Quantifying the Inflationary Impact of Fiscal Stimulus Under Supply Constraints (No. 1050, February 2023)

Summary

Federal Reserve Bank of New York Staff Reports, no. 1050, Quantifying the Inflationary Impact of Fiscal Stimulus under Supply Constraints, by Julian di Giovanni, Şebnem Kalemli-Özcan, Alvaro Silva, and Muhammed A. Yıldırım, dated February 2023. The abstract states that the paper quantifies the contribution of fiscal policy to U.S. inflation over the December 2019-June 2022 period, and that model calibrations show aggregate demand shocks explain roughly two-thirds of total model-based inflation, with fiscal stimulus contributing half or more of the total aggregate demand effect. The introduction states that headline inflation reached 9 percent per annum at its peak in June 2022 and that the 2021 Biden fiscal package totaled 15% of GDP. It then describes the model.

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

About this paper. The authors' model attributes roughly two-thirds of model-based inflation from December 2019 to June 2022 to aggregate demand shocks, and half or more of that demand effect to fiscal stimulus. In this model fiscal support is a large contributor to inflation, not the only cause. Blanchard and Bernanke, in What Caused the US Pandemic-Era Inflation?, find that most of the inflation surge that began in 2021 "was the result of shocks to prices given wages," shocks that included sharp increases in commodity prices reflecting strong aggregate demand.

Issuer: Source documents
Document type: Quantifying The Inflationary Impact Of Fiscal Stimulus Under Supply Co
Case: Quantifying The Inflationary Impact Of Fiscal Stimulus Under Supply Co

Cite this document

“Quantifying the Inflationary Impact of Fiscal Stimulus Under Supply Constraints (No. 1050, February 2023).” Original at www.newyorkfed.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/quantifying-the-inflationary-impact-of-fiscal-stimulus-under-supply-co/, SHA-256 a054dc37f0c858c5452c87d7e8d42f92080602e095be165f73b93212d3bf93a4.

Full text

Quantifying the Inflationary
NO. 1050
FEBRUARY 2023 Impact of Fiscal Stimulus
Under Supply Constraints
Julian di Giovanni | Şebnem Kalemli-Özcan | Alvaro Silva |
Muhammed A. Yıldırım
Quantifying the Inflationary Impact of Fiscal Stimulus under Supply Constraints
Julian di Giovanni, Şebnem Kalemli-Özcan, Alvaro Silva, and Muhammed A. Yıldırım
Federal Reserve Bank of New York Staff Reports, no. 1050
February 2023
JEL classification: E23, E31, E32

Abstract

This paper builds on Baqaee and Farhi (2022) and di Giovanni et al. (2022) to quantify the
contribution of fiscal policy to U.S. inflation over the December 2019-June 2022 period. Model
calibrations show that aggregate demand shocks explain roughly two-thirds of total model-based
inflation, and that the fiscal stimulus contributed half or more of the total aggregate demand effect.

Key words: inflation, fiscal policy, supply constraints

_________________

Giovanni: Federal Reserve Bank of New York (email: julian.digiovanni@ny.frb.org). Kalemli-Özcan,
Silva: University of Maryland (emails: kalemli@umd.edu, asilvub@umd.edu). Yıldırım: Harvard
University, Koç University (email: muhammed_yildirim@hks.harvard.edu).

This paper presents preliminary findings and is being distributed to economists and other interested
readers solely to stimulate discussion and elicit comments. The views expressed in this paper are those of
the author(s) and do not necessarily reflect the position of the Federal Reserve Bank of New York or the
Federal Reserve System. Any errors or omissions are the responsibility of the author(s).
To view the authors’ disclosure statements, visit
https://www.newyorkfed.org/research/staff_reports/sr1050.html.
1 Introduction

U.S. headline inflation has hit levels not seen for several decades, reaching 9 percent per annum at
its peak in June 2022, before declining to approximately 7 percent per annum by the end of 2022.
In contrast, inflation was below 2 percent before the 2020 COVID-19 pandemic.
A priority that has been at the top of the minds of both policymakers and academics alike
has been to quantify the relative importance of the key factors in driving the observed inflation,
particularly the relative importance of supply bottlenecks vs. consumer demand, as the U.S. and
world economies struggled with supply-demand imbalances arising from the COVID-19 health shock
combined with stimulative policies.
The literature thus far has found differing results, ranging from one-third to two-thirds con-
tributions from supply factors (with the remaining being demand). Shapiro (2022a,b) takes an
econometric approach while di Giovanni et al. (2022) and Ferrante, Graves and Iacoviello (2022)
use quantiative models.
Though these papers provide important early evidence on the different channels that drove the
surge in inflation, none of them take a stand on the inflationary impact of specific policy actions.
In particular, the 2021 Biden fiscal package totaled 15% of GDP and has been blamed by some for
today’s high inflation (Blanchard, Domash and Summers, 2022).
In this paper, we explicitly measure the impact of the fiscal stimulus on inflation over the Dec-
2019 to June-2022 period. We follow our previous work and use the framework developed in Baqaee
and Farhi (2022) in order to quantify the impact of different shocks on inflation. Importantly, unlike
our previous work, we feed aggregate demand shocks into the model that vary depending on whether
the fiscal impulse is included or not. Doing so allows us to (1) quantify the impact of aggregate
demand in driving inflation, and (2) run a counterfactual scenario that omits observed government
spending as part of the aggregate demand shock. This second scenario allows us to gauge the
importance of the fiscal package’s impact on inflation.
Our baseline results show that over the Dec19-Jun22 period, aggregate demand shocks explained
roughly two-thirds of total model-based inflation, and that the fiscal stimulus contributed half
or more of the total aggregate demand effect. The range for the impact of fiscal stimulus vary
depending on how we detrend the data in constructing the empirical shock series. Since the fiscal
packages came in a discrete fashion as bursts of government spending, such sensitivity is expected.
Section 2 presents a brief description of the model. Section 3 describes the data and methodology
we use to construct the shocks that we feed into the model. Section 4 presents the main results.

1
2 Model

We build on previous work (di Giovanni et al., 2022) to quantify the sources of inflation using a
multisector macro-network model in the spirit of Baqaee and Farhi (2022).

Inter-temporal Allocation. There are two periods: the first period corresponds to the pandemic and
the second one represents the post-pandemic (i.e., the future). We denote the future quantities with
an asterix (∗) in the subscript. There are two types of consumers. Ricardian consumers optimize
their budget across two periods to smooth out their consumption such that their intertemporal
consumption decisions optimize:
y β y∗1−β ,

where β captures the Ricardian consumers’ time preferences. We assume we are at the zero-lower-
bound for the interest rate. Hence, household spending and income are related with each other:

I + I∗ = py y + py∗ y∗ .

Hand-to-mouth consumers, on the other hand, cannot borrow against their future income and
spend only their current income. The share of Ricardian consumers is denoted by ϕ.

Within-Period Consumption. We assume that there are N sectors. Within each period, the con-
sumers allocate their budgets across the sectors with a Cobb-Douglas utility:
N
X
ln y = αi δi ln ci , (1)
i=1

where ci is the consumption in sector i, αi is the consumption share during the non-Covid period
such that N
P
i=1 αi = 1, and δi is the shift in sectoral consumption during the pandemic such that
PN
i=1 αi δi = 1.

Production. Each sector i uses the intermediate inputs from other sectors (input from sector j to
sector i is denoted by xij ), sector specific labor (Li ) and sector specific capital Ki . The output of
sector i (yi ) is given by:
"
γ−1 γ−1
γ θ−1
γ−1 θ
γ γ
yi = ωiL Li + ωiK Ki
θ
  ε θ−1  θ−1
N ε−1 θ
X ε−1
+ ωij x ε  , (2)

ij 
j=1

where ωiL (ωiK ) determines the labor (capital) share, and ωij captures the intermediate input
shares. ε dictates the inter-industry substitution between inputs, γ controls the substitution be-
tween labor and capital and θ determines the substitution between the factors and input bundle.

2
Equilibrium. For normalization purposes, we take py∗ = 1 and y∗ = 1. The equilibrium is achieved
through adjustment of prices, wages and rental rents of capital such that good markets clear
(yi = ci + nj=1 xji ), capital markets clear (Ki = Ki∗ ), producers maximize their profits and
P

consumers optimize their consumption.
For labor, during pandemic, some workers are unable to work due to COVID-related reasons.
Let’s denote the pre- (post-) pandemic level of labor in industry i with Li∗ . During pandemic,
number of available workers in industry i shrinks to L̄i ≤ Li∗ . Moreover, the workers will not
accept a wage below their pre-pandemic levels. Denoting the wage of workers in industry i with
wi , the wage levels satisfy wi ≥ wi∗ , i.e., wages don’t go below their equilibrium levels absent the
pandemic.

3 Data
3.1 Detrending methods

We implement two detrending procedures to estimate the shocks the model requires. The model
needs sectoral demand and supply shocks and an aggregate demand shock. In the first procedure,
for sectoral shocks at monthly frequency, we compute the average annual growth rate between
2015-2019 for sectoral total hours worked and sectoral consumption expenditure for each of the 66
sectors separately. For quarterly nominal GDP, we do the same for the period 2010-2019. Then
for each sector for consumption and labor, and for aggregate nominal GDP, we take the deviations
from these constant average growth rates during our analysis period to get at our shocks.
The second procedure estimates the following linear regression for each time series Yt , at sector
or at aggregate level:
ln Yt = β0 + β1 t + εt ,

where β0 and β1 are estimated parameters, t is a linear-trend, and εt is an error term. We then
compute the trend variable as
Ybt = βb0 + βb1 t.

The shocks we feed in are then the residuals:

shockt = ln Yt − ln Ybt .

To get a sense of how these detrending procedures look like in practice, Figure 1 plots these trends
for three aggregate time series together with actual data. Panels (a) and (b) plot the aggregate
demand shock, nominal GDP and nominal GDP without government expenditure, respectively,
while panel (c) plots headline inflation. The solid blue lines denote the raw data, the gray dotted
lines denote the constant annual growth trend, and the blue dashed lines denote the log-linear trend.
As can be seen, both methods deliver similar patterns for the three aggregate time series. For the

3
given by CPI from December 2019 and June 2022 was 14.35 percent. The model predicts something
close to this number: 13.17 percent under constant-growth detrending, and 14.18 percent under
log-linear detrending1 . Sub-figures (a) and (c) use nominal GDP as an aggregate demand shock
measure, while in sub-figures (b) and (d), we subtract total government expenditure from nominal
GDP. Sectoral demand and supply shocks are as described above.
As expected, the model delivers higher inflation when feeding in nominal GDP as an aggregate
demand shock relative to the model that excludes government expenditure. The aggregate demand
shocks (orange bars) generate by themselves roughly two-thirds of the total model-based inflation
(blue bars) in figures (a) and (c). Removing government expenditures in figures (b) and (d) drops
the contribution of aggregate demand shocks considerably. Regardless of the detrending method,
aggregate demand explains two-thirds of the model-based inflation when we include government
stimulus. When we exclude government expenditure from nominal GDP, aggregate demand explains
at most half of the model-based inflation, while sectoral supply shocks and sectoral demand shocks
explain the rest (purple and yellow bars, respectively). These latter shocks contribute non-trivially
to aggregate inflation; importantly, their absolute magnitude is not affected when government
expenditure is dropped from the aggregate shock.
These results assume all households are Ricardian, that is ϕ = 1. Figure 4 presents results when
we allow thirty percent (ϕ = 0.7) of the population to be hand-to-mouth consumers. Results are
similar to the Ricardian model, except now predicted inflation is lower. Why is this? Remember
that the model allows for the possibility of unemployment. When consumers are Ricardian and
become temporarily unemployed, their consumption is unaffected as they can substitute future
consumption for current consumption. In contrast, when hand-to-mouth consumers become tem-
porarily unemployed, they reduce their demand for goods in the economy, as they have no income
and no possibility of borrowing. As a result, any shock that causes unemployment now has a lower
effect on prices as hand-to-mouth consumers lose their income, pushing demand down and, given
supply, also prices. This mechanism is precisely what Figure 4 shows: both sectoral demand and
sectoral supply shocks have lower inflationary effects in the hand-to-mouth scenario relative to
the Ricardian scenario. Aggregate demand, in contrast, exhibits the same magnitudes as before.
Recall that in the model, an aggregate demand shock works through intertemporal substitution:
consumers substitute away from future consumption towards current consumption for given prices
and income. Since all good and factor prices are flexible upwards, an increase in aggregate demand
maps one-to-one to increases in good prices, ultimately resulting in inflation. However, the sectoral
demand and supply shocks will impact inflation via the hand-to-mouth consumer constraint as
these shocks will create some unemployment. This can seen by the different impact of these shocks
1
Our model gives results as deviations from trend. To compare these results to actual inflation, we add the trend
under each detrending method to the model’s results, the numbers in the blue bar of panel (a) and (c) in Figure 3,
respectively. Trend was 4.65 percent with constant-growth detrending and 4.86 under log-linear detrending.

5
References

Baqaee, David, and Emmanuel Farhi. 2022. “Supply and Demand in Disaggregated Key-
nesian Economies with an Application to the Covid-19 Crisis.” American Economic Review,
112(5): 1397–1436.

Blanchard, Olivier, Alex Domash, and Lawrence Summers. 2022. “Bad News for the Fed
from the Beveridge Space.” PIIE Policy Brief 22-7.

di Giovanni, Julian, Şebnem Kalemli-Özcan, Alvaro Silva, and Muhammed A.
Yıildıirıim. 2022. “Global Supply Chain Pressures, International Trade, and Inflation.” NBER
Working Paper 30240.

Ferrante, Francesco, Sebastian Graves, and Matteo Iacoviello. 2022. “The Inflationary
Effects of Sectoral Reallocation.” Mimeo, Federal Reserve Board of Governors.

Shapiro, Adam H. 2022a. “Decomposing Supply and Demand Driven Inflation.” FRBSF Working
Paper.

Shapiro, Adam H. 2022b. “How Much Do Supply and Demand Drive Inflation?” FRBSF Eco-
nomic Letter.

File and source

File: quantifying-the-inflationary-impact-of-fiscal-stimulus-under-supply-co.pdf
Size: 1,029,130 bytes
SHA-256: a054dc37f0c858c5452c87d7e8d42f92080602e095be165f73b93212d3bf93a4
Our copy: quantifying-the-inflationary-impact-of-fiscal-stimulus-under-supply-co.pdf
Original: www.newyorkfed.org