China’s reform-era party-state capitalism

One label covers changing arrangements

“Party-state capitalism” is an analytical label, not the Chinese state's constitutional name for its economy. Margaret Pearson, Meg Rithmire, and Kellee Tsai use it for a late-reform configuration in which Party survival and risk management increasingly shape the treatment of state, private, and hybrid firms.¹ Barry Naughton offers a compatible chronology with a different causal emphasis: market-oriented reform dominated from 1978 through the early 2000s, direct industrial intervention began rising around 2006, and the later system might be called a “government-steered market economy.” He cautions against attributing earlier growth to policies introduced only after that growth was well under way.²

The label therefore should not imply one master design. At least three arrangements succeeded one another. Rural households and local governments first received or created room to depart from collective and planned production. State-enterprise restructuring and global trade then joined markets to continuing state control over strategic assets and appointments. Under Xi Jinping, markets and private enterprise remained indispensable, while Party leadership, security, and political discipline became more explicit inside economic governance.¹²

Across all three arrangements, dispersed participants supplied knowledge that a central plan could not generate transaction by transaction. The center retained the power to authorize, recognize, scale, redirect, or close what they built. That combination—not “state” or “market” alone—is the organizational object to explain.

Rural contracts made local knowledge legible

In December 1978, farmers and local cadres in Xiaogang, Anhui, made a secret agreement to allocate production responsibility to households while preserving collective ownership and state procurement obligations. Simon Deakin and Gaofeng Meng count twenty signers, including two cadres, and emphasize that the surviving agreement's authenticity has been questioned even though its basic account is generally treated as accurate. Their reconstruction also places Xiaogang among other local trials rather than presenting it as a solitary invention.³

The institutional change separated a bundle that collective farming had joined. Land remained collectively possessed, while households acquired use, production, and residual-income rights. A joint World Bank and State Council Development Research Center study reports that about 94.2 percent of rural households had entered the household responsibility system by the end of 1983. It summarizes studies attributing 30 to 50 percent of agricultural output growth from 1978 to 1984 to the system, while also identifying procurement-price changes, modern inputs, research, irrigation, and market liberalization as part of the causal account.⁴

This was not autonomous local lawmaking. Sebastian Heilmann describes reform-era experimentation as a hierarchical process in which local initiative and policy variation could generate information, but central authorities selected which trials received protection, formal recognition, and wider diffusion. His cases show learning through experimentation without a transfer of ultimate political authority to the experimenting locality.⁵ The organizational bargain was therefore precise: local actors could sometimes discover a better practice before the center named it policy, yet continued permission depended on higher-level judgment.

That bargain made delegation, decentralization, and responsibility a source of learning and a source of exposure. A household or local cadre could reveal useful information, but had no independent guarantee that a tolerated departure would remain protected. Selection by the center helped successful trials travel; it also made information upward-facing, because local actors had to anticipate what higher officials would recognize as success.

State-enterprise reform retained control and redistributed risk

Late-1990s state-enterprise reform did more than privatize. Chang-Tai Hsieh and Zheng Song document the policy known as “grasp the large, let go of the small”: small state firms were closed or sold, while large firms were corporatized, grouped into conglomerates, and kept under central or local state control. In their industrial-firm data, more than 80 percent of firms under state control in 1998 had closed or been privatized by 2007; among firms still under state control in 2007, legal registration often appeared private, making control harder to infer from the registration label alone.⁶

Corporatization created boards, controlling shareholders, profit measures, and managerial accountability without removing political appointment. Hsieh and Song describe local governments appointing leaders of locally controlled groups, while the Party's Central Organization Department retained appointment authority for central groups. Their estimates find convergence between state and private firms in labor productivity from 1998 to 2007, persistently lower capital productivity in the state sector, and faster total-factor-productivity growth among state firms. The study also warns that its firm data do not settle post-2007 patterns and that welfare conclusions depend on its model.⁶

Firm performance is not worker welfare. Xinping Tian, Jinquan Gong, and Zhe Zhai use China Health and Nutrition Survey observations from 1991 through 2011 to compare workers displaced during the mass state-enterprise restructuring with workers who were not displaced. They find persistent earnings losses, driven mainly by lower post-displacement wages, and greater reemployment in lower-skilled occupations, less profitable firms, private employers, and informal contracts.⁷ Their design follows observable individuals in a longitudinal survey; it does not establish one national loss for every displaced worker.

The distinction matters for measurement, accounting, and control. Closure, sale, profitability, labor productivity, capital productivity, and worker security answer different questions. A reform could improve the measured performance of surviving firms while transferring income risk and welfare obligations to workers, households, and still-developing public systems. Treating one metric as the whole result would erase the reallocation that made the metric improve.

WTO entry connected the bargain to global production

China became the World Trade Organization's 143rd member on 11 December 2001 after accepting the accession protocol and schedules negotiated with existing members.⁸ Membership supplied a rules-based channel into global goods and services markets; it did not make ownership, finance, land, infrastructure, or appointments politically neutral.

The joint World Bank–Development Research Center study describes coastal export growth, the extension of supply chains inland, private and foreign-invested employment, and rural-to-urban migration as mutually reinforcing parts of economic transformation. By 2018, it reports 288 million migrant workers and estimates that rural migrants accounted for more than 40 percent of urban employment. Migration raised wages and remittances for many households, but in 2016 about 68 percent of surveyed migrants remained in informal employment and fewer than 20 percent were enrolled in urban health and social-security schemes.⁴

The Foxconn production system is an editorially selected nested case, not a proxy for every Chinese workplace. It gives the national arrangement a bounded workplace comparison in which brand customers, local government, supplier management, labor, and production metrics can be examined together. The relation is part-to-whole and diagnostic; conclusions from Foxconn remain local to that case unless separately supported.

The gains were far broader than corporate output. The same joint report estimates that the number of people in China below the World Bank's $1.90-per-day international poverty line fell by close to 800 million over four decades, accounting for nearly three quarters of the global reduction since 1981. It attributes the change to both broad economic transformation and targeted poverty programs, notes that many people remained vulnerable at thresholds more appropriate to an upper-middle-income country, and calls for further work on causation, counterfactuals, and hukou restrictions.⁴ Because the study was jointly produced with a State Council research body and draws on official statistics as well as household surveys and background studies, it is strong evidence for the stated measures and an identified institutional account—not an independent attribution of the entire outcome to one governing arrangement.

Constitutional texts place markets inside Party leadership

The 2018-amended state constitution joins provisions often separated in political shorthand. Article 1 names Communist Party leadership as the defining feature of socialism with Chinese characteristics. Articles 6 and 7 make public ownership the mainstay and the state sector the economy's leading force. Articles 8, 11, and 13 recognize household contracting, non-public sectors, and lawful private property; Articles 15 and 16 establish a socialist market economy and legal operating autonomy for state enterprises.⁹ These are formal legal commitments. The official English text does not by itself show how courts, regulators, firms, or citizens can enforce them when commitments conflict.

Company law adds another layer. Its Articles 18 and 19 provide for employee representation and trade-union consultation and require Communist Party organizations in companies, with companies supplying the conditions for their activities.¹⁰ The provisions establish that a corporation's legal organs do not exhaust its recognized organizational structure. They do not measure the practical influence of an enterprise Party organization or the independence of worker representation.

The Party's own 2022 constitution is more explicit about hierarchy. Its General Program says markets should play a decisive role in resource allocation while the Party consolidates the public sector and encourages, supports, and guides the non-public sector. Article 33 assigns state-enterprise Party committees a leadership role and authority to discuss and decide major issues, while also instructing them to support boards and managers in exercising lawful powers. Party organizations in non-public firms are charged with implementing Party policy, overseeing legal compliance, and leading mass organizations.¹¹ As an official participant document, the constitution establishes the Party's rules and claims about itself; it is not independent proof that participation, worker reliance, legality, and leadership are reconciled in practice.

The result is not an absence of corporate governance. It is a dual ordering in which shareholders, boards, managers, contracts, regulators, and market prices operate, while Party organs retain a distinct claim to political direction. The central question for authority, legitimacy, and acceptance is therefore not whether firms have room to act, but who has standing to settle conflicts among commercial judgment, law, Party policy, worker interests, and public purpose.

The Xi-era turn tightened political priority without ending markets

Pearson, Rithmire, and Tsai locate the emergence of party-state capitalism in the late 2000s and identify three linked features: deeper Party-state reach into economic organization, a blurring of state and private sectors, and stronger demands for political alignment from firms. They also emphasize constraints and backlash rather than describing an all-capable state.¹ Naughton's chronology likewise finds a major increase in direct intervention after 2006, but stresses that the effectiveness of targeted industrial policy remained unproven and should not be credited retroactively for earlier development.² The two accounts support the turn toward stronger steering while warning against treating control as either complete or necessarily effective.

Ant Group's planned Shanghai listing supplied a bounded example. On 3 November 2020, the Shanghai Stock Exchange suspended it after regulators interviewed the controlling person and senior managers. The exchange's notice cited changes in the financial-technology regulatory environment and the possibility that Ant no longer met listing or disclosure requirements.¹² That official record establishes the decision and its stated grounds, not whether speech, prudential risk, data governance, political discipline, or some combination caused it. Pearson, Rithmire, and Tsai read the case as both a response to a systemically important nonstate financial firm and a demonstration that private capital remained politically subordinate; they specifically caution against reducing it to Jack Ma's preceding critical speech.¹

Centralization also changed local experimentation. Abbey Heffer and Gunter Schubert's study of contemporary policy cases rejects a simple claim that local experimentation stopped under Xi. They describe “experimentation under pressure”: local officials may still be required to innovate, but do so under stronger top-down steering and political risk, often seeking authorization or choosing variations that visibly answer central demands.¹³ The authors explicitly limit generalization from their two comparative cases, so the concept identifies a mechanism to test rather than a measured national frequency.

Xinjiang shows why capacity cannot be evaluated only through growth or firm performance. A 2019 State Council white paper describes vocational education and training centers as lawful, voluntary, preventive institutions intended to counter terrorism and extremism, teach skills, increase employment, and protect social stability.¹⁴ After reviewing official laws and statements, open-source material, and forty interviews with people having first-hand knowledge, the Office of the UN High Commissioner for Human Rights found that placements involved deprivation of liberty and that large-scale arbitrary and discriminatory detention of Uyghurs and other predominantly Muslim groups may constitute crimes against humanity. It also found indications that labor and employment schemes could involve coercion and ethnic or religious discrimination.¹⁵

The UN assessment states its limits: it lacked official population totals for the centers, did not claim that every open-source document was independently verified, and could not establish the exact extent of every alleged abuse. It also incorporated government information, sent the assessment for factual comment, and appended the government's rejection. Those limits narrow the claims; they do not reconcile the official account with the UN's findings. The first-hand testimony supporting these findings is mediated through OHCHR interviews conducted with protective measures rather than a community-controlled Uyghur record, an important boundary on whose evidence is publicly available.¹⁵ The organizational relevance is specific: public agencies, data systems, training institutions, and employers can be coordinated around a centrally defined security purpose, while the people classified by that system may lack an independent and effective route to contest its premises or remedy its effects.

Capacity and burden cannot be collapsed into one score

The poverty study describes a state able to align central targets, cadre responsibility, local discretion, public finance, state firms, private actors, and large work teams around a defined objective. It also notes that mechanisms effective for a single target face difficulty when objectives require trade-offs and leaves the net merits of managed migration restrictions open for lack of a credible counterfactual.⁴ This distinction is essential: mobilization capacity answers whether organizations can move resources toward a target, not whether the target is complete, the burdens are fairly distributed, or affected people can require correction.

A national benefit claim would combine populations whose experiences are not commensurable. Rural households obtained production rights; many workers and suppliers gained income; displaced state workers lost earnings and security; migrants entered higher-wage labor markets without equal benefits; owners built large firms under changing political permissions; and some ethnic communities were subjected to coercive security policy. Each statement has a different population, period, measure, and source base. “Mixed” is therefore not a claim that harms and benefits cancel. It records that direction changes with subject and that aggregation would conceal who carried which consequence.

The same restraint applies beyond human economic measures. Industrial growth, agriculture, infrastructure, extraction, pollution control, restoration, and clean-energy production are too broad to convert into one ecosystem direction without lifecycle and counterfactual evidence. Effects on nonhuman life and future generations require their own measures, geographies, and time horizons. Leaving those directions unresolved is more accurate than inferring ecological or intergenerational benefit from state capacity—or ecological burden from industrial scale alone.

Comparative relations and research lenses

The Singapore relation is a structural comparison between two economies that combine markets, public capacity, and strategic direction; it is not a claim that their political authority, ownership, labor relations, or scale are equivalent. Soviet planning is a plan-versus-reform comparator, not a claim that post-1978 China simply preserved the Soviet institutional form. Foxconn is the narrower nested case already identified. The benefit for all life relation is evaluative: it asks whether beneficiary accounting extends beyond growth, owners, the state, and present human populations.

Six ideas define the comparison. Authority, legitimacy, and acceptance identifies who may set and contest purpose. Delegation, decentralization, and responsibility tracks local initiative inside central selection. Measurement, accounting, and control separates firm output from transferred burdens. Strategy, competition, and adaptation connects market rivalry to industrial direction. Governance, stewardship, and accountability asks who can review trade-offs. Executive attention, information, and organizational sensing tracks which local signals reach central decision makers.

Supporting lenses distinguish purpose, mission, and institutional legitimacy, coordination, communication, and common understanding, structure, hierarchy, and scale, and decision making, judgment, and bounded rationality. They also test cooperation, incentives, and organizational equilibrium, knowledge, expertise, and professional autonomy, learning, quality, and reliability, innovation, entrepreneurship, and renewal, and organizational ignorance. The national case does not make work design, productivity, and automation or culture, informal organization, trust, and voice defining; those questions require workplace- and community-level cases rather than inference from macro institutions.

The organizational profile is an explicit comparative inference from the cited account. State bureaucracy and market capital are concurrent authority sources; central, divisional, local, and rule-bound loci all make decisions; state, listed, and private ownership coexist. Planning, markets, hierarchy, and metrics coordinate activity, while knowledge moves downward, upward, through specialists, and through embedded practice. Financial, operational, and mission measures coexist with experimentation, market feedback, and doctrinal revision. The corresponding risks—capture, suppressed voice, metric-gaming, externalized harm, and fragility—are hypotheses made inspectable by the historical episodes, not direct measurements of the whole economy.

Protected contradiction remains the governing problem

By 2022, the institutional question was not whether China had chosen planning or capitalism. It had repeatedly changed the relationship between them. Household contracting let local practice teach the center;³⁵ state-enterprise reform joined corporate forms to retained political control;⁶ WTO membership expanded competition and production networks;⁸ Xi-era governance made Party priority more explicit while continuing to rely on firms, markets, experts, and local implementation.¹¹¹¹³

The unresolved problem is protected contradiction. Large-scale coordination needs purpose, finance, standards, implementation, and the ability to move across organizational boundaries. Learning also needs people who can preserve inconvenient evidence, refuse an unsafe target, report a failed experiment, and appeal beyond the official or Party organ whose program they challenge. China's reform-era arrangements produced unusual capacity by combining hierarchy with dispersed initiative. Their hardest test is whether dispersed initiative can remain genuinely informative when the same political organization selects the purposes, recognizes the experiments, judges the results, and controls the ultimate forum of appeal.