The Economic Impact of Special Economic Zones: Evidence from Chinese Municipalities

uotaintro Roadmap Reform Review A Conceptual Framework Data and Identi cation Results Conclusion The Economic Impact of s: Evidence from Chinese Municipalities London School of Economics January 16th, 2009

uotaintro Roadmap Reform Review A Conceptual Framework Data Motivation and Identi cation Research Question Results Conclusion Main Results Motivation What are s (SEZ)s? A demarcated area of land within a country to encourage foreign invested industry and services for exports using special laws and policies. In Chinese context, SEZ means a land using property rights protection, tax breaks and a preferential land policy for foreign investors.

uotaintro Roadmap Reform Review A Conceptual Framework Data Motivation and Identi cation Research Question Results Conclusion Main Results Why study Chinese s (SEZ)s? During the last 30 years, China has been greatly integrated into the world in terms of its FDI, exports and foreign exchange reserve. SEZs are very important institutional foundations which drove the phenomenon and have been underexplored. World Bank (2008), 3000 zones in 135 countries today, accounting for over 68 million direct jobs and over $500 billion of direct trade-related value added. The exact economic impact of SEZs is empirically understudied.

uotaintro Roadmap Reform Review A Conceptual Framework Data Motivation and Identi cation Research Question Results Conclusion Main Results Research Question From 1979, China has gradually created SEZs in its municipalities. I collect a unique and comprehensive municipality level dataset from 1978 to 2007 to study: Is SEZ e ective in boosting FDI related outcome?

uotaintro Roadmap Reform Review A Conceptual Framework Data Motivation and Identi cation Research Question Results Conclusion Main Results Main results Strategy to tackle the endogeneity of SEZ granting sequence: 1 Dif-in-Dif with municipalitiy speci c trend and xed e ects 2 Dif-in-Dif combined with propensity score matching 3 Restrict sample to later SEZs only 4 Placebo test for pre-sez period and diversion e ect Preview of the results 1 SEZs attract municipality FDI, mainly in the form of export-oriented and foreign-invested industrial enterprises, by 58%; 2 SEZs do not crowd out domestically owned investment and capital; 3 SEZs increase municipality TFP growth rate by 0.6 percentage points.

uotaintro Roadmap Reform Review A Conceptual Framework Data Motivation and Identi cation Research Question Results Conclusion Main Results Related Literature 1 Research on SEZs: Case studies: Kung(1985), Willmore(1996), Ge(1999), Park(1997), Rolfe(2004), Aggarwal(2005). Theoretical studies: Litwack and Qian(1998), Xu(2009) 2 Economic Incentives of FDI Location Decision: Devereux et al (2006), Razin and Sadka (2006), Wang(2008), Du et. al(2009) 3 The Economic Bene ts of FDI: As a source of capital: Whalley & Xin(2006), McGrattan & Prescott (2007), Desai, Foley & Hines(2009); As a channel of technology spillover: Coe, Helpman & Ho maister(2008); Liu(2008), Hale & Long(2007), Abraham et al(2009).

uotaintro Roadmap Reform Review A Conceptual Framework Data and Identi cation Results Conclusion Roadmap Experiment Review A Conceptual Framework Data and Identi cation Strategy Regression Results Conclusion

uotaintro Roadmap Reform Review A Conceptual Framework Data Mapand Privilege Identi cation Results Conclusion Expansion of SEZs

uotaintro Roadmap Reform Review A Conceptual Framework Data Mapand Privilege Identi cation Results Conclusion SEZs Granting Sequence Variables 1978 1979-1985 1986-1990 1991-1995 1996-2007 1.Number of municipalities newly granted SEZs 0 30 49 145 76 2. Number of municipalities with SEZs 0 30 79 224 300 3. Total Number of municipalities 326 326 326 326 326 4. Ratio of municipalities with SEZs 0.0 0.09 0.24 0.69 0.92 5. Average Distance to the coast(100miles) - 0.15 1.34 3.75 6.26 6. Average per capita industry output in 1978-806 611 429 263 7. Average per capita number of secondary students 1978-0.064 0.060 0.066 0.057

uotaintro Roadmap Reform Review A Conceptual Framework Data Mapand Privilege Identi cation Results Conclusion Packages o ered to foreign investors within SEZs Property Rights Protection: Protect the foreign investors property when the constitution of China is not amended until 2004. Tax incentive: Reduced rate (15-24%) of corporate income tax. Virtually zero custom duties. Income tax exemptions for foreigners working in SEZs Land and utility bill: >15 years investment free land use fee; or ve years free, ve years half. 20% o in electricity, water bill, & exemption from local levies. Liberal economic and labor laws: Limited restrictions on foreign ownerships. Power to hire and re. Working permit for foreigners who work in the zones.

uotaintro Roadmap Reform Review A Conceptual Framework Data and Identi cation Results Conclusion A Conceptual Framework Foreign Firm Decision: SEZ by in uencing property rights protection, corporate tax rate and land rental rate will attract FDI. Hypothesis 1 LnFDI it = α + η SEZdummy it +X it β + ξ it Capital Formation: SEZ by attracting FDI, may crowd out or not of domestic investment & capital depending on the interaction between FDI and domestic economy. Hypothesis 2 LnDomI it =φ + γ SEZdummy it +X it β + ξ it LnDomK it =φ + γ SEZdummy it +X it β + ξ it Technology Progress: If SEZ brings not only foreign capital, but also more advanced technology, it will increase TFP growth. Hypothesis 3 TFP TFP = λ i +γ SEZdummy it +ε it

uotaintro Roadmap Reform Review A Conceptual Framework Data DatandIdenti cation Results Conclusion Data 326 Municipalities from 1978 to 2007. 1 Index SEZdummy = 1, if the municipality is exposed to the SEZ experiment; SEZdummy = 0, if otherwise. 2 FDI Related Outcome Utilized foreign direct investment; Exports; Industrial output by foreign invested enterprises 3 Growth Accounting Data Gross Domestic Output (Y ) & De ator for GDP and Investment Real Physical Capital Stock(K f + K d ) Labor(L) and Human Capital (H) Labor and Capital Share

uotaintro Roadmap Reform Review A Conceptual Framework Data DatandIdenti cation Results Conclusion Identi cation: Di erence in Di erence speci cation Y ipt = α + β SEZdummy ipt + δ i + γ t + ε ipt (1) Y ipt = α + β SEZdummy ipt + δ i + δ p (t 1977) + γ t + ε ipt (2) Y ipt = α + β SEZdummy ipt + δ i + δ i (t 1977) + γ t + ε ipt (3) i, Municipality; p, Province; t, Year. Endogeneity of selection: Time varying factors municipality speci c trend; Time invariant factors municipality xed e ect. Other Chinese macro events/reforms - year xed e ect.

uotaintro Roadmap Reform Review A Conceptual Framework Data DatandIdenti cation Results Conclusion Identi cation: Dif-in-Dif with Matching For dif-in-dif estimates to be unbiased, the timing of the SEZ experiment should be random. Then late treated municipalities are valid control groups for early ones. But is the timing of SEZ granting random? No. A possible identi cation strategy is to assume that given a score of relevant selection criteria X, outcomes are independent of treatment assignment, i.e. SEZ timing is random conditional on P(X ). I use nearest neighbor approach to match early treated municipalities with the closest counterparts that had SEZs later. Propensity score estimation using a probit model PrfD = 1jX g X = (industrial output, education, geographical location) D = 1 is de ned as being granted SEZs by 1992; D = 0 otherwise. Restrict sample to matched 247 municipalities.

uotaintro Roadmap Reform Review A Conceptual Framework Data DatandIdenti cation Results Conclusion a. Probit Regression Results Variable Coe cient (std. error) Industry output78 0.0006*** (0.0002) Secondarystudent78-3.0710 (3.3481) Distance -0.0930*** (0.0197) Log Likelihood -204.11 Pseudo R-squared 0.096 Number of Obs. 326 b. Comparison Before and After Matching Mean Percent Reduction t-test Variable Sample Treated Control Bias in bias t p> t Industry78 Unmatched 565.26 307.43 46.0 4.13 0.000 Matched 565.26 595.22-5.3 88.4-0.42 0.672 Secondarystudent78 Unmatched 0.064 0.060 15.8 1.43 0.153 Matched 0.064 0.065-3.8 76.0-0.34 0.734 Distance Unmatched 2.69 5.06-60.1-5.44 0.000 Matched 2.69 2.91-5.7 90.4-0.64 0.526

uotaintro Roadmap Reform Review A Conceptual Framework Data DatandIdenti cation Results Conclusion Identi cation: Later SEZs only Speci cation Is the result mainly driven by the potential "Special" group which were granted SEZs earlier? After 1990s, big wave in granting, less concern on selection. Restrict sample to municipalities having SEZs post 1990.

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on SEZs on FDI Ln(Per capita Foreign Direct Investment) Full Sample Matched Later SEZ (1) (2) (3) Sample Sample SEZdummy 0.723*** 0.478*** 0.460*** 0.434*** 0.355*** (0.079) (0.056) (0.053) (0.061) (0.061) Municipality FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Province Trend - Yes - - - Municipality Trend - - Yes Yes Yes Observations 9772 9772 9772 7405 7404 Adj R-squared 0.761 0.845 0.891 0.898 0.845

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on SEZs on Exports Ln(Per capita Exports) Full Sample Matched Later SEZ (1) (2) (3) Sample Sample SEZdummy 0.871*** 0.719*** 0.608*** 0.595*** 0.531*** (0.098) (0.080) (0.082) (0.096) (0.092) Municipality FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Province Trend - Yes - - - Municipality Trend - - Yes Yes Yes Observations 9733 9733 9733 7391 7376 Adj R-squared 0.833 0.877 0.922 0.927 0.898

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on SEZs on Industrial Output by Foreign Invested Firms Ln(Per capita Industrial Output of FIEs) Full Sample Matched Later SEZ (1) (2) (3) Sample Sample SEZdummy 0.307*** 0.275** 0.487*** 0.525*** 0.375** (0.137) (0.124) (0.128) (0.130) (0.184) Municipality FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Province Trend - Yes - - - Municipality Trend - - Yes Yes Yes Observations 3667 3667 3667 3055 2604 Adj R-squared 0.906 0.922 0.943 0.946 0.935

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on Placebo Test Panel A Panel B Panel C Ln(Per capita FDI) Ln(Per capita exports) Ln(Per capita FIE output) SEZ(-2) -0.006 0.040-0.123 (0.040) (0.059) (0.108) SEZ(-1) 0.046 0.125* 0.022 (0.047) (0.070) (0.163) SEZ(+0) 0.150*** 0.354*** 0.205 (0.056) (0.087) (0.210) SEZ(+1) 0.376*** 0.507*** 0.631*** (0.068) (0.101) (0.201) SEZ(+2) 0.445*** 0.630*** 0.466* (0.073) (0.112) (0.250) SEZ(3+) 0.880*** 1.024*** 0.765*** (0.083) (0.123) (0.254) Municipality FE YES YES YES Year FE YES YES YES Municipality Trend YES YES YES Observations 9772 9733 3667 Adj R-squared 0.897 0.925 0.944

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on Test for Diversion E ect Within Province

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on Test for Diversion E ect Cross Province

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on SEZs on Domestic Investment Ln(Real Domestic lnvestment) Full Sample Matched Later SEZ (1) (2) (3) Sample Sample SEZdummy 0.044 0.007 0.067** 0.087** -0.017 (0.044) (0.039) (0.033) (0.037) (0.040) Municipality FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Province Trend - Yes - - - Municipality Trend - - Yes Yes Yes Observations 9732 9732 9732 7364 7399 Adj R-squared 0.931 0.942 0.960 0.961 0.960

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on SEZs on Domestic Capital Formation Ln(Real Domestic Capital) Full Sample Matched Later SEZ (1) (2) (3) Sample Sample SEZdummy 0.027 0.012 0.067*** 0.078*** 0.023 (0.038) (0.033) (0.022) (0.025) (0.025) Municipality FE Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Province Trend - Yes - - - Municipality Trend - - Yes Yes Yes Observations 9669 9669 9669 7301 7379 Adj R-squared 0.958 0.967 0.986 0.986 0.986

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on SEZs on TFP Growth Let gross domestic output be a constant returns to scale function of capital and labor inputs(human capital augmented) Y = F (K, H L, t) where the appearance of t, time, denotes the fact that the production function evolves over time due to technology progress. TFP TFP = Y Y To recall Hypothesis 3: θ k K K (H L) (1 θ k ) (H L) TFP TFP = λ i +γ SEZdummy it +ε it The e ect of SEZs on TFP growth will be identi ed by the deviation from each municipality s own growth rate level λ i.

uotaintro Roadmap Reform Review A Conceptual Framework Data Stepand One: Identi cation SEZ Experiment Results on FDI Conclusion Outcome Step Two: SEZ on SEZs on TFP Growth Cont d TFP Growth Capital share Capital share Capital share Using Province average Using National average Using 1/3 (1) (2) (1) (2) (1) (2) SEZdummy 0.035*** 0.006** 0.034*** 0.006* 0.035*** 0.007** (0.002) (0.003) (0.002) (0.003) (0.002) (0.003) Municipality FE Yes Yes Yes Yes Yes Yes Year FE - Yes - Yes - Yes Observations 9440 9440 9440 9440 9440 9440 R-squared 0.071 0.132 0.072 0.132 0.074 0.136

uotaintro Roadmap Reform Review A Conceptual Framework Data and Identi cation Results Conclusion Conclusion, by providing property right protection and economic incentives to foreign investors, attracts foreign direct investment in the form of export oriented industrial enterprises., by bringing FDI, does not crowd out domestic investment and capital stock. SEZ, bringing more advanced FDI, increase municipality TFP growth by 0.6 percentage points.