively more hawkish, agents’ beliefs about future policy remain
unchanged. The reason behind this apparently puzzling result is that by tightening monetary
policy, the monetary authority increases the service of the debt and depresses output. As a
result, the debt-to-GDP ratio increases. Agents expect that the debt will not be stabilized by
future fiscal adjustments, which leads to an increase in fiscal inflation. The monetary authority
raises the interest rate even more to respond to the rising fiscal inflation, exacerbating the
fiscal imbalance and further accelerating the price dynamics.
As the three lines show, this vicious spiral has progressively more pernicious effects on
price stability and the economy as the monetary authority intensifies its response to rising
inflation. This exercise serves as an important warning for policymakers: if fiscal policy is
perceived to be inconsistent with price stability, the monetary authority may fail in its attempt
to rein in inflation, paving the way to fiscal stagflation. Bianchi and Ilut (2017) argue that this
mechanism is behind the failed disinflation of the 1970s, and is the reason why Volcker did not
immediately succeed in bringing inflation down when appointed in August 1979. Monetary
policy changed earlier than fiscal policy. The appropriate fiscal backing came only in the early
1980s.
What can be done when policymakers are confronted with fiscal inflation? Our model
suggests that it is critical to reassure the private sector that the large stock of debt will be
repaid by raising taxes or cutting government expenditure. One way to do so is to announce
a credible plan indicating how the fiscal authority intends to stabilize the existing fiscal
imbalance. Reaffirming that the central bank has the backing of the fiscal authority in its quest
29
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
to reduce inflation also helps, but it is not sufficient if fiscal interventions are not perceived
as consistent with this goal.
In presence of fiscal inflation, tightening monetary policy can be successful only to the
extent that the resulting increase in the cost of servicing debt prompts the fiscal authority to
make the necessary fiscal adjustment, reinstating the credibility of the fiscal framework. In
other words, for monetary policy to be successful, agents need adjust their beliefs about the
future policy mix in response to the increase in interest rates. To illustrate the underlying
mechanism, we revisit the effects of long-term spending shocks when varying agents’ beliefs
about the future policy mix. We keep the response of the central bank to inflation fixed at
its value under the Monetary-led policy mix, but we progressively reduce the probability of a
change to the Fiscally-led policy mix. In this case, the monetary authority can considerably
mitigate and possibly escape the fiscal stagflation scenario.
Figure 6 presents the results. The blue line coincides with the blue solid line of Figure
5, given that we maintain the assumption that agents are certain that the fiscal authority
will not take the necessary fiscal adjustments to stabilize the public debt, and eventually the
monetary authority will revert to a passive policy. As we progressively reduce the probability
of a change to the Fiscally-led policy mix, two important effects arise. First, the inflationary
pressure goes down. Second, fiscal stagflation is mitigated. In the limit, if a change to the
Fiscally-led policy mix is completely ruled out, the inflationary pressure and fiscal stagflation
would almost completely disappear (not shown).8
Summarizing, the massive fiscal interventions implemented during the pandemic succeeded
in raising inflation – after a period of persistently shallow price dynamics and low interest
rates – and in providing quick support to the economy. However, going forward, a clearer
demarcation of the boundaries of fiscal policy might be needed to ensure low and stable
inflation.
5 Conclusions
Historically, movements in fiscal inflation account for changes in trend inflation, while cost-
push shocks determine more transitory movements. Thus, an implicit fiscal limit arises to
the extent that a low and stable inflation target requires fiscal policies that are consistent
with this goal. Absent this fiscal backing, the central bank cannot maintain inflation at the
desired target. We have argued that the rise in trend inflation in the 1960s and 1970s was
a fiscal phenomenon that ended when the monetary/fiscal policy mix changed in the early
1980s. Since then, fiscal inflation has remained modest because of the prevailing policy mix.
8
They would not completely disappear because there is still a probability – albeit a modest one – to move to
the Fiscally-led policy mix from the Monetary-led policy mix in every quarter. See the transition probabilities
in Appendix C.
30
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
In the decade before the pandemic, fiscal inflation has been counteracting deflationary forces
and, in fact, helped in preventing the U.S. economy from slipping into deflation.
Following the COVID pandemic, the United States, like many other countries, has imple-
mented robust fiscal interventions. We have shown that these policy interventions facilitated
the quick rebound observed after the pandemic recession. At the same time, they also con-
tributed to the surge in fiscal inflation. Increasing rates, by itself, would not have prevented
the recent surge in inflation, given that large part of the increase was due to a change in the
perceived policy mix. In fact, increasing rates without the appropriate fiscal backing could
result in fiscal stagflation. Instead, conquering the post-pandemic inflation requires mutually
consistent monetary and fiscal policies providing a clear path for both the desired inflation
rate and debt sustainability.
31
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
References
Abrams, B. A. (2006, December). How Richard Nixon pressured Arthur Burns: Evidence
from the Nixon tapes. Journal of Economic Perspectives 20 (4), 177–188.
Aiyagari, S. R. and M. Gertler (1985). The backing of government bonds and monetarism.
Journal of Monetary Economics 16 (1), 19–44.
Amano, R., T. J. Carter, and S. Leduc (2019). Precautionary pricing: the disinflationary
effects of elb risk. Federal reserve bank of san francisco mimeo.
Barro, R. J. (1974, Nov.-Dec.). Are Government Bonds Net Wealth? Journal of Political
Economy 82 (6), 1095–1117.
Bassetto, M. (2002). A Game-Theoretic View of the Fiscal Theory of the Price Level. Econo-
metrica 70 (6), 2167–2195.
Bassetto, M. and W. Cui (2018). The fiscal theory of the price level in a world of low interest
rates. Journal of Economic Dynamics and Control 89 (C), 5–22.
Bassetto, M. and T. J. Sargent (2020, August). Shotgun Wedding: Fiscal and Monetary
Policy. Annual Review of Economics 12 (1), 659–690.
Benhabib, J., S. Schmitt-Grohe, and M. Uribe (2002). Avoiding Liquidity Traps. Journal of
Political Economy 110 (3), 535–563.
Bianchi, F., R. Faccini, and L. Melosi (2020, May). Monetary and fiscal policies in times
of large debt: Unity is strength. NBER Working Paper Series 27112, National Bureau of
Economic Research.
Bianchi, F., R. Faccini, and L. Melosi (2022). A fiscal theory of trend inflation. mimeo.
Bianchi, F., J. D. M. Fisher, and L. Melosi (2021, April). Some inflation scenarios for the
American Rescue Plan Act of 2021. Chicago Fed Letter (453), 1–8.
Bianchi, F. and C. Ilut (2017). Monetary/Fiscal Policy Mix and Agent’s Beliefs. Review of
Economic Dynamics 26, 113–139.
Bianchi, F. and L. Melosi (2017). Escaping the Great Recession. The American Economic
Review 107 (4), 1030–1058.
Bianchi, F. and L. Melosi (2019). The dire effects of the lack of monetary and fiscal coordi-
nation. Journal of Monetary Economics 104, 1–22.
32
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Bianchi, F., L. Melosi, and M. Rottner (2021). Hitting the elusive inflation target. Journal
of Monetary Economics 124, 107–122.
Blanchard, O. (2021). In defense of concerns over the $1.9 trillion relief plan. PIIE opinion,
18 February 2021.
Burns, A. F. (1979). The Anguish of Central Banking. Per Jacobsson Lecture, Sava Centar
Complex, Belgrade, Yugoslavia.
Clarida, R., J. Galı́, and M. Gertler (2000). Monetary Policy Rules and Macroeconomic
Stability: Evidence and Some Theory. Quarterly Journal of Economics 115, 147–180.
Cochrane, J. H. (1998). A Frictionless Model of U.S. Inflation. In B. S. Bernanke and J. J.
Rotemberg (Eds.), NBER Macroeconomics Annual 1998, pp. 323–384. Cambridge, MA:
MIT Press.
Cochrane, J. H. (2001). Long Term Debt and Optimal Policy in the Fiscal Theory of Price
Level. Econometrica 69, 69–116.
Congressional Budget Office (2022). The budget and economic outlook: 2022 to 2032. May
2022, CBO Publication No. 57950. Washington, DC: Congressional Budget Office.
Davig, T. and E. M. Leeper (2006). Fluctuating Macro Policies and the Fiscal Theory. NBER
Macroeconomics Annual 2006 , 247–298.
Eusepi, S. and B. Preston (2012). Fiscal Foundations of Inflation: Imperfect Knowledge.
Working Paper.
Farmer, R. E., D. F. Waggoner, and T. Zha (2009). Understanding Markov-Switching Rational
Expectations Models. Journal of Economic Theory 144, 1849–1867.
Goodfriend, M. and R. King (2005). The Incredible Volcker Disinflation. Journal of Monetary
Economics 52 (5), 981–1015.
Hall, G. J. and T. J. Sargent (2011). Interest Rate Risk and Other Determinants of Post-
WWII U.S. Government Debt/GDP Dynamics. American Economic Journal: Macroeco-
nomics 3 (3), 192–214.
Hills, T. S., T. Nakata, and S. Schmidt (2019). Effective lower bound risk. European Economic
Review 120, 103321.
Leeper, E. M. (1991). Equilibria Under Active and Passive Monetary and Fiscal Policies.
Journal of Monetary Economics 27, 129–147.
33
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Leeper, E. M., N. Traum, and T. B. Walker (2017, August). Clearing Up the Fiscal Multiplier
Morass. American Economic Review 107 (8), 2409–2454.
Leeper, E. M. and X. Zhou (2021). Inflation’s role in optimal monetary-fiscal policy. Journal
of Monetary Economics 124.
Levin, A. and J. B. Taylor (2013, June). Falling Behind the Curve: A Positive Analysis of
Stop-Start Monetary Policies and the Great Inflation, pp. 217–244. University of Chicago
Press.
Loyo, E. (1999). Tight money paradox on the loose: A fiscalist hyperinflation. John f. kennedy
school of government harvard university mimeo.
Lubik, T. and F. Schorfheide (2004). Testing for Indeterminacy: An Application to U.S.
Monetary Policy. American Economic Review 94 (1), 190–217.
Meltzer, A. H. (2009). A History of the Federal Reserve. Chicago: University of Chicago
Press.
Reis, R. (2016). QE in the Future: the Central Bank’s Balance Sheet in a Fiscal Crisis.
Working Paper 22415, National Bureau of Economic Research.
Romer, C. and D. H. Romer (2009). Do Tax Cuts Starve the Beast? The Effect of Tax
Changes on Government Spending. Brookings Papers on Economic Activity.
Samuelson, R. J. (2008). The Great Inflation and its Aftermath: The Past and Future of
American Affluence. Random House Digital, Inc.
Sargent, T. and N. Wallace (1981). Some Unpleasant Monetarist Arithmetic. Federal Reserve
Bank of Minneapolis Quarterly Review Fall, 1–17.
Sargent, T. J. (2013). Letter to Another Brazilian Finance Minister. Republished in Rational
Expectations and Inflation, 3rd edition: Princeton: Princeton University Press.
Schmitt-Grohe, S. and M. Uribe (2000). Price level determinacy and monetary policy under
a balanced-budget requirement. Journal of Monetary Economics 45 (1), 211–246.
Sims, C. A. (1994). A Simple Model for Study of the Determination of the Price Level and
the Interaction of Monetary and Fiscal Policy. Economic Theory 4, 381–399.
Sims, C. A. (2011). Stepping on a Rake: The Role of Fiscal Policy in the Inflation of the
1970’s. European Economic Review 55 (1), 48–56.
34
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Sims, C. A. (2016). Fiscal policy, monetary policy and central bank independence. In Kansas
City Fed Jackson Hole Conference.
Stock, J. H. and M. W. Watson (2007). Why Has U.S. Inflation Become Harder to Forecast?
Journal of Money, Credit, and Banking 39, 3–34.
Summers, L. H. (2021). Opinion: The inflation risk is real. Washington Post, 24 May 2021.
Taylor, J. B. (2011). Review of Allan H. Meltzer’s A history of the Federal Reserve, VOlume
2, University of Chicago Press, 2009. Journal of Monetary Economics 58, 183–189.
Weise, C. L. (2012). Political Pressures on Monetary Policy During the US Great Inflation.
American Economic Journal: Macroeconomics 4 (2), 33–64.
Woodford, M. (1994). Monetary Policy and Price Level Determinacy in a Cash-in-Advance
Economy. Economic Theory 4, 345–389.
Woodford, M. (1995). Price Level Determinacy without Control of a Monetary Aggregate.
Carnegie-Rochester Series of Public Policy 43, 1–46.
Woodford, M. (2001). Fiscal Requirements of Price Stability. Journal of Money, Credit, and
Banking 33, 669–728.
Woodford, M. (2003). Interest and Prices: Foundations of a Theory of Monetary Policy.
Princeton, New Jersey: Princeton University Press.
35
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Appendix
A Benchmark Model
In what follows, we provide the details for the solution and estimation of the model.
A.1 System of equations
1. Linearized Euler equation:
h i
1 + ΦMa−1 ybt = − 1 − ΦMa−1 R
bt − Et π
et+1 − (1 − ρd ) dt − dξd + Eξd dξd
t t t+1
− ΦMa−1 − ρa at + Et ybt+1 + 1 − ρg + Ma−1 Φ get
+Ma−1 Φ (b
yt−1 − get−1 )
where Ma = exp (γ) and dξd follows a Markov-switching process governed by the transi-
t
tion matrix H d . Please refer to the next subsection for details about how to handle the
discrete shock.
2. New Keynesian Phillips curve:
h i
1 α 1
1−ΦMA −1 + 1−α y bt − 1−ΦM −1 get
π
et = κ ΦM −1
A
yt−1 − get−1 − at )
− 1−ΦMA −1 (b
A
+βEt [e
πt+1 ] + µ
et
υ
where we have used the rescaled markup µ
et = κ 1−υ υ
et
3. No arbitrage condition h i
R
et = Et Rem
t,t+1
4. Return long term bond
R m
et−1,t = R−1 ρPetm − Pet−1
m
5. Government budget constraint:
ebm = β −1ebm + bm β −1 Rbm − ybt + ybt−1 − at − π
et
t t−1 t−1,t
−e e t + g −1 get + tp
τt + tr et
1
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
6. Monetary policy rule
h ih i
∗
R
et = 1 − Zξd ρR,ξtp Ret−1 + (1 − ρ p ) ψ p π
R,ξt π,ξt et + ψ p [b
y,ξt y t − y
b t ] + σ R ϵ R,t
t
h i
+Zξd ρR,Z R et−1 − (1 − ρR,Z ) ψZ log (R) + σZ ϵR,t
t
7. Fiscal rule
h ∗ i
−1
τet = ρτ,ξtp τet−1 + 1 − ρτ,ξtp δb,ξtpebm
t−1 + δe∗ tr t + g g
e yt − ybt∗ ) + στ ϵτ,t
et + δy (b
8. Transfers
e ∗t = ρtr tr e ∗t + (1 − ρtr ) ϕy (b
e t − tr
tr e t−1 − tr yt − ybt∗ ) + σtr ϵtr,t , ϵtr,t ∼ N (0, 1)
9. Long-term component of transfers
e ∗t = ρtr∗ tr
tr e ∗t−1 + σtr∗ ϵtr∗ ,t , ϵtr∗ ,t ∼ N (0, 1)
10. Government purchases (e
gt = ln(gt /g)):
get = ρg get−1 + σg ϵg,t , ϵg,t ∼ N (0, 1) .
11. TFP growth
at = ρa at−1 + σa εa,t
12. Term premium
tpt = ρtp tpt−1 + σtp εtp,t
13. The rescaled markup µt = κ log (ℵt /ℵ) , where ℵt = 1/ (1 − υt ) , follows an autoregressive
process,
µt = ρµ µt−1 + σµ ϵµ,t
14. Output target
ΦMa−1
1 α ∗ 1 ∗
+ y t = g t + y t−1 − gt−1 − a t
1 − ΦMa−1 1 − α 1 − ΦMa−1 1 − ΦMa−1
b e b e
A.2 Model solution
As explained in the main text, the Markov-switching process for the discrete preference shock
dξd is defined in a way that its steady state is equal to zero. In order to solve the model we
t
2
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
implement the following steps:
1. Introduce a variable eξd , always equal to 1, controlling the regime that is in place for the
t
discrete preference shock. Augment the DSGE state vector with this variable, loading
the value of the discrete shock in place at each point in time.
2. Use the aforementioned dummy variable to rewrite all the equations linked to the discrete
preference shock. These are the linearized Euler equation and the linearized Taylor rule.
3. Solve the model using Farmer, Waggoner, and Zha (2009). This returns a MS-VAR:
Set = Te (ξtp , θ, H) Set−1 + R
e (ξ p , θ, H) Q (ξ v , θv ) εt
t t
in the augmented state vector Set .
4. Extract the column corresponding to the dummy variable eξd from the matrix Te and
t
redefine the matrices and the DSGE state vector accordingly. This will return a MS-
VAR with a MS constant:
St = C (ξtp , θ, H) + T (ξtp , θ, H) St−1 + R (ξtp , θ, H) Q (ξtv , θv ) εt
where Q is a diagonal matrix that contains the standard deviations of the structural
shocks and St is a vector with all variables of the model.
Our model allows for non-orthogonality between policymakers’ behavior and a discrete
shock. This allows us to solve a model in which agents take into account that a large preference
shock leads to an immediate change in policy, the zero lower bound, and, potentially, to further
changes. This proposed method is general and can be applied to other cases in which a shock
induces a change in the structural parameters.
A.3 Matrices used in the counterfactual simulations
We here describe the matrices used in the simulations reported in the paper.
A.3.1 Always Monetary-led
In the first counterfactual simulation, policymakers always follow the Monetary-led regime
when out of the zero lower bound. Furthermore, there is only one zero-lower-bound regime
from which agents expect to return to the Monetary-led regime. Therefore, the transition
3
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
matrix used to solve this counterfactual economy is given by:
" #
p d phh 1 − pll
H = 1, H = , H c = H d.
1 − phh pll
where phh and pll are the estimated parameter values.
A.3.2 Active monetary policy following the ARPA shock
In this counterfactual economy, once agents’ beliefs change following the ARPA shock, the
central bank lifts from the ZLB and moves to active monetary policy. However, beliefs are
assumed to be unchanged, meaning that agents still expect that a change to a Fiscally-led
policy mix is now more likely. We further assume that the response of the central bank to the
increase in inflation comes as a surprise. This assumption allows us to simulate the economy
with the same starting point observed in the data as 2021:Q1. If agents could anticipate
this counterfactual behavior of the central bank, the pre-shock period would also be affected.
These assumptions boil down to the following transition matrix:
" #
pZM M pZF M pZC M
phh Hp (1 − pll )
1 − pZM M 1 − pZF M 1 − pZC M
H cc =
.5 .5 0 1 0 0 .
(1 − phh ) .5 .5 0 pll 0 1 0
0 0 0 0 0 1
where ZC denotes the counterfactual regime in which monetary policy coincides with the
Monetary-led Taylor rule, but fiscal policy does not change with respect to the ZLB regimes.
Furthermore, pZC M = pZF M , implying that agents still regard the possibility of a movement
to a Fiscally-led policy mix quite likely.
B Estimation of the DSGE model
This appendix describes the dataset and provides details for the benchmark model.
B.1 Dataset
Real GDP, the GDP deflator, and the series for fiscal variables are obtained from the Bureau
of Economic Analysis. The fiscal series are built using NIPA Table 3.2. (Federal Government
Current Receipts and Expenditures). Government purchases (G) are computed as the sum
4
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
of consumption expenditure (L24), gross government investment (L44), net purchases of non-
produced assets (L46), minus consumption of fixed capital (L47). Transfers are given by the
sum of net current transfer payments (L25-L18), subsidies (L35), and net capital transfers
(L45-L41). Tax revenues are given by the difference between current receipts (L40) and current
transfer receipts (L18). All variables are then expressed as a fraction of GDP. Government
purchases are transformed in a way to obtain the variable gt defined in the model. The series
for the federal funds rate is obtained averaging monthly figures downloaded from the St. Louis
Fed web-site.
B.2 MCMC algorithm and convergence
Draws from the posterior are obtained using a standard Metropolis-Hastings algorithm ini-
tialized around the posterior mode. When working with models whose posterior distribution
is very complicated in shape it is very important to find the posterior mode. In a MS-DSGE
model, this search can turn out to be an extremely time-consuming task, but it is a necessary
step to reduce the risk of the algorithm getting stuck in a local peak. Here are the key steps
of the Metropolis-Hastings algorithm:
• Step 1: Draw a new set of parameters from the proposal distribution: ϑ ∼ N θn−1 , cΣ
• Step 2: Compute α (θm ; ϑ) = min p (ϑ) /p θm−1 , 1 where p (θ) is the posterior eval-
uated at θ.
• Step 3: Accept the new parameter and set θm = ϑ if u < α (θm ; ϑ) where u ∼ U ([0, 1]),
otherwise set θm = θm−1
• Step 4: If m ≤ nsim , stop. Otherwise, go back to step 1
The diagonal matrix Σ is obtained rescaling the posterior mode parameter values with
the parameter c set to obtain an acceptance rate of around 35%. We verify convergence with
the Brooks-Gelman-Rubin potential reduction scale factor using within and between variances
based on three multiple chains used in the paper. The three chains consist of 2, 100, 000 draws
each (1 in every 3000 draws is saved).
B.3 Determining the time of the change in beliefs at the ZLB
For tractability, we fix the sequence of Markov-switching regimes to estimate the model. The
chosen regime sequence is largely based on results obtained in previous work (Bianchi and
Melosi, 2017). For the most recent period, we allow for a regime change while the economy
is still at the ZLB. This regime change involves agents’ beliefs as opposed to a change in the
policy mix. This change in beliefs coincide with the first quarter of 2021, in the aftermath of
5
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
Change in beliefs at the ZLB Likelihood Posterior
2020:Q2 7,328.6 7,386.5
2021:Q1 7,330.5 7,388.5
Table 4 – The table shows the value of the posterior and the likelihood at the posterior mode under
two alternative assumptions about when agents’ beliefs about the future policy mix changed. The
results associated with the highest posterior mode are in bold.
the ARPA shock. We verified that the model returns a worse fit when the change in beliefs
is assumed to coincide with the beginning of the second ZLB episode.
C The Model with the Inconsistent fiscal framework regime
We study a simplified version of the estimated model in which the fiscal framework may
be perceived to be inconsistent with debt stabilization. We achieve this by introducing an
additional regime, which we call the Inconsistent fiscal framework regime. To simplify the
model, we assume that (i ) the Fiscally-led regime is an absorbing state, (ii ) discrete demand
shocks are shut down (dξd ), and (iii ) the ZLB constraint never binds. In the baseline case,
t
the regime with inconsistent fiscal framework is assumed to be followed by the Fiscally-led
policy mix. The new regime is introduced to isolate the consequences of adopting a resolute
anti-inflation monetary policy stance without the necessary fiscal backing. Hence, in that
regime, we assume that the monetary authority responds to inflation as in the Monetary-
led regime (see Table 2) but does not respond to output and the last period’s interest rate.
The quarterly probability that agents start seeing the fiscal framework as inconsistent is one
percent. The Inconsistent fiscal framework regime is assumed to have an expected duration of
twenty months. The probability of moving from the Monetary-led policy mix to the Fiscally-
led policy mix is the same as the estimated one (0.0008). The fiscal authority is committed
to respond to the debt-to-output ratio only in the Monetary-led regime. The monetary and
fiscal policy rules under the Monetary-led and Fiscally-Led policy regimes are identical to the
posterior mode – shown in Table 2. The model parameters that are not affected by changes
in the monetary and fiscal policy mix are also calibrated to the posterior mode.
Assuming that regimes are ordered as follows (1) Monetary-led, (2) Fiscally-led, and (3)
Inconsistent fiscal framework, the transition matrix H reads as follows:
" #
1 − pCF 0.9892 0 0
pM C H̃ p (1 − pCC )
H= pCF = 0.0008 1.0000 0.1500 ,
(1 − pM C ) pCC 0.0100 0 0.8500
where PM C denotes the quarterly probability that the fiscal framework remains consistent
(0.99), PCC stands for the probability that the Incompatible fiscal framework regime will
6
Embargoed until presentation time of 8:00 a.m. Mountain Daylight Time, August 27, 2022
stay in place in the next quarter (0.85), and pCF denotes the quarterly probability that the
Incompatible fiscal framework regime will be followed by the Fiscally-led regime, which is
one in the baseline case. In constructing Figure 6 in the paper, we progressively lower this
parameter to show that fiscal stagflation dissipates as the probability that the Incompatible
fiscal framework regime will be followed by the Fiscally-led regime is reduced. H̃ p is the
transition matrix governing the evolution of the Monetary-led and Fiscally-led policy mixes.
That matrix is calibrated using the estimated transition of the matrix H p (Table 2) defined
in the main text after imposing the restriction that the Fiscally-led led regime is an absorbing
state; that is,
" # " #
pM M 0 0.9992 0
H̃ p = = .
1 − pM M 1 0.0008 1
7
File and source
- File
- inflation-as-a-fiscal-limit.pdf
- Size
- 1,345,646 bytes
- SHA-256
- 8667f15426b5550dfeca60deab13b9551e00bd76a17b059af35840c13b61f219
- Our copy
- inflation-as-a-fiscal-limit.pdf
- Original
- www.kansascityfed.org