You have to love the fact that all four of China’s main private-sector express-delivery companies–the guys shuttling all those Taobao packages around the country–are from the same tiny place. Gabriel Wildau at the Financial Times recently did a nice piece on the phenomenon, and Bloomberg has also picked up on it. It’s one of the most striking examples of the “cluster” phenomenon you can find.

But here’s the thing: the Tonglu cluster, for one of the biggest new-economy growth sectors, is right next to a bunch of other, older clusters. Zhejiang is famous for its specialized light-industry concentrations; a very good World Bank history of these clusters tells the stories of Haining for warp-knit cloth, Qiaotou for buttons, Yiwu for zippers, Yongkang for metalware; Wenzhou for cigarette lighters is another well-known one. Years ago I visited Zhili, a children’s clothing cluster, to write a piece for the WSJ.

At the time it was common to worry about the future of these places because of rising labor costs. But exports from China and Zhejiang have in fact held up better than many expected. And the Tonglu example suggests that Zhejiang’s real competitive advantage is not low labor costs, but a highly flexible and entrepreneurial economy that can adapt quickly to changing trends. This structure has deep historical roots: Zhejiang (along with Jiangsu) has had one of the least state-dominated economies of any Chinese province for as long as statistics have been kept.

If the coastal provinces are transitioning successfully from export-oriented industry to high technology and services, then patterns of regional advantage are being entrenched rather than overturned by structural changes in the economy. Enrico Moretti, in his The New Geography of Jobs, argued that “the knowledge economy has an inherent tendency toward geographical agglomeration.” China certainly seems to demonstrate the point: Just six provinces–Beijing, Shanghai, Guangdong, Jiangsu, Zhejiang, and Shandong–account for about 60% of national R&D spending. The same six provinces account for about 40% of national investment in information technology and software. The clusters of the future seem likely to appear not very far from the clusters of the past.