Le système d'approvisionnement chinois : clusters industriels et spécialisation

Start with the puzzle in the opening. A buyer asks for 5,000 toy cars and receives a full quote in three days. If one factory had to make molds, print stickers, assemble, pack and arrange freight, the queue alone would take weeks. The quote is fast because the work is split across firms that sit close to each other. The seller's job is not to own every machine but to coordinate a network.

The textbook's §1.3 observation is that traditional commerce is often local, while ecommerce lets even narrowly reachable small businesses access wider markets. Applied to supply, this means a cluster no longer sells only to buyers who can visit the market. A foreign buyer can fill an entire product line from one market or one platform, and the cluster can respond without building a new factory for every order.

Why does specialisation lower cost? A plant that only molds invests in molds and molding machines, runs them continuously and becomes good at that step. A plant that tries to do everything must buy equipment for every step and tolerate idle capacity in each. Specialisation also shortens learning time: workers repeat a narrow task and make fewer mistakes. The cluster adds a second saving: proximity. Parts, samples, packaging and trucks travel minutes rather than days, so the sample-to-shipment cycle compresses.

But a cluster is not automatically efficient. Every handoff is a chance for delay, mismatch or a quality dispute. Someone must specify the standard, check the sample, sequence the steps and consolidate the shipment. In many clusters this coordinator is a trader, a stall, or a seller on a platform. This is why "cluster" and "one big factory" are different: the cluster has flexibility, but it also has coordination costs that a single factory internalises.

Textbook §2.1 gives a map for the supply system. The first dimension is goods flow: selling domestic goods abroad is exporting ecommerce; bringing foreign goods into the domestic market is importing ecommerce. The second is parties: B2B, B2C or C2C. The third is ownership: a third-party platform or a self-operated channel. The fourth is category breadth: a comprehensive platform carries many categories; a vertical platform concentrates on one. These dimensions are not alternatives; they are four questions asked of the same transaction.

Use the dimensions in order. Direction tells you which customs and tax rules apply. Parties tell you who the end user is and how the decision is made. Ownership tells you who controls the listing, the customer relationship and the data. Breadth tells you how the buyer searches and compares. A single order can be export, B2B, third-party platform and vertical at once, and each label predicts a different cost or risk.

The Chinese evidence needs careful handling. Customs data show that Guangdong, Zhejiang, Fujian, Jiangsu and Anhui together contribute most cross-border ecommerce exports (SRC-020), and that the export share of China's cross-border ecommerce has risen from about 58% in 2018 to about 80% in 2025 (SRC-038). Those are statistical statements with a year and a scope. Yiwu's market and Chenghai's toy cluster are structural illustrations: Yiwu shows how a specialised market aggregates an enormous variety of small-goods suppliers; Chenghai shows a toy supply chain with molding, assembly and accessory firms close together, and is described by the Shantou government as a provincial "cross-border ecommerce + industrial cluster" pilot, producing a large share of China's plastic toys and selling to more than 140 countries (SRC-044).

The difference between a representative case and a ranking matters. A case can show how something works; it cannot show how large a category is relative to all others. "Yiwu is famous" is a statement about visibility. "General merchandise ranks first in exports" is a statement about magnitude. Only the second needs comparable statistics, and the first cannot support it. Keep the claim types separate, or the analysis becomes a slogan.

Finally, §1.8 distinguishes ecommerce (online buying and selling), ebusiness (running the whole company through internet and intranet) and digital commerce (digitally conducted activity with data flows). For a supplier, joining a platform is only one part of ebusiness. Internal systems, electronic documents, inventory visibility and quality records determine whether the seller can actually coordinate a cluster. The supply advantage is therefore organisational as much as geographical.