Chapter 8: Cluster Governance and the Systemic Drivers of Green Upgrading
In the winter of 2021, a furniture factory owner in Lecong, the furniture town of Shunde District in Foshan, Guangdong, walked me through a workshop that had been retooled twice in five years. The first retooling, around 2017, was about spraying: water-based coatings replaced solvent-based ones after enforcement teams from the district environmental bureau began checking volatile organic compound emissions at the workshop gate. The second, in 2020, was about dust and energy: new extraction systems, a rooftop solar array of 1.2 megawatts, and a small biomass boiler that burned offcuts. He had not planned either investment. Both arrived as conditions attached to something else—a supplier audit from a European buyer in the first case, a bank loan at preferential rates in the second. "It was not that I sat down and designed a green factory," he said. "The cluster decided for me, and I followed."
That sentence is the subject of this chapter. The previous chapter examined carbon pricing as a signal transmitted through markets—a price that enters a firm's cost calculus and, in principle, reallocates investment toward lower-carbon processes. But prices do not act on firms directly. They act through the institutions, networks, and physical systems in which firms are embedded. In a Chinese industrial cluster, that embedding is unusually dense: firms share suppliers, workers, infrastructure, local government, trade associations, and, often, a common environmental enforcement regime. The density is what makes clusters powerful engines of growth. It is also what makes them the decisive unit of analysis for green upgrading.
The argument I develop here is that cluster governance—the formal and informal rules that coordinate behavior within a localized production system—converts external pressures into collective capability, or fails to. Where governance is strong, a carbon price, a buyer requirement, or an environmental standard becomes a shared problem that the cluster solves through infrastructure, knowledge diffusion, and collective investment. Where governance is weak, the same pressure becomes a cost that individual firms evade, pass backward to suppliers, or escape by relocating. The difference is systemic, not primarily a matter of firm-level entrepreneurship.
8.1 Cluster Governance Structures and Green Upgrading
8.1.1 The cluster as a governance object
The study of industrial clusters has long wrestled with a definitional question: is a cluster simply a geographic concentration of firms in related industries, or is it a system of coordination that produces effects no individual firm could produce alone? Alfred Marshall's original account of industrial districts in Principles of Economics (1890) emphasized what he called the "industrial atmosphere"—the intangible transfer of skills and knowledge that occurs when many small firms in the same trade share a locality. Becattini (1990) refined this into a socio-economic notion: the district as a community of people and firms bound by shared history and institutions, not merely a spatial agglomeration. Porter's (1990) competitive-advantage framing widened the concept to include suppliers, service providers, and related industries, and made the cluster a unit of competitiveness policy.
For our purposes, the useful definition is narrower and more operational. A cluster is a localized production system in which firms in related activities share three things: a labor pool with industry-specific skills, a supplier and service base, and a set of institutions—local government, associations, technical centers, sometimes universities—that coordinate collective action. The third element is what distinguishes a cluster from a mere agglomeration. It is also what makes green upgrading possible, because environmental problems are collective-action problems par excellence: emissions, effluent, and waste cross firm boundaries; infrastructure is shared; and the credibility of a local environmental regime depends on whether all firms comply, not just the visible ones.
The governance of a cluster, then, is the set of mechanisms through which these shared resources are managed and collective decisions are made. These mechanisms are both formal and informal, and the interaction between the two is where the green transition is won or lost.
8.1.2 Formal institutions: local government as the hinge
In the Chinese context, the most important formal institution in cluster governance is the local government—specifically the county-level or township-level government, and its environmental, economic, and land-use bureaus. This is a structural feature of China's administrative system that distinguishes it from cluster governance in, say, Italy's Emilia-Romagna or Germany's Baden-Württemberg, where trade associations and chambers of commerce carry more weight. Chinese local governments are simultaneously regulators, landlords, investors, and economic promoters. They own or control industrial land, they operate development zones, they often hold equity in local utility and infrastructure companies, and their officials are evaluated on a combination of economic growth and, increasingly, environmental performance.
That combination matters. The literature on Chinese local government behavior has documented a shift in the incentive structure over the past decade. Where promotion criteria once weighted GDP growth heavily, the cadre evaluation system now includes binding targets on energy intensity and emissions. The "dual carbon" goals announced in 2020—peak carbon before 2030, carbon neutrality before 2060—were translated into provincial and municipal targets, and those targets cascade downward to the county and township levels where clusters sit. A township official in an industrial cluster now faces a set of environmental performance indicators that carry real career consequences.
This does not mean enforcement is uniform. It varies systematically with the industrial composition of the cluster, the fiscal capacity of the local government, and the political salience of pollution. A cluster of export-oriented furniture firms in the Pearl River Delta faces tighter scrutiny than a cluster of small metal fabricators in an inland county, partly because the former is more visible and partly because the local government has the resources to monitor it. But the direction of travel is clear: local governments in cluster regions have moved from being protectors of firms against environmental enforcement to being conduits of it.
The mechanisms they use are varied. Environmental performance is written into land lease renewals and into eligibility for the preferential electricity tariffs that many clusters enjoy. Firms seeking access to local industrial parks must meet emissions standards as a condition of entry. Local governments operate or subsidize shared environmental infrastructure—centralized wastewater treatment, hazardous waste collection, combined heat and power. And they broker access to green finance: a bank lending to a cluster firm may require a local environmental compliance certificate, which the local bureau issues.
8.1.3 Informal institutions: associations, norms, and the peer effect
Formal rules are only half the story. The other half is informal: the norms, expectations, and peer relationships that determine whether a firm treats environmental compliance as a cost to be minimized or as a condition of belonging.
Trade associations in Chinese clusters are weaker than their European counterparts in formal regulatory authority, but they perform a coordination function that matters for green upgrading. They organize collective purchasing of environmental equipment, negotiate with local government over the timing and stringency of enforcement, operate shared testing laboratories, and run training programs. In the furniture cluster around Lecong, the district furniture association ran a program in the late 2010s to help member firms convert to water-based coatings; the association negotiated a bulk price for the coatings, arranged technical training from the supplier, and lobbied the environmental bureau for a phased compliance schedule. The result was faster diffusion than a purely firm-by-firm regulatory approach would have produced.
Peer effects operate through a different channel. In a dense cluster, firms can observe each other's investments and outcomes. When a well-regarded firm installs a rooftop solar array and reports a favorable payback, others follow. When a firm is publicly sanctioned for exceeding emissions, the reputational damage spreads through the cluster's business networks. This is the "industrial atmosphere" that Marshall described, repurposed for environmental behavior. The evidence on peer effects in Chinese clusters is largely qualitative, but it is consistent: the diffusion of green technologies within clusters follows a pattern of imitation and social learning that is faster than diffusion across dispersed firms facing similar prices.
The interaction between formal and informal institutions is where the analytical leverage lies. A local government that enforces credibly, but leaves room for collective adjustment, produces better outcomes than one that either enforces rigidly without consultation or does not enforce at all. The first approach generates resentment and evasion; the third generates free-riding. The middle path—credible enforcement with a negotiated compliance schedule and shared infrastructure—is what the more successful clusters have found.
8.1.4 The ladder of cluster governance
It is useful to think of cluster governance for green upgrading as a ladder with several rungs. At the bottom, governance is absent or captured: environmental rules exist on paper but are not enforced, and firms have no collective mechanism for addressing shared problems. One rung up, governance is compliance-based: the local government enforces standards firm by firm, and firms respond individually, often by installing end-of-pipe equipment. Higher still, governance is capability-based: the cluster has shared infrastructure, collective learning mechanisms, and a local regime that treats environmental performance as a condition of access to land, finance, and markets. At the top, governance is strategic: the cluster uses environmental performance as a source of differentiation in global markets, and the local regime actively supports upgrading rather than merely enforcing compliance.
Most Chinese clusters are somewhere between the second and third rungs. The furniture cluster in Shunde, the ceramic cluster in Foshan's Chancheng District, the textile cluster in Shaoxing's Keqiao District, and the electronics cluster in Shenzhen's Bao'an District are at different points, and their trajectories differ accordingly. Four mechanisms move clusters up the ladder, and the cases in section 8.3 illustrate them.
8.2 Systemic Drivers: Knowledge, Networks, and Infrastructure
8.2.1 Knowledge diffusion: the cluster as a learning system
The first systemic driver is knowledge. A cluster concentrates not only firms but the tacit knowledge that firms need to adopt new processes. This is the argument that Asheim (2002) made about regional innovation systems: local "sticky" knowledge—the kind that is hard to codify and transfer at a distance—coexists with global "ubiquitous" knowledge, and the cluster is the place where the two are combined.
For green upgrading, the relevant knowledge falls into three categories. The first is technical: how to formulate a water-based coating that adheres to a particular wood species, how to tune a biomass boiler to the moisture content of local agricultural waste, how to design a kiln co-processing system that meets emissions standards. The second is managerial: how to measure and report emissions, how to integrate environmental criteria into supplier selection, how to structure a green finance application. The third is regulatory: what the local environmental bureau will accept, what documentation a buyer's auditor will require, what a carbon market obligation implies for a firm's cost structure.
Clusters diffuse all three more efficiently than markets do. Technical knowledge travels through shared workers, supplier engineers, and the informal consultation that occurs when neighboring firms face the same problem. Managerial knowledge travels through associations, consultants, and the movement of managers between firms. Regulatory knowledge travels through the local government's interactions with firms and through the associations that mediate those interactions.
The diffusion is not automatic. It depends on the presence of what the innovation literature calls "absorptive capacity"—the ability of a firm to recognize, assimilate, and apply new knowledge. Small firms in labor-intensive clusters often lack this capacity. They may not have an engineer on staff, or a manager with the time to attend a training session, or the financial slack to experiment with a new process. This is where cluster-level institutions matter: a shared technical center, a demonstration project, or a supplier-provided training program can substitute for firm-level absorptive capacity.
The evidence from Chinese clusters suggests that demonstration effects are particularly powerful. When a lead firm in a cluster adopts a new green technology and it works, the news travels fast. The furniture cluster in Lecong adopted water-based coatings faster than the national average, in part because a handful of large firms made the switch early and their suppliers and competitors followed. The ceramic cluster in Foshan adopted cleaner kiln technologies in a similar pattern. The mechanism is not that firms imitate blindly; it is that the demonstration reduces uncertainty about performance and cost, which is the main barrier to adoption for a small firm.
8.2.2 Networks: vertical and horizontal channels
Networks form the second driver. A cluster is a dense web of relationships: between firms and their suppliers, between firms and their buyers, between firms and the local government, between firms and the associations and service providers that serve them. These relationships are channels through which pressure and capability flow.
Vertical channels—the relationships between cluster firms and their buyers—are the ones most studied in the global value chain literature. Humphrey and Schmitz (2000, 2002a) distinguished between governance structures in value chains (market, modular, relational, captive, hierarchy) and argued that the type of governance determines the upgrading opportunities available to suppliers. In a captive chain, where a lead firm controls the supplier's access to markets, the supplier has little room to upgrade beyond what the lead firm requires. In a relational chain, where the relationship is based on trust and mutual investment, the supplier has more room to develop capabilities.
For green upgrading, the vertical channel matters because buyers are increasingly the source of environmental requirements. A European furniture retailer that requires its Chinese suppliers to report carbon footprints, use certified wood, and meet VOC standards transmits those requirements down the chain. The supplier that can meet them gains access to the buyer's higher-value segments; the supplier that cannot loses the order. This is the "compliance vs. capability" distinction in its most concrete form: a buyer requirement can be a cost to be absorbed or a capability to be built, depending on whether the supplier has the organizational capacity to learn from it.
Horizontal channels—relationships among firms at the same position in the chain—matter for a different reason. They are the channels through which collective action occurs. Firms in a cluster can jointly invest in shared infrastructure, jointly negotiate with local government, jointly purchase environmental equipment, and jointly develop new processes. The transaction costs of collective action are lower in a cluster than among dispersed firms, because the firms already know each other, already share suppliers and workers, and already have a history of interaction.
The interaction between vertical and horizontal channels is where the most interesting dynamics occur. A buyer requirement transmitted vertically can trigger horizontal collective action, as firms realize that meeting it individually is more expensive than meeting it together. The Lecong furniture cluster's water-based coating conversion is an example: the buyer requirement came from European retailers, but the solution was organized horizontally through the association, which negotiated bulk pricing and shared training.
8.2.3 Infrastructure: the physical layer of cluster governance
The third driver is infrastructure. Clusters concentrate physical systems: energy supply, water, wastewater treatment, waste collection, transport, and, increasingly, digital connectivity. The design and operation of these systems determine the marginal cost of green upgrading for individual firms, and they are themselves a form of collective governance.
Consider energy. A cluster with a centralized combined heat and power plant can offer its firms lower-carbon heat than individual boilers, and can absorb biomass or waste heat that individual firms could not use. A cluster with a strong grid connection can support rooftop solar at scale; a cluster with a weak connection cannot. A cluster with a district energy system can decarbonize heating at the system level, which is cheaper than decarbonizing each building. The infrastructure is not neutral: it shapes what individual firms can do.
Waste and water follow the same logic. A cluster with a centralized wastewater treatment plant that meets higher standards can allow its firms to discharge into it at lower cost than if each firm had to treat to the same standard on-site. A cluster with a hazardous waste collection system can solve the small-quantity problem that plagues small firms: the cost of properly disposing of a few kilograms of hazardous waste is prohibitive if each firm has to arrange its own pickup, but manageable if the cluster aggregates.
The Chinese experience with cluster infrastructure is mixed. The development zones and industrial parks that house many clusters were built with shared utilities, and some have been upgraded with environmental infrastructure—centralized treatment, cogeneration, waste-to-energy. But the upgrading has been uneven. Parks built in the 1990s and 2000s often have infrastructure that is adequate for conventional pollution control but not for the higher standards that the dual carbon goals imply. Retrofitting is expensive and raises coordination problems: who pays, who operates, who bears the risk of underutilization.
8.2.4 The systemic-driver framework
It is useful to bring the three drivers together into a single framework. Green upgrading in a cluster depends on the interaction of knowledge, networks, and infrastructure, mediated by governance. Knowledge determines what firms can do; networks determine what they are asked to do and what support they receive; infrastructure determines the cost of doing it. Governance determines whether the three are aligned.
The framework has a cumulative-advantage logic. A cluster that starts with strong knowledge diffusion, dense networks, and adequate infrastructure will upgrade faster than one that starts with weak endowments, because the three drivers reinforce each other. Knowledge diffusion is easier when networks are dense; networks are denser when infrastructure brings firms together; infrastructure is more valuable when firms have the knowledge to use it. This is why cluster green upgrading tends to be path-dependent: the clusters that move up the ladder early tend to keep moving, and the ones that lag tend to stay behind.
The policy implication is that interventions at a single point in the system are less effective than interventions that address the system as a whole. A subsidy for green equipment, without the knowledge to select and operate it and the infrastructure to support it, produces low utilization. A training program, without the finance to act on it, produces frustration. A new infrastructure investment, without the governance to operate it, produces white elephants. The clusters that have upgraded most successfully are those where the local government, the associations, and the lead firms have coordinated across all three drivers.
8.3 Case Studies of Cluster Green Upgrading
Four Chinese clusters, taken from different points on the ladder, with different factor intensities and different trajectories, are examined below. The cases are not a representative sample; they are selected to illustrate the mechanisms. Each case is presented with its governance structure, its systemic drivers, and the evidence on outcomes—including the limits of that evidence.
8.3.1 Case 1: Lecong furniture cluster, Shunde, Foshan
The Lecong furniture cluster is the largest furniture production and distribution center in China, with several thousand firms, a wholesale market that attracts buyers from across the country and abroad, and a supply chain that extends from timber import through board production, hardware, coatings, and finished furniture. It is a labor-intensive cluster with a significant share of small and medium firms.
The cluster's green trajectory has been driven by two pressures. The first is VOC regulation. Furniture coating is a major source of volatile organic compounds, and the Guangdong provincial government designated the cluster as a key control area in the 2010s. Enforcement began with large firms and gradually extended to smaller ones, with a phased schedule that allowed firms to convert to water-based or UV-cured coatings. The second is buyer requirements. European and North American retailers, and increasingly domestic brands, require suppliers to meet formaldehyde and VOC standards and to document the origin of wood.
The governance response has been a combination of formal enforcement and informal coordination. The district environmental bureau set emission standards and monitored compliance, but it also negotiated a compliance schedule with the furniture association that gave firms time to convert. The association organized bulk purchasing of water-based coatings, arranged training with coating suppliers, and operated a shared testing laboratory that firms could use to verify that their coatings met standards. The local government provided subsidized loans for equipment conversion and made compliance a condition of continued access to the industrial park.
The outcome, on the evidence available, is that the cluster converted to water-based coatings faster than the national average. A 2019 survey by the China National Furniture Association found that water-based coating penetration in the Lecong cluster was substantially higher than in furniture clusters in inland provinces. The conversion was not uniform: large firms converted earlier and more completely, and some small firms exited or relocated. By the association's own count, the share of member firms using water-based coatings on their main production lines rose from roughly one in five in 2016 to more than half by 2019, while the cluster's reported VOC emissions fell by about 40 percent over the same period, according to district environmental bureau data.
The limits of the evidence are worth noting. The survey data are self-reported, and the definition of "water-based" varies. The environmental outcome—measured VOC emissions—is monitored at the cluster level, not firm by firm, so it is difficult to attribute the improvement to specific governance interventions. And the conversion was driven by a combination of enforcement and buyer pressure that cannot be disentangled.
8.3.2 Case 2: Foshan ceramic cluster, Chancheng District
The ceramic cluster in Foshan is one of the largest in the world, with a history that goes back to the Song dynasty and a modern industry that grew rapidly in the 1990s and 2000s. It is a capital-intensive cluster: kilns, presses, and glazing lines require substantial investment, and the cluster's firms range from large groups with national brands to small producers of specialized products.
The cluster's green trajectory has been driven by a combination of air quality regulation and energy costs. Ceramic kilns are major consumers of natural gas and coal, and major emitters of particulate matter, sulfur dioxide, and nitrogen oxides. The Foshan municipal government began tightening enforcement in the 2010s, and the "blue sky" campaigns of 2017-2018 accelerated the process. Firms that could not meet the new standards were shut down; those that could invested in cleaner kilns, waste heat recovery, and, in some cases, process substitution.
The governance response has been more top-down than in Lecong. The municipal government designated ceramic production as a "restricted" industry in certain areas and forced relocation to designated industrial parks with centralized gas supply and pollution control. The parks provided shared infrastructure—natural gas pipelines, wastewater treatment, solid waste collection—that individual firms could not have afforded. The larger firms invested in their own cleaner technologies, including the use of biomass and waste-derived fuels in kilns.
The outcome is a substantial reduction in the cluster's emissions, but also a substantial reduction in the number of firms. The Foshan ceramic industry consolidated: many small producers closed or moved to inland provinces, and the remaining firms are larger, cleaner, and more capital-intensive. This is the "relocation-upgrade nexus" that Chapter 4 examined for furniture, in a capital-intensive setting. The cluster moved up the ladder, but the ladder itself changed: it is now a smaller, more concentrated cluster with higher environmental performance and higher barriers to entry.
The evidence on outcomes is stronger here than in Lecong, because the municipal government publishes air quality data and because the consolidation is visible in firm-level data. But the attribution problem remains: the improvement in air quality reflects many factors, including the relocation of production out of the district, the switch to natural gas, and the closure of non-compliant firms. Disentangling the contribution of cluster governance from these other factors is not straightforward.
8.3.3 Case 3: Keqiao textile cluster, Shaoxing
The Keqiao textile cluster in Shaoxing, Zhejiang, is the largest textile printing and dyeing center in China. It is a labor-intensive and pollution-intensive cluster: dyeing and finishing consume large quantities of water and discharge wastewater with high chemical oxygen demand and color. The cluster's firms range from large integrated mills to small dyehouses.
The cluster's green trajectory has been driven by water regulation and by the "five-water governance" campaign that Zhejiang province launched in 2014. The campaign set strict standards for industrial wastewater discharge and gave local governments authority to shut down non-compliant firms. Keqiao's response was to build a centralized wastewater treatment system that serves the cluster, to require firms to pre-treat their effluent to standards that the centralized plant can handle, and to consolidate small dyehouses into larger units.
The governance response has been a hybrid of top-down enforcement and market-based coordination. The local government set discharge standards, operated the centralized treatment plant through a state-owned company, and used the plant's capacity as a lever: firms that did not pre-treat to standard could be cut off. The textile association organized training and negotiated compliance schedules. The cluster also developed a market for treated wastewater reuse, with the centralized plant selling reclaimed water back to firms at a price below the cost of fresh water.
The outcome is a substantial reduction in water pollution and water consumption. The Keqiao model has been cited by the Ministry of Ecology and Environment as a case study in cluster-level wastewater management. But the model has limits: it depends on the centralized plant's capacity, on the local government's willingness to enforce, and on the firms' ability to pay for pre-treatment. Small firms with thin margins have struggled, and some have exited.
8.3.4 Case 4: Shenzhen electronics cluster, Bao'an District
The Shenzhen electronics cluster is a technology-intensive cluster with a global reach. It includes contract manufacturers, component suppliers, design houses, and a dense ecosystem of service providers. It is not a traditional pollution-intensive cluster, but it has significant environmental footprint through energy consumption, electronic waste, and the embodied carbon of components.
The cluster's green trajectory has been driven by a different set of pressures: global buyer requirements (Apple, Dell, HP, and others have supplier codes of conduct that include environmental criteria), local air quality regulation, and the city's own carbon peak and neutrality targets. Shenzhen was one of the first Chinese cities to launch a carbon emissions trading system, in 2013, and the cluster's larger firms are covered by it.
The governance response has been market-oriented and network-based. The local government operates the carbon market and provides technical support for firms to measure and report emissions. The larger firms have their own sustainability teams and require their suppliers to report environmental data. The associations and service providers offer training and consulting. The cluster has a dense ecosystem of environmental service firms—consultancies, testing labs, energy service companies—that has grown up around the compliance and reporting requirements.
The outcome is a cluster that is further up the ladder than the others in this chapter, but with a different profile. The environmental gains are less about local pollution and more about energy efficiency, renewable energy procurement, and supply chain management. The cluster's firms are increasingly integrated into global corporate sustainability regimes, and their environmental performance is measured and reported in formats that global buyers recognize. The governance challenge is different: it is less about enforcing standards on unwilling firms and more about coordinating a complex web of voluntary commitments and reporting requirements.
8.3.5 Cross-case patterns
Four patterns emerge from the cases.
The driver of green upgrading varies with the cluster's position in global value chains. The furniture and electronics clusters are driven primarily by buyer requirements; the ceramic and textile clusters are driven primarily by local environmental regulation. In practice, both drivers operate in all four clusters, but the balance differs.
The governance response varies with the cluster's factor intensity. Labor-intensive clusters rely more on associations and collective action; capital-intensive clusters rely more on local government investment in shared infrastructure; technology-intensive clusters rely more on markets and service providers. This is consistent with the factor-intensity framework developed in Chapters 4 through 6.
The outcomes are uneven and the evidence is imperfect. All four clusters have improved their environmental performance, but the improvement is confounded by relocation, consolidation, and changes in the broader economy. The attribution of improvement to cluster governance is plausible but not proven.
The clusters that have moved up the ladder most successfully are those where formal and informal institutions are aligned. In Lecong and Keqiao, the local government enforced credibly but left room for collective adjustment; the associations organized the adjustment; and the infrastructure supported it. In Foshan, the alignment was more top-down, and the outcome was more consolidation and less collective capability. In Shenzhen, the alignment was market-based, and the outcome was more integration into global regimes and less local collective action.
8.4 The Limits of Cluster Governance
The cases support the argument that cluster governance is a systemic driver of green upgrading. But the argument has limits, and it is worth stating them explicitly.
Scale sets the first limit. Cluster governance works best for problems that are localized and shared: VOC emissions from a furniture cluster, wastewater from a textile cluster, energy use in an industrial park. It works less well for problems that are global and diffuse: the embodied carbon of traded goods, the emissions of upstream suppliers in other regions, the cumulative effect of many clusters on the climate. The cluster is a unit of collective action, but it is not a unit of global carbon accounting. The emissions that a cluster avoids may simply be relocated to another cluster, or to another country.
Equity sets the second limit. Cluster governance can produce upgrading, but the gains are not evenly distributed. In Foshan, the consolidation of the ceramic cluster benefited the large firms that could invest in cleaner technologies and hurt the small firms that could not. In Keqiao, the centralized wastewater system benefited firms that could afford pre-treatment and hurt those that could not. The cluster moves up the ladder, but some firms fall off. This is not an argument against cluster governance; it is an argument for pairing it with social protection and adjustment assistance.
Politics sets the third limit. Cluster governance depends on a local government that is willing and able to enforce environmental rules, and on associations that are willing and able to organize collective action. Both conditions are variable. In clusters where the local government is captured by incumbent firms, or where the associations are weak, the governance response is weak. The cases in this chapter are selected from the more successful end of the distribution; the less successful clusters are less documented, and the evidence on them is thinner.
Measurement sets the fourth limit. The outcomes of cluster governance are difficult to measure, because the counterfactual is difficult to construct. Would the Lecong furniture cluster have converted to water-based coatings without the association's bulk purchasing and training? Would the Keqiao textile cluster have reduced water pollution without the centralized treatment plant? The evidence suggests yes in both cases, but the magnitude of the governance contribution is uncertain. The measurement problem is not unique to cluster governance, but it is acute because the interventions are bundled.
8.5 Implications for Policy and Practice
The analysis in this chapter has several implications for policy and for firm strategy.
For policy, the first implication is that cluster-level interventions are more effective than firm-level interventions when the problem is shared. A subsidy for individual firms to install green equipment is less effective than a cluster-level program that combines equipment subsidies with shared infrastructure, training, and collective purchasing. The Chinese experience with cluster upgrading programs—the "low-carbon industrial park" pilots, the "green factory" certification scheme, the "circular economy" pilots—suggests that the programs that bundle interventions across the three drivers produce better outcomes than the ones that focus on a single driver.
The second implication is that local government capacity matters. The clusters that have upgraded most successfully are those where the local environmental bureau has the technical staff to monitor and enforce, the fiscal capacity to invest in shared infrastructure, and the political support to resist capture. Building that capacity is a precondition for cluster green upgrading, and it is not something that can be substituted by market mechanisms or by voluntary agreements.
The third implication is that associations and service providers are part of the infrastructure. The furniture association in Lecong, the textile association in Keqiao, and the environmental service firms in Shenzhen are not peripheral actors; they are the mechanisms through which knowledge diffuses and collective action occurs. Policy that strengthens these organizations—through funding, training, and formal recognition—strengthens the cluster's capacity to upgrade.
For firm strategy, the implication is that location matters. A firm in a well-governed cluster has access to knowledge, infrastructure, and collective action that a firm in a poorly governed cluster does not. This is a source of competitive advantage that is not captured by conventional firm-level analysis. It also means that firms considering relocation should weigh the governance quality of the destination cluster, not just its land and labor costs. The evidence from the furniture and ceramic clusters suggests that firms that relocated to inland clusters with weaker governance often found that the environmental and infrastructure costs offset the savings in land and labor.
8.6 Conclusion: The Cluster as the Unit of Green Upgrading
The argument of this chapter can be stated compactly. Green upgrading in Chinese manufacturing is not primarily a story of individual firms responding to prices. It is a story of clusters—localized production systems with shared infrastructure, dense networks, and formal and informal institutions—converting external pressures into collective capability. The clusters that have moved up the ladder are those where local government, associations, and lead firms have aligned the three systemic drivers: knowledge, networks, and infrastructure. The clusters that have lagged are those where the alignment has failed.
This is not a counsel of optimism. The limits of cluster governance are real: it works better for local pollutants than for global carbon, it distributes gains unevenly, it depends on political conditions that are variable, and it is difficult to measure. But within those limits, the cluster is the unit at which green upgrading can be governed, and the evidence from Lecong, Foshan, Keqiao, and Shenzhen suggests that governance matters.
The next chapter turns to the risks and unintended consequences of the low-carbon transition. Cluster governance, as we have seen, can produce upgrading, but it can also produce stranded assets, displaced workers, and relocated emissions. The politics of managing those consequences is the subject of Chapter 9.
Note on sources: Cluster governance concepts draw on Marshall (1890), Becattini (1990), Porter (1990), and Humphrey and Schmitz (2000, 2002a, 2002b). The distinction between formal and informal institutions follows North (1990) and Scott (2008). Evidence on Chinese local government behavior draws on the cadre evaluation literature and on the author's field interviews in Shunde, Foshan, Shaoxing, and Shenzhen between 2019 and 2023. Cluster-level environmental data are from municipal environmental bureau disclosures and from the China National Furniture Association's 2019 survey. The characterization of the four cases is based on interviews, published case studies, and the author's earlier research on cluster internationalization. The limits of the evidence are discussed in section 8.4.