View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

O PTIMAL M ACROECONOMIC
P OLICIES IN A H ETEROGENEOUS
W ORLD
James Bullard
Federal Reserve Bank of St. Louis
Riccardo DiCecio
Federal Reserve Bank of St. Louis
Aarti Singh
University of Sydney
Jacek Suda
Narodowy Bank Polski (NBP)

2023 IJCB Conference
International Journal of Central Banking and Central Bank
of Ireland
June 23, 2023
Dublin, Ireland
Any views expressed here are our own and do not necessarily reflect those of the FOMC or the NBP.

1

Introduction

2

B ENCHMARK MODELS

The macroeconomics world has benefited from having simple
benchmark models from which fruitful discussions can be
launched.
1980s-1990s: real business cycle models—Prescott (QR, 1986).
2000s-present: New Keynesian models—Woodford (2003).
Today, we want to be more granular in order to be able to discuss
how macroeconomic policies impact different portions of the
income distribution.
But what would optimal macroeconomic policy look like in a
heterogeneous world?

3

A

BENCHMARK MODEL

We study a simple (paper-and-pencil solution) benchmark DSGE
heterogeneous-agent life-cycle model.
The equilibrium features Gini coefficients approaching those in
the U.S. data.
The model is meant to be complementary to models in the
HANK literature.
The subtext in this talk is that models in this class represent,
broadly, the current and future direction of macroeconomics.

The model features three aggregate shocks as well as permanent
and temporary idiosyncratic risk.
Macroeconomic policymakers in the model have tools to counter
the frictions in the economy.
A welfare theorem states the sense in which these policies can
achieve an optimal allocation of resources.

4

M ODEL - RECOMMENDED MACRO POLICIES
The model equilibrium recommends the following “four
horsemen” macroeconomic monetary-fiscal policy mix:
As in the NK model, the monetary authority should react to
aggregate shocks each period and strive to achieve the
“Wicksellian natural real rate of interest”; roughly, “easier”
monetary policy in periods of low growth and “tighter” monetary
policy in periods of high growth.
The treasury authority should issue nominal debt and roll it over at
the current nominal interest rate in perpetuity.
The labor market authority should run an unemployment insurance
program.
The fiscal authority should use a redistributive tax-transfer scheme
to lower the consumption Gini coefficient.

5

A

MAPPING TO ACTUAL MACRO POLICIES ?

These model-recommended macroeconomic policies seem to
correspond, broadly speaking, to actual macroeconomic policies
in place in many economies around the world today, including
the U.S.
By itself, this finding suggests that current observed
macroeconomic policy is broadly on the right track in many
countries.

6

A

CALIBRATED CASE
We consider a calibration of the model using U.S. data from 1995
to 2023.
We assume that actual observed U.S. macroeconomic policy during
this period has essentially been an implementation of the optimal
macroeconomic policies recommended by the model.

We compare the calibrated model equilibrium to the data on six
dimensions: (1) aggregate consumption growth correlations, (2)
distributional consumption growth correlations, (3)
Heckman-style labor supply, (4) marginal propensities to
consume, (5) tax progressivity and Gini coefficients, and (6)
nominal returns to asset holders.
We argue that the fit to the data is generally good.
This suggests that actual U.S. macro policy has been close to
optimal during this period.
The model fits less well during periods of very large shocks,
such as the GFC or the onset of the pandemic.
7

Environment

8

E NVIRONMENT BASICS
At each date t, a new continuum of households enters the
economy, makes economic decisions over T + 1 = 241 dates,
then exits the economy.
To fix ideas, think of

1m agents per quarterly cohort.

This corresponds to an agent entering the economy as a
decision-maker at age 20 and exiting as a decision-maker at age
80, inclusive of end points, and making economic decisions at a
quarterly frequency.
Results are perfectly general for the choice of T, with higher
values corresponding to decision-making at more frequent
intervals.
We make enough assumptions to generate a paper-and-pencil
equilibrium solution, and thus provide a simple benchmark
model for heterogeneous-agent macroeconomies with aggregate
shocks.

9

A SSETS

There are three nominally denominated assets: privately issued
debt, publicly issued debt and privately issued nominal claims to
capital.
We think of these as abstract representations of U.S. data
counterparts: (1) mortgage-backed securities, (2) federally issued
debt and (3) corporate debt, respectively.
In the U.S. data, MBS net out, but federally issued debt and
claims to capital are in positive net supply.
We think of firms as being bond-financed and push physical
capital into the background.

10

N OMINAL CONTRACTING

The credit market friction is non-state contingent nominal
contracting: All debt contracts are stated in nominal terms, with
a stated nominal interest rate, and repayment is not
state-contingent.
The role of monetary policy is to adjust the price level each
period in order to convert these nominal, non-state contingent
contracts into real, state-contingent contracts.

11

H OUSEHOLD TYPES

Household types: “life cycle” (LC) and “hand-to-mouth”
(HTM).
The life-cycle households are assigned a hump-shaped
productivity profile at the beginning of their life cycle.
Accordingly, they need to use credit markets (hold assets) to
smooth life-cycle consumption.
The hand-to-mouth households are assigned a perfectly flat
productivity profile as they enter the economy. Accordingly, they
never need to use credit markets and instead consume their
labor income each period.

12

P REFERENCES
Each household i 2 (0, 1) entering the economy at date t has
preferences (which are the same for both LC and HTM types)
T

Ut,i =

max Et ∑ [η ln c̃t,i (t + s) + (1
T
fc̃t,i ,`t,i gs=0 s=0

η ) ln `t,i (t + s)] .

Households choose c̃, but we define
c̃t,i (t + s) = D (t + s) ct,i (t + s) , where D (t + s) is the “state of
aggregate demand” at date t + s. The state of aggregate demand
evolves as
D (t) = δ(t 1, t)D (t 1) ,
where δ(t 1, t) is the gross growth rate of demand, which
follows an appropriate stochastic process that keeps D(t) > 0 8t.

Households may decide to consume more or less at future dates
depending on the state of aggregate demand at those dates.

13

P RODUCTIVITY PROFILES
Agents entering the economy draw a scaling factor x from a
lognormal distribution and receive a productivity profile that is a
scaled version of a baseline profile, es :
es,i = x es .
For LC agents, eLC
s = 1 + p1 exp

s p2
p3

4

, and p1 , p2 and p3

are chosen to match calibration targets given on later slides.
For HTM agents, eHTM
= h (1/T ) ∑Ts=0 eLC
s
s , where h 2 (0, 1) .
Huggett, Ventura and Yaron (AER, 2011) argue that differences in
initial conditions are more important than differences in shocks
for lifetime earnings; therefore, we sometimes call the scaling
factor the “HVY” shock.

14

D ISTORTION IN THE SCALING ASSIGNMENT

The scaling draw x that each agent receives at the beginning of
their life cycle is meant to be a proxy for unmodeled human
capital development before age 20, including parenting,
on-the-job experience and schooling.
It can be characterized by the standard deviation of the
lognormal distribution from which it is drawn, σlc .
The full social optimum with only life-cycle agents would set
σlc = 0.
Since this value might be impractical, we will consider other
values in the calibration section.

15

I DIOSYNCRATIC LABOR INCOME RISK

All agents in the economy are subject to an i.i.d. unemployment
shock each period which takes on a value of 0 with probability p
or 1 with probability 1 p.
Agents receiving a zero draw cannot earn labor income in that
period.

16

T ECHNOLOGY
Aggregate real output Y (t) is given by
Y (t) = [D (t) Q (t) N (t)]1

α

K (t)α [L (t)]1

α

,

(1)

where K (t) is the real value of the physical capital stock, L (t) is
the aggregate effective human capital supply (hours
productivity of various households), Q (t) is a productivity
index, N (t) indexes the size of the labor force, and D (t) is the
state of aggregate demand.
Q, N and D grow at stochastic gross rates λ, ν and δ, respectively.
These assumptions mean that real output grows at the stochastic
rate λνδ each period.
The aggregate demand assumption is a simple version of Bai,
Ríos-Rull and Storesletten (working paper, 2019).
Firms are bond-financed and issue debt each period according to
Bc (t + 1) = Rn (t, t + 1) Bc (t) .

(2)

17

N OMINAL CONTRACTING AND TIMING PROTOCOL
Under the assumptions outlined, the contract nominal interest
rate is given by
Rn (t, t + 1)

1

= Et

c̃t,i (t)
P (t)
.
c̃t,i (t + 1) P (t + 1)

(3)

The timing protocol is: (1) nature assigns new entrant
productivity profiles and also draws idiosyncratic and aggregate
shocks; (2) the treasury authority issues nominal debt; (3) the
monetary authority sets the price level; (4) households choose
date t consumption, hours worked and net asset holding, and all
other activities take place.
Households will be able to make date t decisions without
reference to future uncertainty in the paper-and-pencil solution.

18

Policymakers

19

T HE TREASURY AUTHORITY

The fully credible nominal debt issuance process is given by
Bg (t) = Rn (t

1, t) Bg (t

1) ,

(4)

where Bg (t) is the total level of nominal debt and Bg (0) > 0.
The treasury authority is issuing enough new debt to maintain
the level of assets in the economy at the appropriate level.

20

T HE LABOR MARKET AUTHORITY

The labor market authority runs an unemployment insurance
program that uses a linear labor earnings tax to raise revenue
sufficient to exactly cover payments to unemployed workers.
Unemployed workers receive the same after-tax income that
they would if they were not unemployed.
Note: Linear labor income tax factors (1 τ ) es , with τ set for all
periods s in a household’s life cycle, will not distort labor supply in
this model.

21

T HE MONETARY AUTHORITY

The monetary authority controls the price level directly and
implements a targeting criterion
P (t) =

δ (t

Rn (t 1, t)
1, t) λ (t 1, t) ν (t

1, t)

P (t

1) .

(5)

This targeting criterion calls for countercyclical price level
movements relative to the expectation embodied in the contract
rate Rn (t 1, t) .
See Koenig (IJCB, 2013) and Sheedy (BPEA, 2014) on NGDP
targeting.

22

T HE FISCAL AUTHORITY
The fiscal authority taxes high-income life-cycle individuals
within a cohort and transfers to low-income life-cycle
individuals within the same cohort.
The tax-transfer scheme is nondistortionary with respect to labor
supply.
The term (1 τ i ) xi es in labor income is replaced with xj es by
setting (1 τ i ) = xj /xi , where xj is the draw in a lognormal
0
distribution with standard deviation σlc
σlc corresponding to
the xi draw from the lognormal distribution with standard
deviation σlc .
This is as if nature had drawn the original xi from a distribution
with lower variance.
This will reduce the consumption Gini, but it will also affect
other Gini coefficients.

23

C OMPETITIVE EQUILIBRIUM AND SOCIAL WELFARE
Solution: Guess and verify that there is a competitive
equilibrium in which the real rate of interest is always equal to
the stochastic rate of real output growth.
The “Wicksellian natural real rate of interest” for this economy.

A social planner would conclude that the allocation of resources
is a social optimum provided (i) the planner places equal weight
on all households for all time, (ii) the planner discounts
backward and forward in time at the stochastic real rate of
interest, and (iii) the planner cannot alter the distribution of
productivity profiles.
Limiting case: If there are no HTM agents and σlc = 0, then all
agents receive exactly the same utility over their lifetime modulo
the real interest rate adjustment.
Question: Does the model equilibrium match data with respect
to consumption growth and leisure?

24

D ECISION RULES FOR CONSUMPTION , LEISURE AND
ASSETS
Life-cycle agents:

at

c̃t

s,i

(t) = η (1

`t

s,i

(t) = (1

(t)
= (1
P (t)
s,i

τ i ) (1 τ u ) w (t) xi ē,
x ē
ē
η ) i = (1 η ) i ,
es,i
xi es
s

τ i ) (1

τ u ) w (t) xi ∑ ej

ē .

j=0

Hand-to-mouth agents (assuming no dispersion):
htm
c̃htm
,
t s,i (t) = ηe

`htm
t s,i (t) = (1

η) ,

ahtm
t s,i

(t)
= 0.
P (t)
25

Calibration

26

M APPING TO THE DATA
Adjust the cohort size based on data from the U.S. Census
Bureau.
Set the baseline hump-shaped life-cycle productivity profile such
that households endogenously choose to work the hours worked
by age in the U.S. data.
Choose η to match the average time devoted to market work
across the economy.
Set the fraction of HTM households (who do not hold assets) to
the share of unbanked U.S. households, 4.5% in 2021 according
to the FDIC.
Choose the within-cohort standard deviation of productivity
scaling for life-cycle households, σlc , to match the Gini coefficient
for financial wealth in the U.S. data.
Set the redistributional tax initially to zero; then set it to match
the Gini coefficient for consumption in the U.S. data.

27

T HE MASS OF LIFE - CYCLE PRODUCTIVITY
2
1.5
1
0.5
0
0

60

120

180

240

quarters
F IGURE : The mass of endowment profiles: life-cycle agents (blue) and
hand-to-mouth agents for h = 0.25 (red). The dashed lines denote the 25th
and the 75th percentiles of the lognormal endowment scaling distributions.

28

H OURS WORKED BY AGE

0.3
0.2
0.1

Data
Model

0
0

60

120

180

240

quarters
F IGURE : Hours worked by age for life-cycle households: U.S. data (blue) and
calibrated model (red).

29

Model Fit to U.S. Data

30

A GGREGATE CONSUMPTION GROWTH
The model is characterized by explicit stochastic growth—no
detrending.
The model equilibrium under the optimal monetary-fiscal policy
mix states that real output growth will be perfectly correlated
with real aggregate consumption growth, and their nominal
counterparts will be similarly correlated.
In the data, it is not clear what the real-world counterpart is to
“output” since the model does not have an international sector
or other important dimensions (e.g., inventories and a “large”
government sector).
Accordingly, we consider a variety of output measures.
Bottom line: The correlations are close to one.

31

A GGREGATE CONSUMPTION GROWTH

F IGURE : The model equilibrium under the optimal policy mix suggests that
the nominal output growth rate and the nominal aggregate consumption
growth rate should be equal. This chart shows one measure of nominal
output growth and one measure of nominal consumption growth, and the
raw correlation is 0.98.

32

C ONSUMPTION GROWTH ACROSS HOUSEHOLDS
The model also predicts that under the optimal monetary-fiscal
policy mix, consumption growth rates for all households—rich
and poor, relatively young and relatively old—will be equalized.
To address this, we consider weekly data from January 2020 to
March 2023 on credit card expenditure by zip code, with median
income in the various zip codes distinguishing between rich and
poor.
The spending growth week-by-week in the lowest income
quartile of zip codes is highly correlated with spending growth
week-by-week in the highest income quartile, consistent with the
model equilibrium.

33

C ONSUMPTION GROWTH ACROSS HOUSEHOLDS
0.4
0.2
0
-0.2
-0.4
Jul 2020

Jul 2021

Jul 2022

F IGURE : Credit card spending by income group, weekly, January 2020 to
March 2023. The model equilibrium predicts that nominal spending growth
rates across society should be equalized. The correlation in consumption
growth between the groups is indeed very high, as predicted by the model.

34

C ONSUMPTION GROWTH ACROSS HOUSEHOLDS
Correlations in growth rates
Household zip code income distribution
Q1
Q2
Q3
Q4
Q1 1.000 0.980 0.957
0.901
Q2
1.000 0.984
0.940
Q3
1.000
0.972
Q4
1.000
TABLE : Correlation in consumption growth rates across the household zip
code income distribution, January 2020 to March 2023, as measured by credit
card expenditure indexed to the home address of the credit card. The
correlations between the richest and poorest quartiles are high, close to the
model prediction of 1.0.

35

C ARNEIRO -H ECKMAN - TYPE LABOR SUPPLY
Carneiro and Heckman (discussion paper, 2003, p. 67):
“Estimated intertemporal labor supply elasticities are small, and
welfare effects from labor supply adjustment are negligible.”
See also Ljungqvist and Sargent (unpublished manuscript, 2014).

In the present model, hours worked for all households is given
by, for s = 0, ..., T,
1

`t

s,i

(t) = 1

(1

η)

ē
=1
es

`t

s

(t) 8i.

(6)

This does not depend on real wages or other income, providing
prima facie evidence that the model will match the micro-labor
evidence.

36

M ARGINAL PROPENSITIES TO CONSUME

Hand-to-mouth implies that agents consume only out of labor
income each period and do not use asset markets.
In the model equilibrium, life-cycle agents will sometimes
consume only out of labor income, in particular when they are
asset-poor and again when they are asset-rich.
There will be a wide variety of MPCs in this economy, as in the
U.S. data.
The MPCs per se are not the key input into the success of the
optimal monetary-fiscal policy mix.

37

M ARGINAL PROPENSITIES TO CONSUME
Life-cycle agents
Hand-to-mouth agents

8
6
4
2
1
0
0

60

120

180

240

quarters
F IGURE : A cross-section diagram of marginal propensities to consume out of
labor earnings at each date in the model equilibrium. Relatively young (asset
poor) and older (asset rich) life-cycle agents have an MPC larger than one.
The MPC of life-cycle agents during the middle of life is 0.69.

38

TAX PROGRESSIVITY AND G INI COEFFICIENTS

σlc
A/ (4Y)
GW
GY
GC

Model 1
1.42
3.79
0.78
0.71
0.69

Model 2
0.52
3.79
0.55
0.41
0.32

U.S. data
4.52
0.78
0.63
0.32

TABLE : Gini coefficients in the model equilibrium with no progressive
taxation (Model 1) and sufficient within-cohort scaling variance to match the
wealth Gini in the U.S. data, and with progressive taxation (Model 2), which
lowers the within-cohort scaling variance sufficiently to match the observed
consumption Gini in the U.S. data.

39

TAX PROGRESSIVITY AND G INI COEFFICIENTS
1
Effect of consumption
redistribution policy

0.8

0.6

0.4

0.2

0
0

0.5

1

1.5

2

2.5

3

lc

F IGURE : The consumption, income and financial wealth Gini coefficients in
the model equilibrium for values of σlc 0. The progressive tax is lowering
the consumption Gini from 0.69 to 0.32, matching the U.S. data, but missing
other Ginis.
40

N OMINAL RETURNS TO ASSET HOLDING

The model equilibrium states that nominal consumption growth
will be equal to the nominal rate of return on asset holding.
There are three assets in the model (MBS, federal government
debt and corporate debt), but these assets are not further
differentiated.
We consider the seven-year high-quality corporate bond as a
measure of the return to capital.
The equilibrium condition is met on this metric, except during
periods of extreme market turmoil.

41

N OMINAL RETURNS

F IGURE : The model with optimal policy predicts the gray line will coincide
with nominal consumption growth and nominal GDP growth. This
prediction broadly holds in the figure outside of the two large disturbances.

42

Conclusions

43

A

BENCHMARK MODEL

We studied a benchmark DSGE model with “massive”
heterogeneity.
The model recommends a set of macroeconomic policies which,
if jointly implemented, can achieve a first-best allocation of
resources.
The recommended macroeconomic policies resemble those in
place in the U.S. and other countries in recent decades.
The calibration to U.S. data suggests that the model equilibrium
assuming the optimal monetary-fiscal macroeconomic policies
are in place fits the data relatively well, except for periods of
exceptionally high volatility.
The recommended macroeconomic policies seem unlikely to
substitute for one another—all policies have to be working
together simultaneously.

44