Breaking the Coordination Failure of Industrial Chains:
Digital Ecosystem Coupling Governance and Firm Value Creation

Revision note (4 November 2026): this version keeps the original findings and case evidence, and adds the research questions, falsifiable propositions, construct operationalization, design principles and validity safeguards; citation style is normalized and the original's stronger causal claims are downgraded to what the evidence supports. Case data and quotations are unchanged.

1 Introduction

The competitive paradigm of global industrial chains is shifting fundamentally: from competition among individual firms to systemic competition among industrial chains, supply chains and entire industrial ecosystems. China's policy agenda explicitly calls for "enhancing the resilience and security of industrial and supply chains", and the crux is efficient, agile, credible coordination between upstream and downstream firms. Yet many important Chinese industries are caught in a "digitalization paradox": firms invest heavily in internal systems such as ERP and MES, while data still cannot cross organizational boundaries and coordination remains difficult, producing one data silo after another. This coordination failure shows up as the bullwhip effect and excess inventory from demand information that distorts at each tier; as blame-shifting and trust crises from opaque quality data and weak traceability; and as an inability to warn of and respond jointly to disruption because no one has a global view. Beyond wasted resources, it threatens the safety of the chain itself at critical moments.

Two routes have been proposed. The first is technological integration: point-to-point connection of specific systems through APIs, EDI and similar standards. Such solutions are expensive, scale poorly, and are often "connected but not communicating", because they cannot resolve the underlying misalignment of meaning and commercial rules. The second is platform centralization: a chain leader or a third party builds an industrial internet platform and acts as a centralized coordinator. This route runs into a digital version of the Coasian challenge: the platform owner accumulates both data and rule-making power, which invites data monopoly, unfair distribution and reluctance among smaller firms, and in the end suppresses rather than promotes the ecosystem[1].

These difficulties point to a question that management research has not fully answered: in an industrial network that has no legal authority and consists of multiple heterogeneous firms with divergent interests, can a governance order be built that is widely accepted and sustainable — one that protects the rights of each party, especially data sovereignty and commercial autonomy, while significantly lowering transaction costs and motivating firms to create and share network value? This is, in substance, a re-examination of the Coase theorem and of Williamson's transaction-cost theory for the digital economy[2-3], and an open question for ecosystem strategy and platform governance[4].

The paper therefore poses three research questions.

To answer them, the paper integrates complex adaptive systems theory, transaction-cost economics and digital governance theory into one analytical frame. It proposes information potential difference as the construct that decomposes the micro-foundations of coordination failure. It develops a theory of digital ecosystem coupling governance, whose core is a digital, incentive-compatible rule protocol that reshapes inter-firm relations from a high-cost, low-trust state of loose coupling into a low-cost, high-trust state of adaptive tight coupling. Following design science research, it constructs and instantiates the Digital Ecosystem Coupling Operating System, and offers exploratory evidence from a deep process analysis of two cases: automobile manufacturing and a chemical industrial park.

The contribution is twofold: theoretically, it moves beyond the binary of technological connection and central control to propose a third paradigm, rule-based ecosystem coupling, with falsifiable propositions; practically, it gives Chinese firms, especially chain leaders, a roadmap from governance design to technical implementation. One qualification must be stated at the outset: this is theory building with exploratory cases. Its conclusions aim at analytic generalization, not statistical generalization, and the propositions remain to be tested on larger samples.

2 Literature review and research gaps

2.1 The evolution of industrial coordination: from supply-chain coordination to ecosystem value co-creation

Research on industrial coordination is tightly coupled to the evolution of inter-firm theory. Early work was rooted in supply-chain management: in an explicit supplier–manufacturer chain, contract design, information sharing and vendor-managed inventory coordinated logistics, information and capital flows to minimize total cost and improve response. That paradigm rested on relatively stable hierarchical or bilateral relations, grounded in principal–agent and contract theory.

As the division of labour grew more complex and modular, networked forms became dominant and the focus moved from the "chain" to the "network" and then to the "ecosystem". Ecosystem strategy research argues that value is co-created by interdependent firms through complementary innovation, and that the health of an ecosystem depends on a keystone firm's ability to create shared value by shaping architecture and setting standards[4]. Yet existing work attends mostly to keystone strategy — platform openness, subsidies — and says little about the micro-foundations of governance: how the many niche firms are mobilized and coordinated, and how their contributions are fairly rewarded. Where no clear, widely accepted keystone exists, coordination failure is especially severe.

2.2 Digital platform governance: achievements and inner tensions of centralized coordination

Digital platforms are a revolutionary organizational form for coordinating multilateral markets and lowering transaction costs, and their governance has been a research focus for a decade. The literature examines allocation of control rights over complements, data governance rules, pricing and revenue sharing, and dispute resolution, and shows that successful platform governance requires a balance between control and openness[1].

Transplanting that model into industrial coordination meets a fundamental tension, because an industrial internet platform is usually controlled by one dominant firm. Two problems follow. First, conflict of interest: the platform owner is simultaneously participant and rule-maker, and is naturally suspected of using the rules for private gain, which undermines legitimacy. Second, data sovereignty and privacy: firms, especially core component suppliers, treat production, quality and process data as core assets and resist uploading them to a platform controlled by a competitor or third party. That distrust is the deep reason many industrial platforms are praised but poorly adopted. A more decentralized paradigm, better able to express procedural justice, is needed.

2.3 Sharing data elements: the source of value and the governance dilemma

In the digital economy data is both the "blood" of coordination and a source of value, and promoting its safe, efficient flow within industrial chains is national strategy. Research explores technical and institutional solutions: privacy computing (federated learning, secure multi-party computation) and blockchain aim at "data usable but invisible, controllable and measurable"; data trusts and data cooperatives attempt to build trusted intermediaries between providers and users.

Even so, market-based allocation of data elements remains early-stage, facing an unwillingness, fear and incapacity to share. Studies tend to focus on a single technology or arrangement and lack a top-level governance frame that fuses technological innovation, institutional design and concrete business scenarios. The value of data sharing must be realized in specific collaborative work — joint scheduling, quality traceability — while the rules of that work need trusted data as input. This triangular dependency of business, data and rules demands a systemic solution.

2.4 Integrating the gaps

The literature shows three interrelated gaps in addressing industrial coordination, as Table 1 sets out.

The paper fills these intersecting gaps: governance-rule design at the core, elimination of information potential difference as the critical path, emergence of ecosystem value as the end.

3 Theoretical foundations and constructs

3.1 Information potential difference: transaction costs made concrete in the digital dimension

Transaction-cost theory holds that market exchange is not free: search, negotiation, contracting and monitoring all cost something, and when those costs are too high firms substitute internal hierarchy for market exchange[2-3]. Extending the logic to digital industrial ecosystems, this paper proposes that information potential difference is the core root and concrete expression of high market transaction costs inside industrial networks.

Definition. Information potential difference refers to the systemic friction between industrial organizations — arising from heterogeneous technical standards, ambiguous semantics and conflicting commercial institutions — that obstructs the credible, smooth flow of data elements and the realization of value. It is a multi-level construct whose three dimensions map onto the three cost categories of transaction-cost theory, as Table 2 shows.

Because of this potential difference, firms that recognize the gains from coordination still prefer the status quo for fear of high ex ante and ex post transaction costs, and fall into a collective action dilemma. Eliminating it therefore requires a hybrid institutional and technical scheme that lowers transaction costs inside the network. It should be stressed that this is not merely a technical problem: technological potential difference can be eased with standards and adapters, but semantic and institutional differences involve power, liability and the distribution of gains, and can only be eased through rules negotiated among the parties.

3.2 Digital ecosystem coupling governance: a rule-system design aimed at adaptive tight coupling

To break the coordination failure caused by information potential difference, the paper proposes digital ecosystem coupling governance. This is not a strategy for how a single firm should act, but a governance theory for how an industrial ecosystem as a collective should design the rule system of its interaction. Its core proposition: through digital means, convert the tacit commercial conventions and the dispersed, bilateral contract terms of an industrial network into an explicit, machine-executable, transparent and shared set of digital rule protocols.

The target state moves inter-firm relations from an uncertain, high-friction state of loose coupling to a predictable, low-friction state of adaptive tight coupling. "Adaptive" means rules must allow members to retain strategic flexibility and autonomy, and must evolve with technology, markets and policy. "Tight coupling" does not mean rigid integration; it refers to automated, precise, credible coordination at the key interaction points on which consensus has been reached. Effectiveness shows up as collective efficiency (lower total transaction costs), coordination resilience (capacity to absorb uncertainty) and innovative vitality (emergence of new value).

The theory's boundary conditions must be stated first. In task type, it applies to knowledge-intensive, non-standardized networks with high cross-domain coordination needs, not to highly standardized assembly work. In technology, it depends on data standards, smart contracts and trusted execution environments. In power structure, it requires the chain leader or park authority to be willing to delegate part of rule-making; otherwise the rule system degrades into a new instrument of control. Outside these boundaries, different governance designs are needed.

3.3 Three value-creating mechanisms and their causal chains

Digital ecosystem coupling governance creates value through three interlocking mechanisms, each with an observable causal chain: mechanism design → mediating process → performance outcome.

1. Semantic consensus: building and maintaining industry-level data ontologies, term dictionaries and process standards gives the ecosystem a "common law". Causal chain: shared semantics → less ambiguity and re-confirmation → shorter coordination lead time, less rework.
2. Rule codification: smart contracts encode negotiated commercial rules — delivery windows, quality standards, compensation — as executing programs. Causal chain: codified rules → execution on trigger, auditable trail → lower monitoring and enforcement cost, shorter dispute latency.
3. Intelligent benefit allocation: trusted infrastructure immutably records each party's contribution — capacity, quality, data — and a transparent preset algorithm distributes value. Causal chain: measurable contribution plus transparent allocation → less opportunism, stabler expectations → higher participation and persistence.

The mechanisms are not independent: semantic consensus is the precondition of codification, codification is the precondition of measurable contribution, and the fairness of allocation determines whether the first two can be sustained. This interdependence implies that introducing a single mechanism may yield limited returns, and that only combinations produce systemic effects.

3.4 Propositions

The paper states five falsifiable propositions. The unit of analysis is the industrial network or ecosystem, not the individual firm.

P1–P3 can be tested with cross-sectional or panel data, P4 requires interaction terms or configurational analysis, and P5 requires an operational definition of adaptability (time to launch a new scenario, success rate of exception handling). The two cases below provide preliminary evidence that the mechanisms exist; they do not constitute a statistical test of the propositions.

4 Research design and method

4.1 Methodological choice

To secure both construction and robustness, the paper combines design science research (DSR) with an interpretive multiple-case study: the former builds and evaluates the artifact, the latter explains how the mechanisms work and under what conditions. Together they yield a reusable governance architecture and process-level evidence of the mechanisms[5].

4.2 Design science: process and design principles

The core artifact is the Digital Ecosystem Coupling Operating System, an IT artifact intended to eliminate information potential difference and realize coupling governance. The study follows the classic DSR process: (1) problem identification — define coordination failure and its root; (2) solution design — build the theory and the four-layer architecture; (3) demonstration — show feasibility through two cases; (4) evaluation — gather qualitative and quantitative evidence of utility[5].

Five design principles, derived from the theory and the case settings, guide the architecture and its evaluation:

• DP1 Standardized access: provide connectivity as an industry-level public good, so that connection cost falls from per-pair custom development to one integration reused many times (environment interface layer).
• DP2 Semantics first: establish data ontologies and term standards before connecting systems; do not force integration while meaning is inconsistent (connection engine).
• DP3 Explicit, codifiable rules: write tacit conventions and bilateral clauses into rules that all parties can see, review and version, then select the automatable subset (rule coordination engine).
• DP4 Measurable contribution: every act of participation leaves a verifiable record, which becomes the basis of allocation (value allocation center and trusted records).
• DP5 Separation of interface and governance: technical interfaces stay neutral while governance rules are revised by members, so that a platform owner cannot convert interface capability into governance power (whole architecture and governance committee).

4.3 Case selection and data collection

Cases were selected by theoretical sampling to cover different power structures and reproduce the theoretical logic, not to represent a population. The two cases form a contrast: one has a clear chain leader and a hierarchical structure; the other has no equity ties and is geographically clustered.

• Case A (coordinated manufacturing in an automobile chain): a large domestic automobile group and its network of more than one hundred multi-tier suppliers — a hierarchical chain ecosystem led by a chain leader.
• Case B (integrated governance of a chemical park): a national-level chemical park in East China with dozens of independent firms — a cluster without equity ties, where coordination faces the harder challenge of no authority.

Data were collected through triangulation across semi-structured interviews, archival documents, direct observation and operational data (Table 3). A common interview protocol covered coordination pain points, motives, the rule design process and participation experience; archives were used to trace how rules formed and evolved; operational data supplied quantitative evidence.

4.4 Analysis and validity safeguards

Analysis combined pattern matching with time-series analysis. Themes emerging from the cases — disputes over rules, concerns about data access, manifestations of efficiency gains — were repeatedly compared with and used to revise the constructs and mechanisms, producing a mechanism–indicator–evidence mapping. The sequence from problem diagnosis through rule design and implementation to operational optimization was then reconstructed chronologically, producing an antecedent–intervention–outcome evidence chain.

Four safeguards were applied.

• Triangulation: each conclusion is supported, where possible, by interviews, archives and operational data together; conflicts are flagged and traced back in the analysis.
• Rival explanations: for every observed effect, at least one alternative explanation is stated (managerial tightening, scale effects, industry conditions), with a reason why the mechanism account is more persuasive.
• Chain of evidence: the path from raw material to coding to conclusion is retained, and key conclusions can be traced to specific documents or interview segments.
• Process rather than cross-section: the order of rule design, implementation and effect is presented as a time series to reduce the risk of reverse causality.

(Note: for space and confidentiality reasons the coding manual and interview transcripts are not reproduced here; anonymized material is available on reasonable request.)

4.5 Research ethics and data note

The study involves internal firm data and personal interviews. All interviews were conducted with informed consent; informants are identified by role, not name; operational data were anonymized before analysis, retaining relative changes and orders of magnitude only; sensitive indicators of identifiable firms are not disclosed. Quotations in the case narratives were confirmed by informants as usable for research publication.

5 Operationalizing the theory: system architecture

To turn the governance theory into practice that can be deployed and reused, the paper designs the Digital Ecosystem Coupling Operating System. The architecture is not only a technical blueprint but the digital environment in which governance rules are encoded, deployed and run. Its four layers serve one objective — lowering transaction costs and enabling value creation — and correspond one-to-one with the design principles in section 4.2.

5.1 Environment interface layer: an interconnected digital infrastructure (DP1)

This layer systematically lowers the connection cost produced by technological potential difference. Its design philosophy is to supply an industry-level public good: standardized, modular access tools and protocols. Core components include adapters for more than 300 industrial protocols, standard connectors for mainstream commercial software, and intelligent sensing for non-standard systems. It is, in effect, a standard-gauge digital railway and a universal socket for the ecosystem, so that firms can connect at very low marginal cost. One supplier's IT manager said: "Connecting to a new customer used to take two months of interface development. Now it is like plugging in a USB stick; our implementation cost has fallen by more than 70%." That figure is a single-party estimate and was not independently verified; it should be read as directional evidence.

5.2 Connection and governance layer: a digital legislature and judiciary (DP2, DP3)

This is the core of governance and the carrier of the three value-creating mechanisms.

1. The connection engine embodies semantic consensus. It embeds an extensible industry ontology library and performs semantic alignment and conversion as data flows: a "work order number" at firm A and a "production order number" at firm B map to a unified "production task identifier", securing a common language.
2. The rule coordination engine is the heart of codification. Business rules agreed by members become deployable, monitorable smart contracts. In the automobile case, clauses of the supplier quality agreement — "if the batch defect rate exceeds 0.5%, enhanced inspection is triggered" — are codified, and inspection data firing the condition trigger the contract, alerting the parties without human judgement. In the park case, the cross-firm emergency linkage plan is encoded as an event-driven response contract.
3. The intelligent scheduling center is a public brain that optimizes decisions on consensus rules and global data. Using semantically consistent data, it runs demand forecasting, capacity balancing and risk-propagation simulation, offering scheduling advice and risk warnings as public decision support.

5.3 Enablement layer: governance public goods and services (DP4)

This layer packages collective capability into services that member firms can call directly, increasing the system's attractiveness and legitimacy.

• Industry cockpit: tailored panoramic views for managers in different roles, reducing search and monitoring costs.
• Value allocation center: the practical expression of intelligent benefit allocation. Immutable performance data recorded by the rule engine drive automatic computation of each party's contribution and settlement under a preset revenue-sharing model; in supply-chain finance, trusted order and logistics data support automatic receivables financing.
• Compliance automation: environmental, energy and safety requirements are built in as rules, so data are collected, verified and reported automatically.

5.4 Application and ecosystem layer: a proving ground for innovation (DP5)

On top of shared rules, trusted data and public capabilities, participants — third-party developers included — can assemble vertical applications like Lego bricks: collaborative R&D platforms, equipment-sharing markets, carbon-footprint tracking. Stable, fair governance at the base eases the trust and sharing problems that block innovation, so applications multiply. The separation of interface and governance (DP5) is the precondition for this openness: if interface capability were monopolized by one party, the ecosystem layer would degrade into an accessory app market.

6 Case analysis and findings

The narratives below draw on interviews, archives and operational data. Roles are identified by position; firm names are withheld.

6.1 Case A: the automobile chain — from chain-leader control to rule-based co-governance

6.1.1 Background and challenge

The group has annual capacity above one million vehicles and more than 500 direct and indirect suppliers. Its traditional model of monthly forecasts plus weekly orders, with lagging and distorted information, kept inventory high, and quality disputes with suppliers were constant because traceability was poor.

6.1.2 Designing the governance rules

The project did not begin with technical connection. The OEM led 15 core suppliers in a working group on data standards and coordination rules; over three months they produced the Automotive Parts Coordination Data Standard V1.0 and a Basic Agreement on Electronic Supply-Chain Coordination. The agreement defined 22 codifiable business rules covering order confirmation, shipment notice and quality objection handling. A supplier vice-president recalled: "The arguments were fierce, especially over liability for quality problems and deduction rules. But because we argued it through together, everyone accepted the rules once they were written as code." The detail supports the argument in 3.2 that rules must be agreed: acceptability comes from participation, not from authority.

6.1.3 Implementation and effects

Codification produced results quickly. For goods receipt inspection, the rule requires the supplier to send an electronic waybill and quality self-inspection data with each shipment; on arrival, scanning the receipt triggers automatic comparison; matching data trigger scheduled put-away, mismatches freeze the goods and notify both quality teams. Average receipt time fell from 4 hours to 0.5 hours. Over six months of piloting, average inventory turnover for the three tier-one suppliers fell from 52 days to 34 (34.6%), and quality disputes fell by 65%. The OEM's supply-chain director said: "It is not me managing them any more; it is the rules managing all of us. Efficiency is up and conflict is down."

These changes point in the direction predicted by P1, P2 and P3. But they cover only three pilot suppliers and six months, during a period of stable industry demand, so they are directional evidence at best; ruling out managerial tightening and scale effects requires longer observation and a comparison group.

6.2 Case B: the chemical park — from administrative instruction to algorithmic contingency plans

6.2.1 Background and challenge

The park is densely built with many major hazard sources. Traditional safety management relied on self-inspection and committee supervision, and emergency response ran on phones, walkie-talkies and experience — slow and risky.

6.2.2 Designing the governance rules

The committee's core task was a digital charter for coordinated emergency response. The difficulty was converting the actions in each firm's plan into cross-firm digital instruction flows. After several rounds of tabletop exercises, the event "leak monitoring exceeds threshold" was encoded as a smart contract triggering a complex workflow: it alerts the leaking firm, firms within 1,000 m downwind, the park fire service and the medical center with wind data and plan instructions, and plots rescue routes from live traffic data while controlling smart barriers.

6.2.3 Implementation and effects

After go-live the system passed a real test: a sensor in a tank farm detected a minor leak, and within 3 seconds the system raised the alarm, triggered the plan and dispatched instructions. The park emergency center director compared: "It used to take at least 15 minutes from the alarm to phoning the firms, the fire service and the environmental bureau, and then each of them starting their own plan. Now, when the alarm sounds, every responsible person already has an action card with specific instructions." Over the first year, average response time for level-3-and-above warnings fell from 128 minutes to 52 — nearly 60%. A safety director at a foreign-invested chemical firm said: "This system makes us feel we are not fighting alone. The rules are clear, the response is automatic, and that is more reassuring than any administrative order."

The case supports P2 relatively strongly: the reduction occurred mainly in the "notify and start" step, which is precisely the part replaced by code. It supports P1 and P3 weakly — the park was authority-led, and the degree of consensus and the allocation mechanism are both weaker than in Case A.

6.3 Cross-case comparison

The cases contrast in power structure: A has a clear chain leader and rules negotiated between the leader and core suppliers; B has no equity ties and rules pushed by the committee on the basis of shared safety interests. The contrast suggests a pattern: the more the legitimacy of governance rules depends on negotiation, the more complete the semantic consensus and allocation mechanisms; the more it depends on administrative push, the more readily codification runs ahead of allocation. This helps explain why Case B shows a strong efficiency effect while its participation persistence remains to be observed.

In the automobile case, once connected suppliers passed a threshold, live trusted order and capacity data gave rise to new services — a capacity-sharing marketplace, order splitting and subcontracting. The observation is consistent with the predicted coordination emergence, but longer observation is needed to rule out short-term reallocation of resources.

7 Discussion

7.1 Findings

Three findings stand out. First, information potential difference is a useful construct for understanding coordination failure: it unifies technological, semantic and institutional obstacles under one transaction-cost frame, and explains more than single-factor accounts such as "systems do not talk" or "there is no platform". Second, in a network with no single authority, a negotiated rule system can partially substitute for hierarchical authority as the basis of coordination; the acceptability and enforceability of rules depend on whether they were co-authored, whether they were codified, and whether allocation is transparent. Third, the architecture works through a combination of mechanisms: semantics, rules and allocation presuppose one another, and single-point improvements yield limited returns.

It bears repeating that these findings rest on two exploratory cases and aim at analytic generalization: they show under what conditions, through what mechanisms, governance rules may lower transaction costs — not that they necessarily do so in general. Statistical testing of the propositions remains future work.

7.2 Theoretical contributions

1. Deepening and operationalizing ecosystem strategy. Existing work stresses architecture and value sharing but is thin on how they are achieved[4]. This paper makes rule-system design the mechanism of architectural governance and specifies three operational, testable mechanisms, connecting macro ecosystem strategy with micro inter-firm behaviour.
2. Extending platform governance. Faced with the governance dilemma of centralized industrial platforms, consensus-based distributed governance protocols offer an alternative: delegating part of governance — especially enforcement and monitoring — to encoded objective rules eases the platform-owner paradox[1] and opens space for decentralized autonomous organization in complex industrial settings.
3. A contemporary reading of transaction-cost theory. Applying the theory to digital industrial ecosystems, the paper proposes information potential difference as a multidimensional expression of digital transaction costs, and shows through cases how smart contracts and trusted records reduce coordination, monitoring and enforcement costs by changing information structures and automating execution[2-3]. This both confirms the theory's continuing power and marks its boundary: where rules themselves must be repeatedly negotiated, governance cost returns in another form.

7.3 Managerial implications

• For chain leaders and ecosystem orchestrators: reframe the role from resource controller to rule advocate and provider of public goods. Success lies less in how much data you control than in whether you can design fair, transparent, incentive-compatible rules and invest in infrastructure that lowers transaction costs for the whole ecosystem.
• For small and medium enterprises: participate in negotiating the rules. In the digital age, aligning with rules secures long-term rights more reliably than aligning with a leading firm, and participation is the most effective defence against rules being used by one party alone.
• For policymakers: shift emphasis from subsidizing individual platforms toward supporting open, public-interest industry-level governance protocols and reference architectures, and toward data standards and coordination rules in key sectors — a digital commons for market-driven ecosystem innovation.

7.4 Limitations

Four limitations apply. First, the number of cases is small and both are successful implementations, which introduces selection bias: failed or abandoned attempts are excluded, so effectiveness may be overstated. Second, effect data are mostly before-and-after comparisons without a control group, so industry cycles and managerial tightening cannot be fully excluded. Third, the paper concentrates on building and validating the framework and does not examine how rules evolve dynamically or how rule-making is contested under different power structures. Fourth, information potential difference is still a qualitative construct without a cross-industry measurement scheme.

7.5 Future research

Four paths follow: (1) large-sample empirical work, developing scales for information potential difference and coupling governance and testing their relation to chain performance; (2) quasi-experimental designs using comparable chains without a governance platform as controls, to identify causal effects; (3) computational experiments simulating ecosystem evolution under different rule designs, particularly the inverted-U predicted by P5; (4) process studies tracing rule revisions to see who gains or loses rule-making power, and under what conditions.

References

1. Tiwana, A. Platform Ecosystems: Aligning Architecture, Governance, and Strategy[M]. Morgan Kaufmann, 2014.
2. Coase, R. H. The nature of the firm[J]. Economica, 1937, 4(16): 386–405.
3. Williamson, O. E. The Economic Institutions of Capitalism[M]. New York: Free Press, 1985.
4. Adner, R. Ecosystem as structure: An actionable construct for strategy[J]. Journal of Management, 2017, 43(1): 39–58.
5. Hevner, A. R., March, S. T., Park, J., & Ram, S. Design science in information systems research[J]. MIS Quarterly, 2004, 28(1): 75–105.