Agglomeration Business Globalization and Productivity An Empirical Study on Taiwanese Firms Yih Luan Chyi Yi Lee Eric S Lin ShihYing Wu Department of Economics National Tsing Hua University ID: 204418
Download Presentation The PPT/PDF document "Industrial" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Industrial Agglomeration, Business Globalization and Productivity: An Empirical Study on Taiwanese Firms
Yih
-Luan
Chyi
,
Yi
Lee,
Eric S.
Lin, Shih-Ying
Wu
Department
of Economics
National Tsing Hua University
November, 2009Slide2
OutlineIntroductionLiterature ReviewTheoretical Model
Empirical Specification & Results
Conclusion
2009/11/30
Seminar_Chung Cheng Univ
2Slide3
Approved Taiwanese Outward FDI (1990-2006)2009/11/30
Seminar_Chung Cheng Univ
3Slide4
Kernel Distribution of TFP, by Year2009/11/30
Seminar_Chung Cheng Univ
4Slide5
MotivationStylized Fact: the starting period of FDI outflows to China outpacing those to other destinations coincided with a rightward shift of the entire productivity distribution of Taiwanese firms, 2001~2006
Supporting literature of outward FDI boosting multinationals performance; for example, firms tend to perform FDI for the benefit of cross-country knowledge spillovers (among others,
Branstetter
, 2006)
Apart from them, many empirical studies confirm connection between agglomeration and productivity.
Does FDI in China improve firm productivity or any other factor contribute to it?
2009/11/30
Seminar_Chung Cheng Univ
5Slide6
Research GoalMarshallian externalities
localized industry-level knowledge spillovers
Taiwanese high-tech clusters with knowledge spillovers for firms in close geographical and technological proximity.
labor pooling
Number of employees in HSP had increased more than ten times during 1986 to 2004.
input-output linkages
Integrated Circuit Industry in Hsinchu Science Park
Research Goal: investigating
agglomeration and FDI effects on productivity after controlling firm attributes
2009/11/30
Seminar_Chung Cheng Univ
6Slide7
Literature Review of agglomeration economies: MechanismsOriginated from Alfred Marshall (1890); recent studies (Duranton
and
Puga
, 2004) suggest:
similar firms have better chances of sharing suppliers; thick labor markets contribute to easing firm-level shocks or matching jobs;
localized firms are more likely to benefit from technologies and innovations of others.A shared prediction: “the concentration of firms and workers in space makes them more productive." (
Duranton
,
Gobillon
,
Puga
, and Roux, 2009, p. 3)
2009/11/30
Seminar_Chung Cheng Univ
7Slide8
Literature Review of agglomeration economies: Two strands of Empirical studiesFirms and workers in larger cities generally are more productive. (Henderson, 2003; Combes
,
Duranton
, and
Gobillon, 2008)
doubling city size raises productivity by 2~8% (Rosenthal and Strange, 2004; Combes,
Duranton
,
Gobillon
, and Roux, 2007)
District effect: firms located in an industrial district can benefit from agglomerative advantages, (
Bagella
and
Becchetti
, 2000;
Fabiani
et al., 1999)
Cainelli
(2008): firm productivity growth mainly attributes to its memberships of an industrial district and product innovations.
Unanimous results on the positive effects of agglomeration on firm performance using various econometric specifications and data sets.
2009/11/30
Seminar_Chung Cheng Univ
8Slide9
Literature Review of FDI: Proximity-concentration TradeoffBrainard (1997) provides a mixed equilibrium where some firms undertake FDI and others export. The results support the proximity-concentration hypothesis, but no role of firm heterogeneity.
Helpman
et al. (2004) add a role for productivity differences into this proximity-concentration trade-off.
Given certain assumptions on fixed costs, relative production costs and transportation costs, the most productive firms will prefer FDI over exporting.
Empirical finding: the ratio of industry exports to FDI sales is positive correlated to plant scale economies, but negative correlated to trading costs and the dispersion of productivity.
2009/11/30
Seminar_Chung Cheng Univ
9Slide10
Literature Review of FDI: Technology SourcingFosfuri and Motta (1999): a technologically less advanced firm tend to undertake FDI to acquire technology from its more efficient competitor located within the same country.
Empirical supportive evidence of Technology sourcing
Pottelsberghe
de la
Potterie
and Lichtenberg (2001): investing in R&D intensive countries improves a country's productivity by using 13 OECD countries’ data . Branstetter
(2006): a positive relationship between the number of Japanese subsidiaries in the U.S. and the extent of U.S. patent citations.
Grith
et al. (2006): U.S. R&D stock had a strong impact on the total factor productivity(TFP) of U.K. firms with lead inventor in the U.S.
Yeaple
and Chung (2008): country-industries with similar technical profiles are more attractive to U.S. outward FDI.
2009/11/30
Seminar_Chung Cheng Univ
10Slide11
Literature Review of FDI: Technology Transferring costKeller and Yeaple
(2009): a theoretical model focusing on the role of technology transferring costs on firms' foreign operation.
MNEs can choose to export the intermediates to their foreign affiliates or producing them in their foreign subsidiaries.
Industries with higher technological complexity are less likely to produce abroad.
Chang and Lu (2009): how the risk of FDI failure affects the dynamics of FDI entry for Taiwanese firms; assuming the success probability of FDI as a non-monotonic function of firm productivity where more productive firms face higher risk of investment failure and loss more if failure.
the negative effect of investment failure will dominate the positive effect of a larger market share on variable profits for the most productive firms.
predicts that the least productive and the most productive firms will not choose FDI. The risk of FDI failure decreases with the number of FDI firms, and therefore the productivity range for FDI firms increases over time.
2009/11/30
Seminar_Chung Cheng Univ
11Slide12
Model-Set upHeterogeneous firms are entailed with different productivity level, r
.
Firms can acquire external knowledge (
f
g) from other firms in the local area. (agglomeration)
Firms engage in R&D. Firms can offer goods to a foreign market through exporting or FDI.
If exporting, firms need to pay an iceberg transportation costs,
t
.
If FDI, there is an extra fixed investment costs,
f
l
, to set up a plant in country
l
.
Countries differ in their factor input prices.
2009/11/30
Seminar_Chung Cheng Univ
12Slide13
Model-Set up: Factors affecting Productivity
Agglomeration economies: firms acquire external knowledge (
f
g
) from other firms in the local area; firm located in
g
entailed productivity level
q
=
r
∙
f
g
.
Technology sourcing effect: firms gain from foreign knowledge spillovers (
q
g
, where 0 <
g
< 1) through FDI.
Technology transferring cost or risk of technology diffusion in the foreign country (
q
-
b
, where 0 <
b
< 1).
R&D investment: firms raise their productivity by investing in research and development
.
Exporting firms’ productivity levels:
q
x
q
∙
rd
d
FDI firms’ productivity levels:
q
f
q
[1+(
g
b
)]
∙
rd
d
2009/11/30
Seminar_Chung Cheng Univ
13Slide14
Model- Demand, Supply Demand: CES Utility function; inversed d
emand for product
i
in
l
Supply:Production function:
one input-labor,
q
i
=
q
i
L
i
constant MC,
MC =
w
l
/
q
, if producing in location
l
.
MC =
w
l
t
/
q
, if producing in location l and
exporting to a foreign country.
Three stage decisions:
Undertaking FDI or not,
Optimal R&D level,
Deciding the production output level.
Backward induction is used.
2009/11/30
Seminar_Chung Cheng Univ
14Slide15
Model- Firm Behaviors, deciding the production output levelOptimal pricing rule for CES-induced demand function:
Firm
i
’s
profit of selling products in location
l:
Profits for exporting firms (
p
x
) and FDI firms (
p
f
):
(Eq. 3, pp. 12)
2009/11/30
Seminar_Chung Cheng Univ
15Slide16
Model- Firm Behaviors, optimal R&D levelCost of R&D: C(rd) = rd
h
C
l
Profit functions for exporting and FDI firms
(Eq. 4, pp. 13)Optimal R&D investments (Eq. 5, pp. 13)
Profit under optimal R&D level:
P
l
=
f
(
B
l
,
w
l
,
q
,
C
l
)
∙A (Eq. 6, pp. 14)
2009/11/30
Seminar_Chung Cheng Univ
16
P
l
>0
if
h
-
d
(
s
-1)
>0
P
l
increases in market size (
B
l
) and
q
;
P
l
decreases in costs (production, transport and R&D)Slide17
Model- Firm Behaviors, FDI or exportFirms choose production locations for serving foreign market by comparing profits of
NonFDI
and FDI.
Curvature of FDI
2009/11/30
Seminar_Chung Cheng Univ
17Slide18
Comparing Profits for Exporting and FDI Firms: (g -b) > 0
2009/11/30
Seminar_Chung Cheng Univ
18
Appendix 2:
g
↑
or
b
↓
shifting
P
FDI
leftward
Appendix 1: existence and uniqueness of
q
NSlide19
Comparing Profits for Exporting and FDI Firms: (g -b) < 0
2009/11/30
Seminar_Chung Cheng Univ
19
Appendix 1: sign of
DP
(
q
+
)
is undetermined
Appendix 2:
g
↑
or
b
↓
shifting
P
FDI
upward
q
+Slide20
Model PredictionA firm’s incentive to undertake FDI depends on relative magnitude of two effects: knowledge spillovers and technology transferring costs.
Gains from knowledge spillovers are larger than losses on technology replication, then more productive firms will undertake FDI. (consistent with
Helpman
et al. (2004))
The technology loss is larger than the externalities gain, only firms with intermediate productivity perform FDI.
Least productive firms cannot afford the fixed cost;
Most productive firms confront larger efficiency losses relative to less productive firms, which overcome the gains from saving on variable costs. (consistent with Chang and Lu (2009)).
2009/11/30
Seminar_Chung Cheng Univ
20Slide21
Data and MeasuresTwo data sources:Firm data: TEJ, 2001, 2002
Plant-level data: TSCMO, 2000
Agglomeration index: Holmes indicators
H
mg =
j
mg
L
j
,
mg
(sum of all employments of industry
m
in location
g
).
Productivity measure:
Tornqvist
multilateral TFP
TFP index for firm
i
in year
t is calculated as:2009/11/30
Seminar_Chung Cheng Univ
21Slide22
Summary Statistics2009/11/30
Seminar_Chung Cheng Univ
22
Variable
NonFDI
FDI
TFP
2001
-.572
-.449
TFP
2002
-.562
-.433
ln
(firm age)
2.618
2.882
ln
(R&D)
8.920
9.159
ln
(capital)
13.43
13.72
ln
(# employees)
5.551
5.952
ln
(Cluster)
8.55
8.46
#
firms
180
421Slide23
2009/11/30Seminar_Chung Cheng Univ
23
Model
1
Model 2
Model 3
OLS
2SLS
OLS
2SLS
OLS
2SLS
TFP
2001
0.367(0.1264)***
0.3755(0.1260)***
0.3482(0.1237)***
0.3471(0.1235)***
-
0.9405(0.5001)*
-
0.9033(0.5076)*
ln
(firm age)
0.1026(0.0355)***
0.1245(0.0374)***
0.089(0.0347)**
0.1013(0.0359)***
0.0843(0.0292)***
0.1095(0.0323)***
ln
(R&D)
0.0094(0.0051)*
0.0102(0.0054)*
0.005(0.0043)
0.0053(0.0043)
0.0072(0.005)
0.0084(0.0053)
ln
(capital)
-
0.0812(0.0205)***
-0.0794(0.0233)***
-
0.149(0.0454)***
-
0.1492(0.0534
)***
-
0.0605(0.0197)***
-
0.054(0.0247)**
ln
(# employees)
0.1208(0.0597)**
0.1181(0.0896)
ln
(Cluster)
0.0422(0.0157)***
0.041(0.0155)***
0.0372(0.0142)***
0.0369(0.0146)**
0.1214(0.0384)***
0.1185(0.0391)***
FDI
0.0931(0.0361)**
0.0632(0.0362)*
0.1039(0.0361)***
FDI_
hat
-
0.005(0.3086)
0.0654(0.27)
-
0.1077(0.3091)
TFP
2001
×
Cluster
0.1557(0.0615)**
0.1538(0.0631)**
TFP
2001
×FDI
TFP
2001
×
FDI_hat
R
2
0.47
0.47
0.49
0.49
0.53
0.53
Obs.
578
578
578
578
578
578
Robust standard errors in parentheses (Bootstrap errors in parentheses of 2SLS)
* significant at 10%; ** significant at 5%; *** significant at 1%
TFP
i
2002
=
b
0
+
b
1
TFP
i
2001
+
b
2
Cluster
i
+
b
3
FDI
i
+
b
4
R
&
D
i
+
Z
i
’
g
+
e
iSlide24
2009/11/30Seminar_Chung Cheng Univ
24
Model
4
Model 5
Model 6
OLS
2SLS
OLS
2SLS
OLS
2SLS
TFP
2001
-
1.2793(0.4348)***
1.0938(0.5593)*
-
1.2316(0.4130)***
-0.9516 (0.7814)
-
1.2207(0.4036)***
-0.9497(0.7926)
ln
(firm age)
0.0918(0.0304)***
0.0501(0.0254)*
0.0827(0.0310)***
0.0953 (0.0314)***
0.088(0.0310)***
0.1009(0.0333)***
ln
(R&D)
0.0077(0.005)
-
0.0009(0.0054)
0.0046(0.0044)
0.0041(0.0041)
ln
(capital)
-
0.0674(0.0189)***
-
0.0186(0.0154)
-
0.1169(0.0394)***
-
0.1255(0.0533)**
-
0.1247(0.0390)***
-
0.1287(0.0581)**
ln
(# employees)
0.0866(0.0506)*
0.1148(0.0926)
0.0825(0.0500)*
0.0992(0.0933)
ln
(Cluster)
0.1376(0.0348)***
0.0389(0.0317)
0.1304(0.0337)***
0.1117 (0.0482)**
0.1317(0.0331)***
0.1134 (0.0646)*
FDI
0.2396(0.0967)**
0.2164(0.0905)**
0.2229(0.0888)**
FDI_
hat
-
0.5178(0.2912)*
0.0059(0.4097)
0.0702(0.4256)
TFP
2001
×
Cluster
0.1787(0.0531)***
0.0476(0.0499)
0.1716(0.0513)***
0.1483(0.0790)*
0.1689(0.0502)***
0.1453 (0.0869)
TFP
2001
×FDI
0.2627(0.1451)*
0.2602(0.1379)*
0.2663(0.1350)**
TFP
2001
×
FDI_hat
-
1.3415(0.3643)***
0.099(0.3286)
0.1156 (0.2691)
RD_Low
-0.0254(0.0130)*
-
0.0236(0.0169)
RD_Med
0.0017(0.0065)
0.0011(0.0059)
RD_High
0.02(0.0071)***
0.0192(0.0078)**
R
2
0.56
0.57
0.58
Obs.
578
578
578
578
578
578Slide25
2009/11/30Seminar_Chung Cheng Univ
25
Explanatory
Variables
First stage FDI estimation
TFP
0.0674
0.0666
0.0663
0.0676
(0.0358)*
(0.0349)*
(0.0352)*
(0.0356)*
ln
(R&D)
-0.0032
-0.0039
-0.0033
-0.0035
(0.0061)
(0.006)
(0.0061)
(0.0061)
ln
(Cluster)
-0.0091
-0.0115
-0.0087
-0.0088
(0.0185)
(0.0187)
(0.0187)
(0.0187)
ln
(Capital)
-0.0845
-0.087
-0.0866
-0.0847
(0.0249)***
(0.0246)***
(0.0247)***
(0.0249)***
ln
(Foreign Wage)
-0.2377
-0.1424
0.3243
(0.1044)**
(0.1139)
(0.5908)
ln
(Foreign Employment)
0.1148
0.094
0.1057
(0.0448)**
(0.0465)**
(0.0484)**
ln
(
Diposal
Income)
-0.4783
(0.5845)
ln
(employment)
0.2197
0.2273
0.2245
0.2222
(0.0329)***
(0.0326)***
(0.0329)***
(0.0332)***
Constant
3.3596
0.3888
1.8457
1.6416
(1.0007)***
(0.3891)
(1.2182)
(1.2726)
Observations
578
578
578
578
R-squared
0.16
0.16
0.16
0.16
Robust standard errors in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1%Slide26
ConclusionThis paper examines impacts of industrial agglomeration and foreign direct investment on the total factor productivity of Taiwanese firms.
Main contributions:
“Holmes” indicator of industrial agglomeration
A theoretical model of FDI decisions to identify causal links between global knowledge sourcing and total factor productivity.
2009/11/30
Seminar_Chung Cheng Univ
26Slide27
Conclusion: Our findingsAfter firms' attributes are controlled, firm productivities are positively affected by local industrial agglomerations.
FDI in China by Taiwanese firms has
insignificant
effect on
next period’s
TFP.
More productive Taiwanese firms may suffer from efficiency losses, which outweigh benefits from knowledge spillovers as predicted by our theoretical model.
FDI interacting with TFP
in some cases has
negative and significant influences on future TFP
2009/11/30
Seminar_Chung Cheng Univ
27