Effect of global value chain participation on firms' productivity improvement: a PSM based empirical study

XI Yanle1 HE Lifang1

(1.School of Business Administration of Zhongnan University of Economics and Law)

【Abstract】Whether participating in the global value chain is a preferred choice of firms to improve productivity is a topic of realistic concern and theoretical value. Based on Chinese Industrial Enterprise Database and Customs Database between 2000 and 2006, this paper uses PSM and a variety of productivity measures to investigate in detail into this question to provide micro empirical evidence for Chinese companies to step out of the "value depression" and extend to both ends of the smiling curve. The results show that participation in global value chain has a durable and progressive effect on firms' TFP, but it is likely that the driving force of this effect is a simple "improvement" of labor productivity through non-technological innovation factors such as in the production process and the mode of organization and management rather than a real improvement of capital productivity and a real deepening of capital through technological innovation factors; the productivity effect is different for different participating paths, among which firms who first export and then import intermediate inputs to further participate in global value chain gain more productivity than the other two paths; the productivity effect of participating in the global value chain exists both before or after the event, which may be explained by preparations before the participation.

【Keywords】 global value chain; productivity; participating path; prior effect; propensity score match (PSM);


【Funds】 2015 Youth Project of National Natural Science Foundation of China (71503271) 2014 Humanities and Social Sciences Youth Project of Ministry of Education (14YJC790137)

Download this article

(Translated by ZHANG Liang)


    [1]. ① According to Baldwin & Yan’s (2014) defining approach, this article uses the integration with global value chain of the enterprises who are engaged in intermediate products import and export at the same time as the sign. Certainly, this defining approach is a little rough but there are still no better identifying approaches besides this. [^Back]

    [2]. ② In the specific research process, this article uses Kernel matching method because this method uses all control group enterprises' weighted average value to construct treatment group enterprises’ matching enterprises in order to improve data utility rate. In the realistic phenomenon of big number gap between control group and treatment group enterprises, the advantages of this method are more obvious. [^Back]

    [3]. ③ These literature includes Eliasson et al. (2009), Baldwin & Yan (2014) and so on. [^Back]

    [4]. ④ This article chooses Levinsohn & Petrin’s (2003) LP method as the main TFP measurement approach aiming for making up the “simultaneity” errors and “sample selection” errors among traditional Solow residual errors in order to further overcome sample loss caused by OP method. [^Back]

    [5]. ⑤ Due to the limited length of the article, the authors did not list the detailed introduction of matching variables. Please ask the author if necessary. [^Back]

    [6]. ⑥ Please refer to Lu et al. (2014) for the matching method of Chinese Industrial Enterprise Database and Customs Import and Export Database. [^Back]

    [7]. ⑦ According to the method of Tang & Zhang (2012), we regard the enterprises with “foreign trade,” “foreign economics,” “industrial trade,” and “scientific trade” in their names as trade brokers. [^Back]

    [8]. ⑧ Due to the limited length of the article, the authors did not list the statistical results of enterprises’ integration with global value chain from 2000 to 2006. Please ask the author if needed. [^Back]

    [9]. ⑨ Integration experience before 2000 will disturb our judgment about integration behaviors’ productivity effects. Based on the current data, about 7.9%, 2.5%, and 0.9% enterprises will reintegrate one, two, or three years after exit. Therefore, we require that the enterprises not integrated for three consecutive years before integration in order to control enterprises’ accidental exit chance lower than 1% or at least partially trade off the productivity effect left by previous integration practices. [^Back]

    [10]. ⑩ Due to the limited length of the article, the authors did not list PSM and balance test results here. Please ask the author if necessary. [^Back]

    [11]. ⑪ Non-technological innovation factors mean the innovation factors that are not related with technological changes. According to Schumpeter’s argument, “innovation” means introducing unprecedented new combination of production factors and conditions in production system which might be innovation related to technological changes (such as new technology development and new application of original technology) or those related with organizational management pattern and external institution environment. [^Back]


    Chen, Y., Zhang, R. & Cao, L. Finance & Trade Economics (财贸经济), (3) (2012).

    Zhang, J., Li, Y. & Liu, Z. Management World (管理世界), (12)(2009).

    Baldwin J.R.,Yan B.,(2014) Global Value Chains and the Productivity of Canadian Manufacturing Firms,Statistics Canada.

    Le Bris F.,Disdier A. C.,Jaud M.,(2013) “Linking Firm’s Intermediate Input Imports and Export Performances,”European Trade Study Group Fifteenth Annual Conference University of Birmingham.

    Sharma C.,(2014)“Imported Intermediate Inputs,R&D,and Productivity at Firm Level:Evidence from Indian Manufacturing Industries,”The International Trade Journal 28(3),246-263.

    Sharma C.,Mishra R. K.,(2015) “International Trade and Performance of Firms:Unraveling Export,Import and Productivity Puzzle,” The Quarterly Review of Economics and Finance 2(1),1-14.

    Tang H.,Zhang Y.,(2012)“Quality Differentiation and Trade Intermediation,”Development Working Paper No.340.

This Article


CN: 11-1692/F

Vol , No. 12, Pages 39-50

December 2015


Article Outline


  • Introduction
  • 1 Literature review
  • 2 Model and methodology
  • 3 Data processing explanations and descriptive statistics
  • 4 PSM estimate results
  • 5 Conclusion and policy recommendations
  • Footnote