From Local Experimentation to Central Compliance: How the Centralization of China's Industrial Policy Is Rewriting the Logic of Global Supply Chains

An Often Overlooked Trade Variable: How Policy Is Made

International trade analysts studying China's industrial policy are accustomed to focusing on policy content—subsidy scale, export tax rebates, capacity targets, technology catalogs, export control lists. But a brief released in September 2025 by the Stanford Center on China's Economy and Institutions (SCCEI) moves the focus one step earlier: by whom, at what level, and under what incentive structure policy is made.

The significance of this perspective for supply chain research lies in this: policy content determines industrial direction, while the policy formation mechanism determines the degree of fit between industrial direction and real-world conditions. And differences in that degree of fit, along the chain of production geography, supporting investment, logistics radius, and unit cost, ultimately show up in export structure and commodity demand.

The brief draws on the NBER working paper (w34219) by Luo, Wang, and Yang. Its sample covers 3.7 million public policy documents from 2004 to 2020, drawn from the Peking University Law database and prefecture-level city government work reports. Through keyword identification, it identifies about 116,000 independent policies, and through text matching tracks the full life cycle of each policy—its origin, diffusion path, and whether it was adopted by the central government.

A Bottom-Up Policy Market

The study's most counterintuitive finding is that the vast majority of China's policies do not originate in Beijing. During the sample period, about 80% of policies originated at the prefecture-level city level. Among locally initiated policies, 68% spread to an average of four other prefecture-level cities within three years, while 32% were never replicated; 24% eventually entered the central government's view and were elevated to national pilots or directives. At any point in time, about 63% of a local policy portfolio is a purely "bottom-up" product. Even when localities adopt central directives, they often rewrite and localize them rather than copying them word for word.

The driver of this mechanism is officials' promotion incentives. The study shows that local officials who proposed policy innovations had about an 8% higher probability of promotion before 2013. Once such an official leaves office, the diffusion speed of their policies falls by about 40% and does not recover. At the same time, competition creates friction: for two prefecture-level city party secretaries with similar résumés who are both eyeing provincial positions, the probability of adopting each other's policies is 1–2% lower. Shanghai's successfully operated license plate auction system was not adopted by Beijing and ultimately gave way to a lottery—a typical footnote to this friction.

Reading this passage from the perspective of trade and industrial geography yields an important judgment: China's manufacturing clusters are, over the long term, the result of layer upon layer of decentralized decisions—local officials promote local industrial experiments to seek promotion, successful practices are imitated by neighboring prefecture-level cities, and industrial supporting facilities therefore spontaneously cluster in specific regions. This explains why the same industry within China often forms highly segmented, mutually responsive industrial clusters, rather than being rolled out all at once according to a unified national plan.

After 2013: Compliance Replaces InnovationThe research’s turning point is clear. After 2013, central policies rose from about 30% to more than 40% of local policy portfolios, an increase of roughly 40%. The implementation speed of central policies also increased markedly: before 2013, a typical central policy reached about 10 prefecture-level cities on average; afterward, it reached nearly 30. By contrast, the three-year diffusion radius of locally initiated policies remained essentially flat at about five prefecture-level cities. The probability that localities “copied word for word” central policies doubled.

Incentives shifted in tandem. Before 2013, officials who promoted bottom-up policy innovation had an approximately 8% higher probability of promotion; after 2013, this association disappeared, replaced by compliance: faster and more complete implementation of central directives was associated with an approximately 8% higher promotion probability. In policy areas covered by central leading groups established after 2013, local experimentation slowed markedly or even reversed, while obedience to central directives rose sharply.

The Cost of Mismatch: From Ghost Wind Farms to Export Opportunity Costs

The study’s core calculation concerns the match between policies and local conditions. Relative to existing supply chains and prior private investment, top-down industrial policies were 18%–22% less well matched than bottom-up policies. The most intuitive example is wind power: central directives pushed wind power projects into provinces with poorer wind resource endowments, leaving behind large numbers of “ghost wind farms”—installed capacity exists, but generation is far below expectations.

The researchers estimate on this basis that such mismatch causes about RMB 400 billion in lost industrial output per year, about RMB 32 billion in lost exports, and roughly 750 fewer patents. Centralization did bring some coordination benefits by reducing bureaucratic competition frictions, but the study concludes that costs outweighed benefits by a ratio of more than four to one.

From a supply chain and commodities perspective, this estimate is worth unpacking. Wind turbines and supporting equipment depend heavily on steel, copper, rare-earth permanent magnet materials, and fiberglass or carbon fiber. A mismatch between installation siting and wind resources means upstream materials demand is locked into inefficient assets: when generation falls short of expectations, it in turn affects the pace of grid investment and subsequent equipment order cycles, creating a temporal misalignment in demand signals. For global copper, rare earth, and specialty steel markets, this kind of mismatch changes not the total volume of demand but its cadence and certainty—and cadence is precisely the variable to which pricing and inventory strategies are most sensitive.

A caveat on scope: the above figures are partial-equilibrium estimates from a working paper based on a specific identification strategy. They do not equal a decline in China’s total exports and should be understood as opportunity costs caused by specific policy mismatches, not as a forecast for macro trade. The paper is an NBER working paper and has not yet undergone peer review.

The Divergence Between Security Goals and Efficiency GoalsResearch has not concluded that "central policies are comprehensively inferior to local policies." An important caveat is that central policies are not more accurate than local ones in selecting high-growth industries; local governments are often on par with or even better than the center at supporting high-growth sectors; but the center does have different emphases across dimensions—it places more emphasis on national security (for example, industries affected by U.S. export controls) and environmental goals.

The implication of this for multinational buyers is fairly direct: China's industrial policy orientation is partially shifting from "growth maximization" toward "maximization of security and strategic priorities." When policy objectives are multiple and priorities are set at the national level, the availability, prices, and delivery cadence of certain inputs will be more determined by policy ranking, rather than purely by market supply and demand. For firms dependent on Chinese intermediate goods, suppliers' capacity decision-making logic may shift from "following orders" to "following policy catalogues"; the response functions of the two are not the same.

Possible Transmission Chains: Several Analytical Hypotheses to Be Tested

Integrating the above evidence into the global trade framework, several transmission pathways that need ongoing tracking can be proposed. They are currently analytical hypotheses, not established facts:

First, rising policy consistency may accelerate the synchronization of scale deployment. When local autonomous adaptation space narrows and implementation speeds up, capacity expansion in specific segments is more likely to occur synchronously nationwide, thereby affecting global prices and the number of trade remedy cases in that segment.

Second, declining local adaptation may push up supporting and logistics demand. If new capacity is disconnected from the existing supply chain base, demand for transportation, warehousing, and inland distribution of intermediate inputs needed to bridge the gap will rise accordingly, and unit transport costs and inventory turnover structures will change accordingly.

Third, narrowing local autonomy may change the negotiation structure for foreign investment implementation. Local governments' policy flexibility in attracting investment tends to shrink, and the source of certainty for project terms shifts from local to higher levels. Changes in project implementation cycles and the predictability of terms deserve attention.

Fourth, rising security priorities will adjust the import and export flows of critical goods. In fields such as critical minerals, energy equipment, and semiconductor equipment, higher priority for domestic allocation may change the cross-border flow rhythm of related categories.

Significance for Ports, Logistics, and Regional Trading Systems

The implication of this study for ports and shipping is not about a specific cargo type, but about the geographic stability of cargo flows. When capacity layout is driven more by national-level policy consistency rather than local conditions and existing supporting facilities, cargo source locations may be more concentrated and easier to predict during the planning period, but once policy direction changes, the magnitude of cargo volume migration may also be larger. For port investment and route network planning, this combination of "high certainty, low elasticity" requires stress-testing models different from previous scenario assumptions.At the regional level, greater policy consistency means that China's capacity deployment within the RCEP framework may be more synchronized and exhibit greater economies of scale, which will have a structural impact on competition and the industrial absorption landscape in Southeast Asian manufacturing. This judgment is likewise an analytical inference and needs to be continuously verified with regional trade data and FDI flows.

Long-term judgment: Policy production efficiency itself becomes a supply chain variable

Over the past decade, global supply chain "de-risking" discussions have mainly revolved around tariffs, export controls, industrial subsidies, and friend-shoring. This study points to a more fundamental dimension: an economy's policy production efficiency—that is, its ability to match policies with local conditions—is also a component of supply chain reliability.

The study concludes that China is trading some growth and policy innovation for tighter economic control. For the global trading system, this means that forces in two directions exist simultaneously: strengthened central coordination capacity may accelerate the scaling-up of certain strategic industries; while declining local adaptation brings resource misallocation and transmits to global markets through export prices, capacity utilization, and upstream commodity demand.

For supply chain researchers, the next task is not to judge whether centralization itself is good or bad, but to incorporate "policy source structure" into the monitoring framework—when evaluating suppliers, site selection, and sourcing strategies, look not only at what a policy says, but also at where the policy grew from, whether it is copied verbatim or rewritten locally, and what incentives are driving its implementation. In trade analysis, the production function of policy is becoming as important as policy itself.