Professional Documents
Culture Documents
doi:10.1093/cjres/rsz012
Advance Access publication 5 October 2019
We analyse one of the most important policy experiences for industrial clustering in
Southern China—the Specialised Towns programme—that has transformed some Chinese
clusters into the backbone of global production chains. We offer a long-term, detailed over-
view of the policy programme and of Guangdong’s specialised towns, classifying them as
endogenous or exogenous according to their features, and investigate their contribution to
local growth and rebalancing. This analysis of the Specialised Towns programme contrib-
utes to the international debate on revisiting industrial policy and suggests that the discus-
sion should conceive them as articulated processes to reach long-term societal objectives.
© The Author(s) 2019. Published by Oxford University Press on behalf of the Cambridge Political Economy Society.
All rights reserved. For permissions, please email: journals.permissions@oup.com
Barbieri et al.
of the province’s growth is its outward orienta- come from the ST of Chenghai (Jankowiak,
tion. In 2017, Guangdong accounted for nearly 2017) and 20% of smartphones manufactured
30% of China’s national exports1 and 15% of its globally every year comes from Dongguan City
total foreign direct investment (FDI).2 It had a (where 93% of the total townships participate
402
Industrial policy in China
clusters, the Chinese government restruc- The article is organised as follows. After dis-
tured it to support, drive and accelerate clus- cussing the international literature useful to in-
ters growth (Rubini et al., 2015). terpret the STs’ experience, we describe data
3 . Central role of the Department of Science and methodology. Next, we offer a detailed
403
Barbieri et al.
scholars have also warned that without proper key elements, in particular the existence of a
design and implementation, cluster policies community of firms and people giving rise to
might even produce negative consequences the industrial atmosphere (Becattini, 1987). In
(Brakman and van Marrewijk, 2013). Within Guangdong, in fact, clusters often result from
404
Industrial policy in China
Hesse-Biber and Johnson, 2015) by using both technologies and upgrading production. They
qualitative and quantitative data. also encourage cooperation among the town’s
To analyse the STs, we built a novel data- firms, private and public engineering centres
base, containing information about the year and public research institutions in innova-
405
Barbieri et al.
economy, including industrial production (Di at the centre of Guangdong industrial growth
Tommaso et al., 2013). In the Spark Plan zones, since the beginning of China’s open-door policy
firms could benefit from subsidies and loans (Figure 2(a)) (Zheng et al., 2016).
(Barbieri et al., 2010). Some DSTGG experts The concentration of STs in the PRD re-
realised that most of them tended to specialise flected policymakers’ effort to rationalise this
in specific sectors—a process now known as area’s fast—and often chaotic—growth (Chen,
“one city, one product” (Su and Sun, 2016; 2007; Enright et al., 2005). Such economic ex-
Wang and Yue, 2010), which rests at the foun- pansion must be related with the career incen-
dation of the ST programme. tives given by national rules to local officials.
From 2000 to 2003, the second phase—a Indeed, up to the beginning of the 2000s, local
pilot development of the programme—took officials were mainly promoted according to
place, introduced by two formal documents: their ability to produce economic improve-
“Plan for the ST Technological Innovation: ments (mainly GDP growth) at the local level
Pilot Test” and “Managerial Methods for (Li, 2014). To this aim, local officials often used
the ST Technological Innovation: Pilot Test” transactions in the land market, which indeed
(DSTGG, 2008). Most of the STs recognised in boomed between 1988 and 1992. Land con-
this phase are in the Pearl River Delta (PRD), veyance fees generated by such transactions
406
Industrial policy in China
were used for infrastructural financing and to encourage development of lagging areas
upgrading, ending in GDP growth (Ahmad (Barbieri et al., 2012; Sarcina et al., 2014). The
et al., 2018; Cao and Zhao, 2011). Hence, this new STs were established in the so-called Great
phase of the STs programme served the pro- PRD and on the coast and peripheral areas
motion of industrial growth and upgrading, through 2006 (Figure 2(b)), with rural bases and
while rationalising industrial development mountainous territory through 2009 (Figure
(Barbieri et al., 2009). The sectoral specialisa- 2(c)). They had different productive peculi-
tion of these first towns reflected the features of arities, such as being focussed on agriculture-
Guangdong’s—and the PRD’s—manufacturing related, resource-based activities or services
systems at the time, mainly based on low-tech (Figure 3).
production (Figure 3). Some official documents12 highlighted the
In the third phase (2003–2008), the STs pro- double path behind the policy implementa-
gramme extended to some non-PRD cities tion of this phase and recognised the strategic
and, along with upgrading, started to be used value of STs in Guangdong’s development
407
Barbieri et al.
100%
5
26 34 56
90%
80%
69
60%
34 96
50%
39
40% 19
13 46
30% 17
6 62
20% 102
46
10% 9
8 17 29
0% 2
2003 2006 2009 2015
(Barbieri et al., 2010; Lin, 2006; Su and Sun, 1. to further decentralise the economic man-
2016). They stressed how provincial and local agement of STs to local authorities;
institutions started to collaborate in clus- 2. to optimise the industrial and urban struc-
tering processes to avoid sectoral overlap and ture of STs;
to coordinate the regional innovation system. 3. to restore public accounts and realise infra-
This is reflected also in the establishment of structural investments;
the Guangdong Provincial ST Development 4. to promote quality upgrading of traditional
Promotion Association (POTIC) (Su and Sun, industries and
2016). Furthermore, many of these documents 5. to increase GDP, innovation potential and,
pointed to the diffusion of good practices re- more generally, competitive capacity (Tsai,
lated to knowledge diffusion and protection and 2013).
territorial branding. Finally, they emphasised
collaboration between local actors and univer- Similar to this is the “One Town, One Institute”
sities and research institutions. initiative (2012), aimed at founding a third-
Since 2008, a fourth phase has begun, with level degree institution in each township (Su
a new emphasis on endogenous growth and and Sun, 2016). Finally, the initiatives issued
innovation, in line with the whole change in by the provincial agencies stress two elem-
post-crisis Chinese policy orientation. The ents to empower endogenous sources: (i) pro-
most important measure in this framework is moting micro, small and medium enterprises
the “One Town, One Policy” (2010), incentiv- (SME) and establishing dedicated public
ising each town to find its own developmental innovation services, and (ii) enhancing the
path while supporting the coordinated eco- quality and innovative performances of those
nomic development of Guangdong Province towns specialised in locally rooted productions
(DSTC, 2011a, 2011b). This implies the fol- (Di Tommaso et al., 2013; DSTGG, 2017; Su
lowing actions: and Sun, 2016).
408
Industrial policy in China
Nowadays, almost all Guangdong coun- To identify the two groups, we referred to
ties host at least one ST, and while a high Becattini’s (2015) contribution about industrial
number is still in the PRD (particularly in districts as originated by the historical profile of
Dongguan), many others are also localised in the production atmosphere of the hosting areas.
Sectors
409
Barbieri et al.
45 PRD Non-PRD
40
35
Shaoguan
Dongguan
Jiangmen
Heyuan
Jieyang
Qingyuan
Foshan
Huizhou (PRD)
Zhaoqing (PRD)
Shantou
Yangjiang
Guangzhou
Huizhou (non-PRD)
Meizhou
Shanwei
Yunfu
Zhanjiang
Zhaoqing (non-PRD)
Zhongshan
Chaozhou
Zhuhai
Maoming
Exogenous STs Endogenous STs
Figure 4. Geographical distribution of endogenous and exogenous specialised towns, year 2015.
Source: Authors’ elaboration.
endogenous sources (Supplementary Figure 2. RQ2: Can the policy contribute to territorial
A1). Their geographical distribution confirms rebalancing within the province?
different specialisation paths between PRD
and non-PRD (Figure 4). Endogenous town- We use data at the most detailed territorial
ships are mainly concentrated in non-PRD level—that is, counties and districts—which
inner and mountainous areas, testifying to a is yet above the township one. We mainly
larger policymaking investment to promote use panel fixed-effects models to test for
these prefectures by focussing on the local robust correlations between the policy
market and existing productions. The predom- and local economic performance. Given
inance of exogenous specialisation in the PRD, the structure of the data and considering
conversely, clearly indicates its exogenous eco- that policy implementation continuously
nomic growth path, based on foreign capital changes through time and involves the great
and national migrations, although with some majority of Guangdong counties, there are
notable exceptions that founded their eco- obstacles in building proper counterfactuals
nomic growth on traditional productions. to explore causality. 14 However, to address
some causal interpretation, we add a ro-
bustness check based on latest techniques
Empirical analysis to estimate dose-response effects, particu-
We now turn to investigate the two empirical larly fit for analysing non-binary treatments
research questions: as in our case. We use the Cerulli (2015)15
fixed-effect dose-response method, which,
1. RQ1: Is the ST programme associ-
compared with usual General Propensity
ated with improved territorial economic Scores methods (Bia and Mattei, 2008;
performances? Hirano and Imbens, 2004), allows to include
410
Industrial policy in China
non-treated units in the analysis and to con- 3 years.20 Summary statistics for the relevant
sider increasing treatment intensity across variables are in Supplementary Table A1.
time. Our outcome variables are industrial To analyse the association between STs
and agricultural output in county i at time t and territorial rebalancing (RQ2), we divide
411
Barbieri et al.
intermediate doses. These results suggest a posi- suggest that the relation with industrial output
tive and significant relation between STs and is flat and non-significant up to 60% of the dose
output, when STs are sufficiently numerous (about five towns), and then it significantly in-
(50–60% of the dose, corresponding to about 20 creases.This relation becomes U-shaped including
towns). the total number of STs as a control (Figure 6(b)):
Figure 6 reports the dose-response related to for small amounts of endogenous STs, the rela-
endogenous towns. The first results (Figure 6(a)) tion is decreasing—and negative—becoming then
412
Industrial policy in China
1-year lag effect on Industrial output 3-year lag effect on Industrial output
Figure 5. Dose-response function of number of STs on industrial and agricultural output.
positive and increasing. For agricultural output local policymakers and scholars about spe-
(in both cases), the relation between endogenous cific STs cases. Specifically, PRD’s STs have
towns and output is positive and increasing, par- followed two different paths. One is based on
ticularly for doses’ amounts above 60% (again, FDI and exports that generated some of the
five towns). most important national production and ex-
Our interpretation of the results comes from port hubs. An example is Dongguan, whose
the combination of the panel fixed-effects es- specialisation has been mainly linked to FDI-
timations and the dose-response models. First, led ICT production (Wang and Lin, 2008;
we observe significant and positive results Zhou et al., 2011; our interviews). The second
in both models only for endogenous towns. development pattern, conversely, has been
Particularly, for industrial output, these re- more based on local firms and traditional
sults are relevant for PRD counties (Table 2) products for internal demand. This is the
and for large numbers of endogenous towns case, for example, of Foshan, whose indus-
(Figure 6). The results concerning the PRD trialisation was based more on local forces
can be read in the light of its economic de- than foreign capital but becoming, however,
velopment history (Di Tommaso et al., 2013; a core Guangdong’s manufacturing centre
OECD, 2010; Zeng, 2010). They are further (Di Tommaso et al., 2013; NBS, 2010; Zeng,
reinforced by the fieldwork interviews to 2010; our interviews). The positive relation
413
Barbieri et al.
Figure 6. Dose-response function of number of endogenous towns on industrial and agricultural output.
between endogenous towns and industrial marginal, non-PRD areas, engaged in agriculture-
output seems to reflect the performance of related and resource-based productions.
this second group of clusters.
We also find positive association between The effects of specialised towns and
industrial output and large numbers of en- endogenous specialised towns on
dogenous townships. Excluding Foshan, all territorial rebalancing (RQ2)
the counties hosting numerous endogenous
Table 3 reports the subgroup analysis results.
townships are in few non-PRD peripheral
The only persistent positive association is re-
counties. This would suggest that, for high lated to t-1 and t-3 RATIO_ENDOG for areas
numbers of endogenous clusters, positive ef- with initial medium-low levels of output (all out-
fects could be observed also in remote areas. side the PRD). Coupled with previous evidence
The evidence related to agricultural output, of the dose-response, this result reinforces that,
consistent across the models, shows a possible also for some non-PRD areas, large amounts
positive effect of endogenous towns. This is of endogenous towns might promote industrial
linked with the pattern of specialisation of more development. Therefore, although PRD may
414
Industrial policy in China
Figure 6. Continued
particularly benefit from the establishment of en- 1. The evolution of Guangdong clusters
dogenous towns, under some circumstances, such calls for attention on the relationship
types of clusters might be desirable also for mar- between policies’ objectives and tools
ginal areas in a territorial rebalancing perspective. rather than only on the tools. Indeed,
we have highlighted how the same tool,
that is, specialised clustering promotion,
Final remarks was used for accommodating different
Guangdong ST programme is perhaps one evolving goals (for example, from mere
of the largest-scaled experiments of policy- industrialisation to territorial rebalan-
supported territorial industrial transformation cing). The ST experience reinforces the
worldwide. It has contributed to the province’s statement made in a number of academic
widespread industrial growth, shaping its geog- contributions: industrial policy should no
raphy of production and placing some of its longer be viewed as a mere discussion on
towns at the centre of global production chains. the best tools to achieve well-established
We believe this experience draws some useful aims. Conversely, it is a complex process
lessons for the international debate about in- entailing considerations on a society’s
dustrial policy. long-term broad aims (Andreoni and
415
Barbieri et al.
Table 3. Effects of specialisation and endogenous clusters on rebalancing (Dependent Variable – ln_INDOUT).
Chang, 2016; Andreoni and Scazzieri, 3. Concerning clustering policies, our evidence
2014; Chang, 1994; Cimoli et al., 2009; Di suggests that using different types and com-
Tommaso and Schweitzer, 2013). In this binations of specialisation and localisation
view, industrial policy is not just a set of can produce different results, corresponding
instruments to achieve fixed purposes. It to different policy aims. Additionally, STs
should be rather conceived as a process experience confirms that, as for mature
that drives structural economic and social economies (Bailey et al., 2015), also for
change. developing countries, an approach to sec-
2. China’s ability to implement industrial
toral specialisation and upgrading tailored to
policies is founded on a well-defined in- local contexts might improve local economic
stitutional architecture, with agencies and performances better than exogenous-driven
personnel, whose functioning has been growth. It appears to us that policymakers
changing, dedicated to design, experiment might have followed this path, passing from
and apply policies. In particular, career in- a place-neutral attitude in the beginning to a
centives have been lately aligned with new more place-based perspective later.
long-term industrial policy goals, shifting
from mere economic growth to social and Finally, the STs programme is not free from
environmental issues.21 This would suggest important limitations, which also draw general
that revitalised industrial policies must in- policy implications and set the line for future
clude investments in government capacity research. Indeed, environmental and social
building. sustainability issues have become increasingly
416
Industrial policy in China
serious in Guangdong, also given its fast in- in developing areas, such as the garment cluster in
dustrialisation. These aspects have recently Nairobi (Van Dijk and Rabellotti, 2005) or the knife
generated growing concerns and some first cluster in Pakistan (Nadvi, 1999).
policymaking reactions. Accordingly, it seems 8
The debate is wide and cannot be treated here.
emerging countries, an aspect that has been main institutions that we visited.
neglected so far. For each yuan granted from the province, a city
11
5
http://www.hktdc.com/web/featured_suppliers/ that acknowledge the difficulties in making proper
dongguan/index.html. policy evaluation of clustering initiatives (Brakman
6
Even if there are other areas in China character- and van Marrewijk, 2013; Vicente, 2018).
ised by the presence of industrial clusters (such as 15
We use the fixed-effects version of the model.
Zhejiang Province), these are not the target of a spe- Compared to its ordinary least squares (OLS) ver-
cific policy programme. sion (Filippetti and Cerulli, 2018), the results are
7
The industrial clusters category includes very het- equivalent to that obtained by adding individual
erogeneous realities, ranging from agglomerations of fixed effects to a pooled OLS estimation.
high-tech companies in advanced countries (such as The main variable on which we analyse the possible
16
Sophie-Antipolis in France, the Cambridge region in effects of agglomerations is industrial output. With ref-
the UK or the Silicon Valley in the USA) to groups erence to the first part of the empirical analysis, how-
of SMEs specialised in traditional sectors located ever, we also observe the effects on agricultural output
417
Barbieri et al.
for two reasons: first, one-fourth of the specialised on the Agenda? Industrial Policy revisited confer-
towns specialise in agricultural products; second, many ences (Bath, 26 April 2018 and Cambridge, 12–13
of the inner or peripheral areas count on its agricultural July 2018) for the insightful comments and sugges-
sector for a large part of their total output. tions on the paper. We would also like to thank Paolo
418
Industrial policy in China
Becattini, G. (1987) Mercato e Forze Locali - Il endogenous treatment, The Stata Journal, 15:
Distretto Industriale. Bologna: Il Mulino. 1019–1045.
Becattini, G. (2015) La coscienza Dei Luoghi. Roma: Chang, H. J. (1994) The Political Economy of
Donzelli. Industrial Policy. New York: Macmillan.
Becattini, G., Bellandi, M. and De Propris, L. Chen, X. (2007) A tale of two regions in China: rapid
419
Barbieri et al.
online at: http://www.zhyz.gov.cn [accessed 18 Hildreth, P. and Bailey, D. (2014) Place-based eco-
March 2009, in Chinese]. nomic development strategy in England: filling the
DSTGG (2017) A data analysis of the development missing space, Local Economy, 29: 363–377.
of specialized towns (in Chinese), Internal Report. Hirano, K. and Imbens, G. W. (2004) The propensity
Eisingerich, A. B., Bell, S. J. and Tracey, P. (2010) score with continuous treatments. In A. Gelman
420
Industrial policy in China
Lyu, L. and McCarthy, L. (2015) “Logalization”: dynamics: China. In J. D. Wright (ed.) International
local–global processes and the Shiling leather Encyclopedia of the Social and Behavioural
industrial district in Guangzhou, China, Asian Sciences, 2nd ed, pp. 207–212. Amsterdam: Elsevier.
Geographer, 32: 37–57. Sarcina, A., Di Tommaso, M. R. and Bonnini, S.
Marshall, A. (1890) Principles of Economics. London: (2014) China: industrialization, growth and ter-
421
Barbieri et al.
Wang, J. and Yue, F. (2010) Cluster development Vice Director of Department of Science
and the role of government: the case of Xiqiao and Technology of Guangdong, Journal of
textile cluster in Guangdong. In D. Z. Zeng (ed.) Guangdong Science and Technology, 2: 15–18 (in
Building Engines for Growth and Competitiveness Chinese).
in China: Experience with Special Economic Zones Zheng, G., Barbieri, E., Di Tommaso, M. R. and
422