Grasp the Large, Let Go of the Small:

Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Grasp the Large, Let Go of the Small:
The Transformation of the State Sector in China
Chang-Tai Hsieh
Zheng (Michael) Song
Chicago Booth
Chicago Booth, December 2, 2013
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
The State Sector in China
Fifteen years ago: A moribund state sector ...
Counterfactuals
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
The State Sector in China
Fifteen years ago: A moribund state sector ...
A decade later: Full of powerful and giant state-owned
enterprises (SOEs)
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Grasp the Large, Let Go of the Small
Let Go of the Small
Closure and privatization of small SOEs
Grasp the Large
Corporatization of the remaining SOEs
Creating new SOEs
Counterfactuals
Conclusion
Baoshan Iron & Steel in the 1980s
Iron & Steel in the 1980s
Ministry of y
Metallurgical Industry
Baoshan
Iron & Steel
Iron & Steel
Baoshan Iron & Steel in the 2000s
Iron & Steel in the 2000s
Public Listed since 2000
Board of Directors
E al ated b SASAC
Evaluated by SASAC
…….
Corpora za on ≠ Privatization
BaoSteel
Group
Corpora za on ≠ Privatization
State Ownership ≠ Monopoly
Hebei
Steel
Beijing Steel
Wuhan Steel
Jiangsu Steel
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
This Paper
We …rst document the facts regarding “Grasping the Large,
Letting Go of the Small” in the Chinese industrial sector.
We interpret these facts through the lenses of a standard
model of monopolistic competition with heterogeneous …rms.
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Main Findings
TFP of corporatized and privatized SOEs is converging to
that of private …rms.
The performances are highly unequal between large and small
SOEs.
Labor productivity of corporatized and privatized SOEs is also
converging to that of private …rms.
The gap of capital productivity remains unchanged.
Our stories for TFP growth and falling labor distortions
Welfare implication: Less clear
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Data
Annual Survey of Industries from 1998 through 2007
Cover all industrial …rms identi…ed as SOE or as private …rms
with sales above 5 million RMB, which account for above 90%
of the total industrial output in China.
The survey has 137,716 and 319,183 …rms in 1998 and 2007,
respectively.
1998
2007
SOEs
Number VA Share
50,615
55%
19,255
34%
Introduction
Facts
The Model
TFP and Distortions
Our Story
Exit Rates: Before and After
Annual Exit Rates (%)
SOEs Private Firms
1991-1995
0.9
11.6
1998-2007 13.2
12.0
Counterfactuals
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Let Go of the Small: Exit Rates 1998-2007 by Initial Size
40
SOE
Priv ate
35
30
25
20
15
10
5
0
10
20
30
40
50
60
70
perc entiles by 1998 v alue-added
80
90
100
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Size, Labor and Capital Productivity
Panel A: Y 1998
0.5
In c .
Ex it
In c .
Ex it
0.4
0.3
SOE
SOE
Priv a te
Priv a te
0.2
0.1
0
-2
10
-1
10
0
10
10
1
10
2
Panel B: Y-L 1998
0.5
0.4
0.3
0.2
0.1
0
-2
10
-1
10
0
10
10
1
10
2
Panel C: Y-K 1998
0.5
0.4
0.3
0.2
0.1
0
-2
10
-1
10
0
10
10
1
10
2
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Incumbent Firms
To facilitate comparison, we focus on incumbent …rms
observed in both 1998 and 2007.
# Corporatized/Privatized SOEs: 7,556/4,952
# Private …rms: 28,128
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Let Go of the Small: Privatization Rates 1998-2007 by
Initial Size
50
45
40
35
30
25
20
15
10
5
0
0
10
20
30
40
50
60
70
percentiles by 1998 v alue-added
80
90
100
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Distributions of the Incumbent
Panel A: Y 1998
Panel B: Y 2007
0.5
0.5
Co rp . SOE
Pri v . SOE
Pri v a te
0.25
0
-2
10
0.25
-1
10
0
1
10
2
10
10
0
-2
10
Panel C : Y-L 1998
0.5
0.25
0.25
0.25
1
4
10
0
0.05
Panel E: Y-K 1998
0.5
0.25
0.25
0.25
1
1
2
10
10
0.25
1
4
10
4
10
Panel F: Y-K 2007
0.5
0
0.05
0
10
Panel D : Y-L 2007
0.5
0
0.05
-1
10
4
10
0
0.05
0.25
1
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Input and Output Growth of the Incumbent
Panel A: Employ ment Grow th
5
C orp. SOE
Priv . SOE
5
Panel C : Value-Added Grow th
5
0
0
0
-5
-5
-5
-10
Panel B: C apital Grow th
-10
-10
0
50
100
0
50
100
0
50
100
per c entiles by 1998 v alue-added per c entiles by 1998 v alue-added per c entiles by 1998 v alue-added
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Entry
New …rms: birth year > 1998 and observed in 2007
New corporatized SOEs: 4,224 (26%)
New privatized SOEs: 1,238 (15%)
New private …rms: 198,204 (68%)
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Among New Firms
Panel A: Y 2007
0.5
Co rp . SOE
Pri v . SOE
Pri v ate
0.4
0.3
0.2
0.1
0
-2
10
-1
10
0
10
10
1
10
2
Panel B: Y-L 2007
0.5
0.4
0.3
0.2
0.1
0
-2
10
-1
10
0
10
10
1
10
2
Panel C: Y-K 2007
0.5
0.4
0.3
0.2
0.1
0
-2
10
-1
10
0
10
10
1
10
2
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Summary
Exit SOEs: low Y, Y-L and Y-K relative to incumbent SOEs
Incumbent SOEs:
1998: low Y-L and Y-K relative to incumbent private …rms
2007: similar Y-L but low Y-K
New SOEs: high Y, similar Y-L and low Y-K relative to new
private …rms
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
The Model
A standard model of monopolistic competition with
heterogeneous …rms (e.g., Hsieh and Klenow, 2009)
Firms are index by i in industry s, i 2 f1, 2,
, Is g. Each
…rm produces di¤erentiated goods by a Cobb-Douglas
technology with constant returns to scale:
Qsi = Asi Ksiαs L1si
αs
.
Households maximize a CES aggregate of Qsi ,
Is
Qs =
∑
i
1 η
Qsi
!11η
, 1/η
0.
(1)
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Equilibrium Conditions
Denote Ysi
implies
Psi Qsi value-added. Firms’pro…t maximization
MRPLsi
= (1
MRPKsi
= αs (1
αs ) (1
η)
η)
Ysi
= 1 + τ Lsi ws ,
Lsi
Ysi
= 1 + τ Ksi (r + δs ) .
Ksi
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
TFP
We can back out TFP, Asi , by
Asi =
Ysi
Ys
η
1 η
Ysi
Ksi
αs
Ysi
Lsi
1 αs
(2)
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Parameterization
Benchmark:
1 αs : using the average labor income share of private …rms
in the corresponding two-digit industry adjusted by Qian and
Zhu (2012).
1/η: 7 (corresponding to markups of 1.17).
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
TFP of Incumbent Firms
Panel A: TFP 1998
0.5
Co rp . SOE
Pri v . SOE
Pri v a te
0.4
0.3
0.2
0.1
0
-2
10
-1
10
0
10
TFP
1
10
2
10
Panel B: TFP 2007
0.5
0.4
0.3
0.2
0.1
0
-2
10
-1
10
0
10
TFP
1
10
2
10
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Relative TFP by Initial Size
Panel A: R elativ e TFP
of Inc umbent C or p. SOE
Panel B: R elativ e TFP
of Inc umbent Priv . SOE
1
1
1998
2007
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0
20
40
60
80
100
perc entiles by 1998 v alue-added
0.2
0
Panel C : R elativ e TFP
of C orp. SOEs in 2004 and 2007
1
20
40
60
80
100
perc entiles by 1998 v alue-added
Panel D : R elativ e TFP
of Pr iv . SOEs in 2004 and 2007
1
2004
2007
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0
20
40
60
80
100
perc entiles by 2004 v alue-added
0.2
0
20
40
60
80
100
perc entiles by 2004 v alue-added
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
TFP of Exit and New Firms
Panel A: TFP of Firms in 1998
0.5
In c .
Ex i t
In c .
Ex i t
0.4
SOE
SOE
Pri v a te
Pri v a te
0.3
0.2
0.1
0
-2
10
-1
10
0
10
log R elativ e TFP
1
10
2
10
Panel B: TFP of N ew Firms in 2007
0.5
Ne w Co rp . SOE
Ne w Pri v . SOE
Ne w Pri v a te
0.4
0.3
0.2
0.1
0
-2
10
-1
10
0
10
log R elativ e TFP
1
10
2
10
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Distortions
We can back out the relative τ Lsi and τ K
si by
1 + τ Lsi ∝
Ysi
,
Lsi
1 + τK
si ∝
Ysi
.
Ksi
The average distortions in each industry:
1 + τ Ls (J ) ∝
Ys ( J )
,
Ls (J )
1 + τK
s (J ) ∝
Ys ( J )
.
Ks (J )
τ Ls = 0 for incumbent private …rms and τ K
s = 0 for incumbent
corporatized SOEs.
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Labor and Capital Distortions
Panel A: Av erage τ
L
Panel B: Av erage τ
0
K
2
Co rp . SOEs
Pri v . SOEs
Pri v . Fi rm s
Pri v . SOEs
1.8
-0.1
1.6
1.4
-0.2
1.2
1
-0.3
0.8
0.6
-0.4
0.4
0.2
-0.5
1998
2000
2002
y ear
2004
2006
0
1998
2000
2002
y ear
2004
2006
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Distortions for Exit and New SOEs
Exit SOEs
New Corp. SOEs
1998
τL
τK
-0.55 -0.06
2007
τL
τK
0.34
0.10
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Key Findings on TFP and Distortions: A Summary
TFP of SOEs grew at a faster rate than that of private …rms.
Disparity across …rms.
Labor distortions on SOEs have been disappearing.
Capital distortions on private …rms remained stable from 1998
through 2007.
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Caveats
Heterogeneous markups:
No big e¤ect on TFP growth if markups are time-invariant
Downward biased TFP growth of privatized SOEs if
privatization reduces market power
Restructuring of state assets:
Downward biased TFP, labor and capital productivity growth
of incumbent SOEs
Arti…cially high-performing new SOEs
Broadly de…ned “Grasp the Large”
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
China-Style Coporatization
Improvement in corporate governance
Fostering competition: State ownership 6= state monopoly
Most industrial ministries were dismantled in the late 1990s.
Entry of private …rms.
Increasingly …erce competition among SOEs
Active entry at the …rm level (2% entry rate for SOEs)
Accountability: The number of centrally controlled
conglomerates dropped from 189 in 2003 to 124 in 2010.
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Market Shares
VA Share of SOEs
Mining of Coal
Extraction of Petroleum
Processing of Petroleum
Chemical
Ferrous Metals
Non-Ferrous Metals
Transport Equipment
Communication Equip.
Electric and Heat Power
1998
81.6
99.9
87.2
48.6
78.5
53.6
61.3
37.0
72.2
2007
64.7
90.7
48.7
22.0
45.1
32.5
46.1
7.8
82.2
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
TFP Growth and Entry
Panel A: TFP Grow th of C orp. SOEs
Panel B: TFP Grow th of C orp. SOEs
0.2
0.2
0.15
0.15
0.1
0.1
0.05
0.05
0
0.05
0.1
0.15
0.2
0
0
Panel C: TFP Grow th of Priv . SOEs
0.2
0.2
0.15
0.15
0.1
0.1
0.05
0.05
0
0.05
0.1
0.15
0.2
0
0
Panel E: TFP Grow th of Priv ate Firms
0.2
0.2
0.15
0.1
0.1
0.05
0.05
0.1
0.15
av erage entry rate of priv ate firms
0.02
0.03
0.04
0.05
0.01
0.02
0.03
0.04
0.05
Panel F: TFP Grow th of Priv ate Firms
0.15
0
0.05
0.01
Panel D: TFP Grow th of Priv . SOEs
0.2
0
0
0.01
0.02
0.03
0.04
av erage entry rate of SOEs
0.05
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Labor Distortion: Overhead Labor
Denote fsi the …xed overhead labor. Then,
Qsi = Asi Ksiαs (Lsi
fsi )1
αs
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Labor Distortion: Overhead Labor
Denote fsi the …xed overhead labor. Then,
fsi )1
Qsi = Asi Ksiαs (Lsi
αs
The FOC is
MRPLsi
(1
αs ) (1
η)
Ysi
Lsi
fsi
= 1 + τˆ Lsi ws ,
which gives
(1
αs ) (1
η)
Ysi
=
Lsi
1
fsi
Lsi
1 + τˆ Lsi ws .
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Labor Income Share
Rewrite the FOC as
1 + τ Lsi ws Lsi
(1
=
Ysi
{z
}
|
αs ) (1
1
fsi
L si
η)
.
Labor Income Share
Change in fsi /Lsi would manifest itself in labor income share.
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Evidence: Labor Income Shares in the 1998-2007 Balanced
Panel
0.5
C orp. SOEs
Priv . SOEs
Priv ate Firms
0.45
0.4
0.35
0.3
0.25
0.2
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Assumptions
Within industry resource reallocation
No redundant workers and exogeneous interest rates
Robustness checks
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Does TFP Growth of a Firm Reduce Aggregate Output?
Exogenous interest rate: dYs /dAsi > 0 if and only if
Ys /Ls
1
<
Ysi /Lsi
(1
αs (1 η )
.
αs ) (1 η )
Endogenous interest rate: dYs /dAsi > 0 if and only if
(1
αs ) (1
η)
Ys /Ls
+ αs (1
Ysi /Lsi
η)
Ys /Ks
< 1.
Ysi /Ksi
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Experiments
Benchmark: Set TFP and distortions of exit, incumbent and
new SOEs to their initial values.
Let Go of the Small: Exit SOEs + Incumbent privatized SOEs
Grasp the Large: Incumbent corporatized SOEs
Entry: New SOEs
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Results: Aggregate
8
Benc hmark
Data
7
6
5
4
3
2
1
0
1998
1999
2000
2001
2002
2003
y ear
2004
2005
2006
2007
Conclusion
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Welfare Gains
Let Go of the Small
Grasp the Large
Entry
Overall
Open Economy
RW (0%) RW (20%)
1.8
12.6
(4.7 )
Closed Economy
RW (0%) RW (20%)
0.1
6.0
7.7
10.2
-3.7
-2.4
5.8
5.4
0.8
0.7
13.4
27.9
-1.9
5.1
(19.9 )
(6.3 )
Introduction
Facts
The Model
TFP and Distortions
Our Story
Counterfactuals
Conclusion
Conclusion
The transformation of the state sector over 1998-2007
Reduces the gap of TFP and labor productivity between SOEs
and private …rms.
Welfare gains are sensitive to model speci…cations.
Secret weapons:
Fostering competition
Tolerating unemployment
About the objective function of China’s government