the eﬀects of monetary policy · 2020. 10. 25. · 0.30 0.35 recession yl y yd ct cnd gini...

Motivation Data & Estimation Model Results Conclusion

The Effects of Monetary PolicyShocks on Inequality in Japan1

Masayuki Inui Nao Sudo Tomoaki Yamada

27 November, 2020@Keio U.

1The views expressed are those of authors and do not necessarily reflect those of theBOJ/BIS.

M. Inui, N. Sudo and T. Yamada The Effects of Monetary Policy Shocks on Inequality in Japan 1/45

1/45


Table of contents

Motivation

Data & Estimation

Model

Results

Conclusion


2/45


Motivation: growing interest on inequality

Impacts of monetary easing on inequality have attractedincreasing attention recently

• Cohan (2014): “Mr. Bernanke’s extraordinary QE program, startedin the wake of the financial crisis, has only widened the gulfbetween haves and have-nots.”

• Krugman (2014): “The belief that QE systematically favors thekinds of assets the wealthy own is wrong or at least overstated.”

• Bernanke (2015): “Monetary policy is a blunt tool which certainlyaffects the distribution of income and wealth, although whetherthe net effect is to increase or reduce inequality is not clear.”


3/45


Existing empirical studies

Empirical observations are mixed• Coibion et al. (2017): Income and consumption inequality acrossU.S. households respond counter-cyclically to monetary policyshocks◦ Interest rate ⇑ → Inequality ⇑

• Mummtaz and Theophipoulou (2016): The same result holds inthe U.K.

• Saiki and Frost (2014): The opposite is true when using Japanesedata◦ Interest rate ⇓ → Inequality ⇑

• Domanski et al. (2016): Unconventional monetary policy mayhave widened wealth inequality, in particular through an upsurge instock prices


4/45


Questions and Findings

Question 1:

• Does a monetary policy affects inequalities?⇒ Yes, MP shocks matter in earnings inequality. But thedirection is different from the US/UK and it depends on thesample period. The effects disappear recently.

Question 2:

• Does the unconventional monetary policy (QE) matter for therelationship between monetary policy and inequality?⇒ Probably, no. No regime switch. More continuous changes ineconomic environment such as labor market.

Question 3:

• What kind of transmission mechanism works?⇒ One possiblity: labor market flexibility.


5/45


What we do

This paper

• Study the distributional effects of monetary policy, using themicro-level data on Japanese households.

• Draw the broad picture of the effects:

1. focusing on inequality measures of income, consumption, and wealthbased on the micro-level data

2. using both the theoretical model and the data3. using a fairly long-span data sample: periods of conventional and

unconventional monetary policy regimes


6/45


Potential transmission channels

1. Earnings heterogeneity channel ← Japan: This paper◦ the response of earnings to a monetary policy shock differs

2. Job creation channel◦ job creation/destruction following a MP shock

3. Income composition channel ← US: Coibion et al. (2017)◦ the income composition of different income types differs

4. Portfolio channel◦ the size and composite of asset portfolio differs

5. Saving redistribution channel ← Doepke and Schneider (2006)◦ a transfer from lenders to borrowers by a subsequent rise in inflation

caused by monetary easing


7/45


I. Data & Estimation


8/45


FIES

Family Income and Expenditure Survey (FIES)• Monthly survey on household income and expenditures◦ January 1981 – December 2008◦ The number of observations: 8,000◦ Panel data: 6 months◦ Two-or-more household members only

• Single household available only after 2002

• Focus on full-time employees (household head): 25–59◦ Self-employed, retirees and unemployed are excluded

• Construct quarterly series of variables on economic inequality◦ for time series analysis


9/45


FIES (cont.)

Definition of Variables

1. Earnings yL◦ Sum of labor income of all household members

2. Total income y: yL + capital income + private transfer

3. Disposable income yD: y + public transfers - taxes/premiums

4. Nondurable expenditures cND◦ Housing, purchasing cars and other durables are excluded

5. Total expenditure cT: cND + durables (housing excluded)

• Equivalized by OECD equivalent scale

DETAILS


10/45


Time path of inequality measures

Variance of log

1980 1985 1990 1995 2000 2005 20100.10

0.15

0.20

0.25

0.30

0.35

Recession yL y yD cT cND

Gini coefficient

1980 1985 1990 1995 2000 2005 20100.20

0.22

0.24

0.26

0.28

0.30


P90-P10 ratio

1980 1985 1990 1995 2000 2005 20102.50

3.00

3.50

4.00

4.50



11/45


Estimation methodology

Estimate the impulse responses of inequality measures to a monetarypolicy shock, using the LLP by Jorda (2005):

Yt+h︸︷︷︸inequality at t+h

−Yt = αh + Πh(L)Mt + εt+h︸︷︷︸innovation

Mt =

∆TFPt

Factort∆Rt

⇐ Monetary policy variable

• Advantages of LLP:◦ Robustness to model misspecifications:⇒ choice of explanatory variables and the number of lags

◦ Flexibility of model specifications


12/45


Monetary policy instrument

1980 1985 1990 1995 2000 2005 2010 2015

-8

-6

-4

-2

0

2

4

6

8

10

12

Annualized, %

Ueno (2017)Krippner (2015)JGB-2Y

• Baseline: 1981Q1-1998Q4, Shadow rate: after 1999Q1


13/45


Monetary policy instrument

1980 1985 1990 1995 2000 2005 2010 2015

-8

-6

-4

-2

0

2

4

6

8

10

12

Annualized, %

Ueno (2017)Krippner (2015)JGB-2Y

• Baseline: 1981Q1-1998Q4, Shadow rate: after 1999Q1


14/45


Responses of inequality measures: Baseline

0 10 20

-0.005

0.000

0.005

0.010

Variance of log

Earnings

0 10 20

-0.005

0.000

0.005

0.010

Pre-tax income

0 10 20

-0.005

0.000

0.005

0.010

Disposable income

0 10 20

-0.005

0.000

0.005

0.010

Expenditure

0 10 20

-0.005

0.000

0.005

0.010

Consumption

0 10 20-0.004

-0.002

0.000

0.002

0.004

Gini

0 10 20-0.004

-0.002

0.000

0.002

0.004

0 10 20-0.004

-0.002

0.000

0.002

0.004

0 10 20-0.004

-0.002

0.000

0.002

0.004

0 10 20-0.004

-0.002

0.000

0.002

0.004

0 10 20Quarter

-0.050

0.000

0.050

0.100

P90-P10

0 10 20Quarter

-0.050

0.000

0.050

0.100

0 10 20Quarter

-0.050

0.000

0.050

0.100

0 10 20Quarter

-0.050

0.000

0.050

0.100

0 10 20Quarter

-0.050

0.000

0.050

0.100


15/45


U.S. Economy

• Figure 4 in Coibion et al. (2017)


16/45


Channel through job creation

0 5 10 15 20

Quarter

-15

-10

-5

0

(1) # of unemployed HH heads

0 5 10 15 20

Quarter

-0.003

-0.002

-0.001

0.000

0.001

0.002(2) The adjusted Gini of earnings

Adjusted Gini


17/45


Responses of inequality: 1981-2008

0 10 20

-0.005

0.000

0.005

Variance of log

Earnings

0 10 20

-0.005

0.000

0.005

Pre-tax income

0 10 20

-0.005

0.000

0.005

Disposable income

0 10 20

-0.005

0.000

0.005

Expenditure

0 10 20

-0.005

0.000

0.005

Consumption

0 10 20-0.003

-0.002

-0.001

0.000

0.001

0.002

0.003

Gini

0 10 20-0.003

-0.002

-0.001

0.000

0.001

0.002

0.003

0 10 20-0.003

-0.002

-0.001

0.000

0.001

0.002

0.003

0 10 20-0.003

-0.002

-0.001

0.000

0.001

0.002

0.003

0 10 20-0.003

-0.002

-0.001

0.000

0.001

0.002

0.003

0 10 20Quarter

-0.050

0.000

0.050

P90-P10

0 10 20Quarter

-0.050

0.000

0.050

0 10 20Quarter

-0.050

0.000

0.050

0 10 20Quarter

-0.050

0.000

0.050

0 10 20Quarter

-0.050

0.000

0.050

• Red line: 1981-1998, Black line: 1981-2008


18/45


Changes in responses of earnings inequality overtime

1996 1998 2000 2002 2004 2006 2008End point of sample period

-0.002

0.000

0.002

0.004

0.006

0.008

0.010

0.012(1) Variance of log


-0.001

0.000

0.001

0.002

0.003

0.004(2) Gini coefficient


-0.020

0.000

0.020

0.040

0.060

0.080

(3) P90-P10 ratio


19/45


Empirical observations

1. Impact of expansionary monetary policy on income inequality isprocyclical, arises mainly from procyclical response of earningsinequality. ⇒ Earnings heterogeneity channel matters inJapan

2. Once-prevailing distributional effects diminish during 2000s

3. Possibility that a decline in earnings inequality due to job creationchannel counters its rise from earnings heterogeneity channel

4. Transmission of income inequality to consumption inequality is lessthan one-to-one


20/45


Accounting for observations

To address to these observations, we conduct three additionalanalyses:

1. Two-sector DSGE model with attached labor inputs◦ illustrate how the distributional effects of monetary policy change with

the structure of the economy

2. Industry-level aggregate data sets

3. Micro-level data on households’ financial assets and liabilities◦ check whether the model’s predictions accord with the data


21/45


II. Model(under revision)


22/45


Literature: Theory

Models of inequality:• Monetary policy shock:◦ Guerrieri and Lorenzoni (2012), Gornemann et al. (2016), Auclert

(2019), McKay et al. (2016)

• Fiscal policy:◦ Oh and Reis (2012), McKay and Reis (2015)

• Income inequality and Pareto distribution:◦ Gabaix et al. (2016)

• Business cycle and heterogeneity:◦ Challe et al. (2017)

• Mechanism behind earnings inequality in HANK⇒ idiosyncratic risks (stochastic, but exogenous)


23/45


Our model

• Two production sectors (X and Z )◦ each of which has final goods firms and intermediate goods firms◦ consumption composite consists of goods X and goods Z

• Two households (X and Z )◦ each of which has two types of members: “attached” and “mobile”◦ “Attached” supplies its labor inputs to one of two sectors◦ “Mobile” can supply its labor inputs to both sectors


24/45


Objectives of households

• Each type of households (X and Z ) maximizes:

Us,t = Et

[∞

∑q=0

βqu(Cs,t+q,Cs,t+q−1,Ns,t+q,Hs,t+q)

]

u(·, ·, ·, ·, ) = log(Cs,t+q − bCs,t+q−1)− θN

1+ηs,t+q

1+ η− ϕ

H1+ηs,t+q

1+ η

◦ N: attached labor input, H: mobile labor input◦ β: discount factor, C : consumption, b: habit


25/45


Budget constraint

• For s = X ,Z , the budget constraint:

Cs,t +Bs,t

Pt≥

Ws,t

PtNs,t +

WtPt

Hs,t

+(

ΠX ,t+ΠZ ,t

Pt

)γΠs +

(RX ,tKX+RZ ,tKZ

Pt

)γKs

+Rt−1Bs,t−1Pt

+ κB

(Bs,t

Pt

)2

,

◦ γΠs : share of dividends, γKs : share of capital stock◦ κB : adjustment costs of bond holding

• Capital stock is fixed at some level


26/45


Firm’s price setting and gross output

• In sector X , the intermediate firm i choose the prices to solve:

maxPX ,t (i)

E

[∞

∑q=0

βt+q Λt+q

Λt

Πt+q,X (i)

Pt+q

]

• subject to

ΠX ,t+q,X (i) = PX ,t+q(i)xt+q(i)−MCX ,t+q(i)xt+q(i)

− κX2

(PX ,t+q(i)

PX ,t+q−1(i)− 1

)2

PX ,t+qXt+q,

xt(i) = ANX ,t(i)αµUX ,t(i)

α(1−µ)KX ,t(i)1−α


27/45


Rest of the model

• Production technology of gross output:

Xt =

[∫ 1

0xt(i)

1− 1ε di

] εε−1

• The aggregator constructs the composite of consumption goodsusing the aggregate technology

Ct = Xρt Z

1−ρt

• Monetary policy:

logRt = ρn logRt−1 + (1− ρn)φ log πt + ϵR,t


28/45


Setup

• Assume that ρ > 1− ρ and κX > κZ

• Other parameters are symmetric

• Then◦ X receives higher earnings than Z at steady state◦ Monetary policy affects two sectors differently around steady state


29/45


Expansionary monetary policy shock

0 5 10 15 20-0.010

0.000

0.010

0.020

0.030

0.040

%

(1) Value-added

Sector XSector Y

0 5 10 15 20-0.025

-0.020

-0.015

-0.010

-0.005

0.000

%

(2) Relative price

0 5 10 15 200.000

0.010

0.020

0.030

0.040

%

(3) Earnings

Household XHousehold Y

0 5 10 15 20Quarter

0.000

0.010

0.020

0.030

0.040

%

(4) Consumption

Household XHousehold Y

0 5 10 15 20Quarter

0.000

0.005

0.010

0.015

0.020

0.025

0.030

%

(5) Earnings inequality

0 5 10 15 20Quarter

0.000

0.005

0.010

0.015

0.020

0.025

0.030

%

(6) Consumption inequality


30/45


Labor market flexibility: µ ∈ {0.4, 0.6, 0.8}

0 5 10 15 200.00

0.01

0.02

0.03

0.04(1) Earnings of household X

highmidlow

0 5 10 15 200.00

0.01

0.01

0.02

0.02

0.03

0.03(2) Earnings of household Z

highmidlow

0 5 10 15 200.00

0.01

0.02

0.03

0.04

0.05

0.06(3) Earnings inequality

highmidlow

0 5 10 15 20-0.01

0.00

0.01

0.02

0.03

0.04

0.05(4) Consumption of household X

highmidlow

0 5 10 15 20-0.01

0.00

0.01

0.02

0.03(5) Consumption of household Z

highmidlow

0 5 10 15 200.00

0.01

0.02

0.03

0.04

0.05

0.06(6) Consumption inequality

highmidlow


31/45


Labor market flexibility in Japan

1985 1990 1995 2000 2005 2010 2015Year

10

12

14

16

18

20

22

24

26

%

(1) Proportion of temporary workers

Labor force surveyMonthly labor surcey

1985 1990 1995 2000 2005 2010 2015Year

3.0

3.2

3.4

3.6

3.8

4.0

%

(2) Labor turnover ratio


32/45


Transmission to consumption inequality

• Why didn’t consumption inequality increase as much as incomeinequality?◦ Responses of consumption of households are all weak?◦ Or other possible explanations?


33/45


Responses of consumption and the MPC

0 5 10 15 20-0.010

-0.005

0.000

0.005

0.010

0.015

0.020

0.025(1) Consumption responses

0-20%80-100%

0 5 10 15 20-0.02

-0.02

-0.01

-0.01

0.00

0.01

0.01

0.02

0.02

0.03

0.03(2) MPC responses

0-20%80-100%

• by income quintiles


34/45


Transmission to consumption inequality

• Why is the MPC conditional on monetary policy shocks higher inthe low income quintile?

1. Distributional effects via financial asset holdings2. Difference in consumption behavior across households


35/45


Distribution of financial assets matters

0 5 10 15 200.00

0.01

0.01

0.02

0.02

0.03

0.03(1) Earnings inequality

highmidlow

0 5 10 15 200.00

0.01

0.01

0.02

0.02

0.03

0.03(2) Consumption inequality

highmidlow

0 5 10 15 20-0.01

0.00

0.01

0.02

0.03

0.04

0.05(3) Consumption of household X

highmidlow

0 5 10 15 20-0.01

0.00

0.01

0.02

0.03(4) Consumption of household Z

highmidlow

• γKX∈ {0.4, 0.5, 0.6}


36/45


Conclusion

Empirically and theoretically study how monetary policy shocksare transmitted to inequality.• Findings:◦ Distributional effects of monetary policy were once present, but have

become statistically insignificant during the 2000s◦ Transmission of income inequality to consumption inequality is minor

• Takeaways:◦ Labor market flexibility is central to the dynamics of income inequality

after the monetary policy shocks◦ Distributions of financial assets and liabilities don’t play a significant

role


37/45


Thank you!


38/45


Appendix Figures


39/45


Responses of macro variables

0 5 10 15 20-2

0

2

4

6(1) GDP

0 5 10 15 20-5

0

5

10

15(2) Investment

0 5 10 15 20-1

0

1

2

3(3) Consumption

0 10 20-0.4

-0.2

0

0.2

0.4

0.6

0.8(4) Price Level

0 5 10 15 20Quarter

-10

0

10

20(5) Stock Price

0 5 10 15 20Quarter

0

1

2

3(6) Labor Income

0 5 10 15 20Quarter

-2

0

2

4

6(7) Capital Income

0 5 10 15 20Quarter

-10

-5

0

5

(8) Unemployed


40/45


Cross-firm heterogeneity to earnings inequality:baseline

0 5 10 15 20-0.004

-0.002

0.000

0.002

0.004

0.006

0.008(1) Value-added

0 5 10 15 20-0.003

-0.002

-0.001

0.000

0.001

0.002

0.003

0.004(2) Earnings per employee

0 5 10 15 20-0.001

0.000

0.001

0.002

0.003

0.004(3) Earnings per working hour


41/45


Cross-firm heterogeneity to earnings inequality:1981-2008

0 5 10 15 20-0.004

-0.002

0.000

0.002

0.004

0.006

0.008(1) Value-added

0 5 10 15 20-0.004

-0.003

-0.002

-0.001

0.000

0.001

0.002

0.003

0.004

0.005(2) Earnings per employee

0 5 10 15 20-0.005

-0.004

-0.003

-0.002

-0.001

0.000

0.001

0.002(3) Earnings per working hour


42/45


Data Details


43/45


Data Details

Definition of Variables• Labor income y :◦ sum of monthly labor income of household members, which include

household head, his/her spouse and other household members

• Nondurable expenditure c :◦ food; repair and maintenance of houses; fuel, light and water charges;

domestic utensils, non-durable goods, and services; clothing andfootwear; medical care; transportation and communication, excludingpurchase of vehicles and bicycles; education; culture and recreation,excluding recreational durable goods; and other consumptionexpenditure, excluding remittance

RETURN


44/45


Adjusted Gini

Adjusted Gini coefficient

G ∗ ≡ ∑Ni=1 ∑N

j=1 |xi − xj |2N ∑N

i=1

= GN

N+

N −N

N

• Assumption: earnings of unemployed = 0

RETURN


45/45

the eﬀects of monetary policy · 2020. 10. 25. · 0.30 0.35 recession yl y yd ct cnd gini...

Documents