Monetary Policy According to HANK Greg Kaplan Ben Moll Gianluca - - PowerPoint PPT Presentation

Monetary Policy According to HANK

Greg Kaplan Ben Moll Gianluca Violante

European Central Bank, 5 November 2015

HANK: Heterogeneous Agent New Keynesian models

• Framework for quantitative analysis of aggregate shocks and

macroeconomic policy

• Three building blocks
• 1. Uninsurable idiosyncratic income risk
• 2. Nominal price rigidities
• 3. Assets with different degrees of liquidity
• Today: Transmission mechanism for conventional monetary policy

1

HANK: Heterogeneous Agent New Keynesian models

• Framework for quantitative analysis of aggregate shocks and

macroeconomic policy

• Three building blocks
• 1. Uninsurable idiosyncratic income risk
• 2. Nominal price rigidities
• 3. Assets with different degrees of liquidity
• Today: Transmission mechanism for conventional monetary policy

1

HANK: Heterogeneous Agent New Keynesian models

• Framework for quantitative analysis of aggregate shocks and

macroeconomic policy

• Three building blocks
• 1. Uninsurable idiosyncratic income risk
• 2. Nominal price rigidities
• 3. Assets with different degrees of liquidity
• Today: Transmission mechanism for conventional monetary policy

1

Why HANK?

• VAR evidence: sizable effects of monetary shocks on C
• Consumption response to a change in real rates
• Textbook Representative Agent New Keynesian (RANK) model
• Direct response

is everything

• Pure intertemporal substitution (RA Euler Equation)

2

Why HANK?

• VAR evidence: sizable effects of monetary shocks on C
• Consumption response to a change in real rates

dC dr = ∂C ∂r

• direct response to r

+ d Y dr

• GE effect on inc

× ∂C ∂Y

• direct response to Y
• Textbook Representative Agent New Keynesian (RANK) model
• Direct response

is everything

• Pure intertemporal substitution (RA Euler Equation)

2

Why HANK?

• VAR evidence: sizable effects of monetary shocks on C
• Consumption response to a change in real rates

dC dr = ∂C ∂r

• direct response to r
• >95%

+ d Y dr

• GE effect on inc

× ∂C ∂Y

• direct response to Y
• <5%
• Textbook Representative Agent New Keynesian (RANK) model
• Direct response ∂C

∂r is everything

• Pure intertemporal substitution (RA Euler Equation)

2

Why HANK?

• Both theory and data suggest ∂C

∂r is small

• 1. Macro: empirically, small sensitivity of C to r
• 2. Micro: many hand-to-mouth hh for whom ∂c

∂r ≈ 0

• 3. Micro: many wealthy hh for whom ∂c

∂r < 0

• Implication: RANK parameterized to be consistent with data

⇒ small effects of monetary policy shocks on C

• Reconcile small effects in NK model with sizable effects in data?

3

Why HANK?

• HANK ingredients deliver realistic distributions of ∂c

∂r and ∂c ∂Y direct response to direct response to inc GE effect on inc

RANK: >95% RANK: <5% HANK: <25% HANK: >75%

• HANK generates

as large as in data even though is small.

4

Why HANK?

• HANK ingredients deliver realistic distributions of ∂c

∂r and ∂c ∂Y

dC dr = ∂C ∂r

• direct response to r
• +

∂C ∂Y

• direct response to inc

× d Y dr

• GE effect on inc
• RANK: >95%

RANK: <5% HANK: <25% HANK: >75%

• HANK generates dC

dr as large as in data even though ∂C ∂r is small.

4

Why does this matter?

• Much more nuanced view of monetary policy
• HANK: to understand C response to monetary policy,

watch labor demand, investment

• Not true in RANK model

5

Literature and contribution

Combine two workhorses of modern macroeconomics:

• 1. New Keynesian models with limited heterogeneity

Campell-Mankiw, Gali-LopezSalido-Valles, Iacoviello, Challe-Matheron-Ragot-Rubio-Ramirez

• micro-foundation of spender-saver behavior
• 2. Bewley models with sticky prices

Oh-Reis, Guerrieri-Lorenzoni, Ravn-Sterk, Gornemann-Kuester-Nakajima, DenHaan-Rendal-Riegler, Bayer-Luetticke-Pham-Tjaden, McKay-Reis, McKay-Nakamura-Steinsson, Huo-RiosRull, Werning, Luetticke

• assets with different liquidity Kaplan-Violante
• new view of individual earnings risk Guvenen-Karahan-Ozkan-Song
• Continuous time approach Achdou-Han-Lasry-Lions-Moll

6

Building blocks

Households

• Face uninsured idiosyncratic labor income risk
• Consume and supply labor
• Hold two assets: liquid and illiquid

Firms

• Monopolistic competition for intermediate producers
• Quadratic price adjustment costs à la Rotemberg (1982)

Assets

• Liquid assets: nominal return set by monetary policy
• Illiquid assets: real return determined by profitability of capital

7

Households

max

{ct,ℓt,ch

t ,dt}t≥0

E0 ∫ ∞ e−(ρ+λ)tu(ct, ℓt, ht)dt s.t. ˙ bt = r b(bt)bt + (1 − ξ)wztℓt−T (wztℓt)−dt − χ(dt, at) − ct−ch

t

˙ at= r a(1 − ω)at+ξwztℓt+dt ht= ch

t + νωat

zt = some Markov process bt ≥ −b, at ≥ 0, ch

t ≥ 0

• ct: non-durable consumption
• dt: illiquid deposits
• bt: liquid assets
• χ: transaction cost function
• zt: individual productivity
• T: labor income tax
• ℓt: hours worked
• ch

t : rentals

• at: illiquid assets
• ht: housing services
• ξ: direct deposits

8

Households

max

{ct,ℓt,ch

t ,dt}t≥0

E0 ∫ ∞ e−(ρ+λ)tu(ct, ℓt, ht)dt s.t. ˙ bt = r b(bt)bt + (1 − ξ)wztℓt−T (wztℓt)−dt − χ(dt, at) − ct−ch

t

˙ at= r a(1 − ω)at+ξwztℓt+dt ht= ch

t + νωat

zt = some Markov process bt ≥ −b, at ≥ 0, ch

t ≥ 0

• ct: non-durable consumption
• dt: illiquid deposits (≷ 0)
• bt: liquid assets
• χ: transaction cost function
• zt: individual productivity
• T: labor income tax
• ℓt: hours worked
• ch

t : rentals

• at: illiquid assets
• ht: housing services
• ξ: direct deposits

8

Households

max

{ct,ℓt,ch

t ,dt}t≥0

E0 ∫ ∞ e−(ρ+λ)tu(ct, ℓt, ht)dt s.t. ˙ bt = r b(bt)bt + (1 − ξ)wztℓt−T (wztℓt)−dt − χ(dt, at) − ct−ch

t

˙ at= r a(1 − ω)at+ξwztℓt+dt ht= ch

t + νωat

zt = some Markov process bt ≥ −b, at ≥ 0, ch

t ≥ 0

• ct: non-durable consumption
• dt: illiquid deposits (≷ 0)
• bt: liquid assets
• χ: transaction cost function
• zt: individual productivity
• T: labor income tax
• ℓt: hours worked
• ch

t : rentals

• at: illiquid assets
• ht: housing services
• ξ: direct deposits

8

Households

max

{ct,ℓt,ch

t ,dt}t≥0

E0 ∫ ∞ e−(ρ+λ)tu(ct, ℓt, ht)dt s.t. ˙ bt = r b(bt)bt + (1 − ξ) wztℓt − T (wztℓt) − dt − χ(dt, at) − ct − ch

t

˙ at = r a (1 − ω) at + ξwztℓtdt ht = ch

t + νωat

zt = some Markov process bt ≥ −b, at ≥ 0, ch

t ≥ 0

• ct: non-durable consumption
• dt: illiquid deposits (≷ 0)
• bt: liquid assets
• χ: transaction cost function
• zt: individual productivity
• T: labor income tax
• ℓt: hours worked
• ch

t : rentals

• at: illiquid assets
• ht: housing services
• ξ: direct deposits

8

Households

• Adjustment cost function

χ(d, a) = χ0 |d| + χ1

• d

a

• χ2

a

• Linear component implies inaction region
• Convex component implies finite deposit rates
• Recursive solution of hh problem consists of:
• 1. consumption policy function
• 2. deposit policy function
• 3. labor supply policy function

joint distribution of households

9

Households

• Adjustment cost function

χ(d, a) = χ0 |d| + χ1

• d

a

• χ2

a

• Linear component implies inaction region
• Convex component implies finite deposit rates
• Recursive solution of hh problem consists of:
• 1. consumption policy function c(a, b, z; w, r a, r b)
• 2. deposit policy function d(a, b, z; w, r a, r b)
• 3. labor supply policy function ℓ(a, b, z; w, r a, r b)

⇒ joint distribution of households µ(da, db, dz; w, r a, r b)

9

Firms

Representative final goods producer: Y = (∫ 1 y

ε−1 ε

j

dj )

ε ε−1

⇒ yj = (pj P )−ε Y Monopolistically competitive intermediate goods producers:

• Technology:
• Set prices subject to quadratic adjustment costs:

Exact NK Phillips curve:

10

Firms

Representative final goods producer: Y = (∫ 1 y

ε−1 ε

j

dj )

ε ε−1

⇒ yj = (pj P )−ε Y Monopolistically competitive intermediate goods producers:

• Technology: yj = Zkα

j n1−α j

⇒ m = 1

Z

( r

α

)α ( w

1−α

)1−α

• Set prices subject to quadratic adjustment costs:

Θ ( ˙ p p ) = θ 2 ( ˙ p p )2 Y Exact NK Phillips curve:

10

Firms

Representative final goods producer: Y = (∫ 1 y

ε−1 ε

j

dj )

ε ε−1

⇒ yj = (pj P )−ε Y Monopolistically competitive intermediate goods producers:

• Technology: yj = Zkα

j n1−α j

⇒ m = 1

Z

( r

α

)α ( w

1−α

)1−α

• Set prices subject to quadratic adjustment costs:

Θ ( ˙ p p ) = θ 2 ( ˙ p p )2 Y Exact NK Phillips curve: ( ρ − ˙ Y Y ) π = ε θ (m − ¯ m) + ˙ π, ¯ m = ε−1

ε 10

Investment fund sector

• Receive illiquid assets from households:

Ap = (1 − ω) ∫ adµ

• Two sources of income:
• 1. Rent illiquid asset as capital with utilization
• 2. Dividends from ownership of intermediate firms
• Investment fund optimization implies illiquid asset return

11

Investment fund sector

• Receive illiquid assets from households:

Ap = (1 − ω) ∫ adµ

• Two sources of income:
• 1. Rent illiquid asset as capital with utilization u

[ru − δ(u)] K

• 2. Dividends from ownership of intermediate firms
• Investment fund optimization implies illiquid asset return

11

Investment fund sector

• Receive illiquid assets from households:

Ap = (1 − ω) ∫ adµ

• Two sources of income:
• 1. Rent illiquid asset as capital with utilization u

[ru − δ(u)] K

• 2. Dividends from ownership of intermediate firms

qK = (1 − m)Y

• Investment fund optimization implies illiquid asset return

11

Investment fund sector

• Receive illiquid assets from households:

Ap = (1 − ω) ∫ adµ

• Two sources of income:
• 1. Rent illiquid asset as capital with utilization u

[ru − δ(u)] K

• 2. Dividends from ownership of intermediate firms

qK = (1 − m)Y

• Investment fund optimization implies illiquid asset return

r a = max

u

(ru − δ(u)) + q

11

Monetary authority and liquid assets

• Taylor rule

i = ¯ r b + ϕπ + ϵ, ϕ > 1

• Fisher equation r b = i − π
• Two participants in bond market:

Households: Bh = ∫ bdµ Government: Bg = −¯ gY

12

Government

• Progressive tax on labor income:

T (wzℓ) = −τ0 + τ1wzℓ

• Steady state government budget constraint

G − r bBg = ∫ T (wzℓ (a, b, z)) dµ

• Out of steady state:
• 1. τ0 adjusts residually
• 2. G adjusts residually
• 3. Bg adjusts for first n years, then τ0 adjusts

13

Summary of market clearing conditions

• Liquid asset market

Bh + Bg = 0

• Illiquid asset market

K = (1 − ω)A

• Labor market

N = ∫ zℓ(a, b, z)dµ

• Goods market:

Y = C + H + I + G + χ + Θ + borrowing costs

14

Calibration

Three particularly important aspects, relatively unique to paper:

• 1. Measurement and partition of asset categories
• liquid vs illiquid
• productive vs non-productive
• match agg balance sheet of households in Flow of Funds
• 2. Adjustment cost function χ (d, a)
• target key aspects of (a, b) distribution in SCF, e.g. no of HtM
• 3. Continuous time household earnings dynamics

15

Wealth distributions: Liquid wealth

$ Thousands

200 400 600 800 1000 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

← Pr(b = 0) = 0.29 Liquid wealth distribution

0.2 0.4 0.6 0.8 1

• 0.2

0.2 0.4 0.6 0.8 1

Liquid wealth Lorenz curve

Model 2004 SCF

• Top 10% share: Model: 87%, SCF 2004: 89%
• Top 1% share:

Model: 36%, SCF 2004: 51%

• Top 0.1% share: Model: 7%, SCF 2004: 21%

16

Wealth distributions: Illiquid wealth

$ Millions

2 4 6 8 10 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16

Illiquid wealth distribution

0.2 0.4 0.6 0.8 1

• 0.2

0.2 0.4 0.6 0.8 1 1.2

Illiquid wealth Lorenz curve

Model 2004 SCF

• Top 10% share: Model: 59%, SCF 2004: 61%
• Top 1% share:

Model: 19%, SCF 2004: 25%

• Top 0.1% share: Model: 4%, SCF 2004: 7%

17

MPC heterogeneity

Amount of transfer ($)

200 400 600 800 1000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

↑ Quarterly MPC out of $500 = 27% Fraction of lump sum transfer consumed

One quarter Two quarters One year 20 10 Liquid Wealth ($000)

Quarterly MPC $500

• 10

Illiquid Wealth ($000)

500 0.5 0.6 0.1 0.2 0.3 0.4 1000

18

Channels for monetary policy

Innovation ϵ < 0 to the Taylor rule: i = ¯ r b + ϕπ + ϵ

• All experiments: ϵ0 = −0.0025, i.e. −1% annually

19

Channels for monetary policy

Innovation ϵ < 0 to the Taylor rule: i = ¯ r b + ϕπ + ϵ

• All experiments: ϵ0 = −0.0025, i.e. −1% annually

Quarters

5 10 15 20

Deviation (pp annual)

• 1.5
• 1
• 0.5

0.5

Taylor rule innovation: ǫ Liquid return: rb Inflation: π

Quarters

5 10 15 20

Deviation (%)

0.1 0.2 0.3 0.4 0.5 0.6

Output Total Consumption Total Investment

19

Channels for monetary policy: consumption

dC = ∂C ∂r b dr b + ∂C ∂w dw + ∂C ∂r a dr a XXX XXX

Quarters

5 10 15 20

Deviation

• 1
• 0.5

0.5

Liquid return: rb (pp annual) Iliquid return: ra (pp annual) Real wage: w (%)

Quarters

5 10 15 20

Deviation (%)

• 0.1

0.1 0.2 0.3 0.4 0.5

Total Response

20

Channels for monetary policy: consumption

dC = ( ∂C ∂r b + ∂C ∂τ0 ∂τ0 ∂r b ) dr b + ( ∂C ∂w + ∂C ∂τ0 ∂τ0 ∂w ) dw + ∂C ∂r a dr a XXX Transfers adjusts: partly direct effect r b ↓ on govt debt... partly indirect

Quarters

5 10 15 20

Deviation

• 1
• 0.5

0.5

Liquid return: rb (pp annual) Iliquid return: ra (pp annual) Real wage: w (%)

Quarters

5 10 15 20

Deviation (%)

• 0.1

0.1 0.2 0.3 0.4 0.5

Total Response

20

Channels for monetary policy: consumption

dC = ( ∂C ∂r b + ∂C ∂τ0 ∂τ0 ∂r b ) dr b + ( ∂C ∂w + ∂C ∂τ0 ∂τ0 ∂w ) dw + ∂C ∂r a dr a Intertemporal substitution channel: direct effects from r b ↓

Quarters

5 10 15 20

Deviation

• 1
• 0.5

0.5

Liquid return: rb (pp annual) Iliquid return: ra (pp annual) Real wage: w (%)

Quarters

5 10 15 20

Deviation (%)

• 0.1

0.1 0.2 0.3 0.4 0.5

Total Response Direct: rb

20

Channels for monetary policy: consumption

dC = ( ∂C ∂r b + ∂C ∂τ0 ∂τ0 ∂r b ) dr b + ( ∂C ∂w + ∂C ∂τ0 ∂τ0 ∂w ) dw + ∂C ∂r a dr a Direct effect through transfers: r b ↓ on govt debt

Quarters

5 10 15 20

Deviation

• 1
• 0.5

0.5

Liquid return: rb (pp annual) Iliquid return: ra (pp annual) Real wage: w (%)

Quarters

5 10 15 20

Deviation (%)

• 0.1

0.1 0.2 0.3 0.4 0.5

Total Response Direct: rb Direct: τ0

20

Channels for monetary policy: consumption

dC = ( ∂C ∂r b + ∂C ∂τ0 ∂τ0 ∂r b ) dr b + ( ∂C ∂w + ∂C ∂τ0 ∂τ0 ∂w ) dw + ∂C ∂r a dr a Labor demand channel: indirect effects from w ↑

Quarters

5 10 15 20

Deviation

• 1
• 0.5

0.5

Liquid return: rb (pp annual) Iliquid return: ra (pp annual) Real wage: w (%)

Quarters

5 10 15 20

Deviation (%)

• 0.1

0.1 0.2 0.3 0.4 0.5

Total Response Direct: rb Direct: τ0 Indirect: w

20

Channels for monetary policy: consumption

dC = ( ∂C ∂r b + ∂C ∂τ0 ∂τ0 ∂r b ) dr b + ( ∂C ∂w + ∂C ∂τ0 ∂τ0 ∂w ) dw + ∂C ∂r a dr a Labor demand channel: indirect effects from w ↑

Quarters

5 10 15 20

Deviation

• 1
• 0.5

0.5

Liquid return: rb (pp annual) Iliquid return: ra (pp annual) Real wage: w (%)

Quarters

5 10 15 20

Deviation (%)

• 0.1

0.1 0.2 0.3 0.4 0.5

Total Response Direct: rb Direct: τ0 Indirect: w Indirect: τ0

20

Channels for monetary policy: consumption

dC = ( ∂C ∂r b + ∂C ∂τ0 ∂τ0 ∂r b ) dr b + ( ∂C ∂w + ∂C ∂τ0 ∂τ0 ∂w ) dw + ∂C ∂r a dr a Portfolio reallocation channel: indirect effects from r a ↑

Quarters

5 10 15 20

Deviation

• 1
• 0.5

0.5

Liquid return: rb (pp annual) Iliquid return: ra (pp annual) Real wage: w (%)

Quarters

5 10 15 20

Deviation (%)

• 0.1

0.1 0.2 0.3 0.4 0.5

Total Response Direct: rb Direct: τ0 Indirect: w Indirect: τ0 Indirect: ra

20

Channels for monetary policy: consumption

dC = ( ∂C ∂r b + ∂C ∂τ0 ∂τ0 ∂r b ) dr b

• 24%

+ ( ∂C ∂w + ∂C ∂τ0 ∂τ0 ∂w ) dw + ∂C ∂r a dr a

• 76%

Quarters

5 10 15 20

Deviation

• 1
• 0.5

0.5

Liquid return: rb (pp annual) Iliquid return: ra (pp annual) Real wage: w (%)

Quarters

5 10 15 20

Deviation (%)

• 0.1

0.1 0.2 0.3 0.4 0.5

Total Response Direct: rb Direct: τ0 Indirect: w Indirect: τ0 Indirect: ra

20

Monetary policy transmission mechanism

RANK model:

• Rise in C from intertemporal substitution

HANK model:

• Two (small) direct effects:
• 1. Reduction in r b triggers portfolio reallocation and increases I
• 2. Lower interest on govt debt lowers T or increases G
• …trigger (large) indirect effect:
• Rise in labor demand increases labor income → C boom

21

Final thoughts and road ahead

• HANK: framework for quantitative analysis of monetary policy
• Consistency with (y, b, a) and MPC distributions ⇒

monetary policy transmission different from standard NK models

• to understand C response: watch labor demand, investment
• Allows for analysis of distributional effects of monetary policy
• Road Ahead
• Forward guidance and unconventional monetary policy
• Fiscal stimulus according to HANK (fiscal policy)
• Perturbation methods for HANK models

estimation, inference

22

Final thoughts and road ahead

• HANK: framework for quantitative analysis of monetary policy
• Consistency with (y, b, a) and MPC distributions ⇒

monetary policy transmission different from standard NK models

• to understand C response: watch labor demand, investment
• Allows for analysis of distributional effects of monetary policy
• Road Ahead
• Forward guidance and unconventional monetary policy
• Fiscal stimulus according to HANK (fiscal policy)
• Perturbation methods for HANK models

estimation, inference

22

Final thoughts and road ahead

• HANK: framework for quantitative analysis of monetary policy
• Consistency with (y, b, a) and MPC distributions ⇒

monetary policy transmission different from standard NK models

• to understand C response: watch labor demand, investment
• Allows for analysis of distributional effects of monetary policy
• Road Ahead
• Forward guidance and unconventional monetary policy
• Fiscal stimulus according to HANK (fiscal policy)
• Perturbation methods for HANK models

⇒ estimation, inference

22