The Network Firm as a Political Coalition

Theory and contribution

Our proposed model of the network firm as a political coalition retains key elements of March’s original Carnegie School formulation, including especially the ideas that the executive is usefully seen as a political broker, and that the composition and goals of the firm are more bargained than given. Our findings are thus easily to be put into dialogue with and compared to competing approaches to the study of organizational networks and organizational politics. The key novelty is that we marry those elements to an alternative theory of coalitional formation and bargaining that “cross-fertilizes” concepts and mechanisms between analyses of organizations and of social movements, and that thus does not rely on the same simplifying assumptions characteristic of those studies to date. We endeavor, specifically, to explain the mobilization of power and other resources in the pursuit of interest—the natural hallmark of political analysis—without rendering those interests as “fixed in the way conceived by rational choice models” (Kaplan, 2008, p. 744). Instead, we trace how relevant interests—and consequently the range of viable coalitions—are shaped, mediated, and sometimes even transformed by: (i) the frames actors use to “cast issues in particular light and [to] suggest possible ways to respond to those issues” (see also Benford & Snow, 2000; Campbell, 2005, p. 49); (ii) as well as by actors’ embedding in specific social, organizational, and institutional contexts (Campbell, 2005, p. 49).

Frames are conventionally defined as the “metaphors, symbols, and cognitive cues actors use to cast issues in particular light and [to] suggest possible ways to respond to those issues” (Campbell, 2005, p. 48). They have diagnostic and prognostic components, where the former identify problems and assign blame while the latter link problems to claimed solutions. They tend empirically to be “fabricated out of bits and pieces of already available repertoires and cultural artifacts” (Campbell, 2005, p. 53)—which, in our case, means in bricolage from resources circulating in and around the world of technical and financial professionals, managerial consultants, and the like (Barley & Kunda, 1992; MacDuffie, 2013). Thus construed, frames are “schemata of interpretation” (Goffman, 1974, p. 21) all but necessarily deployed when actors find themselves forced to “make sense of ambiguous signals from the environment” (Kaplan, 2008, p. 732). They have been shown heterogeneous not just between industries and firms but within them (Engler, 2015; Kaplan, 2008; Schulze, MacDuffie, & Täube, 2015). They shape—but do not uniquely determine—“how individual actors see the world and perceive their own interests” both because a multiplicity of frames tend normally to be in contention and because they are deployed in social milieu in which people regularly “suggest what is going on to others” (Kaplan, 2008, p. 732).

Kaplan (2008) has in prior art grafted a theory of framing into a political/coalitional model of organizational strategy-making. She observes that organizational actors regularly “attempt to transform their own cognitive frames into the organization’s predominant collective frames” and that “framing contests” therefore arise in organizations when the frames actors deploy prove incongruent. And she shows that those contests turn on the relative ability of organizational actors “to establish the legitimacy of their frames and of themselves as claimsmakers or to realign frames to influence how others see issues” (Kaplan 2008, p. 730). This approach, though, offers us only a partial solution to our problem as it is focused entirely on goings-on internal to a single and well-bounded organization. Our own case, by contrast, is populated by actors who are for the most part “multiply embedded.” We argue across our case narrative that participants to important framing contests include more than just persons formally employed by Fiat proper who are embedded both in their own “internal” corporate networks and in the “external” networks of “the host milieu[x]” in a variety of territories (Becker-Ritterspach & Dörrenbächer, 2009, p. 209). Relevant personnel in that external network are employed by suppliers embedded in the internal networks of their own employers, some of which are themselves subsidiaries of powerful multinational firms riven by their own political fights.

Again, prior art gives guidance. Becker-Ritterspach and Dörrenbächer (2009, p. 204) observe that “studying politics of intrafirm competition [in the contemporary MNC] involves … asking who are the relevant and key actors and how their relevance is constituted.” There are, also, studies of how MNC subsidiaries “balance ‘internal’ embeddedness within the [MNC] network with their ‘external’ embeddedness in the host milieu” (e.g., Meyer, Mudambi, & Narula, 2011, p. 235). But again, that guidance has limits. Geppert and Dörrenbächer (2014, p. 231) are right to observe that partisans of a business network approach have yet to explore in depth “how power is socially enacted and how political maneuvering is grounded in micro-level interactions” across networks. Our proposed solution to this problem—and so too to our need for a window into the ways in which “framing practices” in the network organization affect and are affected by goings-on that run not just within but also across the boundaries of particular organizations—sits in the attention we give in our analysis to actors’ “relational” embedding, to the ways in which variation in social and business ties affect actors’ knowledge of the motives, trustworthiness, and capabilities of others (Granovetter, 1985; Krippner & Alvarez, 2007; Uzzi, 1996). That solution, we argue, allows us to identify “relevant actors” and to explain when and why particular frames do or do not resonate with consequence across the vertical network forms of organization that are our object of study.

Our claimed contribution in this article consists therefore in a studied reassembly of existing constructs in application to a kind of case otherwise outside the useful scope of more standard analyses of organizational politics. We document the utility of the model in two main ways, both regularly identified as standards for good theory-building in case study research on organizations (Bacharach, 1989; Eisenhardt & Graebner, 2007). The first is principally undertaken in the next section, where we use the model—to draw on an oft-used metaphor—as a “lens” to show: (i) how the conceptual imagery of the political coalition, coupled with an analysis of the ways in which those contests are shaped by patterns of relational embedding, allows us to distill an exceedingly complex empirical reality into a plausible and potentially falsifiable narrative account of the evolution of the particular network that revolves about Fiat; while (ii) confirming that our motive constructs are also general enough to “[serve] the investigator as a measuring rod to ascertain similarities as well as deviations” between our case and others (Brady & Davies, 2004; Coser & Merton, 1971, p. 223). We show also that our model identifies some of the reasons and ways frames and practices travel—or fail to travel—within and across the multinational vertical network form that is our object of study (see, e.g., Kostova, 1999; Kostova & Roth, 2002, on the transnational transfer of organizational practice).

The second “way” is taken up most explicitly, but not exclusively, in the discussion and conclusion. We show that our approach in fact explains the evolution of the network in question better than do alternative approaches that either offer an alternative view of organizational politics, or do not focus on organizational politics per se but presume broadly to explain the evolution of vertical network forms of organization. These include “managerial” accounts less convinced that cases like ours have the “character of polities whose ‘citizens’ may engage in collective action” (Davis & Zald, 2005, p. 335) and so give causal primacy to the changing qualities and (relatively autonomous) decisions of leadership teams (e.g., Clark, 2011; Smith & Tushman, 2005). There are also approaches that retain the essence of the simplifying assumptions discussed above by giving causal primacy to dynamics at the macro- or field level in accounts that, for example, highlight the salience of “institutional mimesis” (i.e., Garcia-Pont & Nohria, 2002) and/or—in an alternative political approach—predict that broader shifts in conceptions of control should best explain changes in “corporate profit and accumulation strategies” (Dunford, 2009, p. 145; Fligstein, 1990, 2001).

A first narrative moment: frame divergence

The questions we asked in our early interviews explored the reasons Fiat outsourced so radically (in keeping with our research question at the time). Answers fit in some ways with the claims of those who give causal primacy to “field-level” concepts to explain organizational behavior, in the sense that many in the company had embraced a set of ideas that were much discussed in the business and managerial press and that had diffused rapidly across the industry in those years. ³ Those ideas have since been identified by MacDuffie (2008, 2013) as a distinctive and powerful cognitive frame (“modularity-as-frame”) rooted in three core beliefs. ⁴ Those beliefs—which play heavily across the rest of our account—hold that design rules can minimize interdependencies between groups of components (“modules”) thereby enabling producers simultaneously to: (i) generate flexibility through “rapid innovation at the level of modules, each proceeding independently and at its own pace”; (ii) reduce “costs, from standardization of interfaces that minimize coordination costs during product development, plus standards for modules that generate a larger number of potential suppliers and allow switching among suppliers more frequently”; and (iii) as a result more easily rely on assembly plants dispersed globally without, however, decentralizing R&D or design (Baldwin & Clark, 2000; MacDuffie, 2008, p. 9; MacDuffie, 2013).

The centrality at Fiat of this family of frames is manifest in a 1998 article by Roberto Testore (1998, p. 3)—then Fiat CEO—describing the company’s “178 world car project” as one “destined to beat all previous output records in the company’s history” precisely by using modular concepts in “co-design” with “world class” suppliers. And while such public pronouncements are of course highly strategic and must be read as such, it was clear from other interviews and internal documents that the company in fact made very substantial changes to both structure and organizational routine in light of that frame. It, or its manifestations in practice, surfaced regularly in conversations not just about the 178, but about other models as well. For instance, a top Fiat manager told us that Fiat’s main concern at the time was “to reduce [their] assets and overall development costs and, at the same time, leverage external sources of innovation.” Modularity, in this context, aimed to allow Fiat to retain decision-making, engineering, and design responsibility at its Turin headquarters even as manufacturing capacity would decentralize both globally and within Italy (further in the latter than it already had; see Whitford and Enrietti (2005) for a discussion of the company’s prior growth in Italy’s south).

The initial resonance of a modularity frame internal to a decentralizing network helmsman is easily understood. Fiat had lost market share in Italy—from 60% to 45%—between 1988 and 1992. Future demand in Fiat’s core markets was projected to grow slowly, while the advent of global benchmarking surveys made clear that the company’s costs were in fact high, its quality standards low, and its technology lagging. A global revolution in car-making in response to the Japanese “invasion” was ongoing. And so it was easy to find enough overlap in the “prognostic” side to align the interests of a broad coalition in support of the automaker’s coupling of a “world car” approach with a heavy reliance on modular concepts in the design of the company’s new models. A finance division charged with managing investors expected to be able to push risk onto suppliers. The purchasing division hoped to grow and acquire centrality. ⁵ Marketers could hope to target new niches by mixing-and-matching modules. Technical divisions got the promise of help from suppliers then investing heavily in their own technical infrastructure. Subsidiaries—then just foothold facilities in Poland, Argentina, and Brazil—got new investment in assembly facilities and promises of local R&D facilities going forward (Balcet & Enrietti, 2002). And while there was some opposition from unions, there was no real dispute that the company needed to get more competitive (Negrelli, 2004). ⁶

Our interviews revealed ample support across the automaker for the idea—summarized in Testore’s (as CEO) 1998 writings—that Fiat would make a “a major effort to support [key suppliers] in their development [and would] seek to promote a concept of (co-design and partnership).” This was consistent with the spread of “process-based” understanding of inter-organizational trust into actors’ frames across understandings of the drivers of Japanese manufacturing competitiveness. It also fit with reporting at the time in the academic and business press as in, for instance, the findings in Dyer and Chu’s (2000) cross-national study of the determinants and effects of trust in supplier–automaker relationships. There was, for instance, near consensus among interviewees at Fiat that the automaker would be needing new and more systematic “routines for selecting suppliers” and would require the implementation going forward of more institutionalized “processes for responding to supplier problems” (Dyer & Chu, 2000, p. 263). However, even within frames broadly supportive of Fiat’s embrace of modularity there were some differences which, though seemingly largely questions of emphasis, would give rise to a series of ultimately quite consequential framing contests.

Two main coalitions emerged, each concerned that reductions in Fiat’s technical divisions would leave the automaker too dependent on supplier know-how to directly assess tradeoffs between different combinations of cost and performance. Neither set themselves outside the broad logic of modularity as then understood, but their frames did highlight different elements of that logic. One gave particular diagnostic weight to a risk—a “modularity trap”—that had been identified in the business press at the time (Chesbrough & Kusunoki, 2001) and worried that the standardization of interfaces and componentry so constitutive to modularity would leave Fiat ill-positioned to incorporate novel but more “integral” technologies into its vehicles. Its core ideologues sat in a relatively small group in the technical division—around 120 persons—called the “Components Development Platform” (CDP) charged with mitigating those traps by developing underlying engineering “concepts” (a class of solutions to engineering problems) flexible enough to satisfy multiple scenarios. Reductions in Fiat’s asset base meant that the division worked closely with panels of suppliers, generally deploying a prognostic frame that sought flexibility. This was to be got through an underlying “conceptual” regularity across engineering solutions that could speed product development by shaping the ways in which design teams would standardize across models and that would sway suppliers to collaborate in the development of those components by offering a first-mover advantage in future bidding for the production of subsequent iterations as well.

The main opposing coalition was grounded principally—though not exclusively—in the purchasing division. Its partisans drew on a standardization frame that did not so much deny modularity traps as give more weight to alternative risks. They focused in particular on Fiat’s links to a historic supply base that was famously flexible but that had been given no reason by Fiat (in a near monopsony) to conduct independent research, to develop design competencies, or to export substantially (let alone develop a global footprint). Their prognosis drew in part on lessons gleaned globally from the translation of practices from Japanese automakers across the industry. They could not replace their suppliers overnight, nor did they wish to cede privileged access to suppliers that were, at the least, flexible and cost competitive. So, though undermining their own monopsony position, they began to help once-captive suppliers to diversify by pursuing contracts with some of Fiat’s competitors which—they hoped—would generate technological learning from which they too would benefit (Whitford & Enrietti, 2005). They forced some to internationalize for the first time by providing supply contracts for the 178 contingent on co-investment abroad (Whitford & Enrietti, 2005; Zirpoli & Caputo, 2002). And, to support others, they offered subsidized training programs for a selected core group of Italian suppliers (Whitford & Enrietti, 2005).

Concurrently, the purchasing division encouraged foreign MNCs to place subsidiaries in Italy, often by brokering their purchase of existing Italian suppliers. They did so, in the words of a high-level purchasing manager, because “foreign multinational suppliers had developed more advanced R&D structures due to the demanding technology level required by their local customers.” They were generally successful—Balcet, Ferlaino, and Lanzetti (1999) identify a rapid increase in the number of multinational auto suppliers in the region between 1990 and 1995 and show that those multinationals grew more rapidly than the sector as a whole—but they were also concerned that they had minimal leverage with their new multinational partners. They hoped this would change once the rollout of the “world car” project and some highly modular models then in the works for European launch bore fruit. In the meantime, Fiat’s purchasing division elected to “hybridize” (in the words of an interviewee) principles borrowed from Japanese producers by maintaining, when possible, multiple potential suppliers and pitting them against each other in “controlled” (rather than open) competition (Zirpoli & Caputo, 2002); and they frequently demanded, as a condition of contract, that Fiat’s first-tier and system suppliers work with second-tier suppliers selected by the automaker.

Neither practice was per se unique to Fiat. The American assemblers—with the exception of Chrysler—had similarly “hybridized” lessons drawn from Japan (Dyer, 2000; Mudambi & Helper, 1998). The imposition of a working relationship with a lower-tier supplier was common even among Japanese producers, where it has been used to increase “carry-over” of parts between models (Choi & Hong, 2002). But their joint deployment by Fiat’s purchasing department was distinctively marked by the mutual embedding of Fiat and supplier personnel in the Piedmont region. For instance, a supplier we interviewed argued (derisively) that Fiat’s imposition of second-tier suppliers followed a logic of industrial “eugenics.” Fiat had eschewed a “natural selection,” by which he meant that Fiat did not just “leave everyone to do their job until they lose faith in him [for nonperformance].” The purchasing division and their allies had instead managed their fear of collusion among first-tier global suppliers by guaranteeing the survival of suppliers with whom they had very strong trusting relationships but who might otherwise have found it difficult to remain in the market (hence “eugenics”). Ties into those suppliers gave Fiat personnel insight into the otherwise opaque cost structures and strategies of system and module suppliers, as well as into the technologies, margins, and bargaining strategies of suppliers considered potentially adversarial.

This combination was effective on its own terms, in the sense that it helped the purchasing department both to limit the cost of components and systems (at least in the short-to-medium term) and to enlist an ally in a marketing department eager to introduce new functionalities (since they expected modularity to ease system integration and competitive bidding to ensure good pricing). Its artificers were consequently able to highlight the correspondence between diagnosis made and prognosis to give “resonance” to a frame that emphasized the link between modularity and scale economies and that coveted a heightened ability to threaten exit in bargaining (Benford & Snow, 2000, p. 619; MacDuffie, 2013). That same resonance, however, disfavored engineers in the Component Development Platform and others whose diagnostic frame emphasized instead the dangers of modularity traps. Their efforts to develop new ties with the technical centers of suppliers were thwarted with regularity by a dominant coalition too willing—in their view—to embrace technological solutions inconsistent with the engineering concepts the CDP had developed. ⁷ An engineer from the CDP said, for instance:

We were never sure that the platforms [would] then use our engineering solutions. Sometimes we go on for years and the engineering solutions we find never find a space in vehicle developments. It is not a matter of what engineering solution better suits the new product, it is a matter of who has the right to decide on that. … The power equilibrium was in favor of the purchasing department whose task was to lower the price of supply while improving the quality standards of the component purchased.

A second narrative moment: balkanization

By 1998 Fiat was a company transformed, a Fordist behemoth turned into one of the least vertically integrated automakers in the world across the space of just a decade. That year, though, brought a slate of “ambiguous signals from the environment” (Kaplan, 2008, p. 732) that would bring competing coalitions to a head. Fiat’s European market share had been roughly stable through the 1990s but had fallen nearly a quarter between 1997 and 2000, pushing the company’s balance sheet slightly into the red already by 1998. That decline was (almost by definition) associated with the failure of models released in the middle 1990s to meet sales targets. Those models had, however, primarily just been restylings, and the jury was still out on models more fully shaped by organizational routines more reflective of the automaker’s embrace of a logic of modularity. The 178 in particular had only just come on line and had delivered relatively promising sales in Brazil in 1997. And in Europe, heavy investments had been made in a model—the Stilo—set to release in 2000 with an ostensibly promising bevy of options and technologies at a price point below that of the company’s competitors.

This all placed Fiat’s leadership team—or, in Marchian terms, the political broker—under pressure. Their response reflected the sway of the coalition mobilized around what we referred to in the previous section as a standardization frame. The influence of that frame on the company’s strategy is manifest already in Fiat’s 1999 annual report (released June 2000), which highlights “efforts made in the area of product innovation and the aggressive implementation of cost containment measures” and outlines the logic of the company’s decision to sign a strategic alliance with General Motors in March 2000. That alliance, though backed by an exchange of stock and by a “put” option for Fiat, was at its heart an industrial agreement to generate “synergies” that would “amount to two billion Euros by 2005, half of which [would] benefit Fiat Auto directly” (Fiat Group, 2000, p. 9). Those synergies were to be got across Europe and South America only (the alliance excluded the Asia-Pacific area and North America) across two 50/50 joint ventures in purchasing and in powertrain design and manufacturing (which covered between them some 80% of total vehicle manufacturing costs); and from a commitment to develop a series of shared platforms promising to spread fixed costs “over more than 5.5 million cars, more than twice Fiat’s current volume” (Fiat Group, 2000, p. 9).

Fiat also moved the now-shared (with GM) purchasing division to Germany—though most staff remained where they had been—and hired a specialized automotive industry consulting company to analyze each company’s component lists. Those consultants, acting in the name of the joint venture with business cards to match, then rebid components in an effort to use market power to drive down prices. This cut against lower-level buyers’ ability to play personal favorites, therefore shortcircuiting precisely those ties that had been at the base of a “eugenic” approach on the grounds that they would no longer be needed given that higher-level buyers would—with GM’s added volumes—have far more leverage. And, in the hopes of forcing engineers to standardize more components across models and platforms, the Component Development Platform was disbanded on grounds that so many of its engineering solutions had remained on the shelf. Its personnel were dispersed as part of a larger reorganization of the technical division. Its functions—to the extent they were replaced—were folded into an empowered and centralized technical function that, as a 2002 internal presentation made clear, was to be the centerpiece of a structure “born to leverage economies of scale in engineering and design.”

The reorganized technical division—according to that same Fiat’s internal presentation—was to become a “technology content provider” to seven different “product development platforms” which would rely then on relatively standardized “off-the-shelf” technologies mixed and matched across platforms and models. Each platform was assigned to project teams under the control of a lead manager who would direct “vehicle line managers” responsible for the management of performance tradeoffs that might come up in the integration of the many systems and components that make up the different vehicles derived from the different platforms. In aggregate, this structure openly resembled Toyota’s reliance on “heavyweight” product managers (Wheelwright & Clark, 1992). Its logic is captured in the retrospective words of an interviewee from the technology division:

The idea of the system integrator was coupled with the idea of modularity, and interpreted in the sense that the system integrator should have the competence of integrating systems as its core competence. We thought we could be substantially detached from certain component and systems technologies and focus primarily on “architectural” know-how. For this reason we delegated to suppliers the complete development of certain technologies, those that we did not consider as core.

And yet, this vision never quite manifested in routines and practice across the network. Fiat’s “heavyweight” product managers found—to quote one—that their “right to make decisions mattered little without the information and know-how needed to make those decisions well.” Though they could turn to “competence centers” in the centralized technical division, Fiat’s near total delegation of direct component design had led too many personnel expert in those technologies to move, often to employment with the suppliers who had taken on component design responsibilities. The technical divisions, another explained, consequently lost “the ability to manage performance tradeoffs” on their own and therefore to “set performance targets for suppliers … [or to] monitor [their] work.” ⁸ Relational embedding that might have served as an alternative had eroded due to some combination of the joint venture with GM and (again) the “eugenic” approach taken in years just past. The company’s “heavyweight” product managers were isolated as a result in teams populated largely by engineers formally employed by suppliers (though often co-located in Turin). And, in the complaint of one member of those teams, their task demanded the skills of a “magician. Everything was under the control of the manager responsible for the vehicle. It was just impossible to cope with all the aspects of product development, from technical to economic issues.”

Managers and engineers associated with the platform teams responded with a frame that directed the blame away from themselves and towards Fiat’s technical division’s inability to help them manage tradeoffs that had not, in fact, been ironed out in the “pre-development” phase (as was supposed to happen). This led them to ignore a centralized engineering division that, in the words of another Fiat manager, felt to them like “just another engineering supplier that by accident was part of the [same] firm, nothing more, nothing less.” Instead, they mobilized pragmatic local alliances of shared interest with the suppliers they relied on to get their projects out the door. Those suppliers were for their own part unhappy with the Fiat-GM purchasing alliance that, a sales engineer at a global supplier with a longstanding presence in Turin said, had “created not a few problems in Fiat’s supply base, not just because they lost business to GM suppliers but because [GM’s] attitude towards suppliers was one of absolute arrogance, absolute inhumanity, lack of respect on all levels.”

Suppliers continued to engage with Fiat—many saw their options as constrained given their location in Italy and the absence of other assemblers nearby—but they did so across a balkanized collection of platform teams that were themselves only weakly connected to the decision-making center. This further undermined the effectiveness of a purchasing alliance premised on standardization across models and consequent scale. The problem was not that suppliers had no interest in scale; profits for suppliers in the auto industry are generally earned more on the sale of parts than on development. But the structure of working relationships had constrained the development of an alternative coalition. The effects—over time—on Fiat’s products and profitability were profound. A top Fiat technology executive told us in 2006 that “each platform had [in years previously] tended to develop its ‘own’ cars with very few concerns about component and platform sharing.” And, speaking retrospectively, he told us:

When I arrived [in 2005], I reverse engineered many Fiat cars belonging to the same and different segments. Nobody would have said, just by looking at the components and how they were engineered, that the two cars were produced by the same firm. Due to confidentiality reasons, I cannot tell you the exact figures, but we can make a lot of money just by bringing components commonality to a decent level [that is, by somehow doing what the alliance with GM and consequent reorganization was supposed to do but had not]. ⁹

A third narrative moment: counter-mobilization

The depth of Fiat’s difficulties exploded into the public gaze in 2002. The 178 world car project had foundered. Projected to sell a million units across Brazil, Argentina, Turkey, and India, it sold only a third of that in 2001 (the model never sold more than 360,000 units in a year). The Stilo had come out, late, that same year and had also sold just a third of its target. The failure of the alliance with GM to deliver cost savings—due, again, in no small part to the failure to effectively standardize components—had filtered to the balance sheet and, combined with plummeting sales, had led debt to balloon. The risk premium on bonds grew, raising capital costs that led investors to demand that Fiat sell off high-value assets—like the Ferrari and Alfa Romeo brands—amid fears it would not survive. Things were bad enough that a 2003 article in The Economist declared that “even Italians” were spurning Fiat’s cars and that “the firm’s third rescue plan in two years [was] unlikely to be enough” to “avoid going bust.” ¹⁰ Yet between 2003 and 2007, the company’s market share in Italy grew by 13%, global sales by 28% and, most importantly, an annual loss of nearly 1.5 billion euros in 2003 had been turned by 2007 into a profit of more than 800 million.

Popular accounts of that turnaround—of which there are many in the business press (e.g., Clark, 2011)—highlight the 2005 arrival of a new top management team. And while our own arguments do not imply that changes at the top did not matter (a question to which we return in conclusion) our interviews underscore that the contours of the company’s post-crisis evolution are theoretically continuous with its pre-crisis troubles. We have to this point shown that framing contests that might once have remained internal to Fiat had spilled across organizational boundaries. We have emphasized that a particular framing of modularity into Fiat’s organizational structure had interacted with pre-existing patterns of relational embeddedness to balkanize the network into platform teams generating: (i) a shared diagnostic frame critical of the loss of technical know-how in the technical division; and (ii) a prognostic frame that privileged stopgap measures to get product out the door. This had pushed those actually designing vehicles to mobilize support from the supplier engineers who—for reasons endogenous to Fiat’s decision to rely so heavily on them in the first place—were highly competent in the component-specific technologies those teams needed to bring their projects to (relative) fruition.

In this sub-section, we show how local mobilization in liminal spaces in the network, in the places where engineers found themselves endlessly searching for workarounds due to coordination failures, rippled back to the center. The stories we heard were many. Few were significant by themselves, but taken together they evidenced an alternative prognostic frame that would shape and be shaped by a project we will call, pseudonymously, “X.” ¹¹ That project—led by a Vehicle Line Executive of the sort expected to be a “magician”—was marked by local reaction to the hollowing out of competencies in Fiat’s technical division. The team made the decision to take responsibility for the design of far more parts than was typical at Fiat at the time, hoping to learn about interactions between components and systems in order both (i) to react to problems that emerged as the model went into production and (ii) to develop “derivative” models more quickly and at lower cost. The team leader described it as a “big risk because, sincerely, to do it on such an important project was maybe a little bit crazy,” all the more so since they had taken on the design of components like the dashboard that fundamentally affect consumers’ perceptions of vehicle quality, that interact with other very important systems (i.e., safety, HVAC), and that rely on a technology—injection molding—in which Fiat had “over the years lost so many specialists.”

She said, however, that those specialists were not so much gone from the network as gone from Fiat. They had “gone to work for suppliers at the moment of the outsourcing” and her team knew (through relational embedding) which suppliers—and even which engineers at those suppliers—had the expertise required. They were also able to enlist the support of internal embedded allies like a “supplier quality group somewhere between Purchasing [which was then captured by the joint venture with GM] and Quality” that could help them to understand concrete details of suppliers’ capabilities when relational embedding did not suffice. And where a more centralized structure had undermined the engineers of the “Component Development Platform”, from our first “moment”, by limiting their ability to make credible commitments to suppliers, the team in charge of Project X could exploit balkanization at Fiat to provide cooperative supplier engineers with insights into the (otherwise opaque) inner workings of components and systems that would in fact go into production. That insight would help those suppliers to win contracts to supply parts in volume if the project was a success (and it did in fact prove valuable as Project X gave birth to a model that came out on time, on budget, and that sold well).

Those sales were important in and of themselves, as the company was hemorrhaging cash. Project X was more broadly significant, however, for an additional reason. The team found that the learning garnered through more direct component design allowed them to coordinate suppliers’ engineering and design activities more effectively than when they eschewed direct component design and focused their energies just on system integration. This gave essential grist to subordinate groups internal to Fiat unhappy—as a top technology officer told us—that “engineers in the company had just delegated design and engineering to somebody else, usually outside the company.” The new diagnosis, he explained, ran interestingly opposite to frames that had been prevalent across the network just a few years previously: the company had made a mistake by investing just in “architectural” know-how since “you cannot integrate components’ performances you know very little about … If you have never designed a component or a system it is very difficult to understand the subtle interactions with the rest of the vehicle.” But, he continued, the seemingly obvious solution—“We [Fiat] should have reversed our strategy by integrating back competences that we had lost”—was off the table because, “We [Fiat] had two problems … no money and no time.”

Informed by the experience of Project X, this rising coalition drove Fiat to “quasi-insource.” By “quasi,” we mean that Fiat staff did not exactly eschew their heavy reliance on design outsourcing, but did involve themselves in a far broader array of direct component design. This logic was subsequently extended to projects in which Fiat developed “template” models in the hopes of then amortizing those relatively heavy investments in engineering solutions across a series of “derivative” models spun off more quickly and cheaply (see Zirpoli & Becker, 2009). This frame partially reflected the underlying logic of the Components Development Platform disbanded a few years earlier, so it was naturally accompanied by the (re)formation of an organizational unit tasked with transforming the company from a mere “integrator” to instead a “creator” of technologies. As before—though this time due as much to capital constraints as to framed commitment—the unit could neither hire many new personnel nor just expand out from their one remaining area of deep technical expertise (engine technologies) to related technologies in a classic insourcing strategy; they would necessarily rely heavily on supplier personnel located internal to and beholden to their own employers. But there was a key difference: engineers deployed both to that unit and to the platform teams were in a position to form a very different and much more robust set of alliances across the vertical network in question due to the unwinding of the Fiat-GM alliance in 2005.

Many accounts of that unwinding focus on Fiat’s extraction of 1.55 billion euros from GM in the divorce. There is little disputing that the cash was important, but for the longer-term evolution of the network (contingent on surviving) it is more significant that the purchasing division was also released from its poorly performing joint venture and could be co-opted by the rising faction. This was done at the top with the appointment of a high-level manager who had initially made his career on the engineering side of the company; and then in the ranks by dropping the “very rigid” procedures that had denied middle and lower-level staff “creativity” in the performance of their tasks (we are here using descriptors offered us in 2008 by a mid-level interviewee in that department). This allowed that rising faction that was focused on regaining control of key components and systems development (thanks to direct exposure to design and engineering of those components and systems in “template” projects) to make a big bet on a new sort of outsourcing that would leverage strong cross-cutting ties linking Fiat to engineering suppliers like Magna Steyr and Pininfarina (one tie was especially strong: Fiat’s new Chief Technical Officer was previously at Magna Steyr).

Engineering suppliers had historically served a role in the industry, helping assemblers to develop “niche” products with relatively small anticipated volumes. Fiat’s rising coalition bet that the combination of engineering solutions learned through heavy involvement in template models and deep ties into engineering suppliers would allow them to guide these latter in the development of a series of mass derivative models. This engendered new forms of blurring of the boundaries and command structures of organizational units as Fiat personnel were necessarily involved in “triangular” organizational relationships. It meant, specifically, that the responsibility for product performance sat with an engineering supplier but the long-term relationships with suppliers developing components would be managed by Fiat employees. Those relationships did not always run smoothly but did still allow Fiat to cut the lead time from 26 to 24 months for products using a new architecture and by as much as 15 months for derivative projects (these measures vary according to model segments). Compared with the Stilo, costs of a new model development based on a new architecture but in the same market segment were halved while costs for the development of derivative models were reduced by as much as two-thirds.

Epilogue: multinationalization mark II and a future yet to write

Most accounts of the events that led Fiat and Chrysler to ink an alliance in 2009 and to merge in 2014 reference the knock-on effects of the global financial crisis of 2008 and the contingent decisions of Barack Obama and his automotive task force as matchmakers. They describe a desperate Chrysler CEO who “wanted a small car so badly he flew to Italy in a body cast to see Sergio Marchionne” (Vlasic, 2011, p. 281), and lauded the latter as the “outsider to the car industry” whose “unfamiliarity with Fiat and the auto sector ended up giving him an edge” because it had given him the oblique vision he needed to cut deals that had once seemed unthinkable (Meichtry, 2009). The model we are developing here disputes neither the obvious significance of linked global crises that have forced changes across more than just the car industry, nor the acumen and boldness of specific decision-makers at specific times. It does, however, wrestle more concretely with the means by which relevant actors collectively “make sense of ambiguous signals from the environment” (Kaplan, 2008, p. 731) and, as a consequence, to develop a shared strategic response.

We have focused to this point on European players who were—or who at least became—central participants in the framing contests we have identified, highlighting “counter-mobilization” in the periphery of the production network. Here we note also that players outside that European core had not sat static in the interim. Consistent with Kostova and Roth’s (2002) finding that more dependent subsidiary units are less likely to implement organizational practices pushed on them by those parents, the failure of the 178 project to meet its ambitious sales goals had left personnel at subsidiaries to chafe at the expectation that they merely adapt imported technology to local conditions. Athreye, Tuncay-Celikel, and Ujjual (2014, p. 108) report, for instance, that Fiat’s subsidiaries in Brazil had sought to do more than just develop new models on the 178 platform—which was foreseen in the project itself, reflecting the logic of modularity—and had in fact undertaken “clear qualitative upgrade in terms of [their] development” competencies by exploiting their local (“external”) embedding in the host milieu. They had done so by exploiting “strong ties between the two Fiat subsidiaries in Brazil, and interactions with external suppliers [in ways that] allowed more autonomous product development” (Athreye et al., 2014, p. 108; Mudambi & Navarra, 2004). ¹²

Fiat’s project development organization in Brazil—in Betim—had numbered just 150 in the late 1990s when it had been expected simply to tweak platforms with an eye towards demand conditions in the developing world. That small unit, however, became a logical ally of Fiat’s Turinese technical divisions as these latter sought in the wake of Project X to “quasi-insource” on the cheap in collaboration with suppliers, and then to look for help developing derivatives on the basis of the template models. This was abetted, also, because Fiat’s European engineers were already working with suppliers that had their own subsidiaries in Brazil (i.e., they hoped to exploit a parallel “internal” embedding of MNC suppliers’ own corporate networks). Fiat’s technical divisions consequently directed some investments towards South America and tasked the subsidiary with the eventual development of a derivative model on the basis of the Fiat’s “mini” template. Betim’s headcount grew steadily as a consequence, tripling to 450 by 2005 and nearly doubling again (to 850) by 2008 with engineers, also, rotating more regularly through Turin (Ibusuki, 2011, p. 116). And, crucially for purposes here, the subsidiary had by late 2008 made considerable progress towards the 2010 launch of their “100% Brazilian” Fiat Novo Uno.

The project resonated in its success with the prognostic frame that had begun simultaneously to develop in the network helmsman’s European core. Fiat had proved able to rapidly remake its product development processes across a range of derivative models (including iconic ones such as the Fiat 500) and sites not just because the development of engineering “templates” had allowed the automaker to spin models out quickly, in variety, on budget and on schedule. It had also guided the formation of a series of inter-organizational routines across the formation of what we referred to above as “triangular” organizational relationships with complex reporting requirements. Those routines proved effective in the coordination of the development of derivative models across multiple and “cross-”embedded actors at Fiat, at engineering suppliers, and at component and system suppliers in Europe. Their successful and subsequent deployment internal to the MNC, but across local context, in Fiat’s South American operations helped provide Fiat leadership with the cognitive, material, and political resources to convince the core of a sustaining coalition—located not just in the technical divisions but also in key MNC suppliers—that an alliance with Chrysler offered more than just new access to a North American market abandoned by Fiat nearly three decades previously.

A full account of the sudden incorporation into the network in question of a vast range of newly relevant actors embedded in their own histories and relations runs beyond our scope. However, it is worth noting the words in 2011 of a key manager reflecting continuity in frame. Fiat, he said, was aiming to “restore dignity to the technical areas of the company” in reaction to Chrysler’s previous ownership, these “Wall Street types [where], you know, there was not an emphasis on putting resources against these types of activities”—where by “these types” he meant investments in technology and engineering. The “end of their [old Chrysler’s] story” was that they would just

conclude [their] design model and then … find suppliers and ask them to offer it. When you have completely given up the competence, you can’t even judge the feasibility. It is difficult to make a judgment on the offers you get in terms of development cost, investment.

So, perhaps unsurprisingly, there are obvious parallels in a dominant-at-time-of-writing prognostic frame that has seen Fiat-Chrysler’s Auburn Hills outpost first quasi-insource design in projects in ways strikingly similar to practice in Turin in the early 2000s, and second, place staff with close ties to the engineering division in leadership positions in the purchasing division and make quite deliberate efforts to repair frayed relationships with suppliers.