Chapter 5: Labor-Intensive Pathways: Furniture, Compliance Costs and Relocation Dynamics

5.1 The Factory on Dongguan's Periphery

Houjie town sits at the mouth of the Pearl River Delta, an hour's drive south of Guangzhou and a world away from the export processing zones that made it famous. In the late 1990s, Houjie hosted more than 400 furniture factories and their suppliers—foam cutters, veneer mills, hardware platers, lacquer shops—strung along a single stretch of the 107 National Highway. At the peak, the town's furniture exports exceeded USD 1.5 billion annually. The factories were mostly small: 80 to 300 workers, one or two large CNC routers, a spray booth, and a showroom facing the highway. Buyers from Ashley, IKEA, and a dozen European importers arrived each spring and autumn to place orders for the following season. The business model was straightforward: take imported design specifications, source particleboard and medium-density fiberboard (MDF) from domestic mills, assemble with low-cost labor, finish with solvent-based lacquer, and ship through the Port of Shenzhen or Huangpu. Margins were thin—3 to 7 percent net on a good year—but volume was reliable.

I visited Houjie in the autumn of 2023, after the pandemic had reshuffled the industry. The number of registered furniture firms in the town had fallen to just under 200. Several of the large factory compounds were now logistics warehouses; one had been converted into a live-streaming e-commerce base for domestic furniture brands. The surviving factories were not necessarily the largest. They were the ones that had invested in water-based coating lines, dust extraction systems, and—crucially—the measurement and documentation infrastructure that allowed them to prove to buyers what their emissions actually were. The owner of one such factory, a third-generation family business with 160 workers, walked me through his spray line. The water-based lacquer booth had cost RMB 2.8 million to install in 2019, plus another RMB 600,000 for the drying tunnel and ventilation retrofit. His per-unit finishing cost had risen by roughly 18 percent. But he had kept his European buyers, and his order book in 2023 was fuller than it had been in 2019. "The factories that didn't upgrade," he said, "either moved to Vietnam or closed."

That factory—and the hundreds like it across the furniture clusters of Guangdong, Zhejiang, and Hebei—is the entry point for this chapter's analysis of the labor-intensive pathway. Compliance costs are real, and so is relocation; both impose distributional burdens that fall hardest on the workers and small firms least able to absorb them. But in labor-intensive sectors, green governance capability is the mechanism through which compliance costs are converted into either exit or upgrading, and the difference between the two outcomes is largely organizational rather than technological. Firms that can measure, document, and verify their environmental performance—even at a modest level—retain access to the value chains that matter. Firms that cannot, exit. The furniture sector is the clearest case in China's manufacturing economy where this sorting mechanism is visible in real time.

I should flag at the outset what this chapter does not argue. It does not claim that green upgrading is available to all furniture firms, or that the firms that exit are somehow deficient in effort or foresight. Many of the Houjie factories that closed between 2018 and 2023 were well-run businesses caught between rising compliance costs, volatile order books, and a buyer base that was itself consolidating. The chapter also does not treat relocation as a failure. Relocation to Vietnam, Cambodia, or inland Chinese provinces is a rational response to cost differentials, and in some cases it preserves employment that would otherwise disappear. What the chapter does argue is that the green governance capability that enables upgrading is not evenly distributed, and that the firms and clusters that possess it are pulling away from those that do not. That divergence is the central empirical fact of the labor-intensive pathway.

5.1.1 Why Furniture Is the Right Case

Furniture is not the largest Chinese export sector. It is not the most carbon-intensive. It is not the most technologically sophisticated. It is, however, the sector where the interaction between global value chain (GVC) embedding, environmental compliance, and relocation is most legible, for three reasons.

First, furniture is a consumer-facing, design-driven industry with short product cycles and high buyer concentration. The top 20 global furniture retailers account for a substantial share of China's furniture exports, and these buyers have been at the forefront of supply chain environmental requirements. IKEA's IWAY standard, Ashley's supplier code of conduct, and the European Union's Timber Regulation (EUTR) all impose documentation and traceability requirements that go well beyond what Chinese domestic regulation demands. When a buyer requires a supplier to report the formaldehyde emissions of its MDF, the volatile organic compound (VOC) content of its finishes, and the chain of custody of its timber, the supplier must build the measurement infrastructure to comply. That infrastructure is the raw material of green governance capability.

Second, furniture production is fragmented and cluster-based. The sector's cost structure—low fixed capital, high labor content, modular production—means that it agglomerates in dense clusters where suppliers, workers, and buyers co-locate. Houjie, Lecong, and Longjiang in Guangdong; Yuhuan and Anji in Zhejiang; Xianghe in Hebei; and Chengdu in Sichuan are all furniture clusters of varying scale and specialization. This cluster structure means that compliance costs and upgrading pathways are not firm-level phenomena alone; they are shaped by the collective infrastructure—shared testing labs, waste treatment facilities, training institutions—that clusters provide. The cluster is the unit of analysis that connects firm-level compliance to regional upgrading.

Third, furniture is a sector where relocation is not hypothetical. Vietnam's furniture exports grew from USD 2.7 billion in 2010 to over USD 16 billion in 2022, and a significant share of that growth came from Chinese firms relocating or establishing satellite operations. Cambodia, Malaysia, and Indonesia have seen similar, if smaller, shifts. Within China, furniture production has moved inland to Sichuan, Henan, and Hubei, following the broader pattern of coastal-to-interior relocation. The relocation dynamics are observable, measurable, and—crucially—differentiated by the green governance capability of the firms involved.

These three features make furniture a useful case for the book's broader argument. If green governance capability is an endogenous driver of upgrading in a sector this fragmented, this labor-intensive, and this exposed to relocation pressure, then the argument has purchase beyond the high-technology sectors where it is easier to make. The furniture case is the hard case, and that is why it deserves a full chapter.

5.2 Compliance Costs in Labor-Intensive Sectors: What the Numbers Show

5.2.1 The Cost Structure of Furniture Production

To understand compliance costs, one must first understand the cost structure they are being layered onto. A typical mid-sized Chinese furniture factory producing for export has a cost breakdown that looks roughly like this: raw materials (particleboard, MDF, veneer, hardware, foam, fabric) account for 45 to 55 percent of total cost; direct labor for 12 to 18 percent; manufacturing overhead (energy, depreciation, factory rent) for 10 to 15 percent; finishing materials (lacquer, adhesive, solvents) for 8 to 12 percent; and logistics, tariffs, and compliance for the remainder. The exact proportions vary by product category—upholstered furniture is more labor-intensive, case goods more material-intensive—but the general shape holds.

Environmental compliance costs enter this structure through four channels: material substitution, primarily the shift from solvent-based to water-based or UV-cured finishes and from urea-formaldehyde to low-emission adhesives; end-of-pipe treatment, primarily dust extraction, VOC abatement, and wastewater treatment; monitoring and documentation, primarily emissions testing, chain-of-custody certification, and reporting; and energy costs, primarily the electricity and fuel used in drying and heating processes. Each channel has a different cost profile, and each interacts differently with the firm's existing capabilities.

A concrete accounting makes the scale clear. A factory producing 50,000 MDF cabinets per year for the European market might spend, in a baseline scenario, RMB 4.5 million annually on solvent-based lacquer and adhesives. Switching to water-based finishes raises the material cost by 20 to 35 percent, or roughly RMB 1.0 to 1.6 million. The spray booth retrofit—stainless steel lines, increased ventilation, drying tunnels—requires capital expenditure of RMB 2.0 to 3.5 million, depreciated over 7 to 10 years. VOC abatement equipment (activated carbon adsorption or regenerative thermal oxidation) adds another RMB 0.8 to 2.0 million in capital and RMB 200,000 to 500,000 in annual operating costs. Emissions testing and certification, if done quarterly, cost RMB 30,000 to 80,000 per year. Energy costs may rise or fall depending on the technology: water-based finishes often require more drying energy, but modern drying tunnels can be more efficient than the open spray booths they replace.

The total incremental cost, for this hypothetical factory, is on the order of RMB 1.8 to 3.5 million per year in operating costs plus RMB 3.0 to 5.5 million in capital expenditure. Against a revenue base of perhaps RMB 80 to 120 million and a net margin of 4 to 6 percent, this represents a 30 to 60 percent reduction in net profit if absorbed entirely by the firm. That is the compliance cost problem in its starkest form.

5.2.2 How Firms Actually Absorb Compliance Costs

The hypothetical above assumes the firm absorbs the full cost. In practice, firms absorb compliance costs through four mechanisms, and the mix of mechanisms determines whether the firm upgrades or exits.

Pass-through to buyers is the first. Some compliance costs can be passed through in the form of higher unit prices. This is most feasible for firms with differentiated products, long-term buyer relationships, or niche market positions. A factory producing high-end solid wood furniture for a European brand with a sustainability marketing strategy can often negotiate a 3 to 8 percent price increase to cover certified materials. A factory producing commodity particleboard case goods for a big-box retailer competing on price cannot. The pass-through capacity is a function of the firm's position in the value chain, not its technical capability alone.

Productivity offsets are the second. Some compliance investments also improve productivity. Water-based coating lines, for instance, often reduce rework rates because the finish is more consistent; dust extraction systems reduce equipment maintenance costs; energy-efficient drying reduces fuel bills. In the factory I visited in Houjie, the owner estimated that the water-based line had reduced finishing rework by 40 percent, offsetting roughly a third of the incremental material cost. These productivity offsets are real but rarely sufficient to cover the full compliance cost. They matter because they change the marginal calculus, not because they solve the problem.

Scale economies are the third. Larger firms can spread fixed compliance costs—testing labs, certification, abatement equipment—over more units of output. A factory producing 200,000 units per year faces a per-unit compliance cost roughly one-quarter that of a factory producing 50,000 units, assuming similar technology. This scale effect is one of the most powerful forces driving consolidation in the sector. It also explains why the surviving factories in Houjie tend to be larger than the ones that closed, even though the largest factories were not necessarily the most profitable.

Margin compression is the fourth. Firms that cannot pass through, cannot offset through productivity, and cannot spread through scale must absorb compliance costs by accepting lower margins. This is sustainable for a while, especially for family-owned firms with low debt and patient capital. It is not sustainable indefinitely. The firms that closed in Houjie between 2018 and 2023 were, in many cases, firms that had been absorbing costs for years and finally ran out of margin.

The distribution of these four mechanisms across the firm population is the key empirical question. The China Furniture Association's 2022 survey of 312 export-oriented furniture enterprises (conducted in cooperation with the Guangdong Furniture Association and covering firms in Guangdong, Zhejiang, and Hebei) found that 24 percent reported passing through most compliance costs to buyers, 38 percent reported absorbing them through a combination of productivity offsets and margin compression, and 31 percent reported that compliance costs had driven them to reduce export volumes or exit export markets entirely. The remaining 7 percent did not report. The exact proportions vary by cluster and product category, but the sorting pattern is consistent across all three provinces.

5.2.3 The Measurement Problem

There is a deeper problem beneath the cost accounting: most furniture firms do not know what their compliance costs actually are. This is not a trivial observation. It is the core of the green governance capability argument.

Consider what a firm needs to know to manage its environmental compliance. It needs to know the formaldehyde emission class of every board it buys (E0, E1, E2, or the newer ENF standard), the VOC content of every finish and adhesive, the energy consumption of every major process, the water consumption and wastewater discharge of every wet process, and the carbon footprint of its inputs if it is reporting to a buyer with science-based targets. It needs to know these things not once, but continuously, because the requirements change and the inputs change. And it needs to be able to document them in a form that a buyer's auditor or a third-party certifier will accept.

Most small and medium furniture firms in China cannot do this. They buy boards from traders who buy from mills, and the chain of custody is opaque. They buy finishes from distributors who may or may not have the technical data sheets. They do not meter their energy consumption by process. They do not have environmental management systems (EMS) certified to ISO 14001, and even if they do, the certification is often a paperwork exercise rather than a management tool. The result is that compliance costs are not managed; they are experienced as unexpected expenses that appear when a buyer audit fails or a shipment is rejected.

The firms that have built measurement capability—even modest capability, such as a spreadsheet tracking board suppliers and their emission classes, or a relationship with a testing lab that can turn around a VOC test in a week—are in a fundamentally different position. They can anticipate compliance costs, negotiate them with buyers, and target investments where they will have the most effect. This is what I mean by green governance capability: not the ability to comply with a specific regulation, but the organizational capacity to measure, document, and manage environmental performance across the value chain. It is the difference between reacting to compliance costs and governing them.

5.2.4 Evidence from Cluster-Level Data

The firm-level picture is borne out by cluster-level data, though the data are patchy. China's furniture industry association reports that the number of furniture enterprises above designated size (annual revenue above RMB 20 million) fell from roughly 8,000 in 2015 to about 6,500 in 2022, while total industry revenue grew modestly. This consolidation is consistent with the scale-effect story: larger firms are absorbing smaller ones or driving them out.

More telling is the pattern of environmental certification. According to data from the China National Accreditation Service for Conformity Assessment (CNAS), the number of furniture enterprises with ISO 14001 certification grew from about 1,200 in 2010 to over 4,000 in 2020, but the growth has been concentrated in the larger firms and in the clusters with the strongest export orientation. In Houjie, roughly 40 percent of the surviving factories have ISO 14001 certification; in Xianghe, the share is closer to 15 percent. The difference reflects both the buyer base (Houjie exports more to Europe; Xianghe exports more to the US and domestic markets) and the cluster's institutional infrastructure (Houjie has a furniture testing center operated by the provincial quality supervision bureau; Xianghe does not).

The certification data are imperfect—ISO 14001 certification does not guarantee environmental performance, and some certifications are more rigorous than others—but they are a reasonable proxy for the measurement and documentation capability that I have argued is central. The firms with certification are the ones that can demonstrate compliance to buyers; the firms without it are the ones that struggle.

A second data source reinforces the point. The Guangdong Provincial Department of Ecology and Environment's 2021 audit of VOC emissions from the furniture coating sector found that firms with water-based coating lines in place reported average VOC emission concentrations of 30 to 60 mg/m³, while firms still using solvent-based lines averaged 180 to 350 mg/m³. The audit covered 147 firms in Dongguan, Foshan, and Zhongshan. Firms that had installed continuous emission monitoring systems—roughly 35 percent of the sample—were able to provide batch-level data; the rest could provide only periodic test reports. The gap between the two groups is not just a compliance gap; it is a data gap, and the data gap is what makes it impossible for the firms without monitoring to negotiate with buyers or regulators on the basis of their actual performance.

5.3 Relocation Pressures and Responses

5.3.1 The Relocation Calculus

Relocation is not a single decision. It is a family of decisions, each with different drivers and different implications for green governance. I find it useful to distinguish four types: full relocation of production to another country; establishment of satellite capacity in another country while retaining Chinese production; relocation within China, from coastal to interior provinces; and relocation of specific production stages, such as finishing or assembly, while retaining design and marketing in China. Each type has a different relationship to compliance costs and green governance capability.

The standard relocation calculus weighs labor costs, trade policy, logistics, and regulatory burden. Labor costs in Vietnam's furniture sector are roughly 40 to 60 percent of Guangdong's, depending on the skill level. US Section 301 tariffs on Chinese furniture, imposed in 2018 and 2019, added 10 to 25 percent to the landed cost of Chinese exports to the US, creating a strong incentive for tariff-jumping relocation. Logistics costs to the US East Coast are lower from Vietnam than from Guangdong, though the advantage is partly offset by less developed port infrastructure. Regulatory burden—including environmental compliance—is generally lower in Vietnam, though this is changing as Vietnam tightens its own environmental standards and as buyers apply the same supply chain requirements to Vietnamese suppliers that they apply to Chinese ones.

The green governance dimension of relocation is less discussed but important. Firms that relocate to escape compliance costs often find that the compliance costs follow them, because the buyers who demand environmental documentation in China demand it everywhere. A Vietnamese factory supplying IKEA must meet the same IWAY standards as a Chinese factory. The difference is that the Vietnamese factory may have less institutional support—fewer testing labs, less developed supplier networks, weaker environmental agencies—which makes compliance harder, not easier. Relocation can therefore be a trap for firms that relocate to escape compliance rather than to build capability.

5.3.2 Who Relocates and Who Stays

Mid-sized firms with 200 to 800 workers, moderate technical capability, and strong buyer relationships have done most of the relocating in this sector—not the smallest firms, and not the largest. These firms have enough scale to justify the fixed costs of setting up an overseas operation (typically USD 2 to 5 million for a greenfield factory) but not enough pricing power to absorb Chinese compliance costs indefinitely. They also tend to be the firms with the most developed management systems—the ones that can replicate a factory in another country because they have documented their processes.

The smallest firms rarely relocate internationally; they lack the capital and the management depth. They either absorb costs, exit, or move inland. The largest firms often retain Chinese production for higher-value products while establishing overseas capacity for tariff-sensitive or labor-intensive lines. This "China plus one" strategy is now common among the top 50 Chinese furniture exporters.

The firms that stay in China and upgrade tend to share three characteristics: they have moved upmarket into products where compliance costs are a smaller share of value (solid wood, custom furniture, contract furniture for hospitality and healthcare); they have built measurement and documentation capability that lets them negotiate compliance costs with buyers; and they have invested in automation that reduces their exposure to labor costs. These are not necessarily the largest firms, but they are the firms with the most developed green governance capability.

5.3.3 The Vietnam Question

Vietnam deserves separate treatment because it has become the primary destination for Chinese furniture relocation and because the Vietnamese experience illuminates the limits of relocation as a response to compliance costs.

Vietnam's furniture exports grew from USD 2.7 billion in 2010 to USD 16.1 billion in 2022, according to Vietnam Customs data. Chinese-invested firms account for a significant share of this growth, particularly in the wood furniture segment. The relocation has been concentrated in the southern provinces—Binh Duong, Dong Nai, and Ho Chi Minh City—where industrial parks offer ready infrastructure and port access.

The Vietnamese advantage is real but narrower than it appears. Labor costs are lower, but productivity is also lower; the net labor cost advantage is perhaps 20 to 30 percent, not the 40 to 60 percent that raw wage comparisons suggest. Supply chains are less developed: Vietnamese furniture factories import a large share of their hardware, finishes, and even some boards from China, which adds logistics costs and lead times. Port congestion and shipping costs have been volatile. And the environmental compliance burden is rising, both because buyers are applying global standards and because the Vietnamese government has tightened its own regulations, particularly on formaldehyde emissions and wastewater discharge.

The most important limitation, for our purposes, is that relocation does not solve the green governance problem. A Chinese firm that relocates to Vietnam without building measurement and documentation capability will face the same buyer audits, the same certification requirements, and the same compliance costs—in a context where institutional support is weaker. The firms that have relocated successfully are the ones that brought their management systems with them. The firms that have struggled are the ones that relocated to escape, only to find that there was no escape.

I should be careful here not to overstate the case. Relocation to Vietnam has preserved employment and output that would otherwise have been lost, and many Chinese firms have operated successfully in Vietnam for years. The point is not that relocation is a mistake. The point is that relocation is not an alternative to green governance capability; it is a different context in which that capability is tested.

5.3.4 Inland Relocation and the Cluster Question

Relocation within China—from Guangdong to Sichuan, Henan, or Hubei—follows a different logic. The drivers are lower land costs, lower labor costs (though the gap is narrowing), local government incentives, and access to growing domestic markets. Environmental compliance costs are generally lower inland, both because regulation is less strictly enforced and because the industrial infrastructure is less developed, which cuts both ways: less expensive compliance, but also less support for firms that want to upgrade.

The inland relocation experience has been mixed. Some inland clusters, such as Chengdu's furniture industrial park, have attracted significant investment and developed into viable production centers. Others have struggled with infrastructure, supplier networks, and environmental management. The key variable, again, is cluster-level governance capability: the ability of local governments, industry associations, and lead firms to provide the shared infrastructure—testing labs, waste treatment, training—that individual firms cannot provide for themselves.

The cluster dimension is important because it connects the firm-level story to the regional story that Chapter 9 develops in more detail. A firm that relocates inland to a cluster with strong governance capability can upgrade; a firm that relocates to a cluster without it may find itself worse off than it was on the coast, because it has lost access to the coastal cluster's supplier networks and institutional support without gaining equivalent support inland.

5.4 Upgrading Pathways for Furniture Clusters

5.4.1 Four Pathways and Their Preconditions

The evidence from Houjie, Lecong, Yuhuan, and other clusters suggests four distinct upgrading pathways for labor-intensive furniture firms under carbon constraints. Each has different preconditions, different cost profiles, and different implications for the cluster. The table below summarizes them; the prose that follows takes each in turn.

Material substitution is the most accessible pathway for mid-sized export-oriented firms, and it is the one most widely adopted in the coastal clusters. The firm shifts from solvent-based to water-based or UV-cured finishes, from urea-formaldehyde to low-emission adhesives, and from uncertified to certified timber, and it builds the documentation systems to prove these choices to buyers. Finishes and adhesives are a small share of embodied carbon in most furniture, though, and the certification burden falls disproportionately on small firms.

Process automation and energy efficiency is capital-intensive and favors larger firms. The firm invests in automated cutting, CNC machining, and energy-efficient drying to reduce labor content and energy consumption. This reduces exposure to labor cost increases and improves consistency, which in turn makes compliance easier. It also requires access to capital and technical expertise that many small firms lack, and it can increase the firm's fixed cost base, making it more vulnerable to demand volatility.

Product upgrading and design moves the firm from commodity production to design-led, higher-value products—custom furniture, contract furniture, branded collections—where compliance costs are a smaller share of value and where buyers are more willing to pay for sustainability. This pathway requires design capability, brand building, and closer relationships with buyers. It is the pathway most likely to generate sustained competitive advantage, but it is also the slowest and the most demanding of firm-level capabilities.

Circular economy and remanufacturing is nascent in China but growing, driven by extended producer responsibility (EPR) regulations in Europe and by corporate sustainability commitments among large buyers. The firm develops capabilities in furniture repair, remanufacturing, and material recovery, either as a standalone business or as an extension of its primary production. The business model is unproven at scale and requires new logistics and reverse-supply-chain capabilities.

These pathways are not mutually exclusive. The most successful firms combine elements of two or three: material substitution plus design upgrading, or automation plus circular economy. They are also not equally available to all firms. Material substitution is accessible to most mid-sized firms; automation favors larger firms; design upgrading requires capabilities that many firms lack; circular economy is still experimental. The cluster's governance capability determines which pathways are available to which firms.

5.4.2 The Role of Cluster Institutions

The cluster is where the pathways become feasible or infeasible. A firm that wants to substitute materials needs access to certified suppliers, testing labs, and technical advice. A firm that wants to automate needs access to capital, skilled technicians, and maintenance services. A firm that wants to upgrade its design needs access to designers, showrooms, and market intelligence. A firm that wants to develop circular economy capabilities needs access to reverse logistics, sorting facilities, and remanufacturing equipment.

Some of these inputs can be provided by markets; others require collective action. Testing labs are a classic example. A single furniture factory cannot justify the capital cost of a fully equipped emissions testing lab, but a cluster of 200 factories can support one through shared fees or government subsidy. Houjie has such a lab, operated by the Guangdong Provincial Quality Supervision Bureau; Xianghe does not. The difference affects the speed and cost with which firms in each cluster can certify their products.

Industry associations play a similar role. The Guangdong Furniture Association has been active in developing water-based coating standards, organizing training programs, and lobbying for subsidies for coating line retrofits. The associations in smaller clusters are less active, partly because they have fewer resources and partly because their members are more fragmented.

Local governments matter too, though their role is more ambiguous. Some local governments have provided substantial subsidies for environmental retrofits—Dongguan's "replacing solvent with water" program, launched in 2018, offered grants of up to 30 percent of equipment costs for firms switching to water-based coatings. Other local governments have focused on enforcement, closing non-compliant firms without offering much support for upgrading. The difference in approach shapes the cluster's trajectory: enforcement without support drives consolidation and exit; enforcement with support enables upgrading.

5.4.3 The Houjie Trajectory

Houjie's trajectory illustrates how these factors interact. In the late 2010s, the town faced a dual squeeze: rising environmental enforcement from the provincial government, which was under pressure to meet air quality targets, and rising buyer requirements from European and American retailers. The initial response was widespread exit. Between 2017 and 2020, the number of furniture factories in Houjie fell by roughly half.

The firms that survived did so by combining pathways. Most substituted materials (water-based coatings, low-emission boards). Many automated (CNC routers, automated spray lines). Some upgraded products (custom furniture, contract furniture). A few began experimenting with circular economy models (furniture rental for commercial spaces, remanufacturing of office furniture).

The cluster's institutional infrastructure adapted. The provincial testing lab expanded its capacity. The furniture association organized training programs on water-based coating technology. The local government provided subsidies for equipment retrofits and helped firms access credit. The result is a smaller but more capable cluster: fewer firms, but higher average productivity, higher compliance rates, and stronger buyer relationships.

This is not a story of unqualified success. The workers who lost jobs when factories closed did not necessarily find equivalent employment in the surviving factories, which were more automated and required different skills. The small suppliers—foam cutters, veneer mills, hardware platers—that served the closed factories also suffered. The cluster's output is lower than it was at its peak, even if its value added is higher. These distributional consequences are real and should not be glossed over.

But the Houjie trajectory does illustrate the book's central argument: green governance capability, built at the firm and cluster level, is the mechanism through which compliance costs are converted into upgrading rather than exit. The firms that built measurement and documentation capability, that invested in cleaner technologies, and that upgraded their products are still in business. The firms that did not are not.

5.4.4 A Note on What the Cluster Cannot Do

I want to be careful not to overstate the cluster's capacity to solve the compliance cost problem. There are limits.

The first limit is scale. Some compliance costs—particularly the cost of carbon border adjustments or the cost of meeting science-based targets—are simply too large to be absorbed by small and medium firms, no matter how well-organized the cluster. The EU's Carbon Border Adjustment Mechanism (CBAM), which currently covers cement, steel, aluminum, fertilizers, electricity, and hydrogen but may expand to other sectors, would impose costs on furniture exports that are proportional to the embodied carbon of their inputs. For a sector where margins are thin and buyers are price-sensitive, even a modest carbon border cost could be decisive.

The second limit is technology. Some of the pathways I have described—particularly automation and circular economy—require technologies that are not yet mature or affordable at the scale of small furniture firms. Water-based coatings have improved dramatically, but they still perform less well than solvent-based coatings in some applications, particularly for high-gloss finishes. Automated spray lines are expensive and require skilled operators. Remanufacturing requires reverse logistics that do not yet exist at scale.

The third limit is institutional. Clusters can provide shared infrastructure, but they cannot provide the strategic direction that individual firms need. The most successful upgrading trajectories in Houjie were led by firms that had a clear vision of where they wanted to go—into contract furniture, into design-led products, into brand building—and used the cluster's resources to get there. Firms that waited for the cluster to solve their problems did not survive.

These limits are important because they temper the optimism that sometimes surrounds cluster-based upgrading narratives. Clusters are enabling environments, not guarantees. The firms that upgrade are the ones that use the environment, not the ones that wait for it to act on them.

5.5 The Compliance Cost Paradox

Let me now step back and state the chapter's central analytical claim in its starkest form. In labor-intensive sectors, compliance costs are simultaneously a barrier to upgrading and a mechanism of upgrading. They are a barrier because they raise the fixed and variable costs of production, squeezing margins and driving out firms that cannot absorb them. They are a mechanism because they force firms to build the measurement, documentation, and management capabilities that are the foundation of green governance—and these capabilities, once built, become a source of competitive advantage.

This is a paradox, and like most paradoxes it resolves only when you look at the time dimension. In the short run, compliance costs are a barrier: they raise costs, reduce profits, and drive exit. In the long run, the firms that survive have built capabilities that their competitors lack, and those capabilities allow them to access markets, negotiate with buyers, and invest in further upgrading. The short-run barrier is the mechanism of long-run advantage.

I do not want to overstate this. Not every firm that survives builds capabilities; some survive by cutting costs in ways that are not sustainable. Not every firm that builds capabilities survives; some invest in the wrong capabilities or invest too early. The paradox is a tendency, not a law. But it is a strong tendency, and it explains why the furniture clusters that have gone through the compliance shock most thoroughly—Houjie, Lecong, Yuhuan—are now more competitive, not less, than the clusters that have delayed or avoided it.

The policy implication is uncomfortable. If compliance costs are a mechanism of upgrading, then policies that shield firms from compliance costs may delay upgrading. This is not an argument for imposing costs without support; the Houjie experience shows that support—subsidies for equipment, testing infrastructure, training—is essential to help firms make the transition. It is an argument for designing compliance policies so that they push firms toward capability building rather than toward exit. The difference between a well-designed compliance policy and a poorly designed one is not the level of the cost; it is whether the policy creates incentives and support for firms to build the capabilities they need.

5.6 The Relocation Question Revisited

The relocation dynamics I described earlier now look different in light of the compliance cost paradox. Relocation responds to the interaction between cost differentials and capability requirements, and cost differentials alone explain little. Firms relocate when the cost of staying—including compliance costs—exceeds the cost of moving, adjusted for the capabilities they would lose or gain.

This framing explains why relocation has been concentrated among mid-sized firms with moderate capabilities. These firms have enough capability to replicate their operations elsewhere but not enough pricing power to absorb Chinese compliance costs. The smallest firms cannot relocate; they exit. The largest firms can absorb compliance costs or pass them through; they stay and upgrade. The mid-sized firms are the marginal relocators, and their decisions are shaped by the capabilities they have built.

The framing also explains why relocation to Vietnam has not been the escape that some firms hoped for. Vietnam offers lower labor costs, but it does not offer the cluster infrastructure—testing labs, supplier networks, skilled workers—that Chinese clusters have built over decades. Firms that relocate to Vietnam must either bring their capabilities with them or build them anew. The ones that bring them succeed; the ones that don't struggle.

The framing has implications for policy. If relocation is driven by capability gaps as much as cost gaps, then policies that build capability—training, testing infrastructure, access to finance for equipment—can reduce relocation pressure. This is not to say that relocation can be stopped; cost differentials are real, and some relocation is inevitable. But the pace and scale of relocation can be shaped by policies that help firms build the capabilities they need to stay competitive.

5.7 What the Chapter Does and Does Not Claim

I want to close with a clear statement of what this chapter has argued and what it has not.

The chapter has argued that in labor-intensive sectors like furniture, green governance capability is the mechanism through which compliance costs are converted into either upgrading or exit. It has shown, through the Houjie case and cluster-level data, that firms with measurement and documentation capability have been able to absorb compliance costs, negotiate with buyers, and upgrade their products, while firms without that capability have exited or relocated. It has identified four upgrading pathways and shown how cluster institutions shape which pathways are available to which firms. It has argued that relocation is not an alternative to green governance capability but a different context in which that capability is tested.

The chapter has not argued that compliance costs are benign, that all firms can upgrade, or that relocation is a failure. It has not claimed that the Houjie trajectory is replicable everywhere, or that cluster institutions can solve problems that exceed their scale. It has not claimed that the firms that exited were deficient in effort or foresight. And it has not claimed that green upgrading in labor-intensive sectors will be sufficient to meet China's dual carbon goals without complementary policies in other sectors.

The evidence base behind these claims deserves a word of qualification. The firm-level interviews conducted in Houjie, Lecong, and Yuhuan between 2019 and 2023 overrepresent surviving firms, because the firms that exited are harder to interview. I have tried to correct for this by triangulating with association data on firm numbers and with interviews of former factory owners who now work in related businesses. The correction is imperfect. The exit rate I report—roughly half of Houjie's factories between 2017 and 2020—is consistent with association data, but the characteristics of the exiting firms are less certain. My claim that the exiting firms tended to have weaker measurement and documentation capability is based on interviews with surviving firms and with suppliers who served both, not on direct observation of the exiting firms' internal systems. The clustercertification data are imperfect proxies, the local government reports are not independent evaluations, and the association interviews carry the usual biases of interested parties. I have flagged these limitations where they matter. What the chapter offers is not a definitive causal estimate of the effect of compliance costs on firm survival but an interpretive account, grounded in specific firms and clusters, of the mechanism through which compliance costs are converted into upgrading or exit. A larger sample, a longer time series, and better data on firm-level capability would allow sharper claims. The furniture sector will not determine whether China meets its dual carbon goals—its emissions are small relative to power, steel, and cement—but it is where the human dimension of the transition is most visible and where the endogeneity argument is most testable.

5.8 Coda: The Factory Owner's Question

On my last afternoon in Houjie, the factory owner asked me a question I have been thinking about since. We were standing in his showroom, surrounded by sample cabinets in water-based finishes, and he said: "Everyone tells me I should be proud of what I've built. But I look at my neighbors who closed their factories and moved to Vietnam, and some of them are doing better than me. So what did I actually gain?"

I did not have a good answer at the time. The honest answer is that he gained a set of capabilities—measurement, documentation, buyer relationships, technical knowledge—that are valuable but not always visible on a balance sheet. He gained access to markets that require those capabilities, and he gained the option to upgrade further. Whether those gains are worth the costs he incurred depends on how the market evolves, and that is not something either of us can know.

His question points to the limits of the book's argument. Green governance capability is a mechanism of upgrading, but it is not the only mechanism, and it is not sufficient on its own. The furniture sector shows what the mechanism can do in the least favorable terrain—low margins, fragmented firms, price-sensitive buyers—and it also shows where the mechanism runs out: in the firms too small to build capability, in the workers whose skills do not transfer to the automated line, in the clusters whose local governments lack the fiscal capacity to subsidize retrofits. A theory of green upgrading that cannot account for these limits is not a theory of the real economy; it is a theory of the winners.

There is one more thing the Houjie case makes visible, and it concerns the direction of causality that the book's core argument asserts. The standard account treats environmental compliance as an exogenous shock: regulation arrives, firms respond, some survive. What I observed in Houjie is closer to the reverse. The firms that had already invested in measurement systems, supplier relationships, and process documentation—for reasons that had nothing to do with carbon—were the ones that found the new compliance requirements manageable. Their capability preceded the regulation. The regulation selected for it. This is what I mean when I say green governance capability is endogenous: it is produced by the ordinary work of running a modern export-oriented factory, and it becomes decisive only when the compliance environment tightens. The firms that treated compliance as a separate project, bolted on after the fact, almost always arrived late.

The factory owner's 2024 order book offers one concrete data point on how the bet has paid off so far. As of the first quarter, he had booked 14 percent more export volume than in the same period of 2023, at unit prices roughly 6 percent above his pre-upgrade baseline, with two new European buyers added since the water-based line came online. His Vietnamese competitors, meanwhile, were quoting 8 to 12 percent below his prices on commodity case goods but could not match his delivery times on the custom and contract lines. That is not proof the strategy was right. It is the kind of evidence that will accumulate, order by order, over the next few years—and the kind of evidence that a firm without measurement systems would not be able to generate about itself.

5.9 From Firm Capability to Sector-Level Governance

The firm-level story is necessary but not sufficient. A sector in which every surviving firm has built its own measurement system, its own testing relationships, and its own buyer-audit routines is a sector with high private capability and high private cost. The same functions can often be provided more cheaply at the cluster or sector level, and the difference between a sector that organizes them collectively and one that does not is a difference in how much upgrading the compliance shock can buy.

Consider testing. A single VOC test on a finished panel costs a few hundred renminbi at a public lab and takes three to five working days. A factory that runs one test per product family per year spends perhaps RMB 20,000 to 50,000—manageable. But buyer audits often require batch-level or even shipment-level documentation, and the frequency multiplies the cost. A cluster that operates a shared testing facility, staffed and accredited to the same standard as the provincial lab, can cut turnaround to a day and cost by half. Houjie's lab does this for the firms that use it. The firms that do not use it—because they are too small, or because they do not know it exists, or because they have no one on staff who can interpret the results—do not benefit.

The same logic applies to waste. Solvent-based lacquer generates hazardous waste—waste solvent, contaminated filters, sludge from spray booth water curtains. A single factory producing 50,000 units a year might generate 30 to 50 tonnes of hazardous waste annually. The legal disposal cost, through a licensed hazardous waste contractor, is RMB 3,000 to 6,000 per tonne in Guangdong, or roughly RMB 100,000 to 300,000 per year. A cluster that operates a centralized solvent recovery or waste-to-energy facility can reduce this cost substantially, and can also recover some of the solvent value. Without such a facility, each firm bears the full cost, and the incentive to dispose of waste illegally is strong. The enforcement crackdowns that closed factories in Houjie in 2018 and 2019 were, in part, crackdowns on illegal disposal.

These are classic collective goods problems, and they are the reason cluster governance matters. The firms that survive the compliance shock are not just the ones with private capability; they are the ones embedded in a cluster that provides the shared infrastructure—testing, waste treatment, training, supplier networks—that lowers the cost of capability. The cluster is the unit at which the compliance cost paradox is resolved, or not resolved.

This has an implication for how we think about the sector's carbon footprint. The emissions associated with furniture production are not just the direct emissions of the factories; they are the embodied emissions of the boards, the finishes, the hardware, and the transport. A firm that substitutes low-emission boards reduces its own compliance burden but shifts the burden upstream to the board mill, which may or may not have the capability to produce low-emission board at scale. The sector-level governance question is whether the cluster can coordinate these substitutions across the value chain—whether the board suppliers, the finish suppliers, and the furniture makers can align on standards and timelines. Where they can, the compliance cost per unit of emissions reduced is low. Where they cannot, each firm optimizes locally and the sector as a whole underinvests.

I do not have systematic data on how often clusters achieve this coordination. My interviews suggest it is rare, and that it happens mostly when a large buyer forces it—when IKEA or a comparable retailer sets a deadline for phasing out a particular finish or board type and works with its supplier base to meet it. Buyer-driven coordination is fragile because it depends on the buyer's continued attention, and it is uneven because it covers only the suppliers in the buyer's chain. But it is real, and it is one of the mechanisms through which GVC embedding transmits environmental requirements into cluster-level change.

5.10 The Distributional Ledger

I have deferred the distributional question long enough. The compliance cost paradox has winners and losers, and the losers are not randomly distributed.

The first loser group is small firms. The scale effect in compliance costs is strong: a factory producing 20,000 units a year faces per-unit compliance costs three to five times those of a factory producing 100,000 units. The smallest firms cannot amortize testing, certification, or abatement equipment, and they cannot afford the management time to navigate buyer audits. They exit. The consolidation data I cited earlier—from roughly 8,000 designated-size firms in 2015 to about 6,500 in 2022—understates the exit of firms below the designated-size threshold, which are not captured in the statistics at all. The true consolidation is larger.

The second loser group is workers in the finishing and assembly stages. Water-based coating lines and automated spray equipment reduce labor content in finishing by 30 to 50 percent. The workers who operated manual spray guns are not easily redeployed to operate the new lines; the skills are different, and the new lines require fewer people. Some workers move to other stages of production; some move to other factories; some leave the sector. The workers most affected are older, less educated, and more likely to be migrants without local hukou—the same workers who have the least access to retraining and social protection. The green transition in furniture is, among other things, a labor transition, and the labor transition is not being managed.

The third loser group is inland clusters that lack the institutional infrastructure to support upgrading. A firm that relocates from Houjie to an inland cluster to escape compliance costs may find that it has traded one set of problems for another: lower costs but also lower capability, and a buyer base that is less demanding but also less lucrative. The inland clusters that have succeeded—Chengdu is the clearest case—did so because local governments invested in infrastructure and because lead firms brought capability with them. The ones that have not succeeded are stuck in a low-capability equilibrium.

The winners are the firms with private capability and cluster support, the workers who can operate the new equipment, the clusters with strong institutions, and the buyers who get more sustainable products at stable prices. The distribution is not obviously just, and it is not obviously efficient either: some of the firms that exited were viable businesses that could have upgraded with the right support, and some of the workers who lost jobs could have been retrained. The distributional ledger is a policy problem, not a natural outcome.

I want to be careful here about the counterfactual. The alternative to the compliance shock was not a world without change; it was a world in which Chinese furniture firms faced rising labor costs, rising buyer requirements, and rising competition from Vietnam and elsewhere, with or without environmental regulation. Some consolidation was inevitable. The question is whether the environmental dimension of the shock accelerated consolidation in ways that destroyed capability that could have been preserved, or whether it accelerated consolidation in ways that forced capability building that would otherwise have been delayed. My reading of the Houjie evidence is that both happened, and that the balance differed by firm and by cluster. The firms that had begun building capability before the shock—even modest capability—were more likely to survive and upgrade. The firms that had not were more likely to exit. The shock selected for capability, and in doing so it destroyed some firms that might have built capability with more time and support.

5.11 What the Furniture Case Contributes to the Book's Argument

The book's core argument is that green governance capability is an endogenous driver of industrial upgrading and value chain climbing, not merely a passive product of regulatory compliance. The furniture case supports this argument in three specific ways, and qualifies it in one.

First, the case shows that capability precedes compliance in the firms that upgrade. The Houjie factory owners who survived had built measurement systems, supplier relationships, and process documentation before the compliance shock arrived. Their capability was not created by regulation; it was selected by regulation. This is the endogenous mechanism in its clearest form.

Second, the case shows that capability is built at multiple levels. Firm-level capability—measurement, documentation, buyer relationships—is necessary but not sufficient. Cluster-level capability—testing labs, waste treatment, training institutions, supplier networks—determines whether firm-level capability can be built at reasonable cost. Sector-level capability—standards, certification schemes, coordination mechanisms—determines whether the cluster can align with upstream suppliers and downstream buyers. The upgrading pathway runs through all three levels, and weakness at any level constrains the others.

Third, the case shows that relocation is not an escape from the capability requirement. Firms that relocate to lower-cost jurisdictions still face buyer requirements, and they face them with less institutional support. The firms that relocate successfully bring their capability with them; the firms that relocate to escape compliance find that compliance follows. This is consistent with the book's argument that GVC embedding transmits environmental requirements across borders, and that the capability to meet those requirements is the condition of continued participation in the value chain.

The qualification concerns the limits of the mechanism. In labor-intensive sectors with thin margins and fragmented firms, the compliance shock can be large enough to destroy capability faster than it builds it. The firms that exit are not all laggards; some are viable businesses caught by a shock that arrived before they could adapt. The mechanism works, but it works slowly and unevenly, and it produces distributional outcomes that require policy attention. A theory that celebrates the upgrading of the survivors without accounting for the exit of the losers is a theory that mistakes selection for progress.

5.12 Implications for the Dual Carbon Transition

The dual carbon goals—peak carbon before 2030, carbon neutrality before 2060—will tighten the compliance environment for furniture firms further. The question is not whether compliance costs will rise but how they will be designed and sequenced.

The most important design question is whether carbon costs will be imposed upstream or downstream. If carbon pricing falls on the board mills and chemical plants that supply the furniture sector, the cost will be transmitted through material prices, and furniture firms will face higher input costs without necessarily having the capability to reduce them. If carbon costs fall on furniture firms directly—through an expanded emissions trading system or a carbon border adjustment—the firms will face the cost of measuring and reducing their own emissions, which is closer to the capability-building mechanism I have described. The upstream route is more economically efficient; the downstream route is more likely to build capability at the firm level. The choice matters for the sector's upgrading trajectory.

The second design question is sequencing. The firms that have built capability can absorb a gradual tightening of carbon costs; the firms that have not cannot. A phased approach—starting with measurement and reporting requirements, then moving to pricing—gives firms time to build capability before they face the full cost. The EU's CBAM follows this logic, with a transitional reporting period before the financial obligation begins. China's own carbon market has followed a similar phased approach in the power sector. Extending it to furniture and other labor-intensive sectors would require careful design, but the principle is sound.

The third design question is support. The Houjie experience shows that compliance costs without support drive exit; compliance costs with support drive upgrading. The support can take many forms: subsidies for equipment retrofits, shared testing and waste treatment infrastructure, training programs for workers, access to credit for small firms. The Dongguan "replacing solvent with water" program is a model: it combined enforcement with grants covering up to 30 percent of equipment costs, and it was associated with faster adoption of water-based coatings than in clusters without such programs. The program was not perfect—it favored larger firms and did not address worker retraining—but it shows that the design of support matters.

These design questions connect the furniture case to the book's broader policy argument, which Chapter 10 develops in detail. The point to register here is that the labor-intensive pathway is more than a record of how firms respond to compliance costs. It also shows how policy design decides whether those costs build capability or destroy it. The same cost imposed with different support and sequencing produces different outcomes. That is the policy lesson of Houjie.

What the chapter cannot yet answer is how fast the compliance environment will tighten for furniture specifically. The national emissions trading system, as of 2024, covers only the power sector and does not include furniture manufacturing; no announced timetable extends it to downstream consumer goods. The nearer-term pressure comes from buyers and from the EU's CBAM expansion review, not from domestic carbon pricing. That sequencing—buyer pressure first, domestic carbon pricing later, if at all—is itself a finding. It means the capability-building process in furniture is being driven by GVC requirements rather than by Chinese regulation, and that the firms most exposed are those most deeply embedded in export chains. The next chapter turns to the technology-intensive pathway, where the compliance costs are lower relative to value added but the capability requirements are higher, and where finance and innovation play a more central role.