So you want to be like China?
A new paper analyses 3 million Chinese policy documents and comes up with some interesting findings
One of the funniest episodes in recent Nigerian history unfolded in November 2019 during the border closure fiasco. President Buhari had blindsided the nation in August, ordering Nigeria's land borders shut to all goods without warning. The official reason was to stop smuggling and keep criminals out.
The problem was that border closure hadn't appeared even once in his election campaign that year. Only after the shock announcement did officials scramble to justify this out-of-nowhere policy.
Enter Hameed Ali, then Comptroller General of Nigerian Customs and Buhari loyalist, who stepped up to "settle" the debate with this masterpiece of logic:
“So, for anybody to believe that the border closure is the reason for the hardship in the country is completely dumbfounding. There are no statistics to support that claim. The border closure is a win-win for Nigeria”
“Farmers in Abakaliki who could not sell a truck of millet a week, now do so in a day, Go to Kebbi and ask those farmers how business is now. People are beginning to invest in our agricultural system”
“Why do we need to be fed by foreigners? At what point do we begin to truly feel independent? Any country or nation that can not feed itself can not claim independence. Because the crux is that once there is a diplomatic misunderstanding, they will close the gap, what happens then?
“Hence, we must grow Nigeria, we must eat and drink Nigeria. Even China closed its borders to the whole world for 40 years and today it is considered a great nation. Don’t you want to be a great Nigeria?” he added.
If you opposed the border closure, you didn't want Nigeria to be great. Mic drop. Never mind whether China actually closed its borders - this policy was Nigeria's first step toward greatness.
The Great Paper Wall of China
I remembered this absurd episode after stumbling on a fascinating paper by Hanming Fang (University of Pennsylvania) and Ming Li and Guangli Lu (Chinese University of Hong Kong, Shenzhen). In "Decoding China's Industrial Policies," they analysed 3 million policy documents from China's central, provincial, and municipal governments between 2000 and 2022 using AI language models (Google Gemini and ChatGPT).
Why such a massive dataset? These three government levels churn out roughly 100,000 policy documents annually, covering every conceivable industrial policy under the sun. By assembling this treasure trove, the authors identified which policy tools China favours most and which ones actually work.
Due to China's rather opaque language and bureaucracy, policymakers worldwide - particularly in Africa - have taken to invoking "this is what China did" to justify their preferred policies. Given China's spectacular one-generation economic transformation, this argument carries surprising weight. Since China has had a go at virtually everything, anyone can cherry-pick evidence for their pet policy.
The paper cuts through the waffle and systematically maps China's actual industrial policy toolkit over two decades, revealing what really drove success. The findings help us get rid of simplistic narratives - Hameed Ali's border closure thesis included - that attribute China's rise to single magic bullets.
For African policymakers genuinely serious about replicating China's success rather than just borrowing its rhetoric, this paper provides the definitive starting point. No more guesswork about what China "actually" did, here's the proper data.
So what did China do, really?
So what tools has China used in the last 2 decades according to the paper?
Before we get into that, a bunch of caveats. Policy tools are hardly ever used in isolation - subsidies are typically paired with at least one “softer” lever such as R&D credits or labour grants, as such single-tool evaluation studies can be misleading. Central government focuses mainly on regulatory power while cities use cash and land concessions. Further, as industries mature, policy focus shifts from entry (entrepreneurship grants, land breaks) to upgrade (R&D, cluster policies) and finally to supply-chain anchoring, giving a sort of life-cycle view of industrial strategy.
From the history of Shenzen, we already know that coastal cities tend to pilot novel mechanisms (industrial funds, green credits) before they flow up to provinces and ministries. This upends the presumption of China as a top-down system by revealing bottom-up experimentation channels. City governments are first to try new, often costlier tools; provinces follow; the centre is most conservative. Richer cities also lean heavily on those fiscally expensive tools.
The degree to which city policies mirror provincial/central choices however rises markedly from 2013, echoing China’s recent re-centralisation under Xi Jinping who took office in 2012. Yet cities run by officials with strong personal ties to higher-ups stray further from the script, showing that networks - guanxi - still trump rules.
Final important caveat: Because over half the policies omit direct subsidies, relying on budgetary subsidy data grossly understates China’s state activism. And so any praise, or criticism, of Chinese subsidies should bear this in mind.
Fiscal Subsidies
This is the single most common policy tool. It appears in 48% of city policies although only 25% of central policies giving an overall incidence of 41%. We can define fiscal subsidies as direct cash, grants, concessionary loans or utility/land price reductions. These are the single most common instruments in this category. Yet, as caveated above, most policies (59 %) do not contain a subsidy clause, so headline subsidy figures under-state the scale of intervention.
One of my favourite examples of this tool was narrated in Dinny McMahon’s excellent China’s Great Wall of Debt. Shiyan, a Chinese city of 800,000 lacking natural geographic advantages, solved its industrial development problems through massive land creation. In 2006, when Dongfeng's joint venture with Nissan Motors relocated from Shiyan to Wuhan, local authorities realised their constraint wasn't location but available land. Dongfeng had requested 80 acres that the city couldn't provide. To prevent future industrial exodus, Shiyan's government launched an ambitious plan in 2007 to create 400 square kilometres of new, flat land by levelling mountains, more than doubling the city's usable footprint. This massive land reclamation project was described by officials as "the golden key to resolve Shiyan's development impasse." Not long after, Volvo set up in Shiyan.
Why do Chinese authorities like these fiscal subsidies? They provide quick, visible wins that boost GDP and political scorecards. They can also be easily fine-tuned to favour local firms and lure investment in a “race to the bottom” among municipalities. They can also be hidden off-budget which is a sizeable form of local government financing in China.
Market Access and Regulatory Measures
These feature in 42% of central government policies and 35% of policies overall. Under this category the authors find that central government agencies use of licences, quotas, standards and outright entry bans to shape sectors. As China opens up and seeks to maintain its WTO membership due to its export model, the usage of these tools is reducing but it still outstrips every tool other than subsidies.
This tool aligns with the central government’s core competence as it controls the permit system and can impose nationwide technical standards. It is also cheaper than cash outlays and politically useful for curbing bubbles or “blind investment”.
Technology - R&D and Adoption Incentives
This tool features in 24% of policies overall but, unlike market access and regulatory measures which are reducing in usage, this one is seeing increasing usage by Chinese authorities. This is not surprising given the much heralded national upgrading agenda a.k.a “Made-in-China-2025”. It is also less distortionary than blanket subsidies with payments often conditional on patent counts, labs or tech-platform creation.
When the authors decompose effects by tool, technology‐R&D support tops the league table for productivity gains; labour subsidies, entry promotion and even broad tax incentives do little or worse.
Policymakers also like it because innovation metrics weigh increasingly in promotion evaluations for cadres.
Labour Policies
These future in 22% of policies overall (27% in city policies as they have direct control over local labour markets). These include things like wage subsidies, skills-training grants and relaxed hukou quotas to help new entrants. The downside of this tool, which perhaps explains why they are less common is that they are linked with lower average productivity, hinting at a tilt toward labour-intensive technology choices that help achieve social welfare objectives.
As a side-note, “helping‐hand” policies tend to be aimed at low-productivity sectors, whereas “disciplining” policies focus on sectors that are already overheating with high TFP. This explains why supportive tools raise productivity only briefly, while regulatory tools show a persistent drag, according to the authors.
Tax Incentives
This features in 20% of overall policies and, like market access and regulatory measures, it is declining in usage. These take the form of preferential company income tax rates, VAT rebates and accelerated equipment depreciation. They have a low cash-flow cost for governments and can be sunset when needed and made sector-specific. They are also simple to administer through existing tax offices, so enforcement costs are modest and are popular with foreign investors, dovetailing with export-processing zones.
What’s missing?
Given all of the above, what are the surprises in terms of what policy tools do not show up in documents? Firstly, some widely debated practices – forced technology transfer, informal pressure on foreign JVs, or cyber-enabled IP theft – do not leave an official paper trail and therefore cannot be captured by an LLM scan of published regulations. The authors caution that their database “captures what governments say they do; covert or purely administrative practices may still matter but are beyond the scope of text-based measurement” . That caveat itself is very important because many contentious tools are invisible precisely because they are not promulgated policies.
That said, the most interesting policy tool missing in action is that the authors find zero instances of exchange-rate tinkering, monetary easing or other macro-stabilisation tools (the authors adopt a narrow, sector-targeted definition and explicitly rule out “general policies” such as WTO accession, carbon tax or any macro policy that does not bias one sector over another.)
Capital control measures also do not appear in the list of tools. The authors note that while the UNCTAD-MAST template lists “capital controls”, their schema retains only entry subsidies, import/export controls, finance, tax and fiscal subsidies – capital controls are dropped because they are “predominantly national-level trade instruments” and not visible in local policy documents.
The biggest surprise for Nigerian readers comes from the trade file. Tariffs and import bans appear in just 9% of all policies (19% central, 8 per cent provincial, 7% city). Beijing devotes far more energy to deciding who may enter a sector - via licences and standards - than to banning imports.
Consumer-side subsidies are only 5 % of policies while government procurement - another favourite of Nigerian policymakers - a mere 7 %. This is perhaps surprising given all that is often said about electric vehicle subsidies but high-profile consumer-rebate programmes are exceptional, not typical, according to the data. Most demand-side engineering is still done through industrial promotion and cluster policies rather than cash in people’s pocket.
Finally, while fiscal subsidies are indeed the single most common tool (41 %), counting only subsidies will miss most of the story as nearly 60% of policies rely on non-pecuniary levers like licensing, land, labour, R&D mandates, etc. Ignore those, and you get less than half the story.
So you really want to be like China?
If policymakers in Africa are genuinely intent on emulating China’s economic ascent (I assume they are given how impressive the results have been), the first step is to abandon the folklore and attend to the evidence. Fang, Li and Lu’s painstaking review of three million Chinese policy documents makes clear that progress came not from dramatic gestures such as border closures or across-the-board protectionism, but from a disciplined mixture of targeted subsidies, rigorous control of market entry, and carefully sequenced support for research and development as industries evolved.
That approach demands governments prepared to revise, retire and relaunch policies on a regular basis and a political culture that rewards officials for measurable gains in productivity, not for theatrical announcements. The lesson is clear enough. For anyone who genuinely “wants to be China”, it is the patient, technocratic craft of licensing, standards, conditional finance and skills upgrading that merits attention. Anything less is a tribute to the myth rather than an effort to replicate the method.
Replicating China’s model is anything but easy. A sober alternative is to take the L, forego the imitation, and concentrate on benefiting from the competitively priced goods China already supplies. It may sound cynical and unambitious, but it is a clear-eyed appraisal of what is realistically within reach.
Would have been better if the tool addressed policies from 1980. Because that's where Nigeria is still at.
Very interesting, but what are we left with for learning? Can we take lessons from a LLM-based paper? Also - if fiscal subsidies were the main driver, does that help african countries which may not have that tool available?