Spillovers and Path Dependences in the Chinese Manufacturing Industry: A Firm-Level Analysis

Abstract

In China, numerous policy interventions have been undertaken to promote exports of the manufacturing industry. In this paper, we study how the burgeoning export sector has affected the performance of the manufacturing industry. In particular, we focus our attention on two dimensions: efficiency and technological advancement, and the presence of spillovers and path dependences. The main feature of our empirical study is to distinguish between different types of firms according to their output orientation, ownership, and technology intensity. We also recognize that outputs for domestic use and exports may be produced using different technologies. Our results demonstrate the superiority of multi-output firms in both efficiency and technological advancement. Further, these firms generate strong outgoing spillovers to the industry. In contrast, export-only firms are found to be the most inefficient and technologically underdeveloped, while barely generating outgoing spillovers. In terms of efficiency and technological advancement, we also find that foreign- and privately-owned firms and high-tech firms are the best performers. Finally, we find evidence of strong path dependence and weak absorptive capacities for incoming spillovers in the manufacturing industry. We give targeted policy recommendations based on these findings.

Acknowledgements

The data file and the MATLAB program that replicates the analysis are available upon request.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplementary Materials are available for this article which can be accessed via the online version of this journal available at https://doi.org/10.1080/00220388.2019.1605058

Notes

1. According to WTO (2016), over 70 per cent of all exported merchandise is subject to a rebate rate between 15 and 17%. In 2015, China’s VAT rebate totalled 1.29 trillion RMB, representing 17.5 per cent of the total turnover tax revenue of the nation (China Financial Yearbook).

2. In 2016, total merchandise exports from the 219 national SEZs reached 2.7 trillion RMB, or 19.5 per cent of the national total. In the same year, total FDI in the SEZs was 330 billion RMB, or 39.4 per cent of the national total (China Association of Development Zones).

3. For example, the Export-Import Bank of China provides government concessional loans and preferential credit to exporters, and the China Export and Credit Insurance Corporation provides export credit insurance, reinsurance and credit guarantees. To provide an idea of the importance of these measures, in 2014, the Export-Import Bank of China disbursed 178.6 billion RMB in export sellers’ credit and 59.4 billion RMB in export buyers’ credit (WTO, 2016).

4. The prevailing practice in the literature is to define firms with non-zero exports as exporters (Bernard & Jensen, 1995). Recent literature has documented the coexistence of firms of different product-orientation. Lu (2010) reported a U-shaped distribution of export intensity among Chinese firms–that is, a large number of Chinese firms are either dedicated to domestic production or export production, yet many serve both markets. Lu et al. (2010) made a similar observation: firms that are dedicated to domestic production, export production and hybrid production are comparable in numbers among foreign affiliates. Our data indicate that the number of multi-output producers is about three times the number of export-only firms (Table S2 in the Supplementary Material).

5. Yu, Ye, and Qu (2013) and Elliott and Zhou (2013) are two exceptions. Interestingly, Elliott and Zhou (2013) found that controlling for export status has a huge influence on the relative performance of state-owned firms.

6. While previous studies recognize the importance of this distinction (Amiti & Freund, 2010; Jarreau & Poncet, 2012), few studies have related technology intensity to efficiency or technological advancement.

7. A recent theoretical innovation was made by Bustos (2011), who modelled technology choice together with export choice. It was shown that only exporters adopt advanced technology. However, once the technology is chosen, firms produce exports and domestic merchandise using the same technology.

8. Koopman et al. (2012); Ma, Wang, and Zhu (2015); and Upward et al. (2013) reported that a substantial fraction (about 50% in our period of study) of Chinese exporters participate in processing trade, and this type of trade uses a large amount of imported intermediate goods. Consequently, their average share of domestic value added is low (25.4%), while that of non-processing firms is nearly 90 per cent (Koopman et al., 2012). The huge difference in input decomposition indicates that at least some exports are produced using a different technology.

9. Our data indicate that 17.7 per cent of all firms were foreign owned which generated 66.7 per cent of total exports during the period of study.

10. In microeconomic contexts, increasing attention has been devoted to the input-output linkages in efficiency analysis. See, for example, Cherchye, De Rock, and Walheer (2015, 2016); Ding, Dong, Liang, and Zhu (2017); Silva (2018); and Walheer (2018b, 2018c, 2019c). The main motivation is to improve the realism and flexibility of modelling the production process. Recently, these techniques have also been used in macroeconomic contexts. See, for example, Walheer (2016a, 2016b, 2018a, in press).

11. The concept of meta-technology is based on the notion of meta-production function by Hayami and Ruttan (1970). See also Battese, Rao, and O’Donnell (2004), and O’Donnell et al. (2008) for more detail about this methodology.

12. The validity of the chosen returns-to-scale assumption can be tested. We follow the technique discussed in Walheer (2019a) and find that constant returns-to-scale is acceptable for our empirical study. In addition, Equations (17) and (18) can easily be adapted to consider other returns-to-scale assumptions; in practice, it suffices to add constraints for the multipliers.

13. Sector 42 (art ware and other manufacturing) is unclassified and is subsequently dropped from our analysis. Thus, we lose 37,808 observations, which is a small number compared with the size of the full sample (over 2.2 million).

14. This occurs when registered capital is classified as legal person capital, which can be eventually owned by any entity. In the second step, ownership is well defined only if the registration information is clear about the control rights. Specifically, firms registered as state-owned enterprises, state-owned partnerships and state-owned limited liability companies are defined as state-owned; firms registered as collective enterprises and collective partnerships are defined as collective-owned; firms registered as sole proprietorships, private partnerships, private limited liability companies and private joint-stock companies are defined as private-owned; firms registered as (wholly) foreign-owned are defined as foreign-owned. All other registration types are left undefined.

15. Greenaway, Guariglia, and Yu (2014); Walheer (2019b); and Wang and Wang (2015) recently highlighted the importance of foreign capital when studying Chinese firms. Unfortunately, the distinction between foreign and domestic capital is not available in the CIS dataset.

16. We refer to the online appendix of Brandt et al. (2012) for more detail.

17. We may think of factories and machines being used to produce both domestic outputs and exports, but the same employee cannot be used to produce both outputs simultaneously.

18. For example, refer to Cherchye, De Rock, and Walheer (2016); Fukuyama and Weber (2008); and Sahoo, Mehdiloozad, and Tone (2014) for more discussion about considering quality in efficiency analysis. Interestingly, in the trade literature, the quality of an output is usually associated with or measured by its price (unit value). Examples include Bastos and Silva (2010); Bernard, Redding, and Schott (2011); and Hummels and Klenow (2005).

19. The averages are computed by considering the relative importance of the firms. In practice, this is achieved by using the relative output shares as weights when computing the averages. This weighting scheme is also justified theoretically. Refer to Färe and Zelenyuk (2003) for the efficiency ratio, and Walheer (2018a) for the technology gap ratio.

20. In Table 1, this is captured by $TGR=TG{R}^d$ and $ER=E{R}^d$ for domestic firms, and by $TGR=TG{R}^e$ and $ER=E{R}^e$ for export-only firms.

21. Direct evidence usually comes from the customs data, which provide information on the nature of imported intermediates (processing or ordinary imports). Through using these data, one can isolate processing exporters from other exporters, as done, for example, by Dai et al. (2016). However, these data are unavailable in our case.

22. A similar treatment has been used in, for example, Delgado, Fariñas, and Ruano (2002).

23. There, the switchers are defined as domestic firms in 2003 that switch to export-only firms at some point between 2004 and 2007.

24. ER is significant at the 10 per cent significance level.

25. Alternatively, the presence of path dependence and spillovers can be tested using (parametric) econometric methods. We refer to Tsekouras et al. (2016) for more detail about this approach.

26. We thank the editor and an anonymous referee for suggesting these additional tests.

27. The only exception occurs when we study path dependence of the technical efficiency ratio using the common-technology assumption (Table S9). There, we find that the coefficient of multi-output firms is slightly larger than that of domestic firms.

28. These include the Catalog of Industries, Products, and Technologies Currently Particularly Encouraged by the State (2000) and the Catalog for Guiding Industrial Restructuring (2005, 2011 and 2013 editions), all developed by the National Development and Reform Commission.