40
Competition, Uncertainty, and Misallocation Kaoru Hosono (Gakushuin University) Miho Takizawa (Toyo University) Kenta Yamanouchi (Keio University) Summer Workshop on Economic Theory August, 2017 1

Competition, Uncertainty, and Misallocation

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Competition, Uncertainty, and Misallocation

Competition, Uncertainty, and Misallocation

 Kaoru  Hosono  (Gakushuin  University)  Miho  Takizawa  (Toyo  University)  

Kenta  Yamanouchi  (Keio  University)    

Summer  Workshop  on  Economic  Theory  August,  2017

1

Page 2: Competition, Uncertainty, and Misallocation

Composition

1.  IntroducJon  

2.  Literature  Review  

3.  Model  and  SimulaJon  

4.  Data  and  Methodology  

5.  EsJmaJon  Results  

6.  Conclusion

2

Page 3: Competition, Uncertainty, and Misallocation

1. Introduction

Uncertainty

CompeJJon

MisallocaJon

3

•  This  paper  invesJgates  the  effects  of  the  degree  of  compeJJon  on  the  negaJve  impact  of  uncertainty  on  capital  misallocaJon  across  producers.  

•  Using  a  large  dataset  of  manufacturing  plants  in  Japan,  we  find  that  the  impact  of  uncertainty  is  stronger  when  the  product  market  is  compeJJve.

Page 4: Competition, Uncertainty, and Misallocation

•  Under  favorable  producJvity  shock,  plants  adjust  their  inputs  to  the  opJmal  amount.  

•  If  adjustment  is  insufficient  by  some  reasons,  marginal  revenue  product  of  capital  (MRPK)  is  deviated  from  rental  rate  (E  to  A,  not  B).  

•  If  MRPK  is  dispersed  across  firms,  reallocaJon  of  capital  from  low  MRPK  firms  to  high  would  increase  aggregate  output.  

MRPK A Rental Rate E B Productivity shock K

Productivity Shock and Misallocation

4

Page 5: Competition, Uncertainty, and Misallocation

Changes of Productivity and MRPK

5

-4-2

02

4ch

ange

in ln

MR

PK

-2 -1 0 1 2change in productivity

•  If  the  amount  of  capital  is  adjusted  immediately  a]er  the  producJvity  shock,  MRPK  is  constant  and  equal  to  rental  cost.  

•  But  the  changes  of  producJvity  and  MRPK  are  posiJvely  correlated,  so  insufficient  adjustment  is  implied.

Page 6: Competition, Uncertainty, and Misallocation

Uncertainty and Misallocation Larger  uncertainty  deteriorates  allocaJve  efficiency  of  capital  across  firms.    1.  Time-­‐to-­‐build  

•  If  capital  adjustment  requires  Jme-­‐to-­‐build,  uncertainty  directly  affects  on  the  deviaJon  between  actual  and  opJmal  levels  of  capital.  

•  Asker,  Collard-­‐Wexler,  and  De  Loecker  (2014)  

2.  ReducJon  of  investment  •  Many  papers  considers  the  effects  of  uncertainty  on  investment  (e.g.  Dixit  and  Pindyck  (1994)).  

•  But  all  of  those  studies  don’t  focus  on  the  effects  of  allocaJve  efficiency.

6

Page 7: Competition, Uncertainty, and Misallocation

Competition Matters •  The  degree  of  compeJJon  is  closely  related  to  the  effects  of  uncertainty  on  investment.  

• While  the  fierce  compeJJon  theoreJcally  miJgates  the  negaJve  effects  of  uncertainty,  the  results  of  empirical  studies  are  mixed.  

•  In  addiJon,  no  studies  link  the  effects  of  compeJJon  to  uncertainty-­‐misallocaJon  relaJonship.  

• CompeJJon  macers  for  the  effects  of  uncertainty  on  resource  misallocaJon.

7

Page 8: Competition, Uncertainty, and Misallocation

Competition and Allocative Efficiency •  Some  papers  focus  on  the  effects  of  compeJJon  on  allocaJve  efficiency  across  producers.  

•  Olley  and  Pakes  (1996)  •  Collard-­‐Wexler  and  De  Loecker  (2015)  •  Edmond,  Midrigan,  and  Xu  (2015)  

 •  Tougher  compeJJon  promotes  the  reallocaJon  of  resources  from  less  producJve  firms  to  producJve  firms.  

•  In  other  words,  producJvity  determines  the  firm  size  strongly  in  a  compeJJve  industry.  

 •  Most  of  those  papers  assume  immediate  adjustment,  but  the  effect  of  uncertainty  is  strong  under  costly  adjustment  and  Jme-­‐to  build.  

•  The  reducJon  of  uncertainty  complements  compeJJon  policies  to  improve  allocaJve  efficiency.

8

Page 9: Competition, Uncertainty, and Misallocation

This Paper •  This  paper  invesJgate  the  relaJonship  between  the  degree  of  product  market  compeJJon  and  the  adverse  effects  of  uncertainty  on  resource  misallocaJon  across  plants.  

• We  conduct  numerical  simulaJon  for  investment  model  and  reduced  form  esJmaJon,  using  a  large  panel  dataset  of  manufacturing  plants  in  Japan.  

•  The  results  show  that  the  adverse  effects  of  uncertainty  are  stronger  for  industries  with  tougher  compeJJon.  

•  The  effects  of  uncertainty  on  investment  •  The  dependence  of  opJmal  amount  of  capital  on  producJvity  

9

Page 10: Competition, Uncertainty, and Misallocation

Research Questions and Results 1.  Does  greater  uncertainty  worsen  resource  allocaJon  

across  plants  in  Japan?  -­‐  Yes,  uncertainty  increases  capital  misallocaJon  in  Japan.  

 2.  How  is  the  product  market  compeJJon  related  to  the  

adverse  effects  of  uncertainty  on  allocaJve  efficiency?  -­‐  The  adverse  effects  of  uncertainty  are  large  for  the  industries  with  tougher  compeJJon.  

 3.  How  does  the  degree  of  compeJJon  affect  on  the  

adjustment  process  of  capital?  -­‐  The  negaJve  effects  of  uncertainty  on  investment  is  stronger  in  compeJJve  industries  along  both  extensive  and  intensive  margin.

10

Page 11: Competition, Uncertainty, and Misallocation

Composition

1.  IntroducJon  

2.  Literature  Review  

3.  Model  and  SimulaJon  

4.  Data  and  Methodology  

5.  EsJmaJon  Results  

6.  Conclusion

11

Page 12: Competition, Uncertainty, and Misallocation

1. Competition and uncertainty-investment relationship Uncertainty  and  investment  

A.  Convexity  of  MRPK  with  respect  to  producJvity:  posiJve  •  Uncertainty  increases  the  expected  return  from  investment  by  Jensen’s  inequality.  

•  Hartman  (1972),  Abel  (1983)  B.  Real  opJons  theory:  negaJve  

•  It  is  becer  to  wait  and  avoid  a  costly  disrupJon  in  a  highly  uncertain  environment.  

•  McDonald  and  Siegel  (1986),  Pindyck  (1988),  Dixit  and  Pindyck  (1994)  

C.  Empirical  studies:  negaJve  •  Most  but  not  all  studies  find  negaJve  effects  of  uncertainty  on  investment.  

•  Bloom,  Bond,  and  Van  Reenen  (2007),  Bloom  (2014)  

12

Page 13: Competition, Uncertainty, and Misallocation

Theory  A.  Capital  adjustment  cost  

•  Both  of  imperfect  compeJJon  and  asymmetric  capital  adjustment  costs  are  required  for  the  delay  of  investment.  

•  Caballero  (1991)  

B.  OpJon  exercise  games  •  OpJon  values  erode  under  fierce  compeJJon  because  compeJtors  may  preempt  the  opportunity  of  investment.  

• Williams  (1993),  KulaJlaka  and  Peroj  (1998),  Grenadier  (1996,  2002)  

13

1. Competition and uncertainty-investment relationship

Page 14: Competition, Uncertainty, and Misallocation

Evidence  A.  CompeJJon  miJgates  the  negaJve  effects  of  uncertainty.  

•  Guiso  and  Parigi  (1999):  price-­‐cost  margin  •  Bontempi,  Golinelli,  and  Parigi  (2010):  price-­‐cost  margin  •  Bulan  (2005):  industry  concentraJon  raJos  •  Bulan,  Mayer,  and  Somerville  (2009):  #  of  compeJtors  •  Akdogu  and  MacKay  (2008):  HHI  

B.  NegaJve  effect  is  stronger  in  compeJJve  industries.  •  Ghosal  and  Loungani  (1996,  2000):  concentraJon  raJo  

→No  studies  focus  on  the  effects  of  uncertainty  on  allocaJve  efficiency.

14

1. Competition and uncertainty-investment relationship

Page 15: Competition, Uncertainty, and Misallocation

2. Resource Misallocation across producers

15

Baseline  framework  

A.  Perfect  compeJJon  

•  Even  in  the  U.S.,  the  allocaJon  of  inputs  across  plants  are  severely  distorted.  

•  Restuccia  and  Rogerson  (2008)  

B.  MonopolisJc  compeJJon  

•  Compared  with  the  U.S.,  the  allocaJve  efficiency  is  very  low  in  China  and  India.    

•  Hsieh  and  Klenow  (2009)  

Page 16: Competition, Uncertainty, and Misallocation

16

CompeJJon  

•  Increased  compeJJon  reduces  markup  distorJon  across  plants.  

•  Edmond,  Midrigan,  and  Xu  (2015)  

Uncertainty  

•  Higher  Jme-­‐series  volaJlity  contributes  to  larger  cross-­‐secJonal  dispersion  of  MRPK.  

•  Asker,  Collard-­‐Wexler,  and  De  Loecker  (2014)  

→The  effects  of  compeJJon  and  uncertainty  are  not  simultaneously  explored.  

2. Resource Misallocation across producers

Page 17: Competition, Uncertainty, and Misallocation

Composition

1.  IntroducJon  

2.  Literature  Review  

3.  Model  and  SimulaJon  

4.  Data  and  Methodology  

5.  EsJmaJon  Results  

6.  Conclusion

17

Page 18: Competition, Uncertainty, and Misallocation

Model Simple  dynamic  investment  model  • Asker,  Collard-­‐Wexler,  and  De  Loecker  (2014)  • Demand  

• ProducJon  

•  Sales  

• Profit  Si =ΩiKi

βK LiβLMi

βM , ��βX = 1− 1ε

#

$%

&

'(αX

18

Qi = BPi−ε

Qi = AiKiαK Li

αLMiαM , �αK +αL +αM =1

π (Ωit,Kit ) = (βK +ε−1) βL

pL

#

$%

&

'(

βL(βK+ε

−1 ) βMpM

#

$%

&

'(

βM(βK+ε

−1 )Ωit1/(βK+ε

−1 )KitβK /(βK+ε

−1 )

Page 19: Competition, Uncertainty, and Misallocation

Model • Dynamic  process      

• Value  funcJon      • Capital  adjustment  cost  

19

ωit = ρωit−1 +σκ it, �κ it ~ N(0,1)

Kit = δKit−1 + Iit−1

V (Ωit,Kit ) =maxIitπ (Ωit,Kit )−C(Iit,Kit,Ωit )+β V (Ωit+1,δKit+1 + Iit )ϕ(Ωit+1 Ωit )dΩit+1

Ωit+1

C(Iit,Kit,Ωit ) = Iit +1 Iit>0{ } CKF+π (Ωit,Kit )+CK

Q+KitIitKit

"

#$

%

&'

2(

)**

+

,--+1 Iit<0{ } CK

F−π (Ωit,Kit )+CKQ−Kit

IitKit

"

#$

%

&'

2(

)**

+

,--

Page 20: Competition, Uncertainty, and Misallocation

MisallocaJon          Uncertainty      CompeJJon      OpJmal  amount  of  capital  

lnMRPKi = ln(βK )+ si − ki

20

Volatility =σ

ε = 2, 4, 6

Misallocation = SD lnMRPKi( )

kit* = cons+εωit

Simulation

Page 21: Competition, Uncertainty, and Misallocation

21

Simulation result 1 •  Large  uncertainty  increases  capital  misallocaJon.  •  When  the  compeJJon  is  tough,  the  effects  of  uncertainty  are  large.  •  The  amount  of  opJmal  producJon  strongly  depends  on  producJvity  

in  the  case  of  tougher  compeJJon,  so  the  volaJlity  of  producJvity  is  largely  reflected  into  large  dispersion  of  MRPK.  

02

46

8SD

(lnM

RPK

)

0 .5 1 1.5volatility

ε = 2 ε = 4 ε = 6

Symmetric costs

02

46

8SD

(lnM

RPK

)

0 .5 1 1.5volatility

ε = 2 ε = 4 ε = 6

Asymmetric costs

Page 22: Competition, Uncertainty, and Misallocation

22

Simulation result 2 •  In  the  case  of  symmetric  costs,  the  fracJon  of  posiJve  investment  is  

high  under  large  volaJlity  due  to  the  Hartman-­‐Abel  effect.  •  On  the  other  hand,  uncertainty  reduces  the  fracJon  of  posiJve  

investment  when  adjustment  costs  are  asymmetric.  

0.1

.2.3

.4.5

.6.7

.8.9

1Fr

actio

n of

pos

itive

inve

stm

ent

0 .5 1 1.5volatility

ε = 2 ε = 4 ε = 6

Asymmetric costs

0.1

.2.3

.4.5

.6.7

.8.9

1Fr

actio

n of

pos

itive

inve

stm

ent

0 .5 1 1.5volatility

ε = 2 ε = 4 ε = 6

Symmetric costs

Page 23: Competition, Uncertainty, and Misallocation

Composition

1.  IntroducJon  

2.  Literature  Review  

3.  Model  and  SimulaJon  

4.  Data  and  Methodology  

5.  EsJmaJon  Results  

6.  Conclusion

23

Page 24: Competition, Uncertainty, and Misallocation

Data

•  Source:  Census  of  Manufacturing  published  by  METI  

• Coverage:  All  manufacturing  plants  with  more  than  30  employees  located  in  Japan  

 • Period:  1986-­‐2013    •  Industry:  4  digit  (491  industries)  

24

Page 25: Competition, Uncertainty, and Misallocation

Measurement (plant-level variables) Log-­‐linearized  producJon  funcJon:  

•                                                                                         :  TFPR  •  Dynamic  panel  esJmaJon:  Blundell  and  Bond  (1998,  2000)  

     

(Log  of)  MRPK:  

sit = βKskit +βLslit +βMsmit +ηi + yeart + ωit +ζ itωit = ρ ωit−1 +ξit

MRPKit = ln∂Sit∂Kit

"

#$

%

&'= ln(βKs )+ sit − kit

25

ωit =ηi + yeart + ωit +εit

sit = π1kit +π 2kit−1 +π3lit +π 4lit−1 +π 5mit +π 6mit−1 +π 7sit−1 +ηi* + yeart

* +ωit

E xit−sΔωit( ) = 0

E Δxit−s ηi* +ωit( )( ) = 0

s ≥ 3

Page 26: Competition, Uncertainty, and Misallocation

Measurement (industry-level variables) Uncertainty:      

•  VolaJlity  of  producJvity  shocks  MisallocaJon:    

•  Dispersion  in  MRPK  results  in  aggregate  producJvity  losses  from  the  staJc  view  point.  

•  Hsieh  and  Klenow  (2009)  CompeJJon:    

•  Constant  return  to  scale  is  imposed.  • We  also  check  the  method  of  De  Loecker  and  Warzynski  (2012).  

Volatilityst = SDst (ωit −ωit−1)

Misallocationst = SDst (MRPKit )

26

Markups =εsεs −1

=1

βKs +βLs +βMs

Page 27: Competition, Uncertainty, and Misallocation

Estimation method 1.  Greater  uncertainty  reduces  investment  and  results  in  

staJc  misallocaJon.  

 2.  The  impact  of  uncertainty  on  misallocaJon  is  weaker  in  

the  market  where  compeJJon  is  severer.      

3.  The  probability  of  investment  and  the  investment  rate  is  related  to  uncertainty  and  compeJJon.  

27

Misallocationst = βVolatilityst +FEs +ust

Misallocationst = γ1Volatilityst +γ2Volatilityst *Markups +FEs +ust

1{I /K>0.05}it = λ1 lnMRPKit +λ2Volatilityst +λ3 lnMRPKit *Volatilityst +FEi +FEt +uitIit /Kit = λ1 lnMRPKit +λ2Volatilityst +λ3 lnMRPKit *Volatilityst +FEi +FEt +uit

Page 28: Competition, Uncertainty, and Misallocation

Summary statistics

28

Page 29: Competition, Uncertainty, and Misallocation

Capital misallocation

29

•  Capital  misallocaJon  has  increased  within  industry  in  Japan.

Page 30: Competition, Uncertainty, and Misallocation

Change of misallocation by investment status

30

•  MisallocaJon  is  reduced  a]er  investment  among  posiJve  investment  plants.

Page 31: Competition, Uncertainty, and Misallocation

Composition

1.  IntroducJon  

2.  Literature  Review  

3.  Model  and  SimulaJon  

4.  Data  and  Methodology  

5.  EsJmaJon  Results  

6.  Conclusion

31

Page 32: Competition, Uncertainty, and Misallocation

Volatility and misallocation

•  The  measures  of  volaJlity  and  capital  misallocaJon  are  posiJvely  correlated.  

32

Page 33: Competition, Uncertainty, and Misallocation

Volatility and misallocation

•  The  effects  of  volaJlity  on  misallocaJon  are  stronger  in  small  markup  industries.  

33

Page 34: Competition, Uncertainty, and Misallocation

Estimation results 1 •  SD(MRPK)  is  posiJvely  correlated  with  volaJlity  measure.  •  The  effects  of  volaJlity  are  strong  in  the  industries  with  small  markup.  

34

Page 35: Competition, Uncertainty, and Misallocation

Estimation results 2 •  The  results  are  not  changed  even  if  the  measures  of  volaJlity  and  markup  are  changed.  

35

Markup2st =medianβMs

PMstMit / Sit

!

"#

$

%&Volatility2st = SDst (ωit −

ρωit−1)

Page 36: Competition, Uncertainty, and Misallocation

Estimation results 3 •  Large  MRPK  plants  are  more  likely  to  conduct  posiJve  investment.  • When  uncertainty  is  large,  the  investment  is  less  likely  to  be  made  even  if  MRPK  is  large.  

•  The  negaJve  effects  of  uncertainty  are  not  observed  in  the  industries  with  tough  compeJJon.  

•  NegaJve  investment  are  not  affected  by  uncertainty.

36

Page 37: Competition, Uncertainty, and Misallocation

Estimation results 4 •  Large  MRPK  plants  tend  to  conduct  more  investment.  •  Under  large  uncertainty,  the  investment  rate  is  less  affected  by  MRPK.  •  The  negaJve  effects  of  uncertainty  are  observed  in  the  industries  with  tough  compeJJon.  

•  CompeJJon  affects  misallocaJon  both  through  the  opJmal  amount  of  capital  and  through  the  reducJon  of  investment.  

37

Page 38: Competition, Uncertainty, and Misallocation

Composition

1.  IntroducJon  

2.  Literature  Review  

3.  Model  and  SimulaJon  

4.  Data  and  Methodology  

5.  EsJmaJon  Results  

6.  Conclusion

38

Page 39: Competition, Uncertainty, and Misallocation

Summary •  This  paper  invesJgates  the  effect  of  compeJJon  to  the  uncertainty-­‐misallocaJon  relaJonship.  

•  SimulaJon  results  show  that  the  negaJve  effects  of  uncertainty  are  severe  in  the  industries  with  tougher  compeJJon.  

•  These  results  are  confirmed  in  the  esJmaJon,  using  a  large  dataset  of  Japanese  manufacturing  plants.  

•  The  negaJve  effects  of  uncertainty  on  investment  are  strong  in  the  compeJJve  market.  

•  Some  papers  showed  compeJJon  improves  allocaJve  efficiency,  but  our  results  suggest  compeJJon  induces  fragility  to  uncertainty.  

39

Page 40: Competition, Uncertainty, and Misallocation

Future work

40

-.15

-.1-.0

50

.05

.1.1

5.2

.25

.3.3

5M

ean(

lnM

RPK

) in

next

yea

r

1985 1990 1995 2000 2005 2010 2015Year

zero positivenegative

-.2-.15

-.1-.05

0.05

.1.15

.2Mean(lnMRPK

)

1985 1990 1995 2000 2005 2010 2015Year

zero positivenegative

•  While  this  study  sheds  new  lights  on  the  role  of  compeJJon  in  the  uncertainty-­‐misallocaJon  relaJonship,  we  have  not  yet  explored  the  durability  of  uncertainty-­‐driven  misallocaJon.    

•  If  the  major  source  of  such  misallocaJon  is  Jme-­‐to-­‐build,  then  uncertainty-­‐driven  misallocaJon  may  be  short-­‐lived.    

•  We  find  a  relaJvely  small  quanJtaJve  impact  of  volaJlity,  which  may  reflect  the  short-­‐run  effect  of  uncertainty.