Haier's Rendanheyi

Quality discipline preceded decentralization

When municipal official Zhang Ruimin became director of the struggling Qingdao General Refrigerator Factory in 1984, his early program concentrated authority around product quality. Haier's corporate chronology dates the emblematic smashing of 76 defective refrigerators to 1985 and later records the hammer's placement in the National Museum of China. The chronology supplies Haier's own institutional memory; an independent management study supports the broader 1984 turnaround chronology but does not independently reconstruct every detail of the hammer episode.¹²

The sequence matters because Rendanheyi did not begin as spontaneous self-management. Zhang first made standards and market response harder to ignore, then repeatedly changed the structure through which employees met them. Haier says its 2012–2019 strategic stage changed the company from a manufacturer of products into a platform for entrepreneurs.¹ The authority to decentralize was itself centralized.

Rendanheyi, introduced in 2005, links a person or employee (ren) with an order or user need (dan) and their integration. Its practical claim is that people should not wait for a hierarchy to interpret the customer. Teams should find a need, assemble capabilities, make commitments, and bear consequences closer to the user. That is a theory of operating responsibility, not by itself a rule for who owns the corporation or may rewrite its constitution.²

Microenterprises distribute operating authority under retained corporate powers

Haier first moved through strategic business units and cross-functional ZZJYT teams. In 2014 it reorganized around xiaowei, usually small microenterprises. The independent study describes user-facing units and node units that provide services to them. It reports that a microenterprise leader could recruit, contract, budget, choose outside suppliers, seek external partners, and hold an ownership stake; financial indicators and “user value added” determined performance. Headquarters still allocated corporate resources, supplied platforms, and set enterprise strategy.²

This design pushes delegation, decentralization, and responsibility beyond ordinary delegation of a departmental budget. A service team no longer receives demand simply because its box appears on an organization chart; it may have to win another team's commitment. A product team can combine people and resources around an opportunity without routing each choice through a chain of managers. Yet the brand, capital pool, strategic boundary, platform access, accounting rules, and selection environment remain corporate powers.

The ownership record makes the distinction sharper. Frynas, Mol, and Mellahi describe Haier Group as partly publicly owned and technically collective, but also report that exact stakes were opaque, employees received no dividends, and rank-and-file workers had negligible influence over strategy. They attribute the transformation to strong, top-down leadership. The same study says more than 20,000 employees left in the decade after Rendanheyi began and that staff fell from more than 80,000 in 2012 to 60,000 in 2016.²

Contemporaneous reporting in 2014 offers a separate but overlapping view. It described roughly 2,000 self-managed teams, continual formation and dissolution, senior executives retaining great authority, and about 16,000 jobs cut in 2013. Those workforce figures overlap the longer period reported by the management study and should not be added together. The report also relied heavily on Zhang, managers, and selected product stories rather than a representative workforce sample.³

Operating autonomy was therefore substantial and conditional. Some employees could become microenterprise leaders, broaden their judgment, and share upside. Others left, remained in node work, or faced a system in which remuneration and continued participation depended more directly on unit performance. Removing a supervisor from one decision does not settle who may appeal the metric, contract, ownership rule, or dissolution of a team.²

User value became both a learning signal and a selection environment

Rendanheyi tries to shorten the distance between evidence and authority. Teams can use customer complaints, market response, and partner knowledge to revise a product without waiting for a single corporate product committee. Multiple units can test different responses, which connects the design to innovation, entrepreneurship, and renewal and to strategy, competition, and adaptation.

The same mechanism is also a control system. Financial results and user value make teams comparable; internal competition allocates attention, participants, and resources; weak units may contract or dissolve. Metrics can expose an unwanted service or unsuccessful product. They do not automatically represent job security, care work, bargaining power, repairability, data privacy, resource use, or costs borne after an appliance leaves the customer relationship.

Haier's 2020 forum materials present ecosystem microenterprise communities as self-organizing, decentralized, and able to coevolve with users and partners. That is an authoritative statement of the company's later design vocabulary, not independent evidence that every participant could revise platform rules or that every stakeholder won.⁴

Haier-affiliated author Yunjie Zhou gives the model a related ocean-and-iceberg image: microenterprises form and dissolve as user needs change while Haier coordinates an open resource platform. The two-page letter also claims that enrolled employment fell 32 percent from a stated peak while the wider “ecosystem” created more than 1.6 million jobs. It does not define or independently audit that job measure, sample affected workers, or test its universal claims.⁵ The metaphor conveys fluidity; it can also make the person bearing the cost of a melting unit disappear from view.

Platforms answer a genuine coordination problem. Small enterprises can become small silos. Shared services, interfaces, and ecosystem communities let product teams combine manufacturing, logistics, software, suppliers, and customer knowledge. That places Rendanheyi in cooperation, incentives, and organizational equilibrium: participants must have reasons to contribute capabilities that no one team owns. It also raises a governance question. Control over access, data, metrics, brand, resource allocation, and dispute resolution can create a consequential hierarchy even when few managerial titles remain.

GE Appliances turned transfer into a negotiated adaptation

Haier acquired GE Appliances in 2016. A Harvard Business School case follows the U.S. adaptation through mid-2018. After a year spent building standalone systems, Kevin Nolan became chief executive in June 2017 and joined Melanie Cook and Rick Hasselbeck on a three-person Executive Council. Cooking, Refrigeration, Laundry, and Dishwasher became product microenterprises under a locally named “End to End” model, with their leaders reporting to Nolan and holding profit-and-loss responsibility from design through manufacturing and sales.⁶

The delegated choices were concrete. Cooking leader Paul Surowiec said calls previously escalated through further meetings became the product leader's call. Laundry leader Peter Pepe reported that moving more than $100,000 had required central approval before the acquisition, whereas his later threshold was $5 million. Dishwasher leader Cynthia Fanning's team found an existing Haier platform that could be adapted to U.S. accessibility requirements and shipped 18- and 24-inch models in 2018. The new Zoneline unit placed people from engineering, manufacturing, sales, quality, and service into shared reviews of business results.⁶

The Chinese design did not transfer intact. Haier leaders asked why each product unit lacked its own sales force. GE Appliances leaders argued that major U.S. retailers controlled floor space and that one shared sales platform could represent the full product line more effectively. Sourcing produced another boundary decision: research could remain shared while commodity negotiation sat closer to product units. Haier accepted the sales arrangement after what Pepe described as an active debate. Finance, human resources, legal, technology, supply chain, distribution, and other functions also remained platforms.⁶

Those episodes establish local argument and adaptation, not a completed transfer or a causal performance result. The case says some employees remained skeptical about jobs, compensation, and possible co-investment, and concludes that “zero distance” still required more work. It was reviewed by a company designate; one author had spoken for Haier management; and its own disclaimer says a teaching case is not a source of primary data or proof of effective management. Its rich participant detail is most useful for locating decisions and unresolved questions.⁶

Collective bargaining remained a separate source of worker authority

Production workers at Louisville's Appliance Park did not receive their primary voice through product-unit profit responsibility. In November 2016, before End to End began, 3,608 members of IUE-CWA Local 83761 voted on a proposed four-year contract; WDRB reported that 72 percent rejected it. The proposal used lump sums instead of hourly raises, tightened overtime, and set a $12 starting wage for a new lower-paid tier. The report includes statements from workers, the local president, and the company.⁷

That vote cannot show that Rendanheyi caused the proposal: it followed the acquisition, preceded the 2017 product-unit reorganization, and concerned a labor relationship inherited from GE. It does show that many production workers could accept or refuse employment terms through collective bargaining rather than through an internal market. Product-leader autonomy, team participation, and union ratification were different kinds of authority with different constituencies.⁷⁶

Reported growth does not isolate the management model's effect

GE Appliances later described 2016–2021 as a five-year growth period. Its 2021 economic-impact report claimed 15,000 U.S. employees, 16 percent employment growth, and $2.8 billion in purchases from 5,400 U.S. suppliers. It connected microenterprises and closer customer ties to its 2017 strategy, while also saying the U.S. reshoring effort began in 2010, six years before Haier's acquisition.⁸

The report establishes what the company counted and attributed to its strategy; it does not publish a counterfactual, separate Haier ownership from prior investment and market conditions, or independently validate the claimed multiplier effects. The 2018 teaching case likewise observed that implementation was incomplete. Growth, supplier spending, and product examples are therefore evidence compatible with benefit, not proof that Rendanheyi caused it or that workers, customers, suppliers, communities, and owners shared it equally.⁸⁶

Relations and comparative coding

Kyocera's amoeba management and General Motors under Alfred Sloan are comparative institution relations. Kyocera also divides a large enterprise into accountable small units; GM used profit-responsible divisions under a corporate center. The comparisons expose differences in unit scale, fluidity, internal exchange, and retained central powers. Resemblance is not evidence that either institution directly caused Haier's design.

The two interpretive relations ask different questions. Organizational intelligence asks how customer, operating, financial, partner, and worker knowledge gains standing in decisions. Benefit for all life tests whether “user value” includes people and living systems that are not parties to an internal contract. Neither relation attributes a declared purpose to Haier.⁵

Five idea relations define the scoring. Delegation identifies substantial local operating authority under retained platform powers. Structure, hierarchy, and scale identifies the replacement of much hierarchy with microenterprises and interfaces. Cooperation identifies internal contracting and ecosystem assembly. Strategy identifies experimentation, unit selection, and central reconfiguration. Innovation identifies entrepreneurial product and service search. Purpose receives a 2 because user value legitimates the design without fully specifying affected interests; coordination, decision making, measurement, and culture receive 2s because they shape but do not singly define it. Work, knowledge, and learning receive 1s. Governance, executive sensing, and organizational ignorance receive 0s because the cited history raises those problems without making them the institution's defining formal mechanisms.

The profile follows the same record. Market capital, municipal and public institutional origins, and managerial or technical expertise all supplied authority; “founder-owner” is excluded because Zhang was appointed to an existing municipal factory and the ownership record is opaque. Decisions sat at the central executive, divisional, frontline-local, and peer-distributed levels. Public-corporation and partnership-network forms capture listed corporate entities and variable microenterprise or partner stakes without mistaking formal collective ownership for demonstrated worker governance.²

Markets, teams, metrics, and modular interfaces coordinated the system. Knowledge moved bottom-up, in both directions, across peer networks, and through practice. Financial, operational, and behavioral measures joined market feedback, experiments, and continuous improvement. Local iteration, central reconfiguration, and selection among units were distinct adaptation modes. Customers, workers, shareholders, and suppliers were named beneficiaries, while metric gaming, suppressed voice, leader dependence, siloing, and externalized harm remain evidenced risks rather than claims about every unit or period.²⁶

What the evidence can carry

The peer-reviewed management study supplies the broadest independent account of mechanisms, ownership ambiguity, workforce change, leadership, and limited foreign transfer. Its fieldwork relied on senior executives, microenterprise owner-CEOs, consultants, a headquarters visit, a site tour, and documents—not a representative sample of rank-and-file workers, former employees, customers, suppliers, or communities.²

Haier's history, forum language, and Zhou's letter are strongest as participant accounts of institutional memory and design intent. Time and WDRB add contemporaneous reporting on workforce flux and one bargaining episode. The Harvard case preserves unusually specific GE Appliances decisions but was company-reviewed and stopped while transfer remained incomplete. The GE Appliances report supplies later company counts, not independent causal evaluation.

No cited source measures representative customer outcomes, supplier bargaining power, worker experience across unit types, appeal rights, product privacy, repair and durability, material extraction, appliance energy and water use, refrigerants, disposal, effects on nonhuman life, or intergenerational liabilities. Those absences are why several impact fields remain unclear or mixed rather than being inferred from the word “ecosystem.”

The supported conclusion is narrower. Rendanheyi can place meaningful operating judgment closer to products, users, and partners, and GE Appliances shows that local leaders could negotiate how those powers fit a different market. It does not dissolve ownership, platform governance, labor institutions, or central strategy. A defensible judgment about autonomy must therefore ask not only who may act inside a microenterprise, but who may change its measures, contest its contract, survive its dissolution, and require costs outside the customer transaction to count.