How important are inter-city spillovers for FDI? Evidence from Chinese Cities (work-in-progress)

1
How important are inter-city spillovers for FDI?
Evidence from Chinese Cities(work-in-progress)

• Chang Liu
• Sailesh Gunessee
• GEP China,
• University of Nottingham Ningbo.

2
Research Background Motivation

• Wide array of contributions to the literature on
  FDI in China. While weve learnt a lot yet we
  identify some issues
• Spatial relationship quite rare in FDI studies in
  general (Coughlin and Segev, 2000 Hong et al.,
  2008 Chen, 2009)
• omission of spatial interdependence of FDI gt may
  lead to biased estimates gt fail to capture
  third-country effects (see Yeaple , 2003 Ekholm
  et al., 2007 Blonigen, et al. 2007). Thus,
  neighbourhood effects spillovers neglected in
  most previous studies.
• Inter-city spillovers understudied
• Most work have used provincial data or
  combination of firm-provincial data.
• Spatial studies focus on inter-provincial
  spillovers Spillovers better captured in
  smaller areas
• Two notable studies using city-level data (He,
  2002) and firm-city (Head and Ries, 1996) both
  ignore spatial elements
• Regional differences limited consideration given
  due to use of provincial data.

09/05/2016
2
3
Our work Research Objectives

• 1) How important are inter-city spillovers for
  FDI? Is FDI in one city promoted by FDI/Market
  Potential in surrounding cities or at the
  expense?
• 2) Is the influence the same for - i) the
  Hinterland and the Eastern regions.
• Use city-level data to examine relationship
  between FDI and neighbouring market potential
  the spatial lag of FDI (two major sources
  identified by literature). These may also be
  important to tell us about FDI motives (see
  Blonigen et al. 2007 Ledyaeva, 2009).

4
Our work Insights

• Potential insights from this work?
• uncover insights into the importance of
  inter-city spillovers.
• Since 2000 regional policy, to reduce regional
  disparity, has banked on cheap labour costs and
  an even spread of resources under the Great
  Western Development Strategy (GWDS) and Central
  China Rising Strategy (CCRS) to prop up the
  Hinterland. However, these policies are now
  deemed to be failures. The focus now has shifted
  towards developing key cities to form city
  clusters. Our results may tell us about
  importance of spillovers in the hinterland and
  can be seen as relevant to the recent policy
  change.

5
Methodology Data

• Step 1 City level analysis
• Data 200 Chinese cities for 9 years (1999-2007)
  Hinterland Cities 105 Eastern 95
• Selection of cities based on Head and Ries (1995)
  of choosing cities that attracted minimum no. of
  FDI
• China City Statistical Yearbook (various issues)
• Ad Hoc solution for endogeneity used lagged
  explanatory variables
• Spatial weights (W) constructed similar to
  Madariaga and Poncet (2007)
• Carried spatial diagnostic tests spatial lag
  model supported spatial dependence for
  surrounding market potential and spatial FDI
• Estimated using OLS Spatial OLS Spatial 2SLS
  Spatial ML

6
Variables Definitions
FDI Log of Real Realised FDI at the city level.
Market Size Log of real GDP per capita at the city level
Unit Labour Costs Log of average wages to labour productivity
Education Log of share of students enrolment at the third level in citys population
Infrastructure Log of length of highways per sq km
Agglomeration Log of HOOVER coefficient
Policy Ordinal variable with values 0, 1, and 2. 0 no special preference policies (SEZ, ETDZ, FTZ, EPZ) 1 at least one 2 2 or more.
Surrounding Market Potential Distance weighted Log of real GDP per capita of neighbouring cities
Spatial lag of FDI Distance weighted spatial lag of dependent variable
Hypotheses - /- /-
7
Estimation results of the whole China (Whole
Sample)
Traditional OLS Spatial OLS Spatial 2SLS Spatial ML
Market Size 0.304 (0.080) 0.242 (0.077) 0.233 (0.078) 0.501 (0.084)
Unit Labour Costs -0.242 (0.047) -0.135 (0.043) -0.120 (0.043) -0.137 (0.041)
Education 0.265 (0.044) 0.353 (0.039) 0.366 (0.040) 0.387 (0.038)
Infrastructure 0.530 (0.075) 0.205 (0.074) 0.157 (0.083) 0.215 (0.061)
Agglomeration -0.109 (0.081) -0.083 (0.076) -0.079 (0.076) -0.217 (0.071)
Policy 0.933 (0.054) 0.847 (0.051) 0.835 (0.052) 0.683 (0.051)
Surrounding-Market Potential 0.660 (0.106) -0.287 (0.154) -0.427 (0.186) -0.321 (0.085)
Spatially Weighted FDI 0.707 (0.067) 0.812 (0.081) 0.774 (0.052)
Constant 1.819 (1.082) 4.762 (1.151) 5.200 (1.280) 2.201 (0.880)
R2 / Log Likelihood 0.4950 0.5724 0.5707 -2547.0952
F Test /Wald ?2 238.96 242.99 2012.47 1186.42
Observations 1525 1525 1525 1521
8
Estimation results of the eastern cities(Eastern)
Traditional OLS Spatial OLS Spatial 2SLS Spatial ML
Market Size 0.387 (0.141) 0.348 (0.122) 0.338 (0.117) 0.547 (0.110)
Unit Labour Costs -0.188 (0.054) -0.133 (0.050) -0.118 (0.050) -0.135 (0.049)
Education 0.264 (0.051) 0.294 (0.046) 0.302 (0.046) 0.244 (0.043)
Infrastructure 0.314 (0.090) 0.202 (0.081) 0.173 (0.081) 0.254 (0.081)
Agglomeration 0.139 (0.101) 0.044 (0.095) 0.019 (0.097) 0.011 (0.094)
Policy 0.727 (0.061) 0.708 (0.058) 0.703 (0.060) 0.675 (0.058)
Surrounding-Market Potential 0.249 (0.115) -0.517 (0.123) -0.715 (0.165) -0.162 (0.096)
Spatially Weighted FDI 0.554 (0.069) 0.697 (0.098) 0.352 (0.056)
Constant 5.566 (1.394) 7.631 (1.243) 8.164 (1.281) 3.966 (1.073)
R2 / Log Likelihood 0.5836 0.6205 0.6180 -1021.2591
F Test /Wald ?2 145.95 150.73 1217.42 751.85
Observations 733 733 733 732
9
Estimation results of the hinterland cities
(Hinterland)
Traditional OLS Spatial OLS Spatial 2SLS Spatial ML
Market Size 0.075 (0.072) 0.098 (0.083) 0.094 (0.079) 0.482 (0.149)
Unit Labour Costs -0.274 (0.065) -0.169 (0.063) -0.185 (0.064) -0.132 (0.057)
Education 0.456 (0.069) 0.483 (0.065) 0.479 (0.064) 0.447 (0.067)
Infrastructure 0.458 (0.089) 0.146 (0.088) 0.192 (0.095) 0.067 (0.077)
Agglomeration -0.366 (0.124) -0.206 (0.122) -0.230 (0.120) -0.198 (0.112)
Policy 0.685 (0.164) 0.800 (0.121) 0.783 (0.121) 0.855 (0.111)
Surrounding-Market Potential 0.573 (0.164) -0.030 (0.181) 0.060 (0.166) 0.137 (0.128)
Spatially Weighted FDI 0.660 (0.085) 0.562 (0.112) 0.896 (0.062)
Constant 5.035 (1.790) 4.479 (1.871) 4.562 (1.740) -2.425 (2.041)
R2/ Log Likelihood 0.3079 0.3988 0.3968 -1421.0893
F Test /Wald ?2 65.95 66.36 507.55 523.893
Observations 792 792 792 789
10
Initial findings, Discussion and Future
Directions

• How important are inter-city spillovers for FDI?
• Important to an extent but differences across
  regions macro/micro view
• City-results only SMP weak and ve Spatial FDI
  ve sig. (elements regional differences)
• Firm-City results SMP ve (Overall East) and
  insig.(Hinterland) Spatial FDI ve (Overall
  Hinterland) and -ve (East)
• Explanations 1) Datasets/Methods 2) FDI motives

• Future Directions
• Dynamic Spatial Panel Data for city-level
  analysis FDI in Space and Time
• Explore further disaggregated location choice of
  MNEs
• Firm-level analysis pursued with larger sample
  and other multinomial logit techniques
• Paths New Economic Geography Spatial
  Agglomeration and Fiscal Competition
  Environmental Spillovers

11
Appendix FDI motives

• Work by Blonigen et al. (2007) and (later adapted
  by) Ledyaeva (2009) and Hong et al. (2008) have
  used the following classification

FDI Motivation Sign of Spatial Lag Variable Sign of Surrounding Market Potential Variable
Pure horizontal 0 0
Pure vertical - 0
Regional trade platform - /-/0 (depends on local protectionism and sig. of neighbourhood effects)
Complex strategy with agglomeration /-/0 (depends on local protectionism and sig. of neighbourhood effects)
09/05/2016
11
12
Appendix
Figure 1 Regional Disparity of FDI inflows in
China data of the first half year is used in 2009
13
Appendix
Figure 2 Spatial Disparity of FDI inflows in
China
14
Appendix Average Wages and Unit labour costs
Figure 3 Spatial Disparity of Labour Costs in
China
09/05/2016
14
15
Spatial Weights
Appendix

• We follow Madariaga and Poncet (2007) in the
  construction of our spatial weights
• We choose a spatial weighting matrix W that
  depends exclusively on the geographical distance
  dij between cities i and j since the exogeneity
  of distance is unambiguous.
• Distance-based weights are defined as follows
• dij is the distance in kilometres between cities
  i and j. The distance 1,624 km is the cut-off
  parameter above which interactions are assumed to
  be negligible.
• This distance is chosen such that each city
  interacts with at least one other Chinese city.
  This cut-off parameter is important since there
  must be a limit to the range of spatial
  dependence allowed by the spatial weights matrix
  (Abreu et al., 2005)

09/05/2016
15
16
Hoover Coefficient of Specialisation
Appendix

• employment for sector j in city i

employment for sector j in China
Total employment in city i
Total employment in China
Employment Data for 10 sectors Manufacturing,
Real Estate, Primary, Utilities, Construction,
Transportation, Wholesale Retail, Education
Health, Finance Business Services, Other
09/05/2016
16
17
References

• Abreu, M., De Groot, H. L. F., and Florax, R. J.
  G. M. (2005), Space and Growth A Survey of
  Empirical Evidence and Methods. Région et
  Développement 21, 13-44.
• Blonigen, B.A., Davies, R.B., Waddell, G.R., and
  Naughton, H.T. (2007) FDI in space Spatial
  Autoregressive Relationship in Foreign Direct
  Investment. European Economic Review 51 (2007)
  1303-1325.
• Chen, Y-J. (2009) Agglomeration and Location of
  Foreign Direct Investment The Case of China.
  China Economic Review 20 (2009), 549-557.
• Coughlin, C. and Segev, E. (2000) Foreign Direct
  Investment in China A spatial Econometric Study.
  The World Economy 23 (1), 1-23.
• Crozet, M., Mayer, T. and Muchielli et al. (2004)
  How do firms agglomerate? A study of FDI in
  France. Regional Science and Urban Economics
  vol. 34, 27-54
• Ekholm, K., Forslid, R., Markusen, J.R. (2007)
  Export-Platform Foreign Direct Investment.
  Journal of European Economic Association, Vol. 5,
  No. 4, 776-795
• He, C. (2002) Information costs, agglomeration
  economies and the location of foreign direct
  investment in China. Regional Studies 36,
  1029-1036

18
References

• Head, K. and Ries, J. (1996) Inter-City
  Competition for Foreign Investment Static and
  Dynamic Effects of Chinas Incentive Areas.
  Journal of Urban Economics 40, 38-60.
• Hong, E., Sun, L-X., and Li, T. (2008) Location
  of Foreign Direct Investment in China A Spatial
  Dynamic Panel Data Analysis by Country of Origin.
  Discussion Paper 86, The Centre for Financial
  Management Studies, University of London.
• Ledyaeva, S. (2009) Spatial Econometric Analysis
  of Foreign Direct Investment Determinants in
  Russian Regions. The World Economy, Vol.32, Issue
  4, 643-666.
• Madariaga, N. and Poncet, S. (2007) FDI in
  Chinese Cities Spillovers and Impact on Growth.
  The World Economy, Vol.30, Issue 5, 837-862.
• Yeaple, S.R. (2003) The Complex Integration
  Strategies of Multinationals and Cross Country
  Dependencies in the Structure of Foreign Direct
  Investment. Journal of International Economics 60
  (2), 293-314.