Jawn Lam

Network, modular, and virtual organization

Organizational StructuresNetwork and market-like forms13 min read

A firm keeps a few core activities and coordinates the rest through contracts and relationships with other firms that perform them.

What it is

A network organization keeps the few activities it does best and has other firms perform the rest. Often a hub or broker firm decides over the whole and partners over their own links of the value chain; some variants have no hub.

The network keeps a core and contracts out the rest of the value chain

Select a unit or a letter to read it; the shaded field is the firm, and the partners stand outside it.

Contract or priced tradeAdjust to each other directlyMMutual adjustmentOStandardization of outputsThe firm; partners stand outside it

The firmPartner firmsHub or brokerfirmActivity itdoes bestDesign firmManufacturerDistributorSplit by value-chain stepOM
The firmPartner firmsBrokerCoreDesignMakerDistributeSplit by value-chain stepOM
Select a unit or a letter to read it. M, mutual adjustment: People coordinate their own work by informal communication with each other. O, standardization of outputs: The results of the work are specified, such as targets or product dimensions, and the method is left to the unit.Hub or broker firm. Layer: The firm. No single head above it. Decides over the whole and brings the partners together (Miles and Snow 1986); some variants have no hub.Activity it does best. Layer: The firm. No single head above it. The few activities the firm keeps; the others are performed by partner firms.Design firm. Layer: Partner firms. Outside the firm. Split by value-chain step. One link of the value chain, run by its own firm, which decides over its own link.Manufacturer. Layer: Partner firms. Outside the firm. Split by value-chain step. The manufacturing link of the value chain, held by contract.Distributor. Layer: Partner firms. Outside the firm. Split by value-chain step. The distribution link, also held by contract.O, standardization of outputs. The results of the work are specified, such as targets or product dimensions, and the method is left to the unit.M, mutual adjustment. People coordinate their own work by informal communication with each other.
The structure in words
  1. The firm
    • Hub or broker firm, no single head above it. Decides over the whole and brings the partners together (Miles and Snow 1986); some variants have no hub.
    • Activity it does best, no single head above it. The few activities the firm keeps; the others are performed by partner firms.
  2. Partner firms
    • Design firm, outside the firm, grouped split by value-chain step. One link of the value chain, run by its own firm, which decides over its own link.
    • Manufacturer, outside the firm, grouped split by value-chain step. The manufacturing link of the value chain, held by contract.
    • Distributor, outside the firm, grouped split by value-chain step. The distribution link, also held by contract.
  • Hub or broker firm and Design firm: contract or priced trade.
  • Hub or broker firm and Manufacturer: contract or priced trade.
  • Activity it does best and Distributor: contract or priced trade.
  • Design firm and Manufacturer: adjust to each other directly, coordinated by mutual adjustment.
  • Split by value-chain step: coordinated by standardization of outputs.

Different firms do the design, manufacture, and distribution, and a hub or broker brings them together. Contracts and standardized interfaces specify what each partner delivers, and the partners adjust to one another directly.

Source: Miles and Snow 1986; Sanchez and Mahoney 1996; Powell 1990.

Five profile levels rest on theory. Grouping is by function or process, the value chain disaggregated among firms, and centralization is decentralized, market substitutes replacing administrative mechanisms (Miles and Snow 1986). Coordination is by mutual adjustment and standardization of outputs: interfaces embed coordination (Sanchez and Mahoney 1996), and trust does the rest (Powell 1990). The boundary is permeable, the defining move. The mode of exchange is market for Miles and Snow and contested: Powell (1990) calls networks distinct from markets and hierarchies, so clan is nearest, our judgment.

The other six are our judgment: specialization high; formalization medium; layers flat; decision rights spanning limited vertical decentralization to selective vertical and horizontal decentralization, though Mintzberg's (1980) synthesis drew these types for one firm; lateral links by direct contact and liaison roles; and moderate differentiation with some integrating roles.

Where the network, modular, and virtual organization sits on the eleven dimensions

The filled segments are the form's levels; select a row for its reason and sources.

Select a dimension to read why the form sits there, and on what basis.Basis of grouping: Function or process. Basis: Theory. Vertical disaggregation: the functions of the value chain, such as design, manufacture, and distribution, are performed by different firms brought together by a broker. Sources: Miles and Snow 1986.Specialization: High. Basis: Judgment. Each firm keeps the few activities it does best and contracts for the rest; judgment for the band. Sources: Miles and Snow 1986; Jarillo 1988.Standardization and formalization: Medium. Basis: Judgment. Contracts and standardized interfaces between components embed much of the coordination, while trust and adjustment carry the rest; judgment for the band. Sources: Sanchez and Mahoney 1996; Powell 1990.Centralization: Decentralized. Basis: Theory. Vertical disaggregation into independent firms joined by brokers and market substitutes for administrative mechanisms leaves each member firm to decide for itself. Sources: Miles and Snow 1986.Layers and span of control: Flat. Basis: Judgment. The hub keeps a few core activities and few layers; judgment. Sources: Miles and Snow 1986.Decision rights and the locus of authority: Limited vertical decentralization to selective vertical and horizontal decentralization. Basis: Judgment. Partner firms hold the rights over their own activities while the broker holds the rights over the whole; Mintzberg's types are drawn for one firm, so this is judgment. Sources: Miles and Snow 1986; Mintzberg 1980.Prime coordinating mechanism: Mutual adjustment and standardization of outputs. Basis: Theory. Standardized component interfaces embed coordination, a specification of outputs, plus trust and mutual adjustment among partners. Sources: Sanchez and Mahoney 1996; Powell 1990.Lateral mechanisms: Direct contact and liaison roles. Basis: Judgment. Brokers and relationship managers link the firms directly; judgment. Sources: Miles and Snow 1986; Jarillo 1988.Differentiation against integration: Moderate differentiation, some integrating roles. Basis: Judgment. Firms with different specialties are joined by brokers and contracts; judgment. Sources: Miles and Snow 1986; Lawrence and Lorsch 1967a.Mode of exchange and control: Market. Basis: Theory. Miles and Snow put market substitutes in place of administrative mechanisms. Sources: Miles and Snow 1986. View 2: Clan. Networks are a form distinct from markets and hierarchies, coordinated by reciprocity and relationship; clan is the nearest of the three levels, a reading that is judgment. Sources: Powell 1990.Boundary permeability: Permeable. Basis: Theory. Activities are moved across the firm's own boundary to other firms, which is the form's defining move. Sources: Miles and Snow 1986; Powell 1990.

Five levels rest on theory: grouping by function or process, decentralized authority, coordination by mutual adjustment and standardized outputs, permeable boundaries, and the market as the mode of exchange. The other six are judgment. The mode of exchange is contested: Powell (1990) sets networks apart from markets and hierarchies, so on his view clan is nearest, our judgment.

Source: Miles and Snow 1986; Jarillo 1988; Sanchez and Mahoney 1996; Powell 1990; Mintzberg 1980, a synthesis of earlier research, offering propositions without a new test; Lawrence and Lorsch 1967a, a multi-firm comparative field study.

Where it came from

Raymond Miles and Charles Snow (1986) described the dynamic network, built on vertical disaggregation, brokering, and market substitutes, as a forecast, not a test. Hans Thorelli (1986, conceptual) set the interorganizational network between the open market and internalizing an activity. J. Carlos Jarillo (1988, conceptual) called strategic networks a mode of organization that can yield competitive strength. Walter Powell (1990) argued that network forms, typified by reciprocal communication and exchange, are neither market nor hierarchy.

The schools of organization design and the forms' origins, 1900 to now

Each row is a school's era; each square under them is a form at its first year. Select a row or a square to read it; on a form's page, its own square and span are marked, with the schools that treat it.

1900192019401960198020002020Bureaucracy and its criticsClassical administrationHuman relations and System 4Transaction cost economicsCarnegie schoolSocio-technical systemsLabor-managed firmModularity and mirroringStrategy and structureContingency theoryFrench bureaucratic schoolJapanese-firm literatureScandinavian and DutchConfiguration theoryInformation processingLabor processStrategic choiceGerman codeterminationOrganizational economicsInstitutional theoryPopulation ecologyResource dependenceBusiness groupsScandinavian institutionalSocietal effectNetwork perspectiveComputational designNew-forms debateAlgorithmic managementForms by first yearNetwork, modular, and virtual organization, 1986 to 1990Now

The forms by first year

1900192019401960198020002020Network, modular, and virtual organization, 1986 to1990Now

Schools that treat this form: Transaction cost economics and the M-form hypothesis; Modularity and the mirroring hypothesis; Organizational economics: decision rights, authority, and complementarities; Resource dependence; the business-groups literature; Scandinavian institutionalism; the network perspective on organizations; the new-forms debate; and the algorithmic-management literature.

Select a school's row or a form's square to read it.Select a form's square to read it; open a school below.Bureaucracy and its critics1900s to 1950s, revived from the 1990sWhat it claimsWeber held that rational-legal authority exercised through an office hierarchy, written rules, trained officials, and promotion by qualification is technically superior in precision, speed, and reliability. Critics studied its unintended consequences, Michels's iron law sharpest.What it has shownMerton (1940) argued, with no test, that rules can turn from means into ends; case studies followed. The Aston group concluded that the type's dimensions vary independently, and Wagner's meta-analysis (2021) found that many of their interrelationships have weakened.Who disputes itHuman relations and System 4 sets the costs of impersonal rules to motivation and cooperation against the technical superiority Weber stresses.Bureaucracy and its critics in The Schools of Organization DesignClassical administration1910s to 1940s, and in German to the 1970sWhat it claimsThere are general principles of administration, among them division of work, unity of command, a scalar chain of authority, and a limited span of control, and a structure built to them will be efficient in most settings. Taylor added that planners should set work methods.What it has shownThe school offered reasoning and experience, not tests. Woodward (1965) found that the spans and levels of the more successful firms in her sample differed by production system, and Simon (1946) showed that the principles come in pairs that point in opposite directions.Who disputes itStructural contingency theory found that the structure of successful firms varies with technology and environment, where classical writers treat their principles as general.Classical administration in The Schools of Organization DesignHuman relations and System 41920s to 1960sWhat it claimsOutput depends on the social relations of work, the norms of informal groups, and the style of supervision as much as on formal design. Likert's System 4, overlapping work groups joined by linking-pin managers under participative management, outperforms authoritative systems.What it has shownThe founding evidence is the Hawthorne studies, which later reanalyses dispute (Carey 1967; Levitt and List 2011). Likert's support came largely from his own institute's surveys. Wagner (1994) found participation's effects on performance and satisfaction significant but small.Who disputes itLabor process and critical theory reads human relations as a technique of managerial control rather than a change in the distribution of power.Human relations and System 4 in The Schools of Organization DesignTransaction cost economics and the M-form hypothesis1937, then the 1970s to the presentWhat it claimsFirms exist, and are organized as they are, to economize on transaction costs. Hierarchy replaces the market when assets are specific, uncertainty is high, and exchange is frequent, and the multidivisional form should outperform the functional form in large, diversified firms.What it has shownArmour and Teece (1978) found, in United States petroleum firms from 1955 to 1973, a positive relationship between the multidivisional form and profitability while the form was spreading, but not once an equilibrium in organizational form was reached.Who disputes itThe network perspective on organizations: Granovetter argued that the theory undersocializes economic action, which is embedded in networks of personal relations that also produce trust and order.Transaction cost economics and the M-form hypothesis in The Schools of Organization DesignThe Carnegie school and the attention-based view1940s to the presentWhat it claimsPeople in organizations are boundedly rational, so structure matters because it sets the premises of their decisions and channels their attention: who sees which problems, which goals they pursue, and which routines they follow. Ocasio makes structure distribute attention.What it has shownMarch and Simon (1958) drew propositions from the existing literature, and Cyert and March (1963) added process models and simulations. Two studies show hierarchy shaping what is sent up as well as what is chosen (Reitzig and Maciejovsky 2015; Keum and See 2017).Who disputes itOrganizational economics: decision rights, authority, and complementarities models a principal and agents with defined objectives; the Carnegie school, firms as coalitions with unresolved goal conflict.The Carnegie school and the attention-based view in The Schools of Organization DesignSocio-technical systems1950s to 1980sWhat it claimsAn organization is a social system and a technical system that must be designed together. The same technology admits more than one way of organizing work, and groups that own a whole task and control its variances at the source perform better and are more satisfied.What it has shownTrist and Bamforth (1951) found that longwall mechanization broke up self-regulating work groups. A review of 134 experiments found them extremely successful overall, and Stewart (2006) linked team autonomy to performance; studies of socialist brigades and Volkswagen qualify it.Who disputes itLabor process and critical theory: Kelly argued that labor intensification, not joint optimization, accounts for what socio-technical redesign achieved, and that the part of pay incentives was seriously underestimated.Socio-technical systems in The Schools of Organization DesignThe economics of the labor-managed firm and of enterprise ownership1958 to the presentWhat it claimsA firm run by its workers behaves differently from one run for investors: Ward modeled the Yugoslav-style firm as maximizing income per worker. Hansmann argued that an ownership form's success depends on the balance between the costs of market contracting and of ownership.What it has shownBurdín (2014) found a 29 percent lower hazard of dissolution for Uruguayan worker-managed firms than for conventional ones, and Fakhfakh, Pérotin, and Gago (2012) found French labor-managed firms as productive or more so in every industry studied.Who disputes itOrganizational economics (Jensen and Meckling) contrasts the labor-managed firm, in which tradable residual claims are legally prohibited, with cooperatives and partnerships that arise out of voluntary contracting.The economics of the labor-managed firm and of enterprise ownership in The Schools of Organization DesignModularity and the mirroring hypothesis1960s to the presentWhat it claimsComplex systems are built of nearly decomposable parts, and an organization's structure tends to mirror the architecture of what it makes: modular products allow loosely coupled organizations, and tightly coupled products need tightly coupled teams. Conway's law runs in reverse.What it has shownColfer and Baldwin (2016) review the empirical studies and report both support and principled exceptions. Direct tests point both ways: MacCormack, Baldwin, and Rusnak (2012) found open source products more modular than their commercial pairs; dynamic components break the mirror.Who disputes itNo school in the record disputes it head on; it answers information processing, which meets the coordination load inside the organization; mirroring writers locate it in the product's architecture.Modularity and the mirroring hypothesis in The Schools of Organization DesignStrategy and structure1960s to 1980sWhat it claimsStructure follows strategy: firms that diversify by product or geography outgrow the centralized functional structure and, after a lag and usually a crisis, adopt divisions. Chandler drew this from the histories of DuPont, General Motors, Standard Oil of New Jersey, and Sears.What it has shownChandler (1962) is comparative business history of four firms set against a wider survey. Rumelt (1974) related diversification and structure to performance, and Egelhoff (1988), with 34 multinationals, found some of Stopford and Wells's relationships supported and others not.Who disputes itInstitutional theory: in Fligstein's model divisions spread with a product-related or unrelated strategy, presidents from sales or finance, and other firms in the industry changing; strategy was one of three.Strategy and structure in The Schools of Organization DesignStructural contingency theory1960s to the presentWhat it claimsThere is no one best structure. Performance depends on fit between structure and contingencies such as environment, technology, size, and strategy, on an open-system view of the organization. Drazin and Van de Ven separate fit as selection, interaction, and systems.What it has shownBurns and Stalker, Woodward, and the Aston group related structure to context; Drazin and Van de Ven (1985) supported selection and systems fit only. Pennings (1975) found that fit did not explain effectiveness; Schoonhoven (1981) found traditional hypotheses unsupported.Who disputes itStrategic choice: Child argued that contingency theory explains structure by the constraints of context and leaves out the political process by which power-holders choose strategic action, structural forms included.Structural contingency theory in The Schools of Organization DesignThe French bureaucratic school (strategic analysis)1960s to 1980sWhat it claimsA bureaucracy is a system of power relations organized around zones of uncertainty. Impersonal rules and centralization protect members from depending on one another, those who control uncertainties hold power, and, unable to correct its errors, it changes only through crisis.What it has shownLe phénomène bureaucratique (1963; English 1964) rests on two case studies, a clerical agency and a state industrial monopoly. L'acteur et le système (1977; English 1980) is theoretical.Who disputes itSocietal effect and comparative business systems: D'Iribarne argues that national logics of conduct, such as a French logic of honor, explain differences that Crozier attributed to power games.The French bureaucratic school (strategic analysis) in The Schools of Organization DesignThe Japanese-firm literature1960s to the presentWhat it claimsThe large Japanese firm coordinates operating decisions through horizontal information sharing among semi-autonomous shops and uses its ranking hierarchy mainly as an incentive scheme (Aoki). It creates knowledge through middle-up-down management (Nonaka and Takeuchi).What it has shownAoki (1986) is a formal model. Lincoln, Hanada, and McBride (1986) compared 51 Japanese and 55 United States manufacturing plants, and Nonaka and Takeuchi (1995) rest on case studies, chiefly of Japanese firms. Fruin (1994) found the enterprise system defined at three levels.Who disputes itLabor process and critical theory reads shop-floor delegation as control: Sewell and Wilkinson argue that just-in-time and total quality regimes delegate tactical responsibility while centralizing strategic control.The Japanese-firm literature in The Schools of Organization DesignThe Scandinavian and Dutch socio-technical schools1960s to 2000sWhat it claimsThe Scandinavian program sought democracy at work through field experiments with semi-autonomous groups, then through democratic dialogue. The Dutch school's design theory turns complex organizations with simple jobs into simple organizations with complex jobs.What it has shownEmery and Thorsrud reported field experiments in Norwegian industry; Gustavsen (1992) records that they spread less than hoped, his reason for turning to dialogue. Berggren studied work organization in the Swedish auto industry. The Dutch work rests on design cases and argument.Who disputes itThe Japanese-firm literature: defenders of the Toyota system hold that standardized work with employee participation learns faster; Berggren reports Swedish long-cycle plants matching a line plant on productivity.The Scandinavian and Dutch socio-technical schools in The Schools of Organization DesignConfiguration theory1970s to the presentWhat it claimsStructures cluster into a small number of internally coherent configurations whose elements reinforce one another, so organizations move between a few types in large jumps, not small steps. Mintzberg's five configurations each rest on one coordinating mechanism.What it has shownDoty, Glick, and Huber (1993) found that fit to Miles and Snow's ideal types predicted effectiveness while fit to Mintzberg's did not. A meta-analysis of 40 tests (Ketchen and colleagues 1997) found that configuration partly explains performance, more so with broad definitions.Who disputes itStructural contingency theory: Donaldson counters that most organizations sit at intermediate points on a continuum of fits, not in a few configurations, and that studies over time find mostly incremental change.Configuration theory in The Schools of Organization DesignInformation processing1970s to the presentWhat it claimsThe more uncertain the task, the more information must be processed during its execution. An organization either reduces that need, through slack resources or self-contained tasks, or raises its capacity, through vertical information systems or lateral relations up to the matrix.What it has shownGalbraith (1973, 1974) built the framework from cases and from Simon's theory of bounded rationality, and Tushman and Nadler (1978) set it out as an integrating concept for design. Daft and Lengel (1986) argued that structure must reduce equivocality as well as supply data.Who disputes itThe Carnegie school and the attention-based view: Ocasio holds that firm behavior results from how the firm channels and distributes attention, so structure governs where attention goes, not only how much is processed.Information processing in The Schools of Organization DesignLabor process and critical theory1970s to the presentWhat it claimsStructure is above all a system of control over labor. Braverman argued that management deskills work by separating conception from execution; Edwards traced a sequence from simple control by the boss to technical control through machines to bureaucratic control through rules.What it has shownHistorical and theoretical (Braverman 1974; Edwards 1979). Delbridge (1995), observing a Japanese-owned plant in England, found that just-in-time and total quality regimes combined the control of labor with management's economic goals more completely than piece rates had.Who disputes itBureaucracy and its critics: Adler argues that labor process theory ignores the progressive socialization of the labor process, which drives long-term skill upgrading, and disputes deskilling as the trend.Labor process and critical theory in The Schools of Organization DesignStrategic choice1970s to the presentWhat it claimsStructure is the outcome of choices by those with power in the organization, who decide on strategic action, including the structural form, the manipulation of environmental features, and the performance standards to be met. Context constrains but does not determine structure.What it has shownLargely conceptual: Child (1972) built the argument from gaps in contingency findings, and the empirical case rests mostly on the modest share of structural variance contingencies explain. A 2026 review of 42 studies finds upper echelons theory their main theoretical anchor.Who disputes itStructural contingency theory: Donaldson replies that organizations mostly adopt the structures their contingencies dictate, to avoid the loss of performance misfit brings, so choice over structure is at most limited.Strategic choice in The Schools of Organization DesignThe German codetermination literature1970s to the presentWhat it claimsWorks councils and employee board seats form a structural layer through which coordination can become negotiation between management and employees. Defenders argue that it supports long-term employment, training, and cooperation; critics, that it weakens shareholder control.What it has shownJäger, Schoefer, and Heining (2021) found board-level codetermination left wages unchanged and, if anything, raised capital formation. Jirjahn (2011) reports studies pointing to positive works council effects; a meta-analysis found codetermination laws tied to lower productivity.Who disputes itOrganizational economics (Gorton and Schmid) found that firms with equal employee and shareholder representation did worse for their shareholders than firms with one-third employee representation.The German codetermination literature in The Schools of Organization DesignOrganizational economics: decision rights, authority, and complementarities1976 to the presentWhat it claimsDesign is the joint choice of who holds which decision rights, how performance is measured, and how it is rewarded. Rights should be placed where the specific knowledge lies, formal authority differs from real authority, and practices are complements that pay off in clusters.What it has shownIchniowski, Shaw, and Prennushi (1997) found, on 36 steel finishing lines, that lines using a set of innovative work practices achieved substantially higher productivity, consistent with complementarities. Bloom and colleagues found more decentralization where trust was higher.Who disputes itGhoshal argued that management research has shaped managers' worldview with ideologically inspired amoral theories, built on a scientific model that excludes human intentionality and so denies moral considerations.Organizational economics: decision rights, authority, and complementarities in The Schools of Organization DesignInstitutional theory1977 to the presentWhat it claimsFormal structures are often adopted because they are legitimate in the organization's field, as rationalized myths, and then decoupled from the work done. Organizations in a field grow alike through coercive, mimetic, and normative isomorphism, whether or not it is efficient.What it has shownTolbert and Zucker (1983) found that city characteristics predicted early but not later adoption of civil service reform, which they read as institutionalization. Palmer, Jennings, and Zhou (1993) found support for institutional and economic accounts of late M-form adoption.Who disputes itStructural contingency theory: Kraatz and Zajac found little support for neoinstitutional predictions in 631 liberal arts colleges, which grew less alike; the change they studied was curricular, not structural.Institutional theory in The Schools of Organization DesignPopulation ecology1977 to the presentWhat it claimsOrganizations face strong structural inertia, because selection favors the reliability and accountability of stable structures. Change in a population's forms comes mainly from founding and failure, not adaptation, and changing core features resets the liability of newness.What it has shownAmburgey, Kelly, and Barnett (1993) found that the effects of change on the hazard of failure vary by type of change and over time, and that rates of change depend on how often and how recently an organization has changed. Carroll and Hannan (2000) set out the program's methods.Who disputes itFor structural contingency theory, Donaldson argues that ecology understates managers' capacity to adapt structure.Population ecology in The Schools of Organization DesignResource dependence1978 to the presentWhat it claimsOrganizations depend on outside parties for critical resources, and structure follows the management of that dependence: units that cope with critical dependencies gain power, and boundaries are redrawn through cooptation, joint ventures, and mergers to absorb constraint.What it has shownPfeffer and Salancik (1978) assembled studies of boards, mergers, and joint ventures. A meta-analysis of 157 tests (Drees and Heugens 2013) corroborated the main prediction: organizations meet resource dependencies with interorganizational arrangements that raise their autonomy.Who disputes itTransaction cost economics and the M-form hypothesis explains boundaries by efficiency; resource dependence, by power and constraint.Resource dependence in The Schools of Organization DesignThe business-groups literature1978 to the presentWhat it claimsGroups of legally independent firms bound by ownership and social ties are a durable form of their own, neither one firm nor a market. Where institutions function poorly, groups can benefit member firms and can destroy value; they are sometimes paragons, sometimes parasites.What it has shownKhanna and Palepu (2000) found that affiliates of the most diversified Indian groups outperformed unaffiliated firms. A meta-analysis of 141 studies (Carney and colleagues 2011) found that affiliation lowers performance in general, less so where institutions are underdeveloped.Who disputes itOrganizational economics: agency writers read pyramids and cross-holdings as devices that separate control from ownership; Almeida and Wolfenzon offer a financing rationale for pyramids instead.The business-groups literature in The Schools of Organization DesignScandinavian institutionalism1980s to the presentWhat it claimsIdeas of organizing do not diffuse unchanged; they travel and are translated as they are adopted, so a form taken up in one place is not the same form elsewhere. Organizations can also be built of organizations (meta-organizations), which behave differently.What it has shownA discussion paper by Czarniawska and Joerges (1995) contrasts the diffusion of ideas with their translation, drawing on examples such as Swedish municipal reform; the work on meta-organizations (Ahrne and Brunsson 2008) is conceptual.Who disputes itInstitutional theory can read the same processes, seen from a translocal view, as institutional isomorphism, though translation holds that each adopter modifies an idea, so an institution re-created elsewhere is new.Scandinavian institutionalism in The Schools of Organization DesignSocietal effect and comparative business systems1980s to the presentWhat it claimsFirms matched on size, technology, and industry differ systematically by country, because how a society educates, trains, recruits, and promotes its workforce, and its industrial relations, shape how work is divided and supervised, so structure is partly a national product.What it has shownMaurice, Sorge, and Warner (1980) compared closely matched manufacturing units in France, West Germany, and Great Britain and found more pervasive differences in their organization than usually recognized. Whitley (1999) set out a comparative account of business systems.Who disputes itStructural contingency theory stresses culture-free elements of structure; matched comparisons find nationally different shapes of organization.Societal effect and comparative business systems in The Schools of Organization DesignThe network perspective on organizations1985 to the presentWhat it claimsOrganizations and their members are nodes joined by ties, and the pattern of ties shapes what they can do. Economic action is embedded in networks of personal relations (Granovetter 1985), and network forms are a distinct form, not a hybrid of market and hierarchy (Powell 1990).What it has shownPodolny and Page (1998) review the research and find that network forms foster learning, confer status and legitimacy, bring economic benefits, help manage resource dependencies, and give employees considerable autonomy, and they ask for attention to networks' dysfunctions.Who disputes itTransaction cost economics: Williamson treats the hybrid as a third generic form between market and hierarchy; for Powell, network forms are a distinct form, not a mix of the two.The network perspective on organizations in The Schools of Organization DesignComputational organization design and search1990s to the presentWhat it claimsDesign choices interact, so design space is a rugged landscape with many local peaks, and structure affects how well an organization searches it. Decentralizing interdependent decisions can trap units on poor peaks; temporary decentralization, then reintegration, can help.What it has shownMostly simulation, with one field test: Csaszar (2012), with more than 150,000 stock-picking decisions by 609 mutual funds, found that decision-making structure affects the number of initiatives pursued and the omission and commission errors made.Who disputes itNo school in the record disputes it head on; it disputes structural contingency theory, whose bivariate fit, its simulations show, cannot capture how the best design depends on interactions among choices.Computational organization design and search in The Schools of Organization DesignThe new-forms debate1990s to the presentWhat it claimsOrganizations must solve four universal problems (dividing tasks, allocating them, providing rewards, and providing information), and a form is new when it solves them in new ways. Self-managing organizations radically decentralize authority; skeptics hold hierarchy reappears.What it has shownFoss (2003) traced the rise and partial reversal of Oticon's spaghetti organization to the problem of selective intervention. A meta-analysis of 54 studies of 13,914 teams (Greer and colleagues 2018) found that hierarchy within teams on net slightly lowers team performance.Who disputes itBureaucracy and its critics: removing formal structure leaves informal elites whose power is unaccountable (Freeman), and democratic and network organizations are much less hierarchy-free than their labels suggest.The new-forms debate in The Schools of Organization DesignThe algorithmic-management literature2015 to the presentWhat it claimsSoftware does much of the directing, evaluating, and disciplining that managers and rules once did. Kellogg, Valentine, and Christin (2020) call algorithmic control a new contested terrain; Puranam (2021) treats dividing decisions between people and machines as a design problem.What it has shownMostly qualitative field studies of gig work. Wood and colleagues (2019), with 107 interviews in six countries and 679 survey respondents, found that algorithmic management tends to give remote gig workers flexibility and autonomy, but also low pay, isolation, and overwork.Who disputes itLabor process and critical theory reads control at work as contested terrain; Kellogg, Valentine, and Christin take up that view but argue that algorithmic control is distinct from earlier forms.The algorithmic-management literature in The Schools of Organization DesignFunctional structureOrigin: 1911 to 1937.Family: Classical hierarchies.Line and staffOrigin: 1916 to 1943.Family: Classical hierarchies.Multidivisional (M-form) by productOrigin: 1920s practice; named and analyzed 1962 to 1975.Family: Divisionalized forms.Machine bureaucracyOrigin: 1922 (Weber); 1979 (Mintzberg).Family: Classical hierarchies.Divisional by customer or market segmentOrigin: 1937; 2005.Family: Divisionalized forms.Matrix organizationOrigin: 1950s to 1960s practice; analyzed 1971 to 1977.Family: Lateral and multidimensional forms.Co-determined organizationOrigin: 1951 to 1976 (the laws); 1984 to 2022 (the main studies).Family: Forms from other traditions.Socio-technical autonomous work groupsOrigin: 1951 to 1965.Family: Team-based and self-managing forms.Stratified hierarchy (requisite organization)Origin: 1951 to 1989.Family: Classical hierarchies.Cooperative organizationOrigin: 1958 to 1999.Family: Forms from other traditions.Linking-pin organization (System 4)Origin: 1961 to 1967.Family: Team-based and self-managing forms.Divisional by regionOrigin: 1962 to 1972.Family: Divisionalized forms.Adhocracy (innovative organization)Origin: 1968 to 1979.Family: Professional, project, and knowledge forms.J-form (the Japanese firm)Origin: 1968 to 1990.Family: Forms from other traditions.Professional bureaucracyOrigin: 1968 to 1979.Family: Professional, project, and knowledge forms.Holding company (H-form)Origin: 1972 to 1975; 1987.Family: Divisionalized forms.Ambidextrous organizationOrigin: 1976 to 2004.Family: Professional, project, and knowledge forms.Simple structureOrigin: 1979.Family: Classical hierarchies.Hybrid structureOrigin: 1980 to the present.Family: Lateral and multidimensional forms.Missionary (ideological) organizationOrigin: 1980 to 1989.Family: Ideological forms.Network, modular, and virtual organizationOrigin: 1986 to 1990.Family: Network and market-like forms.Transnational and multidimensional organizationOrigin: 1987 to 1989.Family: Lateral and multidimensional forms.Business group (keiretsu, chaebol, family house)Origin: 1988 to 2007.Family: Forms from other traditions.Shamrock organizationOrigin: 1988 to 1989.Family: Network and market-like forms.Federal organizationOrigin: 1989 to 1992.Family: Network and market-like forms.Self-managing organization (less-hierarchical forms)Origin: 1989 to 2017.Family: Team-based and self-managing forms.Boundaryless organizationOrigin: 1990 to 1995.Family: Network and market-like forms.Front-back organizationOrigin: 1990s to 2007.Family: Lateral and multidimensional forms.Professional partnership (P2 form)Origin: 1990.Family: Professional, project, and knowledge forms.The corporate center and shared servicesOrigin: 1991 to 2000.Family: Divisionalized forms.Internal market organizationOrigin: 1993 to 1994.Family: Network and market-like forms.Process (horizontal) organizationOrigin: 1993 to 1999.Family: Lateral and multidimensional forms.Cellular organizationOrigin: 1994 to 1997.Family: Network and market-like forms.Hypertext organizationOrigin: 1994 to 1995.Family: Forms from other traditions.Team-based organizationOrigin: 1995.Family: Team-based and self-managing forms.Project-based organizationOrigin: 2000 to 2006.Family: Professional, project, and knowledge forms.Agile at scaleOrigin: 2001 to 2016.Family: Contemporary and emerging forms.Distributed (all-remote) organizationOrigin: 2002 to 2021.Family: Contemporary and emerging forms.Open-source and community-based organizationOrigin: 2002 to 2007.Family: Contemporary and emerging forms.Algorithmically managed organizationOrigin: 2015 to 2020.Family: Contemporary and emerging forms.Team of teamsOrigin: 2015.Family: Team-based and self-managing forms.

The schools by first year

  1. Bureaucracy and its critics1900s to 1950s, revived from the 1990sWhat it claimsWeber held that rational-legal authority exercised through an office hierarchy, written rules, trained officials, and promotion by qualification is technically superior in precision, speed, and reliability. Critics studied its unintended consequences, Michels's iron law sharpest.What it has shownMerton (1940) argued, with no test, that rules can turn from means into ends; case studies followed. The Aston group concluded that the type's dimensions vary independently, and Wagner's meta-analysis (2021) found that many of their interrelationships have weakened.Who disputes itHuman relations and System 4 sets the costs of impersonal rules to motivation and cooperation against the technical superiority Weber stresses.
  2. Classical administration1910s to 1940s, and in German to the 1970sWhat it claimsThere are general principles of administration, among them division of work, unity of command, a scalar chain of authority, and a limited span of control, and a structure built to them will be efficient in most settings. Taylor added that planners should set work methods.What it has shownThe school offered reasoning and experience, not tests. Woodward (1965) found that the spans and levels of the more successful firms in her sample differed by production system, and Simon (1946) showed that the principles come in pairs that point in opposite directions.Who disputes itStructural contingency theory found that the structure of successful firms varies with technology and environment, where classical writers treat their principles as general.
  3. Human relations and System 41920s to 1960sWhat it claimsOutput depends on the social relations of work, the norms of informal groups, and the style of supervision as much as on formal design. Likert's System 4, overlapping work groups joined by linking-pin managers under participative management, outperforms authoritative systems.What it has shownThe founding evidence is the Hawthorne studies, which later reanalyses dispute (Carey 1967; Levitt and List 2011). Likert's support came largely from his own institute's surveys. Wagner (1994) found participation's effects on performance and satisfaction significant but small.Who disputes itLabor process and critical theory reads human relations as a technique of managerial control rather than a change in the distribution of power.
  4. Transaction cost economics and the M-form hypothesis1937, then the 1970s to the presentWhat it claimsFirms exist, and are organized as they are, to economize on transaction costs. Hierarchy replaces the market when assets are specific, uncertainty is high, and exchange is frequent, and the multidivisional form should outperform the functional form in large, diversified firms.What it has shownArmour and Teece (1978) found, in United States petroleum firms from 1955 to 1973, a positive relationship between the multidivisional form and profitability while the form was spreading, but not once an equilibrium in organizational form was reached.Who disputes itThe network perspective on organizations: Granovetter argued that the theory undersocializes economic action, which is embedded in networks of personal relations that also produce trust and order.
  5. The Carnegie school and the attention-based view1940s to the presentWhat it claimsPeople in organizations are boundedly rational, so structure matters because it sets the premises of their decisions and channels their attention: who sees which problems, which goals they pursue, and which routines they follow. Ocasio makes structure distribute attention.What it has shownMarch and Simon (1958) drew propositions from the existing literature, and Cyert and March (1963) added process models and simulations. Two studies show hierarchy shaping what is sent up as well as what is chosen (Reitzig and Maciejovsky 2015; Keum and See 2017).Who disputes itOrganizational economics: decision rights, authority, and complementarities models a principal and agents with defined objectives; the Carnegie school, firms as coalitions with unresolved goal conflict.
  6. Socio-technical systems1950s to 1980sWhat it claimsAn organization is a social system and a technical system that must be designed together. The same technology admits more than one way of organizing work, and groups that own a whole task and control its variances at the source perform better and are more satisfied.What it has shownTrist and Bamforth (1951) found that longwall mechanization broke up self-regulating work groups. A review of 134 experiments found them extremely successful overall, and Stewart (2006) linked team autonomy to performance; studies of socialist brigades and Volkswagen qualify it.Who disputes itLabor process and critical theory: Kelly argued that labor intensification, not joint optimization, accounts for what socio-technical redesign achieved, and that the part of pay incentives was seriously underestimated.
  7. The economics of the labor-managed firm and of enterprise ownership1958 to the presentWhat it claimsA firm run by its workers behaves differently from one run for investors: Ward modeled the Yugoslav-style firm as maximizing income per worker. Hansmann argued that an ownership form's success depends on the balance between the costs of market contracting and of ownership.What it has shownBurdín (2014) found a 29 percent lower hazard of dissolution for Uruguayan worker-managed firms than for conventional ones, and Fakhfakh, Pérotin, and Gago (2012) found French labor-managed firms as productive or more so in every industry studied.Who disputes itOrganizational economics (Jensen and Meckling) contrasts the labor-managed firm, in which tradable residual claims are legally prohibited, with cooperatives and partnerships that arise out of voluntary contracting.
  8. Modularity and the mirroring hypothesis1960s to the presentWhat it claimsComplex systems are built of nearly decomposable parts, and an organization's structure tends to mirror the architecture of what it makes: modular products allow loosely coupled organizations, and tightly coupled products need tightly coupled teams. Conway's law runs in reverse.What it has shownColfer and Baldwin (2016) review the empirical studies and report both support and principled exceptions. Direct tests point both ways: MacCormack, Baldwin, and Rusnak (2012) found open source products more modular than their commercial pairs; dynamic components break the mirror.Who disputes itNo school in the record disputes it head on; it answers information processing, which meets the coordination load inside the organization; mirroring writers locate it in the product's architecture.
  9. Strategy and structure1960s to 1980sWhat it claimsStructure follows strategy: firms that diversify by product or geography outgrow the centralized functional structure and, after a lag and usually a crisis, adopt divisions. Chandler drew this from the histories of DuPont, General Motors, Standard Oil of New Jersey, and Sears.What it has shownChandler (1962) is comparative business history of four firms set against a wider survey. Rumelt (1974) related diversification and structure to performance, and Egelhoff (1988), with 34 multinationals, found some of Stopford and Wells's relationships supported and others not.Who disputes itInstitutional theory: in Fligstein's model divisions spread with a product-related or unrelated strategy, presidents from sales or finance, and other firms in the industry changing; strategy was one of three.
  10. Structural contingency theory1960s to the presentWhat it claimsThere is no one best structure. Performance depends on fit between structure and contingencies such as environment, technology, size, and strategy, on an open-system view of the organization. Drazin and Van de Ven separate fit as selection, interaction, and systems.What it has shownBurns and Stalker, Woodward, and the Aston group related structure to context; Drazin and Van de Ven (1985) supported selection and systems fit only. Pennings (1975) found that fit did not explain effectiveness; Schoonhoven (1981) found traditional hypotheses unsupported.Who disputes itStrategic choice: Child argued that contingency theory explains structure by the constraints of context and leaves out the political process by which power-holders choose strategic action, structural forms included.
  11. The French bureaucratic school (strategic analysis)1960s to 1980sWhat it claimsA bureaucracy is a system of power relations organized around zones of uncertainty. Impersonal rules and centralization protect members from depending on one another, those who control uncertainties hold power, and, unable to correct its errors, it changes only through crisis.What it has shownLe phénomène bureaucratique (1963; English 1964) rests on two case studies, a clerical agency and a state industrial monopoly. L'acteur et le système (1977; English 1980) is theoretical.Who disputes itSocietal effect and comparative business systems: D'Iribarne argues that national logics of conduct, such as a French logic of honor, explain differences that Crozier attributed to power games.
  12. The Japanese-firm literature1960s to the presentWhat it claimsThe large Japanese firm coordinates operating decisions through horizontal information sharing among semi-autonomous shops and uses its ranking hierarchy mainly as an incentive scheme (Aoki). It creates knowledge through middle-up-down management (Nonaka and Takeuchi).What it has shownAoki (1986) is a formal model. Lincoln, Hanada, and McBride (1986) compared 51 Japanese and 55 United States manufacturing plants, and Nonaka and Takeuchi (1995) rest on case studies, chiefly of Japanese firms. Fruin (1994) found the enterprise system defined at three levels.Who disputes itLabor process and critical theory reads shop-floor delegation as control: Sewell and Wilkinson argue that just-in-time and total quality regimes delegate tactical responsibility while centralizing strategic control.
  13. The Scandinavian and Dutch socio-technical schools1960s to 2000sWhat it claimsThe Scandinavian program sought democracy at work through field experiments with semi-autonomous groups, then through democratic dialogue. The Dutch school's design theory turns complex organizations with simple jobs into simple organizations with complex jobs.What it has shownEmery and Thorsrud reported field experiments in Norwegian industry; Gustavsen (1992) records that they spread less than hoped, his reason for turning to dialogue. Berggren studied work organization in the Swedish auto industry. The Dutch work rests on design cases and argument.Who disputes itThe Japanese-firm literature: defenders of the Toyota system hold that standardized work with employee participation learns faster; Berggren reports Swedish long-cycle plants matching a line plant on productivity.
  14. Configuration theory1970s to the presentWhat it claimsStructures cluster into a small number of internally coherent configurations whose elements reinforce one another, so organizations move between a few types in large jumps, not small steps. Mintzberg's five configurations each rest on one coordinating mechanism.What it has shownDoty, Glick, and Huber (1993) found that fit to Miles and Snow's ideal types predicted effectiveness while fit to Mintzberg's did not. A meta-analysis of 40 tests (Ketchen and colleagues 1997) found that configuration partly explains performance, more so with broad definitions.Who disputes itStructural contingency theory: Donaldson counters that most organizations sit at intermediate points on a continuum of fits, not in a few configurations, and that studies over time find mostly incremental change.
  15. Information processing1970s to the presentWhat it claimsThe more uncertain the task, the more information must be processed during its execution. An organization either reduces that need, through slack resources or self-contained tasks, or raises its capacity, through vertical information systems or lateral relations up to the matrix.What it has shownGalbraith (1973, 1974) built the framework from cases and from Simon's theory of bounded rationality, and Tushman and Nadler (1978) set it out as an integrating concept for design. Daft and Lengel (1986) argued that structure must reduce equivocality as well as supply data.Who disputes itThe Carnegie school and the attention-based view: Ocasio holds that firm behavior results from how the firm channels and distributes attention, so structure governs where attention goes, not only how much is processed.
  16. Labor process and critical theory1970s to the presentWhat it claimsStructure is above all a system of control over labor. Braverman argued that management deskills work by separating conception from execution; Edwards traced a sequence from simple control by the boss to technical control through machines to bureaucratic control through rules.What it has shownHistorical and theoretical (Braverman 1974; Edwards 1979). Delbridge (1995), observing a Japanese-owned plant in England, found that just-in-time and total quality regimes combined the control of labor with management's economic goals more completely than piece rates had.Who disputes itBureaucracy and its critics: Adler argues that labor process theory ignores the progressive socialization of the labor process, which drives long-term skill upgrading, and disputes deskilling as the trend.
  17. Strategic choice1970s to the presentWhat it claimsStructure is the outcome of choices by those with power in the organization, who decide on strategic action, including the structural form, the manipulation of environmental features, and the performance standards to be met. Context constrains but does not determine structure.What it has shownLargely conceptual: Child (1972) built the argument from gaps in contingency findings, and the empirical case rests mostly on the modest share of structural variance contingencies explain. A 2026 review of 42 studies finds upper echelons theory their main theoretical anchor.Who disputes itStructural contingency theory: Donaldson replies that organizations mostly adopt the structures their contingencies dictate, to avoid the loss of performance misfit brings, so choice over structure is at most limited.
  18. The German codetermination literature1970s to the presentWhat it claimsWorks councils and employee board seats form a structural layer through which coordination can become negotiation between management and employees. Defenders argue that it supports long-term employment, training, and cooperation; critics, that it weakens shareholder control.What it has shownJäger, Schoefer, and Heining (2021) found board-level codetermination left wages unchanged and, if anything, raised capital formation. Jirjahn (2011) reports studies pointing to positive works council effects; a meta-analysis found codetermination laws tied to lower productivity.Who disputes itOrganizational economics (Gorton and Schmid) found that firms with equal employee and shareholder representation did worse for their shareholders than firms with one-third employee representation.
  19. Organizational economics: decision rights, authority, and complementarities1976 to the presentWhat it claimsDesign is the joint choice of who holds which decision rights, how performance is measured, and how it is rewarded. Rights should be placed where the specific knowledge lies, formal authority differs from real authority, and practices are complements that pay off in clusters.What it has shownIchniowski, Shaw, and Prennushi (1997) found, on 36 steel finishing lines, that lines using a set of innovative work practices achieved substantially higher productivity, consistent with complementarities. Bloom and colleagues found more decentralization where trust was higher.Who disputes itGhoshal argued that management research has shaped managers' worldview with ideologically inspired amoral theories, built on a scientific model that excludes human intentionality and so denies moral considerations.
  20. Institutional theory1977 to the presentWhat it claimsFormal structures are often adopted because they are legitimate in the organization's field, as rationalized myths, and then decoupled from the work done. Organizations in a field grow alike through coercive, mimetic, and normative isomorphism, whether or not it is efficient.What it has shownTolbert and Zucker (1983) found that city characteristics predicted early but not later adoption of civil service reform, which they read as institutionalization. Palmer, Jennings, and Zhou (1993) found support for institutional and economic accounts of late M-form adoption.Who disputes itStructural contingency theory: Kraatz and Zajac found little support for neoinstitutional predictions in 631 liberal arts colleges, which grew less alike; the change they studied was curricular, not structural.
  21. Population ecology1977 to the presentWhat it claimsOrganizations face strong structural inertia, because selection favors the reliability and accountability of stable structures. Change in a population's forms comes mainly from founding and failure, not adaptation, and changing core features resets the liability of newness.What it has shownAmburgey, Kelly, and Barnett (1993) found that the effects of change on the hazard of failure vary by type of change and over time, and that rates of change depend on how often and how recently an organization has changed. Carroll and Hannan (2000) set out the program's methods.Who disputes itFor structural contingency theory, Donaldson argues that ecology understates managers' capacity to adapt structure.
  22. Resource dependence1978 to the presentWhat it claimsOrganizations depend on outside parties for critical resources, and structure follows the management of that dependence: units that cope with critical dependencies gain power, and boundaries are redrawn through cooptation, joint ventures, and mergers to absorb constraint.What it has shownPfeffer and Salancik (1978) assembled studies of boards, mergers, and joint ventures. A meta-analysis of 157 tests (Drees and Heugens 2013) corroborated the main prediction: organizations meet resource dependencies with interorganizational arrangements that raise their autonomy.Who disputes itTransaction cost economics and the M-form hypothesis explains boundaries by efficiency; resource dependence, by power and constraint.
  23. The business-groups literature1978 to the presentWhat it claimsGroups of legally independent firms bound by ownership and social ties are a durable form of their own, neither one firm nor a market. Where institutions function poorly, groups can benefit member firms and can destroy value; they are sometimes paragons, sometimes parasites.What it has shownKhanna and Palepu (2000) found that affiliates of the most diversified Indian groups outperformed unaffiliated firms. A meta-analysis of 141 studies (Carney and colleagues 2011) found that affiliation lowers performance in general, less so where institutions are underdeveloped.Who disputes itOrganizational economics: agency writers read pyramids and cross-holdings as devices that separate control from ownership; Almeida and Wolfenzon offer a financing rationale for pyramids instead.
  24. Scandinavian institutionalism1980s to the presentWhat it claimsIdeas of organizing do not diffuse unchanged; they travel and are translated as they are adopted, so a form taken up in one place is not the same form elsewhere. Organizations can also be built of organizations (meta-organizations), which behave differently.What it has shownA discussion paper by Czarniawska and Joerges (1995) contrasts the diffusion of ideas with their translation, drawing on examples such as Swedish municipal reform; the work on meta-organizations (Ahrne and Brunsson 2008) is conceptual.Who disputes itInstitutional theory can read the same processes, seen from a translocal view, as institutional isomorphism, though translation holds that each adopter modifies an idea, so an institution re-created elsewhere is new.
  25. Societal effect and comparative business systems1980s to the presentWhat it claimsFirms matched on size, technology, and industry differ systematically by country, because how a society educates, trains, recruits, and promotes its workforce, and its industrial relations, shape how work is divided and supervised, so structure is partly a national product.What it has shownMaurice, Sorge, and Warner (1980) compared closely matched manufacturing units in France, West Germany, and Great Britain and found more pervasive differences in their organization than usually recognized. Whitley (1999) set out a comparative account of business systems.Who disputes itStructural contingency theory stresses culture-free elements of structure; matched comparisons find nationally different shapes of organization.
  26. The network perspective on organizations1985 to the presentWhat it claimsOrganizations and their members are nodes joined by ties, and the pattern of ties shapes what they can do. Economic action is embedded in networks of personal relations (Granovetter 1985), and network forms are a distinct form, not a hybrid of market and hierarchy (Powell 1990).What it has shownPodolny and Page (1998) review the research and find that network forms foster learning, confer status and legitimacy, bring economic benefits, help manage resource dependencies, and give employees considerable autonomy, and they ask for attention to networks' dysfunctions.Who disputes itTransaction cost economics: Williamson treats the hybrid as a third generic form between market and hierarchy; for Powell, network forms are a distinct form, not a mix of the two.
  27. Computational organization design and search1990s to the presentWhat it claimsDesign choices interact, so design space is a rugged landscape with many local peaks, and structure affects how well an organization searches it. Decentralizing interdependent decisions can trap units on poor peaks; temporary decentralization, then reintegration, can help.What it has shownMostly simulation, with one field test: Csaszar (2012), with more than 150,000 stock-picking decisions by 609 mutual funds, found that decision-making structure affects the number of initiatives pursued and the omission and commission errors made.Who disputes itNo school in the record disputes it head on; it disputes structural contingency theory, whose bivariate fit, its simulations show, cannot capture how the best design depends on interactions among choices.
  28. The new-forms debate1990s to the presentWhat it claimsOrganizations must solve four universal problems (dividing tasks, allocating them, providing rewards, and providing information), and a form is new when it solves them in new ways. Self-managing organizations radically decentralize authority; skeptics hold hierarchy reappears.What it has shownFoss (2003) traced the rise and partial reversal of Oticon's spaghetti organization to the problem of selective intervention. A meta-analysis of 54 studies of 13,914 teams (Greer and colleagues 2018) found that hierarchy within teams on net slightly lowers team performance.Who disputes itBureaucracy and its critics: removing formal structure leaves informal elites whose power is unaccountable (Freeman), and democratic and network organizations are much less hierarchy-free than their labels suggest.
  29. The algorithmic-management literature2015 to the presentWhat it claimsSoftware does much of the directing, evaluating, and disciplining that managers and rules once did. Kellogg, Valentine, and Christin (2020) call algorithmic control a new contested terrain; Puranam (2021) treats dividing decisions between people and machines as a design problem.What it has shownMostly qualitative field studies of gig work. Wood and colleagues (2019), with 107 interviews in six countries and 679 survey respondents, found that algorithmic management tends to give remote gig workers flexibility and autonomy, but also low pay, isolation, and overwork.Who disputes itLabor process and critical theory reads control at work as contested terrain; Kellogg, Valentine, and Christin take up that view but argue that algorithmic control is distinct from earlier forms.

The forms by first year

  1. Functional structure: 1911 to 1937
  2. Line and staff: 1916 to 1943
  3. Multidivisional (M-form) by product: 1920s practice; named and analyzed 1962 to 1975
  4. Machine bureaucracy: 1922 (Weber); 1979 (Mintzberg)
  5. Divisional by customer or market segment: 1937; 2005
  6. Matrix organization: 1950s to 1960s practice; analyzed 1971 to 1977
  7. Co-determined organization: 1951 to 1976 (the laws); 1984 to 2022 (the main studies)
  8. Socio-technical autonomous work groups: 1951 to 1965
  9. Stratified hierarchy (requisite organization): 1951 to 1989
  10. Cooperative organization: 1958 to 1999
  11. Linking-pin organization (System 4): 1961 to 1967
  12. Divisional by region: 1962 to 1972
  13. Adhocracy (innovative organization): 1968 to 1979
  14. J-form (the Japanese firm): 1968 to 1990
  15. Professional bureaucracy: 1968 to 1979
  16. Holding company (H-form): 1972 to 1975; 1987
  17. Ambidextrous organization: 1976 to 2004
  18. Simple structure: 1979
  19. Hybrid structure: 1980 to the present
  20. Missionary (ideological) organization: 1980 to 1989
  21. Network, modular, and virtual organization: 1986 to 1990
  22. Transnational and multidimensional organization: 1987 to 1989
  23. Business group (keiretsu, chaebol, family house): 1988 to 2007
  24. Shamrock organization: 1988 to 1989
  25. Federal organization: 1989 to 1992
  26. Self-managing organization (less-hierarchical forms): 1989 to 2017
  27. Boundaryless organization: 1990 to 1995
  28. Front-back organization: 1990s to 2007
  29. Professional partnership (P2 form): 1990
  30. The corporate center and shared services: 1991 to 2000
  31. Internal market organization: 1993 to 1994
  32. Process (horizontal) organization: 1993 to 1999
  33. Cellular organization: 1994 to 1997
  34. Hypertext organization: 1994 to 1995
  35. Team-based organization: 1995
  36. Project-based organization: 2000 to 2006
  37. Agile at scale: 2001 to 2016
  38. Distributed (all-remote) organization: 2002 to 2021
  39. Open-source and community-based organization: 2002 to 2007
  40. Algorithmically managed organization: 2015 to 2020
  41. Team of teams: 2015

Twenty-three of the 29 schools run to the present, and 16 of those began in the 1970s or later. The forms arrive in waves: 14 of the 41 first appear between 1985 and 1994.

Source: Fayol 1949; Weber 1978; Mayo 1933; Trist and Bamforth 1951, an interview-based field study in one industry; Burns and Stalker 1961, a field study of twenty firms; Mintzberg 1979; Child 1972a; Coase 1937; Meyer and Rowan 1977; Hannan and Freeman 1977; Galbraith 1973; Jensen and Meckling 1976; Crozier 1963; Emery and Thorsrud 1969; Yoshino 1968; Streeck 1984a; Kellogg and colleagues 2020, a review of earlier research through one theoretical lens, with no new data; Puranam and colleagues 2014, a conceptual argument; Chandler 1962, several independent case histories by a historian; Pfeffer and Salancik 1978; Simon 1946; Maurice and colleagues 1980, a comparison of matched factories in three countries; Levinthal 1997; Simon 1962; Braverman 1974; Hamilton and Biggart 1988, a comparison of three economies by scholars; Ward 1958; Czarniawska and Joerges 1995; Granovetter 1985.

The form answered the problem of combining specialized firms flexibly when vertical integration is too slow and spot markets too thin. The antecedents are Coase (1937), for whom firms exist because using the price mechanism has costs, and Williamson (1975), on when hierarchy replaces it.

Where it fits

Ten cells are our judgment and two rest on evidence; one of each is contested.

The network, modular, and virtual organization against the twelve contingency factors

Each factor at its three levels, Fits, Partial, or Strain; select a row for its reason, sources, and grade.

Select a factor to read its reason, sources, basis, and grade.Size. A hub can be a microfirm in a district or a large assembler; size changes how many ties must be managed, not whether the form works. Basis: Judgment. Sources: Lazerson 1995; Dyer and Nobeoka 2000; Miles and Snow 1986.Age. Relational capability and trust among partners are built over years, so a young hub has less to coordinate with than an established one. Basis: Judgment. Sources: Lorenzoni and Lipparini 1999; Dyer and Nobeoka 2000.Environmental uncertainty (unpredictability, dynamism). Contracts and standard interfaces coordinate well while the product architecture is stable; when technology moves fast, the hub must keep knowledge and share information beyond what the interfaces carry. Basis: Evidence. Grade: a multi-firm empirical study or several independent case studies by scholars. Sources: Colfer and Baldwin 2016; Cabigiosu and Camuffo 2012; Furlan and colleagues 2014; Brusoni and colleagues 2001; Pennings 1975.Environmental complexity. Contracting out works cleanly when few factors matter, and in complex, multi-technology settings it still works if the hub keeps enough knowledge in house to integrate its partners. Basis: Judgment. Sources: Brusoni and colleagues 2001; Williamson 1975; Pennings 1975.Munificence and hostility. Under threat the hub depends on partners for critical resources it no longer makes, and Burton and Obel lower centralization at high hostility but raise it at extreme hostility, so the works read point both ways. Basis: Judgment. Sources: Pfeffer and Salancik 1978; Burton and Obel 2004.Diversity of products and markets. A hub firm can contract out activities in any business mix; nothing read ties the form to diversity. Basis: Judgment. Sources: Miles and Snow 1986.Strategy type. A broker that can recombine partners moves into new products and markets quickly (Miles and Snow), and Burton and Obel name low centralization, which the network's hub shares, a misfit for a defender; no work tests the network form against strategy type. Basis: Judgment. Sources: Miles and Snow 1986; Burton and Obel 2004.Technology and routineness. Well-specified work can be contracted across firm lines through standardized interfaces; where exceptions are many, partners must share more knowledge than the contract carries, as firms in technologically dynamic industries do. Basis: Judgment. Sources: Sanchez and Mahoney 1996; Colfer and Baldwin 2016; Perrow 1967; Campagnolo and Camuffo 2009; Hickson and colleagues 1969; Mohr 1971. View 1: Routine fits, Mixed partial, Nonroutine partial. Well-specified work can be contracted across firm lines through standardized interfaces; where exceptions are many, partners must share more knowledge than the contract carries, as firms in technologically dynamic industries do. Sources: Sanchez and Mahoney 1996; Colfer and Baldwin 2016; Perrow 1967. View 2: Routine partial, Mixed partial, Nonroutine partial. In 68 Italian manufacturing industries the use of modular organizational forms was driven by labor intensity, industry specificities, and nation-specific factors, diverging from the American results, so the technology of the work does not settle where firms use the form. Sources: Campagnolo and Camuffo 2009.Interdependence. Loosely coupled work goes with loosely coupled organizations across firm lines, a prevalent pattern, but whether the network does better for it is disputed, since supply performance tracked information sharing, not modularity. Basis: Evidence. Grade: several independent quantitative studies or a meta-analysis. Sources: Colfer and Baldwin 2016; MacCormack and colleagues 2012; Cabigiosu and Camuffo 2012; Furlan and colleagues 2014; Sanchez and Mahoney 1996; Brusoni and colleagues 2001; Bidwell 2012. View 1: Pooled fits, Sequential fits, Reciprocal partial. Mirroring is a prevalent pattern: loosely coupled work goes with loosely coupled organizations across firm lines, while tightly coupled work tends to stay in one firm unless relational contracts and shared knowledge carry it across. Sources: Colfer and Baldwin 2016; MacCormack and colleagues 2012; Cabigiosu and Camuffo 2012; Sanchez and Mahoney 1996. View 2: Pooled partial, Sequential partial, Reciprocal partial. The performance of supply relationships depended on information sharing, not on how modular the components were; mirroring across firms did not hold for technologically dynamic components; firms that coordinate loosely coupled supplier networks must know more than they make; and outsourcing decisions tracked transaction characteristics only weakly. Sources: Cabigiosu and Camuffo 2012; Furlan and colleagues 2014; Brusoni and colleagues 2001; Bidwell 2012.Knowledge intensity and professionalization. Firms that keep a few core activities and contract out the rest are found with both trained and less trained workforces, and no work read ties the form to training levels. Basis: Judgment. Sources: Miles and Snow 1986; Quinn and colleagues 1996b.Global integration pressure. Contract networks span countries, but no work read sets the network form against integration pressure across countries. Basis: Judgment. Sources: Miles and Snow 1986.Local responsiveness pressure. Local partners can supply local knowledge, but no work read sets the network form against responsiveness to countries. Basis: Judgment. Sources: Miles and Snow 1986.

Counting each contested cell by its first view, the 36 settings read Fits at 13, Partial at 22, and Strain at 1, the one strain a defender strategy. Two of the twelve cells are contested, and ten rest on judgment.

Source: Lazerson 1995, one case study of an industrial district; Dyer and Nobeoka 2000, one case study of a supplier network; Miles and Snow 1986; Lorenzoni and Lipparini 1999, a longitudinal study of three firms and their supplier networks; Colfer and Baldwin 2016, a review of 142 empirical studies; Cabigiosu and Camuffo 2012, a study of 100 supply relationships at three firms; Furlan and colleagues 2014, a study of supply relationships in one industry; Brusoni and colleagues 2001, a study of one technology in one industry; Pennings 1975, a quantitative field test of the structural-contingency model; Williamson 1975; Pfeffer and Salancik 1978; Burton and Obel 2004; Sanchez and Mahoney 1996; Perrow 1967; Campagnolo and Camuffo 2009, an industry-level study of 68 Italian manufacturing industries; Hickson and colleagues 1969, a multi-organization comparative survey; Mohr 1971, a quantitative field study of technology and structure; MacCormack and colleagues 2012, a matched comparison of five pairs of software products; Bidwell 2012, one case study of a firm's outsourcing decisions; Quinn and colleagues 1996b.

Your own situation against the network, modular, and virtual organization

Set any factor to outline its cell in the strip above; the lines under the switches give the counts, the form's place among the 41, and the misfit rules that fire.

Size
Age
Environmental uncertainty (unpredictability, dynamism)
Environmental complexity
Munificence and hostility
Diversity of products and markets
Strategy type
Technology and routineness
Interdependence
Knowledge intensity and professionalization
Global integration pressure
Local responsiveness pressure

Set: 0 of 12 factors. Fits on 0, Partial on 0, Strain on 0, Contested on 0.

Its place in the explorer's rankingEvery factor is on Any: all 41 forms tie, and the explorer lists them alphabetically.1st of 411st of 41, tied with one other1st of 41, tied with 0 othersSunk by a misfit rule: 1st of 41Sunk by a misfit rule: 1st of 41, tied with one otherSunk by a misfit rule: 1st of 41, tied with 0 others

  • Flagged (its sources disagree): A defender with authority spread wide. Disputed: Burton and Obel 2004; against it, Mintzberg 1980.
  • Sinks the form: An analyzer with authority spread wide. Source: Burton and Obel 2004.

Ranking rule: by count of Fits on the factors the reader has set, then fewest Strains, then alphabetical; a fired misfit rule that no source disputes sinks the form to the bottom when every level of the form's profile meets it, a disputed one, or one that meets the form only through one end of a spanned level, one view of a contested profile cell, or one part of a two-part form, is flagged on the row (with both views where it is disputed), and a misfit between two of the reader's own settings is shown once above the list; a contested cell counts only where its views agree at the reader's level, and otherwise shows as Contested; factors on "Any" count for nothing; with every factor on "Any" the list is alphabetical and says so.

Open these settings in the explorer

The misfit rules that can fire for this form

  • A defender with authority spread wide fires when Strategy type at Defender (exploitation): flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Mintzberg 1980.
  • An analyzer with authority spread wide fires when Strategy type at Analyzer (exploitation and exploration): sinks the form. Source: Burton and Obel 2004.

Of the 39 contingency misfit rules, two can fire for this form: one would sink it and one would only flag it.

Source: Burton and Obel 2004; Burton and colleagues 2020a.

  • Size. Fits at every size, judgment: hubs range from district microfirms (Lazerson 1995) to large assemblers (Dyer and Nobeoka 2000), each one case study.
  • Age. Partial young, fits established and old, judgment on Lorenzoni and Lipparini (1999), a longitudinal study of three firms, and Dyer and Nobeoka, one case study: relational capability and trust take years.
  • Environmental uncertainty. Fits stable and shifting, partial unpredictable, evidence: a review of 142 studies (Colfer and Baldwin 2016), here on its authors' inference for dynamic industries, not its main finding; a study of 100 supply relationships at three firms (Cabigiosu and Camuffo 2012); one sharing authors and the industry (Furlan, Cabigiosu, and Camuffo 2014); and a study of one technology in one industry (Brusoni, Prencipe, and Pavitt 2001). The four overlap and test technological dynamism, not the whole environment.
  • Environmental complexity. Fits simple, partial moderate, fits complex while the hub keeps the knowledge to integrate partners, judgment: Williamson's (1975) premise that contracts are easier to write when few factors matter, and Brusoni and colleagues' condition, a premise, not a test, since their complexity is technological.
  • Munificence and hostility. Partial throughout, judgment: under threat the hub depends on partners for critical resources (Pfeffer and Salancik 1978); Burton and Obel's (2004) rule, in the most developed scholarly fit model, lowers centralization and keeps formalization low under high hostility but centralizes under extreme hostility, and this cell's hostile level spans both.
  • Strategy. Strain defender, partial analyzer, fits prospector, judgment: Burton and Obel call low centralization a misfit for a defender; a broker recombines partners quickly (Miles and Snow 1986), though Burton and Obel's prospector wants the low formalization the network lacks.
  • Technology. Contested, judgment. View one (Sanchez and Mahoney 1996; Colfer and Baldwin; Perrow 1967): fits routine, partial mixed and nonroutine, since exceptions need knowledge a contract does not carry. View two (Campagnolo and Camuffo 2009, an industry-level study of 68 Italian manufacturing industries): partial throughout, since labor intensity, industry, and nation drove modular forms there.
  • Interdependence. Contested, evidence, graded on the review. View one (Colfer and Baldwin; MacCormack, Baldwin, and Rusnak 2012, a matched comparison of five pairs of software products; Cabigiosu and Camuffo; Sanchez and Mahoney): fits pooled and sequential, partial reciprocal, because mirroring is prevalent; our judgment reads product coupling as interdependence and sets the fit at sequential, which the review does not separate from reciprocal. View two (Cabigiosu and Camuffo; Furlan and colleagues; Brusoni and colleagues; Bidwell 2012, one case study): partial throughout, since supply performance tracked information sharing, not modularity. The arrow may also run backward: a conglomerate restructured to mirror its platform (Pan, Chakkol, and Johnson 2026, one firm's case study), and Raveendran, Silvestri, and Gulati (2020, a review ungraded here) argue interdependence arises in part from design itself.
  • Diversity, knowledge intensity, global integration, and local responsiveness. Partial throughout, judgment; nothing read ties the form to them (Miles and Snow 1986; Quinn, Anderson, and Finkelstein 1996).

Dissent-panel limits on fit in general, not this form: Pennings (1975), a quantitative field test of the structural-contingency model, found environment and structure fit did not explain effectiveness; Hickson, Pugh, and Pheysey (1969), a multi-organization comparative survey, found the technological imperative unsupported beyond the structure the workflow touches; and Mohr (1971), a quantitative field study of technology and structure, found no support for effectiveness depending on the consonance of technology and structure. Donaldson (2001) blames method for Pennings's and Mohr's results and reports later studies of the job unpredictability Mohr studied finding positive evidence for contingency theory. Resource dependence, the munificence premise, would manage and change the environment, not only fit it (Pfeffer and Salancik 1978).

What it does well

It reaches capabilities the firm does not own. Lorenzoni and Lipparini (1999), a longitudinal study of three firms and their supplier networks, found that relational capability speeded access to and transfer of knowledge, with effects on growth and innovativeness.

Interfaces can stand in for some coordination: in a study of 100 supply relationships at three firms, modularity weakened the effect of information sharing on performance, acting as an embedded substitute for high-powered integration mechanisms (Cabigiosu and Camuffo 2012). Lazerson (1995), one case study of an industrial district, found Modena's interlinked family microfirms compatible with high living standards and efficiency, a success it credits to technology and marketing, cohesive families, cooperation between business and community, and supportive institutions.

What it costs

It leaves the hub dependent on partners for capabilities it needs, our judgment on Pfeffer and Salancik (1978).

The hub must still know what it no longer makes: Brusoni and colleagues (2001), a study of one technology in one industry, found integrators keeping knowledge beyond what they make. Chesbrough and Teece (1996), a practitioner article, warn that outsourcing everything could undermine a firm's ability to innovate, since the virtual form fits autonomous innovations, not systemic ones.

How it fails

How the network, modular, and virtual organization fails, and the coordination failures behind it

The entry's failure modes at the left, each tied to the Coordination Failures patterns it names at the right; a pattern links its entry there.

Tacit-knowledge evaporationDormant-assumption reactivationJurisdictional gap or overlapCommons overdrawModel-world divergenceLocal optimization against the global constraint
Tacit-knowledge evaporationDormant-assumption reactivationJurisdictional gap or overlapCommons overdrawModel-world divergenceLocal optimization against the global constraint

How many ties the failure has. Highest: The hub forgets how to integrate what it buysInterfaces stop carrying the load when the technology movesContracts down the tiers block adaptationPartners free ride on what the network sharesTies so close the network stops hearing the outsideOutsourcing follows managers' interests, not the workTacit-knowledge evaporationDormant-assumption reactivationJurisdictional gap or overlapCommons overdrawModel-world divergenceLocal optimization against the global constraint

Select a failure to read its place in the network.The hub forgets how to integrate what it buys. Degree 1Interfaces stop carrying the load when the technology moves. Degree 1Contracts down the tiers block adaptation. Degree 1Partners free ride on what the network shares. Degree 1Ties so close the network stops hearing the outside. Degree 1Outsourcing follows managers' interests, not the work. Degree 1Tacit-knowledge evaporation. Degree 1Dormant-assumption reactivation. Degree 1Jurisdictional gap or overlap. Degree 1Commons overdraw. Degree 1Model-world divergence. Degree 1Local optimization against the global constraint. Degree 1
FailureDegree
The hub forgets how to integrate what it buys1
Interfaces stop carrying the load when the technology moves1
Contracts down the tiers block adaptation1
Partners free ride on what the network shares1
Ties so close the network stops hearing the outside1
Outsourcing follows managers' interests, not the work1
Tacit-knowledge evaporation1
Dormant-assumption reactivation1
Jurisdictional gap or overlap1
Commons overdraw1
Model-world divergence1
Local optimization against the global constraint1

Six failure modes name six patterns, one each, so no pattern is shared between modes. Every tie is our reading of the studies each mode cites.

Source: Brusoni and colleagues 2001, a study of one technology in one industry; Furlan and colleagues 2014, a study of supply relationships in one industry; Colfer and Baldwin 2016, a review of 142 empirical studies; Baumann and colleagues 2020, one case study of an aircraft program; Dyer and Nobeoka 2000, one case study of a supplier network; Uzzi 1997, an ethnographic study of 23 firms; Bidwell 2012, one case study of a firm's outsourcing decisions.

The hub forgets how to integrate what it buys. When knowledge follows production out, the hub can no longer specify or combine partners' work: tacit-knowledge evaporation, our reading of the converse of Brusoni and colleagues (2001), a study of one technology in one industry.

Interfaces stop carrying the load when the technology moves. A dormant assumption reactivated, our reading: Furlan and colleagues (2014), a study of supply relationships in one industry, found cross-firm mirroring failing for technologically dynamic components, and Colfer and Baldwin (2016), a review of 142 studies, infer partial mirroring likely superior there.

Contracts down the tiers block adaptation. Baumann, Becker, and Horrmann (2020), one case study, argue that down a supplier hierarchy relational contracts matter more but are harder to establish; in the Airbus A350 program, skip-level ties to suppliers Airbus had no contract with resolved the frictions. A jurisdictional gap, our reading.

Partners free ride on what the network shares. Dyer and Nobeoka (2000), one case study, name the dilemmas a knowledge-sharing network must solve, free riders among them; left unsolved, they are commons overdraw, our reading.

Ties so close the network stops hearing the outside. Past a threshold, which Uzzi (1997, an ethnographic study of 23 firms) offers as a proposition, embeddedness can derail performance by leaving firms vulnerable to shocks or insulated from outside information: model and world divergence, our reading.

Outsourcing follows managers' interests, not the work. Bidwell (2012), one case study of a firm's outsourcing decisions, found only a weak link between transaction characteristics and outsourcing; managers' interests and the distribution of authority shaped what went out: local optimization against the whole, our reading.

The decentralized profile trips two misfit rules, both Burton and Obel (2004) propositions, untested: wide authority in a defender, and in an analyzer (they name it for the analyzer that does not innovate, recommending medium centralization for one that does; the explorer's one analyzer level fires on both). The defender rule is contested: a second view, on Mintzberg's (1980) synthesis, propositions without a new test, placing the decentralized professional bureaucracy in complex but stable environments, holds that wide authority is not ruled out in a stable domain, so the rule flags the form; the undisputed analyzer rule sinks it, though the fit table counts the analyzer partial. Three studies set limits: Burton, Lauridsen, and Obel (2002), a quantitative study of 224 Danish small and medium-sized firms, tested an earlier edition's rules only as a set and found lower return on assets in firms with misfits; Drazin and Van de Ven (1985), a quantitative study of 629 work units in two states' employment security agencies, found the interaction form of fit (one factor, one structure) unsupported and the selection form, the explorer's, supported; and Volberda and colleagues (2012), a survey of 3,259 respondents in 1,904 companies, found contingency and institutional fit complementary, though firms with perfect institutional fit slightly lost performance striving for contingency fit.

What the research found

The deepest evidence concerns mirroring, the claim that organizational ties follow the work's technical dependencies. Colfer and Baldwin (2016), a review of 142 empirical studies, found it prevalent but not universal: firms break it with modular partitions inside or relational contracts across their boundaries, and open collaborative projects did not support it. MacCormack, Baldwin, and Rusnak (2012), a matched comparison of five pairs of software products, found the loosely coupled organization's product more modular in every pair, comparing open source communities with firms, not firm networks. Cabigiosu and Camuffo (2012), a study of 100 supply relationships at three firms, found the mirror holding under stable architecture, and with Furlan (2014), in the same industry, breaking under technological change. Hoetker (2006), a quantitative study of many firms' sourcing decisions in one industry, found modular products make organizations reconfigurable sooner than they let firms move activities out.

Replication is uneven: Campagnolo and Camuffo (2009), an industry-level study of 68 Italian manufacturing industries, replicated Schilling and Steensma's (2001) model with results differing significantly from the U.S. ones, not reported here since unchecked.

The hub-firm evidence is mostly single cases, the most recent Baumann and colleagues (2020), one case study. Ahmadjian and Lincoln (2001), several case studies by scholars, found keiretsu supply relations drifting toward arm's-length contracting and top-down administration.

The schools disagree about what the form is. Powell (1990) calls networks distinct, Williamson (1991, conceptual) a hybrid between market and hierarchy; business-group writers see a form the choice between firm and market leaves out (Granovetter 1995). Granovetter (1985) argued that transaction cost reasoning undersocializes economic action, and Ghoshal and Moran (1996) that its prescriptions are likely wrong for managers, since organizations govern some activities by a logic of their own. Resource dependence explains boundaries by power, transaction cost economics by efficiency (Pfeffer and Salancik 1978; Santos and Eisenhardt 2005, a conceptual review); Bidwell's case finds internal politics shaping one firm's boundary. Whitford and Zirpoli (2016), one longitudinal case study of Fiat Auto's network after radical outsourcing, read the network firm as a political coalition, its politics and shifting boundaries shaping each other. Diefenbach and Sillince (2011, conceptual) argue network organizations are much less hierarchy-free than their label suggests.

The evidence ledger holds 24 studies filed under this form.

In practice: one published case

Toyota and its supplier network, as Jeffrey Dyer and Kentaro Nobeoka (2000) describe them, one case study.

Toyota created and managed knowledge-sharing processes at the level of the whole network. It gave the network a strong identity, set rules for participation and entry, and treated production knowledge as the property of the network. The network was highly interconnected, with strong ties, and built institutionalized routines that move knowledge among suppliers in many directions.

What happened: the network solved three dilemmas of sharing knowledge. It motivated members to take part and share valuable knowledge openly while preventing unwanted spillovers to competitors, it prevented free riders, and it lowered the cost of finding and reaching valuable knowledge. Suppliers learned more quickly after joining, and the network's knowledge-sharing processes at least partly explain the productivity advantages of Toyota and its suppliers.

The authors concluded that the idea of a dynamic learning capability needs to reach beyond the firm's boundaries, and that a network with a strong identity and coordinating rules can be better than a single firm at creating and recombining knowledge.

A caution: the network was chosen because it performed well, and the comparison with other automakers' networks rests on earlier research.

Source: Dyer and Nobeoka 2000. Creating and managing a high-performance knowledge-sharing network: the Toyota case, Strategic Management Journal 21(3), 345 to 367; one case study of a supplier network.

Variants and near relatives

Internal, stable, and dynamic networks. Snow, Miles, and Coleman (1992), illustrated, not tested: one firm owning most or all of the assets; vendors around a large core firm; and a lead firm brokering assets owned by others. Only the last two take activities out of the firm; owned units trading at prices are the Internal market organization's move. The broker assembles firms, where project-based's latent organization draws individuals and teams into recurrent projects.

Modular organization. Standardized interfaces embed coordination among loosely coupled units (Sanchez and Mahoney 1996), carried across firm lines into contract manufacturing, alternative work arrangements, and alliances (Schilling and Steensma 2001); dividing the workforce by employment relationship is the shamrock's move, and the recombinant (modular) multidivisional form moves charters among one firm's divisions.

Virtual corporation. Davidow and Malone (1992), a practitioner book: an ever-varying cluster of activities in a fabric of relationships, made possible by information technology. Lucas's (1996) T-form joins it for its virtual components (its leveling belongs to machine bureaucracy's infinitely flat organization and to algorithmic management), as does the virtual type of Lucas and Baroudi (1994).

Platform orchestration. A firm runs an industry platform for outside innovators (Gawer and Cusumano 2014; Jacobides, Cennamo, and Gawer 2018).

Industrial district, meta-organization, hollow state, and modes of network governance cross legally independent organizations and, like ecosystems, strategic alliances, joint ventures, and franchising, are named, not covered: Becattini's (1990, reprinted 2004) small firms bound by shared norms; Ahrne and Brunsson's (2008) organizations whose members are organizations; Milward and Provan's (2000) state delivering services through contracted third parties; Provan and Kenis's (2008) three governance modes, offered as propositions, two without a lead organization.

ARPA model. Azoulay and colleagues (2019): a flat agency of term-limited program managers funding only outside research, each program a bounded project (the project-based entry), where an adhocracy keeps its experts inside; its authors call empirical evaluation elusive.

Spider's web organization. Quinn, Anderson, and Finkelstein (1996): many nodes linked to many with no order-giving center, unlike machine bureaucracy's infinitely flat form.

Systems structure. Drucker (1974): relations-focused, each unit's objective derived from the whole's, NASA's program and its contractors his model, a very poor fit, he judges, with all his design specifications; the case bounded by one mission is project-based's.

Imaginary organization. Hedberg and colleagues (1997): a leader enterprise treating others' resources as its own, which its authors also offer as a perspective on any enterprise.

The Cellular organization is the nearest neighbor: its cells share one ownership and divide as they grow; the network's partners are separate firms. Two cellular variants sit at the boundary: Blade.org, a designed community of independent firms, and Quinn's starburst, units spun off under some shared ownership. Shared knowledge alone does not separate them: Toyota treated production knowledge as the network's property. Toyota's network, the case here, is also a keiretsu (Ahmadjian and Lincoln 2001): its knowledge-sharing rules are this form's move, and the cross-shareholdings binding a keiretsu are the business group's.

The network beside its nearest neighbor, the cellular form

Select a number to fill what it marks in both drawings; the list under them gives the dimensions on which the two profiles differ.

It takes activities out of the firm altogether and coordinates them by contract, standardized interfaces, and relationship, where the cellular form keeps its self-organizing cells under shared ownership.

Network, modular, and virtual organization

The firmPartner firmsHub or brokerfirmActivity itdoes bestDesign firmManufacturerDistributorSplit by value-chain stepOM11123
The firmPartner firmsBrokerCoreDesignMakerDistributeSplit by value-chain stepOM11123

Contract or priced tradeAdjust to each other directlyMMutual adjustmentOStandardization of outputsThe firm; partners stand outside it

Cellular organization

CellsCellCellCellNew cell,split offBy product or purposeMN11123
CellsCellCellCellNew cellBy product or purposeMN11123

Adjust to each other directlyMMutual adjustmentNStandardization of norms

  1. The network's partners are separate firms outside it, held by contract; the cells sit inside one shared ownership.
  2. Standardized outputs coordinate the network's partners, while shared norms hold across the whole cellular form.
  3. Often a hub or broker decides over the network as a whole; no unit heads the cells, which grow by splitting.

Differs on 6 of 11 dimensions

  • Basis of grouping Function or process; Cellular organization: Product or purpose
  • Specialization High; Cellular organization: Low
  • Standardization and formalization Medium; Cellular organization: Low
  • Decision rights and the locus of authority Limited vertical decentralization to selective vertical and horizontal decentralization; Cellular organization: Selective vertical and horizontal decentralization to vertical and horizontal decentralization
  • Prime coordinating mechanism Mutual adjustment and standardization of outputs; Cellular organization: Mutual adjustment and standardization of norms
  • Mode of exchange and control Market; Cellular organization: Market and clan

The network and the cellular form are both flat, leave decisions to their units, and have those units adjust to one another directly. The network ties separate firms by contract; the cellular form keeps its cells under one shared ownership.

Source: Miles and Snow 1986; Sanchez and Mahoney 1996; Powell 1990; Miles and colleagues 1997, a conceptual article with illustrations; Clarke 1994.

Moving into or out of it

Moving in: Lorenzoni and Lipparini (1999), a longitudinal study of three firms, found managers over time building a specialized supplier network and a narrower set of core competencies. Bidwell (2012), one case study of a firm's outsourcing decisions, found the consequences transaction cost and capabilities theories expect (less authority, weaker controls, and different capabilities), but managers judged them by their own interests.

Moving out: Toyota took a controlling interest in Daihatsu and built its own electronics capability to depend less on Denso, moving those supplies toward hierarchy (Ahmadjian and Lincoln 2001, several case studies by scholars). Running the reorganization falls to Change & Transformation.

The thinking behind it

  • Raymond Miles and Charles Snow (1986; 1992, with Coleman): the dynamic network, which explains how market substitutes replace administrative mechanisms; a forecast, untested.
  • Oliver Williamson (1975; 1991, conceptual): when hierarchy replaces the market, and the hybrid as a third form, which explain why the hub makes some things and buys others.
  • Walter Powell (1990) and Mark Granovetter (1985), conceptual: neither market nor hierarchy, and embeddedness, which explain the contested mode of exchange and why trust coordinates.
  • Ron Sanchez (1996, with Mahoney) and Carliss Baldwin (2016, with Colfer, first author): interfaces embed coordination (conceptual) and ties mirror the work (a review of 142 studies; open collaborative projects did not support it), which explain when contracts can replace close coordination.

Also called

Network organization, modular organization, virtual organization, and hollow corporation. The sources add dynamic network, interorganizational network, strategic network, network forms of organization, and modular organizational forms.

Works cited

On this page, “read” means read by this series: where a line says a work, or part of one, was not read, the series did not read it.

Back to the explorer