A machine prints on continuous plastic film at a printing factory. Photo: IC PHOTO

In county economies dominated by traditional manufacturing and made up of large numbers of micro, small, and medium-sized enterprises (MSMEs), the dense concentration of micro-level business entities within a single industry is often hastily labeled as “inefficient competition” and “overcapacity.” Behind this perception lies a mechanical application of neoclassical perfect competition theory: if numerous enterprises produce “homogeneous” products and market entry is unrestricted, then long-run equilibrium should yield zero economic profit.

The evolution of the printing and packaging industry in Longgang, Wenzhou, Zhejiang Province, over more than four decades directly challenges this prevailing view, revealing a wide gap between statistical illusion and micro-level reality. In this county-level city, which is home to more than 130,000 business entities, including over 40,000 connected to the printing and packaging sector, many MSMEs classified statistically as belonging to the same industry have not fallen into a state of meager profits. Instead, they have given rise to hundreds of highly specialized niches and formed a complex, efficient network of division and collaboration. Based on fieldwork in Longgang, this study attempts to cut through the statistical illusion and uncover the micro-mechanisms behind it. Under specific industrial foundations and spatial constraints, a vast number of business entities have evolved spontaneously into a market network capable of transforming potential competitive pressure into a deep division of labor and collaboration. This process of technical differentiation helps explain the enduring vitality of private MSMEs, while also serving as the core engine through which the entire industrial cluster captures gains from collaboration and achieves endogenous economic growth.

Fieldwork as correction to macro narratives

In conventional industrial policy discourse, the clustering of numerous enterprises is often equated with homogeneous overcapacity. Field research in Longgang, however, reveals a very different micro-level picture: many MSMEs grouped under the same industry in macro-level statistics may, in practice, be highly heterogeneous entities occupying specialized process-based niches.

The findings indicate that the local industrial chain has developed a distinctive spatial configuration. Enterprises in the midstream processing segment are highly concentrated, while upstream raw material suppliers and downstream branding entities remain relatively underdeveloped. This spindle-shaped structure—thick in the middle and thin at both ends—has emerged as different links in the industrial chain, under spatial constraints, have been repeatedly subdivided, differentiated, and recombined according to their physical attributes. The core mechanism lies in the varying sensitivity of different segments to spatial friction. Midstream processes such as proofing, color matching, and die-cutting rely heavily on face-to-face communication to resolve complex information asymmetries, creating the strongest demand for geographic clustering and naturally giving rise to exceptionally high manufacturing density.

Relevant data clearly confirm this process of differentiation. Among the 279 industrial enterprises above designated size, 23.7% are engaged in upstream raw material supply, 28.7% in midstream processing and manufacturing, and 24.4% in downstream finished products. Supporting this structure are tens of thousands of micro-enterprises and self-employed individuals. These capillary-like market actors break the midstream segment into hundreds of non-overlapping process nodes—from specialized processes in packaging decoration and flexible packaging to highly narrow tracks such as self-adhesive labels and anti-counterfeiting labels. Within each highly segmented field, there are specialized “hidden champions.”

Extreme spatial compactness forms the foundation of this process differentiation. Along Century Avenue in Longgang alone, 19 enterprises above designated size are clustered together, while countless micro-enterprises are densely distributed through streets, alleys, and old industrial zones, weaving a highly flexible and responsive division-of-labor network. A company specializing in gravure cylinder manufacturing noted that such narrow specialization would have been impossible in a county-level economy two decades ago, but today it can thrive by drawing on a vast local demand pool made up of more than 40,000 business entities. Within this network, the full production process for a paper wine box—from gray board paper to finished product—or for an eco-friendly bag—from raw materials to bag-making—can be completed within the city. What appears from a macro perspective to be homogeneous clustering is, in reality, a highly complex heterogeneous division of labor within the industry.

Low-friction networks reshaping rules of competition

To understand the vitality of such clusters, it is necessary to move beyond a passive perspective focused on avoiding involution and to recognize the active logic through which gains from collaboration are generated. Spatial agglomeration reshapes the structure of transaction costs, converting high external transaction costs into efficient internal coordination mechanisms.

In Longgang, the marginal cost of inter-firm collaboration has been compressed to an extraordinary degree. Proof confirmation, delivery coordination, and quality feedback can be completed within hours. When a temporary process-related need arises, a suitable partner can be found within half an hour. This convenience generates two forms of benefit.

First, it shifts the battlefield of competition and helps enterprises secure profits. When the cost of external collaboration is extremely low, engaging in price wars in end markets becomes uneconomical. The rational choice for enterprises is to retreat into specific process nodes and secure incremental niches by deepening specialization. Intense co-opetition, in effect, means that micro-level entities abandon low-level cutthroat competition in end markets and instead form specialized complementarities across adjacent process links, thereby locking in stable profits within niche markets.

Second, it enables economies of scale to be realized through networks. Without this high-density division-of-labor network, enterprises would have to secure their supply chains through internal vertical integration, inevitably incurring substantial management costs. In Longgang, however, extremely low coordination costs allow vast numbers of micro-enterprises to achieve, through market networks, efficiencies comparable to or even greater than those produced by vertical integration inside large corporations. More importantly, market-based network collaboration has stronger incentive effects than corporate hierarchy. As independent residual claimants, micro-level entities have far greater incentives to innovate and reduce costs than production or service departments inside large enterprises.

Financial logic further reinforces this structural lock-in. Some enterprises have attempted to relocate certain processes to reduce costs, only to find that hidden costs—such as communication delays and disjointed quality control—rose exponentially in cross-regional collaboration, ultimately forcing them to return. The high marginal coordination cost of relocation creates a strong financial pull back toward the local network, reinforcing its stability.

Midstream thickness as ‘efficiency base’

The Longgang case offers an important corrective for local industrial policy. In industrial policy discourse, the dominant narrative has long relied on macro-level industry classifications to interpret micro-level reality, simplistically treating large-scale, high-density midstream segments as excess capacity that should be reduced. The limitation of this approach lies in mistaking statistical industry homogeneity for actual product homogeneity. Fieldwork reveals a key proposition: the number and density of enterprises in the midstream processing segment are indispensable core assets that support the entire process-based division-of-labor network and generate gains from collaboration. For county economies, what appears to be a crowded cluster of processing enterprises is in fact a strategic asset that enables them to secure core competitiveness across the industrial chain. From ink and printing cylinders to bag-making equipment, and even early exploration of biodegradable materials, these critical links require little external attraction; localized production capabilities can evolve spontaneously on the basis of massive manufacturing density.

This view is also widely shared among Longgang entrepreneurs. Precisely because tens of thousands of enterprises in the surrounding area can provide a full range of differentiated processes within half an hour, they are willing to commit all their resources to a narrow technical track. If this manufacturing density, sustained by countless business entities, were forcibly suppressed in the name of regulation or upgrading, the division-of-labor network and the gains from collaboration on which it depends would collapse almost immediately.

From ‘physical density’ to ‘institutional density’

The spindle-shaped structure built on high-density clustering is not the endpoint of this evolution. As constraints on land, the environment, and other factors tighten, and as marginal returns in the midstream segment diminish, clusters face pressure to move toward the two ends of the smiling curve. Longgang now stands at this critical juncture.

Field research finds that when Longgang enterprises attempt to extend into higher value-added segments such as R&D design, brand services, and testing and certification, they encounter new institutional frictions. For example, products exported to the EU must be sent to Hangzhou, Shanghai, or other cities for testing, while patent applications depend heavily on external agencies. The absence of local high-end producer services has become a new barrier to collaboration, weakening the gains from a deepening division of labor.

Traditional thinking centered on hard investment attraction is therefore facing new challenges. The core elements of high-end services are not tangible machinery and equipment, but the effective use of local tacit knowledge and a trust-based culture grounded in the rule of law; they cannot simply be airlifted in from outside. If external design agencies do not understand the process boundaries within the local supply chain, their solutions will be difficult to implement. If external testing platforms are unfamiliar with the specific characteristics of the local industry, their reports will fail to meet practical needs.

This implies that the key variable in breaking through midstream lock-in is shifting from reducing physical friction to reducing institutional friction. Accordingly, the focus of policy should move away from building hard environments through competition over land, tax incentives, and other preferential policies, and toward constructing soft environments that lower transaction costs and promote market-based division and collaboration. Governments should concentrate on providing quasi-public goods—goods that generate significant positive externalities but that the market fails to supply on its own. These may include jointly built regional platforms for quality testing and R&D pilot trials, efficient mechanisms for rapid intellectual property rights protection, and professional service institutions deeply embedded in the local industrial ecosystem. Only by systematically reducing institutional transaction costs can high-end factors take root in counties and enable clusters to undergo a qualitative transformation from manufacturing agglomeration to an ecosystem empowered by innovation and services.

The evolution of Longgang’s printing and packaging industry cluster shows that, on the basis of traditional industries, extreme spatial clustering can trigger process-based differentiation and thereby transform potential involutionary pressure into powerful gains from collaboration. Its competitive advantage comes from a market network that activates and sustains the specialized division of labor among micro-level entities, along with a corresponding institutional environment. Behind this model of process differentiation and collaborative gains lies an endogenous development path that enhances industrial competitiveness not through external grafting, but by optimizing the business environment in line with local conditions.

Mao Zhuqing and Li Huiwen (professor) are from the School of International Business at Shanghai University of International Business and Economics.