Collective Action at the Front-end of Complex Systems Projects: The Case of Planning Large Infrastructure Projects in the UK This study uses an organizational design perspective to illuminate the front-end of capital projects to develop complex socio-technical systems. The research is grounded on planning activity for four large infrastructure projects in the UK. By modelling the front-end with Design Structure Matrices, the analysis uncovers polycentric structures formed to govern large arenas of collective action. In these structures, the promoter formulates the grand problem but shares the authority to make interdependent strategic choices with independent actors. The findings trace regular slippages in the performance targets to the need to safeguard compromise solutions for emerging disputes rooted in differing interests and visioning. However, context determines the extent the accountability for the front-end outcomes is shared. The main contribution is a contingency model of performance that contains a relationship between the umpiring structure used to resolve internal disputes and the slack resources used to preserve project legitimacy in the public eye. Implications to practice and policy are drawn. Introduction The project front-end is a central concept to the literatures in the management of capital-intensive projects (Morris and Hough 1987, Morris 1994, Flyvbjerg et al. 2003, Dvir and Lechler 2004, Gil and Tether 2011) and in developing complex products and systems more generally (Hughes 1987, Miller et al. 1995, Hobday 1998, 2000, Gann and Salter 2000, Davies and Brady 2000).1 Both research strands cover the projectbased production of one-off, complex socio-technical systems including large infrastructure (e.g., transport, energy systems), defence systems, and IT infrastructure. The project front-end aims to resolve strategic choices and produce a performance baseline. This baseline sets the budget, timescale, and scope; the scope is embodied in drawings and design requirements (Cleland and King 1968). Crucially, the baseline informs the decision to sanction the project, and thus creates normative expectations in the eyes of third parties (Morris 1994). Enterprises that fail to conform to established practices and norms in the environment lose legitimacy to operate and acquire Other terms used to address the phenomena include ‘major projects’ (Morris and Hough 1987), ‘megaprojects’ (Szyliowicz and Goetz 1995), and ‘large engineering projects’ (Miller and Lessard 2000) 1 1 resources (Scott 1987, Suchman 1995). Hence, projects that struggle to meet the initial performance expectations are perceived to be ‘failing’ (Dvir and Lechler 2004). Numerous accounts of complex systems projects where performance baselines were way off the mark have fuelled two views in management and policy studies. One claims that the agent who promotes the scheme and finances it frequently underestimates the cost and schedule targets during the front-end. The reasons offered are: strategic misrepresentation and optimism bias (Wachs 1989, Flyvbjerg et al. 2003), underinvestment in front-end planning (Morris 1994, Merrow et al. 1988, Merrow 2011), and use of rigid buyer-supplier contracts that preclude know-how of implementation to feed into the front-end (Stinchcombe and Heimer 1985). The second view is common too—that complex systems projects cannot be planned reliably because the promoters are hostage to scope creep (Hall 1982, Shapiro and Lorenz 2000), escalation of commitment (Ross and Staw 1986), and ‘shaping’ of the scope by events and institutional interests that are outside the promoter’s control (Altshuler and Luberoff 2003, Szyliowicz and Goetz 1995, Miller and Lessard 2000). Whilst long-standing, this debate has struggled to move forward due to difficulties in acquiring data about the inner workings of the front-end. Most empirical studies have adopted an ‘outside’ view (Flyvbjerg et al. 2003, Miller and Lessard 2000, Dvir and Lechler 2004). Thus, while there is consensus that the front-end is crucial, Pinto and Winch (2015) note recently that “over the last 20 years…the actual processes of shaping remain somewhat mysterious—project shaping remains a black box”. This study expands on the ideas of these researchers, but shifts the focus to a view that illuminates the project front-end from an organizational design perspective—a cognitive lens useful to explore how governance structures allocate decision-making authority and resource control, and resolve disputes (Galbraith 1973, Lawrence and 2 Lorsch 1967, Simon 1969). Importantly, the front-end occurs before the promoter has let major supplier contracts because the promoter is yet to enter into formal agreements with actors that have key stakes on the project and control critical resources. Thus, the front-end occurs before the promoter has simulated an authority hierarchy through principal-principal and buyer-supplier contracts (Stinchcombe and Heimer 1985). This study argues that the project front-end for one-off complex systems is carried on by a consensus-oriented community of production. Consensus refers to the degree to which the collective goals and plans are agreed upon by all involved (Van de Ven 1976). For example, a large infrastructure project cannot forge ahead without land, capital, planning consent, political support, and knowledge of needs-in-use. The control over these resources is distributed across autonomous stakeholders and many resources are not up for sale. Hence at the front-end the promoter has to let many actors participate in strategic choices in exchange for their cooperation; as Miller and Lessard (2000) state, the front-end is about ‘building momentum and commitment’. Put differently, the project front-end is carried on by a community of production where the promoter formulates the problem but problem-solving requires creating a ‘negotiated environment’ (Cyert and March 1963). To this purpose, the promoter decomposes a complex problem into more manageable sub-problems; it then needs to interact with other stakeholders to produce mutually consensual solutions, and move from an individual strategy to a ‘collective strategy’ (Astley and Fombrun 1983). In distributed communities of production, the managers cannot rely on authority hierarchies (March and Simon 1958), markets (Ouchi 1980), or system integrators (Brusoni et al. 2001) to get things done. If governance is ‘flat’, the managers must attend to the concerns of the community members to avoid defections (Rothschild and 3 Russell 1986, O'Mahony and Ferraro 2007). This is, management has agency but cannot exercise it fully and must engage in negotiations (Pfeffer and Salancik 1978). However, the focal problem is markedly different from phenomena that has informed management studies on production communities such as open source software (O'Mahony and Ferraro 2007), private-collective innovation (von Hippel and von Krogh 2003), new business models (Baldwin and von Hippel 2011), standardsetting (King and Lenox 2000), and science communities (Lakahni et al. 2007). Unlike the settings above, the project front-end for complex systems is heavily regulated, and frequently unfolds under externally-imposed deadlines. Strategic choices are also often contentious given the high-stakes riding on the outcomes—if the front-end succeeds, substantial capital gets committed to long-lived assets through decisions that difficult to reverse. As such, the front-end struggles to meet known antecedents of cooperation including a positive history of working relationships, unifying goals, and the presence of a legitimate convener to draw together autonomous actors (Gray 1989a, Ring and Van de Ven 1992, Thomson and Perry 2006); as Rittel and Webber (1973) famously put it, “planning problems are wicked problems”. At the front-end of complex systems projects, Jessop (1997) argues that governance must steer multiple firms, agencies, and organizations that are both operationally autonomous and structurally coupled through reciprocal interdependencies. But we still lack in-depth empirical studies of how governance actually works. This leads to the three questions motivating this study: i) which structure governs the project front-end for a complex system? ii) how are disputes resolved? and iii) what is the impact of front-end governance to project performance? This research adopts a multiple-case study approach with embedded units of analysis (Eisenhardt 1989, Yin 1984). Case studies allow researchers to incorporate 4 contextual and temporal dimensions, and thus are suitable to explore novel ideas (Eisenhardt and Graebner 2007). Embedded multiple-case studies in particular are useful to investigate a holistic question without overlooking operational details (Yin 1984 p.45). This research design was thus suitable to address the holistic issue of project performance by exploring the inner workings of the front-end stage. The sample consists of four large infrastructure projects in the UK. The units of analysis are incidences of front-end disputes that emerged between the promoter and resource-rich stakeholders. Disputes are situations in which people disagree and thus illuminate the “scene of the battle” (Latour 1986) and how people resolve conflicts (Coleman and Ferguson 2014). The analysis reveals that the source of the disputes is a conflation of three factors: i) rilvary over strategic choices; ii) scarcity of resources (time, money) to bridge differences; and iii) the promoter’s reluctance to let prior performance targets slip to preserve the legitimacy of the project in the public eye. The study leads to three main contributions: first, it reveals that the front-end of a complex systems project creates a large arena of ‘collective action’ (Astley and Fombrun 1983, Ostrom 1990). In this structure, the authority to make strategic choices is distributed across nested centres of decision-making power with capacity for mutual adjustment—hence governance is polycentric (Ostrom 1972). Second, the study reveals how disputes over strategic choices are resolved through the intertwinement of deliberative and tough bargaining. It also traces major slippages in the performance targets, during and after the front-end, to the costs of safeguarding compromise solutions and to the extent context creates shared accountability for the outcomes. And third, the study proposes a contingency model of project performance that contains a relationship between: i) the umpiring structure used to resolve internal disputes; and ii) the promoter’s use of slack resources to sustain external legitimacy for the project. 5 The rest of this paper is structured as follows. First, it reviews literature on consensus and governance in complex systems projects. Then it introduces the methods and analysis, and discusses the relationship between front-end governance and project performance. It concludes by highlighting implications to practice and policy. Background: Front-end Governance in Complex Systems Projects Interorganizational collaborations are central to management and policy studies because they are critical to find solutions to complex problems—situations where a single actor does not have all the information-processing capacity and resources to solve the problem (Gray 1989, Fjeldstad et al. 2012, Hargrave and Van de Ven 2006). One stream of this large body of literature looks at distributed communities of production which are not governed by firms, buyer-supplier contracts, or central administrative structures. In settings as diverse as the production of open source software, industry standards, and science, well known structures to govern include robust relational contracts and meritocracy-based authorities (Gibbons and Henderson 2011, O'Mahony and Ferraro 2007, King and Lenox 2000, Tuertscher et al. 2014). The front-end of a one-off complex systems project differs, however, from the aforementioned settings in important ways. First, relational contracts are fragile since regulation forces the promoter to attend to the interests of opponents and work with resource-rich stakeholders who may demand a high price for their cooperation (Cleland 1986, Divr and Lechler 2004, Morris 1994, Winch et al. 1997, Floricel and Miller 2001). Stakeholders are also drawn from different epistemic, institutional, and ideological communities (Hughes 1987, Miller et al. 1995, Hobday 2000). This leads to heterogeneity of ‘legitimating logics’ (Creed et al. 2002) where politics and power can stymie the emergence of a meritocracy-based authority (Miller and Olleros 2000). 6 A second factor complicating the front-end governance are deadlines that are associated with electoral and regulatory cycles—as Engwall (2002) puts it, no project is an island. Context puts pressure to hammer out deals. But building consensus over strategic choices is hard to rush as holding lengthy talks is needed to allow actors to make sense of complex problems and coordinate collective action (Susskind and Cruikshank 1987, Gersick 1994, Thomson and Perry 2006, Weick and Roberts 1993). Time pressure thus amplifies perceptions of risks, which leads to tough bargaining and politics (Ring and Van de Ven 1992)—two mechanisms that create difficulties in fostering cooperation but arise when the resources available are insufficient to develop a common construction of the problem (Lawrence et al. 2002, Gray and Clyman 2003). And third, the participants in the project front-end for a complex system cannot bank on modular designs to circumvent difficulties in striking a consensus. Modularity breaks apart interdependency between tasks, thereby enabling actors to do their work with limited need to interact with one another (Baldwin and Clark 2000). Modular choices thus dampen conflict caused by rivalrous preferences, encourage unpaid contributions, and limit the impact of uncooperative behaviour. In contrast, when autonomous actors must agree one-off design choices, and thus problem-solving is ‘non-decomposable’, the design choices qualify as an Ostrom’s (1990) shared resource—this is, many autonomous actors with rivalrous preferences share the right to participate in the design decision-making process (Gil and Baldwin 2013). Hence the front-end for a complex system project involves ‘indivisible’ or shared strategic choices, and can unravel if the participants refuse to cooperate and compromise. The governance of shared resources is invariably a struggle (Dietz et al. 2003). But even large arenas of collective action are sustainable if governance is: i) polycentric, this is, it is decentralised across nested centres of decision-making and power with 7 capability for mutual adaptation (Ostrom 1972); and ii) robust, this is, there is mutual respect between the high-level authorities and the local claimants. This is the core claim in Ostrom (1990)’s theory of governing large resources that are shared by many autonomous and non-excludable claimants with rivalrous interests. From this perspective, it is thus unsurprising that the governance of the project front-end for a one-off complex system is a struggle. For example, a government committed to build a new railway is expected to formulate a preliminary performance baseline. During the front-end, the promoter and other stakeholders will share difficult strategic choices for key components such as stations, train cars, and route. But bounded rationality makes it hard for the promoter to delineate ex-ante solution spaces that are robust enough to cope with all the emerging preferences (Simon 1981). Hence the initial baseline potentially leads to sub-problems that are intractable. Changes to the baseline are tricky because they can impair the external legitimacy of the front-end and thus internal disputes arise. The next section explains how this study set off to explore how the disputes get resolved, and how a contingent model of performance emerged. RESEARCH DESIGN, SAMPLE, AND METHODS This research uses an embedded multiple-case design (Yin 1984, Eisenhardt 1989). The larger unit of analysis is the project, and the holistic question asks how front-end governance influences project performance. Hence the four cases are treated as independent experiments that confirm or disconfirm emerging theoretical insights in replication logic. In multiple-case designs, embedded units of analysis are useful for focusing the case study inquiry (Yin 1984). This study focuses the inquiry on front-end disputes because dispute resolution is a central task of governance. To advance theory and yield more generalizable and robust insights a diverse and polarised sample was built as recommended for inductive studies (Siggelkow 2007). 8 The sample includes four large infrastructure projects for which I gained exceptional access to the inner workings of the front-end: London Crossrail, a high-capacity underground railway; London Olympic park; Heathrow airport terminal two (T2); and UK’s second high-speed railway (HS2). Table 1 summarises the system-level goal for each case, the organizational structure of the promoter, the front-end outcome, and data sources. Appendix I summarises for each case prior history, context, timescale, and the evolution of the performance baseline during the front-end. --Insert here Table 1 -We built the sample to vary three key attributes of projects to develop complex systems and increase the generalizability of our findings. First, the cases differ in the potential to decompose the grand problem. Figure 1 illustrates this in a stylised way. Modular infrastructure design structure (e.g., Olympic park) Integral infrastructure design structure (e.g., railway system) Hybrid infrastructure design structure (e.g., airport terminal) Figure 1 – Stylised representation of different infrastructure design structures An Olympic park is suggestive of a decomposable system. It comprises a set of discrete assets such as sport venues that are connected by underground utilities. But the utilities are ‘slaves’ designed not to constrain the strategic choices for the high-value assets. In contrast, railway systems are far less decomposable. All stations connect to the same functional components (tracks, control and signalling systems) and must accommodate the same train cars. Hence many strategic choices for different components are interdependent. In turn, an airport is suggestive of a hybrid system. Some components are physically linked, for example, the tunnels and passages that 9 connect the concourses, but other components such as the car park and hotel are not. Decomposability was expected to attenuate the complexity of the project front-end. The sample also varies in the organizational structure of the promoter. The HS2 and the T2 schemes were promoted by single organizations, respectively the UK government and the private owner of the airport, BAA2. In contrast, in the other two cases, coalitions were at the helm of the front-end. Crossrail was promoted by the UK and London governments (although the stations located on private land were privately financed); and four organizations were at the helm of the Olympic park. More governance complications were expected the more fragmented the promoter was. Finally, the sample varies in the time pressure on the project front-end. The London 2012 immovable deadline put massive pressure to make planning decisions. The other schemes did not face immovable deadlines. But pressure was still high due to either the politicians’ will to complete the front-end before elections (the cases of Crossrail and HS2 first phase) or the regulatory cycles in the case of BAA3. Time pressure could be a cause for conflict, but also offer a stimulus to resolve disputes. Units of Analysis To probe deeper into the governance of the project front-end, a unit of analysis was embedded: disputes over strategic choices. These disputes trigger difficult questions: should the project proceed if the participants cannot converge? Should the initial performance targets slip to facilitate convergence? And which actors should directly influence strategic choices? Seeking answers for these questions was essential to illuminate the relationship between front-end governance and project performance. 2 In 2012 BAA changed its name to Heathrow Ltd; I keep to the BAA name for the sake of simplicity In the UK, the gap between general elections cannot exceed five years; and regulated private monopolies operate under obligation to produce a new capital investment plan every five years. 3 10 Data Collection Data collection started in 2011 when, as part of a broad research program to advance understanding of megaproject governance, we got access to the top managers of the Olympic Delivery Authority (ODA), the public agency established in 2005 to develop the Olympic park. This agent reported to a governing body formed by the four promoters; the ODA attended the Olympic board meetings but had no power of veto. Between 2011 and 2014, we leveraged the access to the ODA to, first, access top managers of other stakeholders for the Olympic park; and second, to acquire similar levels of access to the other schemes. In a snowball fashion (Biernacki and Waldorf 1981), for all cases, directors of major project suppliers were also interviewed. In total, 121 formal interviews up to 2-hours long were conducted and tape recorded by myself and doctoral students. Follow-up interviews were conducted to probe deeper into particular issues, double check a verbal account, and bridge gaps in the empirical database. We were not asked to sign non-disclosure agreements regarding interview data, but always asked permission to use verbatim quotes and whether to keep the source anonymous; some respondents gave free rein to use the transcripts, whereas others disallowed the use of particular quotes; we formally committed not to share any project reports which were not in the public domain. To gather extra data and allow for member checks (Lincoln and Guba 1985) the emerging findings were shared with respondents and some were invited to give presentations and stay for lunch. We welcomed a total of 13 guests and for each visit produced detailed hand-recorded notes of the talks and lunch conversations. To improve data accuracy and the robustness of the insights (Jick 1979) the verbal accounts were triangulated against archival sources (Miles and Huberman 1994). The front-end of a large infrastructure project is heavily regulated by UK laws and thus many documents are available online or become available through the Freedom of 11 Information act. Key documents included minutes of board meetings, formal communications, and reports announcing performance targets and corresponding plans. In the case of BAA, we studied capital programs, master plans, and consultation documents. The disputes between BAA and STAR were documented in reports produced by the regulator and formal letters between BAA, STAR, and the regulator. Information in the internal project documents was analysed against reports produced by third-parties. Hence we combed through reports produced by the National Audits Office, Parliamentary committees, government watchdogs, and other thirdparties. Other sources of data were articles and interviews with top managers in professional outlets, e.g., New Civil Engineering, Construction News, and articles in the national press, particularly for disputes that had fallen in the public domain. Methods Following recommendations for inductive reasoning (Langley 1999, Ketokivi and Mantere 2010), we produced detailed accounts for each case. Each account provides a contextualised and chronologic understanding and guards against account bias (Miles and Huberman 1994) and risk of self-aggrandizement in recollections (March and Sutton 1997). Design Structure Matrices’ (DSMs) were used to illuminate the governance structure. DSMs are a tool from design theory that allows representing a complex system into a square matrix by capturing interdependencies between its constituent elements (Steward 1981). The cells along the diagonal of the DSMs represent strategic choices (the decisions are listed to the left of the rows) whereas the off-diagonal entries indicate interdependency between these decisions. If the DSM has an entry in row i, column j, the decision concerning element i has an impact on the decision concerning element j. Hence the decisions represented in the diagonal cells have inputs entering from the top and bottom decisions, and outputs leaving from the left and right sides. 12 DSMs have been used to model the decision structure for producing a detailed design (Sosa et al. 2004, Baldwin et al. 2010). But DSMs have not been used to model the project front-end, and thus an original protocol was developed. The aim of this exercise was not, however, to exhaustively model planning but illuminate governance. Hence, for each case, a DSM was developed to qualitatively capture the interdependencies between strategic choices. The empirical database consisted of 35 front-end disputes which were cognitively filtered out by the respondents. The construction of the DSMs was restricted to the components that lodged these disputes. The codes to interrogate the database were derived from the headlines of a standard project performance baseline. Hence, the codes included goal, cost forecast, and scope. The coding for scope was restricted to three salient design requirements: capacity (which affects viability in use), footprint (which affects land acquisition), and sub-elements at the centre of a dispute. In addition, for each DSM, a companion organizational matrix was produced. This matrix reveals which stakeholders participate in the front-end and the decision-making forums where the disputes are resolved. The DSM analyses revealed a first regularity across the four projects: the front-end is governed by a polycentric structure. However, the DSM analyses cannot reveal how disputes are actually resolved. To address this question, a theoretical sample of nine disputes was then built, and the empirical data was analysed using coding and tabular displays (Strauss and Corbin 1990). The sample varied in terms of: i) the locus of dispute resolution; ii) the extent dispute resolution involved relaxing the solution space and thus the performance targets; and iii) whether the outcome was or not consensual. The data analysis iterated between: i) reviewing transcripts and extracting quotes of dispute-resolution mechanisms; ii) using secondary data to verify interview data; and iii) developing the argument. As we iterated between data and theory (Miles and 13 Huberman 1994), and listened to feedback received from previous drafts, a contingency model of project performance started to emerge. We stopped iterating when theoretical saturation was reached. The next sections present the analysis and then discuss the insights, and the implications to management practice and policy. Analysis The analysis first examines the structure governing the project front-end. We then examine empirical regularities and variance in the capabilities of this structure to: i) resolve disputes; and ii) sustain legitimacy for the project in the eyes of third parties. The Polycentric Governance of the Project Front-end Polycentric governance is an intuitive approach to structure large arenas of collective action such as networks of public agencies that perform similar functions but operate in different local contexts (Ostrom 1972). The basic idea is to give local groups decision-making autonomy insofar they keep to the higher-level rules; and when local disputes surface, problem-solving gets deferred to high-level authorities who commit to work back with the local actors to resolve the local problems. In agreement with collective action literature, and across all cases, the analysis reveals the creation of a nested structure of decision-making forums at the project front-end. Figure 2 illustrates the analysis by showing the design and organizational structure matrices for the four cases. Table 2 exemplifies the protocol that we used to interrogate the empirical database of the disputes in order to construct the matrices. <Insert Figure 2 and Table 2 here> A first point to note is a regularity revealed by the four DSM matrices: invariably, the matrices show densely populated clusters of off-diagonals ‘x’. Each cluster reflects the interdependency between the strategic choices necessary to define key functional components such as a railway station, a sports venue, or an airport concourse. The result is intuitive. For example, in the Olympic park case, a key component was the 14 Aquatics Centre. The choice to set the goal as a ‘massive iconic venue’ is an input for deciding on the cost, footprint, and capacity—these interdependences are captured in Figure 2 and Table 2. But, the latter decisions are inputs to refine the goal. For example: the goal needs to be readjusted if: i) there is not enough land available; ii) the cost is unaffordable; or iii) a massive venue is not viable. As the goal gets readjusted, changes ensue to the other decisions. A level down, the decision to add a controversial ‘stylistic’ diving board is interdependent with the decisions to set the goal and budget. In contrast to the regularity of off-diagonals ‘x’ clusters, the four DSM matrices differ substantially in the degree of interdependency between the component clusters. The Olympic park DSM is sparsely populated off the component clusters because the strategic choices for each venue are technologically independent. The notable exception is the interdependency between the cost decisions: increasing the cost for one sport venue potentially leaves less money to invest in developing the other venues. In contrast, the DSMs for the Crossrail and HS2 cases show high interdependency between the component clusters. For example, in the case of the HS2 DSM in Figure 2, the goals for each station along the route are interdependent—the goals need to be similar to respect equitability. Furthermore, system-wide technical constraints create interdependency between the requirements for the stations. Hence the railway DSMs are densely populated off the component clusters; the DSM for T2 is a hybrid: the cluster of strategic choices for the car park is, apart from the cost decision, independent from the decisions for the concourses and baggage system. But the latter components are connected, and thus the strategic choices are technologically interdependent. In marked contrast, the organizational matrices (represented at the right of the DSMs in Figure 2) are remarkably similar across the four cases. At the highest level, the four matrices show a top governing body whose membership is restricted to the 15 organizational actors that constitute the promoter. The promoter and its agent are omnipresent in all strategic choices since they are accountable for project performance. For example, in the HS2 case, the UK government and its agent, HS2 Ltd., directly influence all strategic choices. Likewise, in the Olympic park case, the four promoting organizations and their agent, the ODA, share control over all the strategic choices. A level down, the four organizational matrices show a fragmented structure of local working groups so-called project boards. Each project board is open both to the local resource-rich stakeholders and to the promoter’s agent, but closed to opponents to the project and resource-poor actors (although consultation reaches all actors). The task of each board is to agree the strategic choices for the component of interest. Table 2 shows evidence of direct influence of resource-rich stakeholders on local strategic choices. For example, in the HS2 case, the London government and Transport for London (who together control the transport strategy for London), and Camden Council (who controls local planning consent) directly influence strategic choices for the Euston station; in turn, the Manchester government, Transport for Greater Manchester, and other local actors participate in the front-end for the Manchester station. Likewise, in the Olympic park case, for each sport venue, different sport bodies and other local actors directly influence the strategic choices. The authority of the local stakeholders is thus delineated to the component that requires their critical resources. In sum, the DSM analysis systematically reveals a degree of decomposability in the front-end for a large infrastructure project. This regularity reflects how different components create different sub-problems; solving each sub-problem requires a different bundle of resources. Yet, not one of the four cases is fully decomposable into a set of strictly modular sub-problems. First, some sub-problems are technologically 16 interdependent; and second, early announcements of performance targets for the project create financial interdependence between the sub-problems. Crucially, we uncover regularity across the four cases from a governance perspective. Irrespectively of the observed variance in system decomposability and in the structure of the promoter, front-end governance is systematically polycentric. At the highest level, the promoter has overarching decision-making authority over all strategic choices. But for each component/sub-problem, the promoter shares the authority over the local strategic choices with a distinct group of local stakeholders. The Resolution of the Front-end Disputes over Strategic Choices We turn now to probe deeper into the resolution of disputes. We ask: is polycentric governance not just masquerading a process of consultation? Were there key differences across the four cases? Table 3 provides a sample of disputes that shows the diversity of the high-level dispute-resolution mechanisms encountered. <Insert Table 3 > As aforementioned, at the onset of the project front-end, the promoter selfformulates a preliminary performance baseline that delineates the local solution spaces. It is the mandate of the promoter’s agent to figure out consensual solutions within those constraints. As shown in Table 3 a major source of disputes is lack of local goal congruence. For example, the Olympic stadium dispute (#1) was rooted in differing visions for the venue in legacy: football versus an athletics venue. Other disputes are rooted in rivalrous preferences for lower-order choices but where the stakes are high. This is the case, for example, of the dispute about adding toilettes to the Farringdon station (#9): the local actors felt strongly about it, but the promoter was adamantly against the idea because in his view toilets were costly to operate and keep safe. These disputes are contained, but problem-solving can still be complicated. 17 Goal disputes are particularly challenging. The disputes about the HS2 stations are telling (#4,5). The local claimants argue that grandiose, well integrated stations are needed to catalyse urban regeneration, whereas the promoter prefers utilitarian and modular designs. To persuade the government to increase the local budgets, the cities commissioned their own designs. But under pressure to keep to prior targets, HS2 Ltd recommended plans to the UK government that the cities had not fully endorsed. It was then up for Cabinet4, a level above, to decide what to do; one official explained: HS2 Ltd, if you like, are the infantry out there; actually doing what they’re told by [central] government. So HS2 Ltd get all the fights, appear to have all the fights, are the bad boys, but they’re really only doing what they’re instructed to do Crucially, our findings show that the promoter has less decision-making power than could be assumed prima facie. De jure the promoter may have power to impose reasonable solutions. And if the promoter believes it has strong evidence to support its case, it may try to force its hand—sometimes it loses, sometimes it wins. For example, the Olympic park promoter successfully proved the user wrong when the latter claimed the diving boards of the Aquatics centre were unsafe (#3). But the Crossrail promoter lost the ‘fight’ for not adding toilettes (#9). In the goal disputes, where evidence is easily contestable, our findings suggest that the promoter prefers to negotiate. This was true both for T2 (‘if something gets talked it gets changed’, said a STAR person) and for government schemes (‘You could…make HS2 its own planning authority [but] that would flout democratic processes’, said a HS2 official). Hence the local working groups play two functions. As shown in Table 3, they create opportunity for the front-end participants to engage in cycles of analytical deliberation and design rework. Each participant will mobilize their own technical teams to produce new evidence for backing up the preferred strategic choices. Teams 4 The UK Government Cabinet includes the Prime Minister and the most senior ministers 18 of technocrats from all parties will also meet regularly to get their points across, find ways to bridge gaps in finance, and search for mutually consensual solutions. Seldom is the case, however, that technocratic discussions suffice to bridge differences in interests—‘this is all the art of possible, isn’t it?’, said one respondent. Hence, high-stakes meetings occur concurrently. They are open to elected leaders and top managers, and are crucial to cut deals. We did not uncover an instance where a dispute was resolved by diktat. But the intertwinement of deliberative and bargaining structures makes front-end governance a struggle. Hence, we turn now to discuss regularities and variance in the mechanisms that keep the front-end stage afloat. Empirical regularities: design safeguards and slippages in performance targets The analysis reveals that investments in design safeguards such as redundant parts and spare capacity were instrumental to resolve goal-related disputes. These allowances enable one vision to rule strategic choices without prohibiting the shift to implement the competing vision at a reasonable cost later on (Gil 2007). As a capitalintensive way to build limited flexibility, investments in safeguards are controversial (Gil and Tether 2011). Still, they systematically enabled to resolve goal disputes. The case of the Olympics aquatics centre [#2] is telling. The Olympics bid pledged a massive, sophisticated venue. But the costs spiralled in planning; local claimants also argued that a massive venue would be unviable in use. Yet, to back down from the bid pledge was tricky—one official said, ‘if you challenge them [internationally-renowned architect] directly they will just walk away’. Complicating matters was a hard constraint on the minimum capacity of the venue for the games. In the end, the dispute was resolved by keeping the aesthetics, shrinking the venue size, and designing safeguards to increase the capacity just for the games. In the process, the cost forecast duplicated, which caused an outcry—‘the games' organisers seem to be willing to spend money like water," said a Parliamentary report. 19 Design safeguards were also handy to resolve disputes between BAA and Star. The airlines craved a modern campus, whereas BAA planned to replace old facilities [#6]. Some airlines also preferred ‘closed’ gates which they deemed more efficient, whereas BAA preferred ‘open’ gates so passengers could move around up to boarding [#7]. Consistent with its preferences, BAA announced a less than £700m new concourse (‘Heathrow East’) to open in 2012. Disputes then ensued; in the end, BAA safeguarded the airlines’ vision by adding basements and tunnels; it also designed in flexible gates. In the front-end, the performance targets slipped three times until BAA, Star, and the regulator settled on a £1bn concourse to open in 2013. In the Olympic stadium dispute [#1], in turn, we trace cost and time overruns to a failure to strike a front-end consensus on a safeguarded design. Facing a rift between front-end participants due to differing visions, the promoter’s agent suggested increasing the budget in 20% (~£100m) to build retractable seating and thus safeguard both visions. But the football aficionados refused to compromise on what they called a ‘jack-of-all-trades’. With time running out, the Olympic board chose to go ahead with a rigid design for the 2012 games. But the dispute was only resolved in 2013; the final cost of the conversion to a dual-use venue then reached £190m in 2015. Consistently, resolving goal disputes required throwing more resources (money and/or time) into the pot until a compromise surfaced. This confirms Flyvberg et al.’s (2003) claim that project promoters suffer from optimism bias; one respondent said: ‘early on, people’s eyes are much bigger than their stomachs…our guidance is ‘no matter what your press people say…don’t be drawn towards providing a spot figure; it’s foolish, you’ve just created a hostage to fortune to yourself’. The promoter dislikes slippages to the local performance targets because, first, they compromise equity; and second, they can compromise the targets for the project as a whole that it announced at the onset of the front-end. Slippages in targets then 20 create a risk of losing public legitimacy. We turn now to examine variance in the use of umpires to overcome what otherwise could become intractable sub-problems. Table 4 summarises the evidence at project level for the four cases. Table 4 –Summary of Mechanisms to Resolve Disputes whilst Sustaining Legitimacy Case Olympic park Project front-end: 2002-07 Umpiring Structure Internal External Yes design sponsors; planning board Yes HS2 Project front-end: since 2009 project sponsors No Yes Slippages in the Performance Targets Completion Date Cost Forecast (*) Immovable Yes But some disputes only temporarily resolved for 2012 Yes 2002, ~ £1.1bn 2007, ~£2.0bn 2004, ~ £4.9bn (~29% of cost 2007, ~ £6.9bn forecast) Final (2013): ~£8.1bn Yes 2010, planning (1st phase) (10/11 prices) 3 years in done by 03/2015 st Parliament 2015, planning (1 phase) 2010, ~ £22.7bn end expected by 2017 st 2012, ~ £26.5bn for 1 Project completion: 2014, ~ £28.2bn phase (as unchanged so far of 2015) Heathrow airport T2 Project front-end: 2005-08 Crossrail Project front-end: 2000-2008 Yes design arbitrator; regulator Yes project sponsors Yes Slack resources (contingency) Substantial Yes Substantial 2010, ~£7bn (30% of cost forecast) 2013, ~ £14.4bn (51% of cost forecast) Limited 2005, ~ £1.3-1.8bn 2008, ~£200m 2006, ~ £2.0bn (8% of cost 2008, ~ £2.4bn forecast) Actual Completion: 2014 Final (2015): ~£2.8bn Yes Yes Yes Substantial 2000, open in 2012 2000, ~£4.7bn 2003, ~ £9.8bn 2006/7, ~£5.0bn 3 years in 2003, open in 2016 2007/8, ~ £10.9bn (46% of cost Parliament 2008, open in 2017 Actual Completion (as of Final (as of 2015) forecast) (2005-08) 2015): 2019 ~£14bn No 2005, open in 2012 2008, open in 2013 (*) final (cash) prices unless indicated otherwise; public budget envelope= cost forecast +contingency Variance in Implementing an Umpiring Structure In sports an umpire acts as a referee and settles disputes between competing players. At the front-end of a complex systems project, promoter and autonomous stakeholders strive to win arguments over strategic choices. It turns out that the presence of an umpire is a pragmatic mechanism to put an end to disputes that could otherwise drag forever. However, the findings suggest variance in implementing this structure: the T2 and the Olympics projects relied only on internal umpires, whereas many disputes were deferred to an outside referee in the Crossrail and HS2 cases. In the T2 case, the regulator was both a front-end participant and the umpire of last resort. The presence of the regulator was reassuring for both parties; as one BAA respondent said, ‘we’re battling all the time ...if the airlines don't like it, then they can 21 bring in a formal dispute’. And indeed, in the dispute about the main concourse (#6), Star wrote several letters to the regulator complaining that BAA was ignoring their needs, a claim BAA found unfair—‘I can never get consensus on almost anything’, said a BAA director. A level below, BAA and Star recruited a retired director to arbitrate less substantial disputes such as the dispute over the type of gates [#7]. Likewise, the Olympic park promoter created an internal umpiring structure in the face of self-imposed deadlines to make decisions—‘it’s continuous negotiation, a tough, tough environment’, one manager said. Hence, at the lowest level, each project board had a ‘design sponsor’ that owned a limited contingency to self-resolve small disputes. A level above, the Olympic board was expected to resolve major disputes. The lack of an external umpire was not an issue in the T2 case since the regulator held both deep project knowledge and authority to resolve disputes. However, the lack of external umpiring had disadvantages and advantages in the Olympics case. On the one hand, it increased the risk of chaos; one official said—‘you’ve got powerful figures all over the place…you can’t govern’ (Norris et al. 2013). To avoid the risk of impasse, the Olympics promoter ditched the prior cost targets set in the bid; it also engaged in a democratic decision-making process but bounded it to rigid deadlines—time became a de facto dispute-resolution mechanism; but even then, the stadium dispute [#1] shows, some issues could only be temporarily settled. On the other hand, the lack of external umpiring enabled the promoter to overlap front-end planning with project delivery works to gain time. And by 2009, as more front-end disputes got finally out of the way, the Olympics promoter published an updated performance baseline so-called ‘blue book’. In marked contrast, the Crossrail and HS2 cases show umpires internal and external to the project. Project sponsors worked as internal umpires in a role similar to 22 the aforementioned design sponsors. In turn, the laws in the environment elected the UK Parliament as the referee of last resort—any organization who was ‘materially affected’ by the promoter’s plans could petition against them. One official explained the reason to petition against the plans for the HS2 London Euston St. (#4): HS2 Ltd. didn’t persuade us that our points were wrong nor did they persuade us that their points were right.…[Petitioning] gives us the ability to correct what we feel is a mistake.…that’s ultimately about making your case that your vision is superior. The presence of an external umpire had multiple effects. The case of the Crossrail Woolwich station (#8) is telling. Despite years of talks the promoter refused to add this station in part to avoid creating precedence as there were more claims for stations. Petitioning gave local actors a last chance to overturn what in their perspective was a flawed decision. After 40 months of deliberations, Parliament ruled that the promoter should safeguard the station in the design, and that the petitioners (a house builder and local government) should incur the construction costs. This ruling seems sensible and no major issues ensued, which suggests that the presence of Parliament stymied mutual cooperation. And indeed, aware of how inefficient it was to rely on Parliament to resolve the disputes, the Crossrail promoter managed to privately cut late deals with two thirds of the petitioners who then withdrew their public petitions. We turn now to analyse variance in the slack resources used to sustain public legitimacy for the project as internal disputes put pressure on performance targets. Variance in the Use of Slack resources Compiling registers of front-end disputes that can potentially cause late cost and or time overruns is an established practice (Cleland and King 1983). Once the risks are quantified in probabilistic terms, the project promoter faces a judgement call: either it sets optimistic performance targets that increase the scheme attractiveness but also the risk of overruns, or it sets more conservative targets; one civil servant said: 23 There’s a bandwidth there…if we push it [project budget] too far we won’t get the project...so there’s that game that goes on to try and find what the [UK] Treasury’s real limits are, and how far can we really push it. …it’s a political decision. In the T2 case, the use of large buffers was ruled out. First, BAA was confident on its ability to parry any public backlash caused by slippages in targets. Second, regulation prohibited to start delivery without resolving first all major disputes; if need be, the front-end could be prolonged. And third, the more capital spending, the higher the levies that BAA would be allowed to charge the airlines to recover the investment; hence both the airlines and regulator were also against the use of slack resources In marked contrast, in the other three cases, elected leaders had no appetite to let the public budget (cost forecast plus contingency) slip multiple times; there was also no confidence that major disputes could be resolved in the front-end, let alone a much larger number of minor disputes. The Olympic park case is a good example. The bid cost forecast (~£4.9bn5) turned out insufficient to meet the bid pledges plus all the emergent local claims—‘it’s like the Olympics will solve all the world’s problems’, said one officer. To get a grip on a chaotic situation, in 2005, the promoter set a 2-4-1 rule: two years to plan, four to build, and one to test. But when 2007 arrived, many disputes remained unresolved. This spurred the promoter to agree a large buffer (£2bn) on top of what was by then a much higher cost forecast (~£6.9bn); one officer said: Treasury were really, really clear...big envelope and never knock on our door for money…actually they were right…we were then able to make decisions…rather than being petrified because we didn’t have enough money to do what we needed to do Table 4 shows a similar pattern for HS2 and Crossrail. In both cases, the promoters faced an outcry at the onset of the project front-end due to slippages of the cost targets. To mitigate the risks of further slippages derailing the project, the promoters ditched the original budgets for way more conservative budget envelopes. 5 Includes £971m (venues); £89m (conversion costs); £640m (Olympic infrastructure); £1040m (nonOlympic infrastructure); £700m (local transport schemes); £766m (land) plus VAT (NAO 2007) 24 The cross-case analysis thus suggests a major difference between the privatelyand the publicly-financed cases. Front-end disputes were in all cases traced to flat governance. But in the latter cases, context puts way more pressure both to rush the project front-end and show to third parties that the project is ‘on target’. Complicating matters, in the government projects, stakeholders operate under the expectation that the promoter will incur the costs of resolving all disputes. The promoter’s use of slack resources is thus a high-order device to address this conundrum: slack resources decouple the internal evolution in scope and cost forecast from the figure (so-called ‘budget envelope’) that the public ‘sees’ and government can announce early on. Slack resources do not automatically resolve disputes—Table 3 shows many disputes that could not be self-resolved. Worth noting, the use of slack resources is scrutinised by watchdogs to reduce moral hazard—‘[contingencies] are there for known risks, not for somebody’s betterment’, said one official. But in the long run slack resources appear to create a self-fulfilling prophecy—London 2012 depleted its contingency and Crossrail and HS2 face similar fates (NAO 2014, Butcher 2015). In sum, the analysis reveals two regularities. First, project front-end governance is polycentric since problem-solving authority for developing a one-off complex system is decentralised. And second, slippages in performance targets are necessary to safeguard compromise solutions. Within this pattern, we observe variance in the mechanisms to get things done. One source of variance is the use of external umpires—an effective but inefficient way to resolve longstanding disputes. There is also variance in the use of slack resources. Substantive slack is advantageous to sustain a public narrative that the project is on target but leads to a self-fulling prophecy. DISCUSSION 25 We return now to our central questions: what is the structure governing the frontend for a complex systems project and how does this structure affect performance? At the front-end, the promoter cannot use contracts to simulate an authority hierarchy; it also lacks authority to empower a ‘heavyweight manager’ (Clark and Fujimoto 1991). And setting up a meritocracy-based authority is difficult when a robust relational contract is yet to emerge. But since the promoter is accountable for performance, it cannot also work as a mere ‘rubberstamping hierarchy’ (Rivkin and Siggelkow 2003). To attenuate managerial complexity, our analysis suggests that promoters decentralise governance by creating a polycentric structure. First, they exploit the decomposability inherent in any complex system (Simon 1962) to formulate a set of more manageable sub-problems. And then, they assemble distinct groups of resourcerich stakeholders committed to find compromise solutions for the shared sub-problems. However, the project front-end is not fully decomposable: some sub-problems are technologically interdependent; other interdependences are driven by preliminary cost and schedule targets. These interdependences do not preclude the local claimants to make claims on strategic choices—local actors invariably try to optimize locally due to self-interest (Knudsen and Levinthal 2007). But decentralized searches of solutions for interdependent problems can bog down (Mihm et al. 2010). This leaves the promoter with a catch-22: if it governs by diktat, it alienates the local actors; if it lets the local performance targets slip, the project front-end can lose legitimacy in the public eye. Complicating matters, the front-end is time-bounded by context—no project is an island, Engwall (2002) observes. Expecting government not to heed to the political calendar as in the case of Crossrail and HS2 is unrealistic. In the other two cases, frontend activity was equally time bounded by context. Under time pressure, it becomes tempting to complement deliberations with a bargaining structure to settle disputes. 26 Our findings corroborate literature on the challenges of integrating knowledge and coordinating work across organizational boundaries to produce complex systems (Hughes 1987, Miller et al. 1995, Hobday 1998, 2000, Gann and Salter 2000, Davies and Brady 2000). They also agree with Morris’ (1994) claim that rushing the front-end in complex systems projects makes it hard to produce reliable performance baselines. Our analysis also qualifies Flyvbjerg et al.’s (2003) claim that promoters are optimistically biased. Hence we trace regular slippages of the performance targets to underestimation of the resources necessary to safeguard compromise solutions, i.e., risk-neutral solutions with higher expected benefits for everyone. However, we did not encounter evidence of deceitful scheming (Wachs 1989, Flyvberg et al. 2003). Rather, across all the cases, the analysis shows that promoters go to great lengths to produce reliable baselines. But resolving disputes requires changes to the promoter’s preferred plans—changes that inevitably compromise the targets (Dvir and Lechler 2004). Still, it is fair to ask why promoters announce targets when strategic choices are still up in the air. The thing is, Stone and Brush (1996) argue, organizations that develop plans for endeavours with ambiguous value have to do it—commitments dampen the ambiguity, and are thus a prerequisite to develop the legitimacy needed to acquire critical resources. Once resource-rich stakeholders show predisposition to volunteer resources to the enterprise, local searches ensue for compromise solutions. BAA, for example, announced targets for the T2 project way before nailing down the plans with Star Alliance, and disputes ensued. But BAA had started the construction of another terminal for a rival alliance and was thus under pressure to announce its plans to guarantee parity. Optimistic announcements were also the norm in the government projects. For example, the first budget for Crossrail was released when the plan was to build a central London train. The news helped to gain momentum 27 for a third attempt to promote the scheme. But the promoter had not yet seen that only a much more costly commuters’ train could garner sufficient political support. Notwithstanding regular evolution in the performance targets, the cases differ both in the mechanisms to resolve disputes and sustain public legitimacy for the front-end. Figure 3 summarises the logic in a contingency model of project performance. Slack Resources for Sustaining Project Legitimacy in the eyes of third parties Limited Substantial Umpiring Structure for Resolving Internal Front-end Disputes Internal Internal and External i)High pressure to resolve all major front-end disputes before starting delivery ii)Opportunity for shared accountability for outcomes iii)Potential for moderate slippages in the internal and external performance targets i) High risk of major slippages of the internal performance targets ii) High risk of major slippages of the targets in the public eye iii) Unavoidable inefficiencies iv) High risk of the project frontend not being sustainable No example: UK public policy Example: Heathrow T2 discourages this scenario i)High risk of a self-fulling prophecyi) High risk of a self-fulling ii)Potential for efficient overlap of prophecy front-end planning and delivery ii) All major front-end disputes iii) Slippages in the internal resolved before delivery starts performance targets hidden from iii)Unavoidable inefficiencies third parties iv) Slippages in the internal performance targets hidden from Example: Olympic Park third parties Example: HS2, Crossrail Figure 3- A Contingency Model of Performance of a Complex System Project Umpiring Front-end Disputes The structures that govern strategic choices in complex systems projects are underexplored—as Pinto and Winch (2015) note, ‘the black box of project shaping needs to be opened up’. In so doing, our empirical study uncovers a large arena of collective action, a structure rife in disputes. An intuitive principle of governing collective action arenas is to create affordable and independent conflict-resolution bodies (Ostrom 1990, Ostrom 2005). If conflicts are resolved through democratic decision-making processes, local claimants are less likely to defect; keeping the participants enfranchised also matters to mitigate the risks of decisions becoming embroiled in inefficient power battles and politics (Gray 1989, Tuertscher et al. 2014). 28 In agreement with this literature, across all cases, we found efforts to build internal umpiring structures to resolve disputes, and thus avoid a front-end stalemate. In the T2 case, for example, it is hard to conceive how BAA and the airlines could have bridged their differences without the arbitrator and the regulator. Facilitating the search for compromises was also the role of the design and project sponsors in the other cases. However, the cases differ in the umpiring structure external to the project. Outside referees bring benefits and costs to collective action. The main benefit is the provision of a forum of last resort to overcome impasse. But external umpires potentially create a negative precondition for parties to cooperate (Reilly 2001); they are also inefficient because they need time to assimilate knowledge before intervening. Hence Ostrom (1990) argues that on balance their presence makes collective action more fragile. The lack of an external umpire was not an issue in the T2 case. The regulator was a knowledgeable referee that both parties respected; the regulator also had the power to prolong the front-end talks if need be. However, the consequences were more nuanced in the Olympics case. On the one hand, the absence of a regulator created opportunity to make efficiencies. Hence, as independent elements of scope got gradually resolved, their implementation could start without being hold back by front-end talks for other elements. But the lack of regulation left no structure to fall back on if the front-end participants struggled to converge. In a project that faced an immovable deadline, letting the cost targets slip then became inevitable irrespectively if the local claims were, in the promoter’s view, reasonable or disproportional to what was at stake. In marked contrast, the HS2 and Crossrail projects could only move into delivery after Parliament resolved all disputes that the front-end participants had failed to selfresolve. This situation also requires a nuanced evaluation of performance. It suggests that context presupposes that, in the absence of a rigid deadline, the front-end is not 29 self-governable. An external referee thus mitigates the risk of chaos ensuing where either nothing ever gets done because the promoter and key stakeholders cannot agree on what to do, or excessive compromises lead to ‘white elephants’ that serve no one. However, the use of external umpires has major costs. As the Crossrail and HS2 cases show, their presence leaves strategic choices in limbo for years. This dismal prospect gives the promoter an incentive to self-resolve disputes. But if the stakeholders are not accountable for the outcomes, the incentive is not mutual. Indeed, if local stakeholders defer dispute resolution to the outside referee, they can increase their gains with limited costs; to avoid external umpiring, the promoter would then have to make concessions. But local slippages of targets can create precedence and impair third parties’ perceptions of project performance. Hence, if the context allows for external umpiring, it gets difficult to avoid this structure notwithstanding how inefficient it is. We turn now to discuss variance in the use of slack resources. The Strategic Use of Slack Resources Built-in contingencies are buffers that create slack, this is spare resources that allow an organization to adapt to internal pressures for adjustment or external pressures for change (Bourgeois 1981). The effects of slack resources to how organizations perform are contingent on the environment and the performance variable of interest (Voss et al. 2008). For example, Cyert and March (1963) argue that slack resources reduce politics and bargaining because more resources neutralise conflict. However, Bourgeois (1981) notes that slack resources also create opportunity for managers to engage in self-aggrandizing activity and thus for self-fulfilling prophecies. The variance in slack resources in our sample has important implications to performance debates. In the government projects, slippages of the performance targets potentially turn projects into political footballs; stakeholders, in turn, act as if government should pick up the tab for scope creep. This puts the promoter between a 30 rock and a hard place. Understandably, it is tempting to use slack to decouple what happens inside the project from what the public eye sees. Hence the promoter built substantial slack midway in the front-end for Crossrail and HS2; and even more prudently, it did the same at the ‘end’ of the front-end in the Olympics case. But slack resources cannot eliminate collective action problems, and thus the findings suggest a self-fulfilling prophecy ensues irrespectively if the external umpires are or not in post. In contrast, BAA management denounced the government’s use of, in their view, ‘over-egged budgets’. The context surrounding the T2 project was, however, very different. Here, major disputes were only expected with the airlines, and by regulation BAA had a guaranteed return on investment. Hence the airlines would indirectly pick up the tab for the extra costs of resolving disputes. Shared accountability thus made it in the interest of all front-end participants to contain cost escalation; it also reduced the need to build in slack resources to sustain a public perception of project ‘success’. The Performance of Complex Systems Projects Legitimate perceptions of project ‘success’ by third parties are rooted in comparisons of the outcomes against the baseline (Morris and Hough 1987, Pinto and Slevin 1987). Legitimacy is about perceived consonance of an entity with established norms, beliefs, value, and practices (Scott 2008, Suchman 1995); and in the world of projects, evaluating performance against initial pledges is well established (Dvir and Lechler 2004). Knowing this, promoters of a complex system go to great lengths to construct a narrative that a project is ‘on target’. But doing so is no more challenging than at the front-end where optimism bias and time pressure lead to regular underestimation of the costs of striking compromises. Setting the boundaries that delineate if a resource-rich stakeholder is external to the project or a front-end participant in its own right is, however, a choice of the researcher. Hence, in studies where resource-rich stakeholders are part of the context, 31 i.e., ‘external’ actors, resolving strategic choices becomes part of a process of ‘shaping’ the scope by the larger environment (Miller and Lessard 2000). This process then lays in part outside the control of the project promoter which informs the claim that complex system projects cannot be planned (Szyliowicz and Goetz 1995). In this study, resource-rich stakeholders such as local authorities and user groups are legitimate participants in the project front-end. These actors contribute resources under their direct control to produce a complex system which they will share in use with the promoter during a long service life. As such, the promoter and other stakeholders constitute a de facto distributed community of production at the front-end. This production community is rife in conflict because rilvary in preferences cannot be damped by homogeneity of logics and/or modularity. Using bargaining can speed up things, but also creates winners and losers and thus ambiguity in the effectiveness of the outcomes (Lundrigan et al. 2015). External umpires guarantee that something gets done in the absence of rigid deadlines, but exacerbate difficulties in producing reliable forecasts. Slack resources, in turn, also amplify slippages in the performance targets. However, this picture of regular slippages in performance targets is less dispiriting from a collective action perspective. The governance of shared resources is a struggle (Dietz et al. 2003). Hence a dose of optimism is needed to believe that shared resources are governable (Ostrom 1990). In the project front-end for complex systems, strategic choices are the shared resource. Hence, if we strike a parallel, front-end participants need optimism to believe that the goal is achievable—‘weren’t we optimistic and we wouldn’t get in this sort of job’, one seasoned manager said. Optimism does not excuse the promoters from the obligation to formulate an initial baseline that is true to what they know at the time. But slippages in the targets that are commensurate with the outcome of open searches for compromise solutions are legitimate self-adjustments. 32 This suggests another angle to add on to the established norms and practices to evaluate the front-end performance for complex systems projects. In collective action literature, positive performance passes by sustaining a shared resource, and as a corollary, sustaining the community that uses the shared resource (Ostrom 1990). In a complex systems project, if the front-end is successful, the shared resource evolves from strategic choices (or a ‘design-in-the-making’) into a physical asset that the frontend participants are happy to share in use; in turn, the community of production evolves into a community in use. Positive performance then requires keeping the frontend participants enfranchised. This suggests that the capability to avoid defections and thus to sustain the community of production is also a measure of performance. CONCLUSION AND RESEARCH OUTLOOK The motivation for this study is a question that has long eluded researchers: are slippages in the performance targets for complex systems projects a consequence of neglecting the front-end or are these projects too complex that they cannot be planned? To shed light on this question, we use an organizational design perspective. Here, we argue that the front-end structure of a one-off complex system project is that of a community of production facing a large collective action problem. Polycentric governance attenuates management complexity by delineating the spaces of shared strategic choice, but cannot eliminate collective action. Hence the promoter’s optimism bias is both a blessing and a curse. It is necessary to believe that the goal is achievable, but fuels underestimation of the corresponding costs and thus exacerbates collective action problems. The trick then becomes to let the targets slip to safeguard compromise solutions whilst sustaining legitimacy for front-end activity in the public eye. Our contingent model of project performance shows different ways to tackle this difficult balancing act that comparative studies want to factor in. Projects where 33 context gives the last word over disputes to an external referee differ from those where front-end participants have skin in the game and incentives to self-resolve the issues; the amount of slack resources also changes third parties’ perception of performance. There are, however, limitations to the generalizability of the insights that suggest opportunity for future research. We derive our insights from using Design Structure Matrices and the headlines of a standard performance baseline to model the structure of the front-end problem for a diverse sample of complex systems projects. The analysis shows that polycentric governance over shared strategic choices occurs irrespectively of: i) technical system decomposability (integral versus modular); ii) promoter structure (coalition vs. solo); and iii) deadline rigidity (immovable vs. flexible). However, our four cases are large infrastructure projects occurring in the UK. Three qualities suggest that these enterprises are arguably more prone to collective action problems than other complex systems projects. First, infrastructure projects are not that technologically complex that strategic choice cannot be comprehended by a large array of autonomous stakeholders; this instils in the stakeholders a sense of legitimacy to claim a right to directly influence strategic choices. Second, the root causes of the disputes over strategic choices for long-lived infrastructure are conflicting interests and needs. Long planning horizons allow for ambiguous interpretations of supporting evidence, and open up the strategic choices for contestation and negotiation if the time to bridge the differences is limited. Hence strategic choices are not about selecting optimal options in a ‘substantive rationality’ fashion but rather the outcome of what Simon (1976) calls ‘procedural rational’ to account for possible agreements under ambiguity about future states of the world. And third, large infrastructure projects are unique in the way they directly impair property rights; many projects require land take and all almost invariably blight 34 property. Hence the project front-end is highly regulated in western-style legal systems where external umpires such as Parliaments, public inquiries, and court systems are default structures to fall back on if front-end disputes cannot be self-resolved. Taken together, these qualities make large infrastructure projects quite different from more technologically complex projects in general, and even from large infrastructure projects undertaken by authoritarian states. It remains therefore indeterminate the extent collective action problems surface in these settings Limitations notwithstanding, our insights have important implications to practice and policy. Normative expectations of performance have long dogged the reputation of large infrastructure projects. This is troubling at a time when infrastructure gaps are augmenting dramatically both in developing and developed countries due to deterioration of existing assets, population growth, urban migration, and climate change (WEF 2015). It is hard to conceive how matters can improve without a deeper understanding of front-end governance and the impact to perceptions of performance. The insights are also relevant to countries such as the UK where policy dodged the reputation problem by building massive slack resources at the project front-end. Slack masks slippages in targets, but the trick seems unsustainable when the country faces some of the highest infrastructure production costs in the world (Treasury 2013). A deeper understanding of front-end governance can allow for more shared accountability for slippages in the performance targets. This, in turn, can create opportunity to shrink slack and mitigate the risk of a self-fulfilling prophecy. In sum, this study advances our understanding of the development of complex systems by illuminating the collective action problem at the front-end. We trace slippages in the performance targets to polycentric governance. We also offer a contingency model of performance that contains a relationship between external 35 regulation and internal use of slack resources. Getting the right balance is tricky, and thus the sustainability of the front-end stage is per se a measure of performance. APPENDIX – Brief Summary of the Sampled Projects, History, and Context Crossrail Project front-end: occurs between 2001 and 2008. Prior History: The idea of building a cross-London railway first gained momentum in the seventies but the UK government dropped the plan after a few years because of cost concerns; the idea was reignited in the nineties but the project front-end again unravelled after five years. The start of the third attempt happened in 2001 when the UK and London governments formed a coalition to promote the scheme. Performance baseline: during the front-end, the goal evolved from a 9km London train to open by 2012 into a 118km high-capacity commuters’ train to open by 2017; the cost target evolved commensurately. Context: Project delivery could not start before the promoter acquired from Parliament the power to force land sales. The front-end unfolded under pressure to submit a proposal to Parliament before the 2005 elections; the Parliament took 3 years to deliberate, and in 2008 the project promoter got authorization to proceed. London Olympic park. Project front-end: occurs between 2001 and 2007. Prior history: The idea of hosting the 2012 Olympics in London emerged in 1995 after the UK lost three contests with different cities. In 2001 the UK government formed a coalition to promote the scheme with the London government and the British Olympic Association (the only entity that could bid for the Olympic games). Performance baseline: In 2002, the International Olympic Committee (IOC) opened the competition; it gave the promoter two years to develop an intermediate bid and six additional months to submit the final bid; the scope and cost forecast kept evolving during the bidding process and afterwards. Facing an immovable deadline, the promoter chose to spend the first two years after winning the contest to plan properly and produce a performance baseline (‘Yellow book’). Context: after London gained the contest in 2005, Parliament rushed to give government the power necessary to force land sales; and LOCOG, a IOC’s watchdog, formally joined the promoter organization. Heathrow Airport T2. Project front-end: occurs between 2005 and 2009. Prior History: The goal of consolidating all operations by Star Alliance, a network of airlines, in a new terminal was announced in 2005; this was the year when BAA, the private airport owner, started the construction of Terminal 5, a facility to consolidate the operations of a rival alliance and scheduled to open in 2008. Performance baseline: The initial goal was to replace the old T2 building with a new building so-called ‘Heathrow East’ to open by 2012. The front-end was initially planned to end by 2008 to coincide with a new regulatory cycle, but was later extended into another year. During the front-end, the goal evolved into a modern campus to be developed in two stages; the first stage would open by 2013 and the second by 2017/8. The first phase opened in 2014; as of 2015, no plan exists to start the second phase. Context: Delivery could not start before the performance baseline was approved by the regulator High-speed 2. Project front-end: occurs between 2009 and 2017 (first phase) and 2009-2020 (second phase) (expected dates as of 2015). Prior history: The idea to develop a new national railway gained momentum in 2008 after the global financial meltdown. In 2009, the UK government created HS2 Ltd, a public agency tasked to carry on the project front-end. Performance baseline: It consists of opening the first phase connecting London and Birmingham (225km) by 2026, and a second phase connecting Birmingham to various Northern cities (306km) by 2032/3. Context: Delivery cannot start before the government acquires from Parliament the power to force land sales. The project front-end unfolded under 36 pressure to submit a proposal to Parliament by 2013 as government hoped to see the proposal approved before the 2015 elections; the proposal was submitted in 2013, but Parliamentary work is not expected to be completed before 2017. The opening targets are yet to change. REFERENCES Altshuler, A.A., D. Luberoff. 2003. Mega-projects: The changing politics of urban public investment. Brookings Institution Press, Cambridge, MA. Astley, W.G., Fombrun, CJ (1983). Collective Strategy: Social Ecology of Organizational Environment’s. Academy of Management Review, 8 (4)576-587. Baldwin, CY, Clark, KB 2000. Design Rules: The Power of Modularity. Vol. I, MIT Press, Cambridge, MA. Baldwin, C, von Hippel, E 2011. Modeling a Paradigm Shift: From Producer Innovation to User and Open Collaborative Innovation. Organization Science, 22(6) 1399-1417 Baldwin, C., MacCormack, A., Rusnak, J. 2014. Hidden Structure: Using networks to map system architecture. Researhc Policy, 43:1381-1397. Biernacki, P., Waldorf, D. (1981). Snowball Sampling: Problems and Techniques of Chain Referral Sampling. Sociological Methods & Research, 10,141-163. Bourgeois, LJ, III. 1981. On the Measurement of Organizational Slack. The Academy of Management Review 6(1) 29-39. Brusoni S, Prencipe A, Pavitt K .2001. Knowledge specialization, organizational coupling, and the boundaries of the firm: Why do firms know more than they make? Administrative Science Quarterly, 46(4)597–621. Butcher, L. 2015. Railways: HS2 Phase 1. Standard Note 316. House of Commons Library Clark, KB, Fujimoto, T. 1991. Product Development Performance. Harvard Business School Press, Boston. Cleland, DI 1986. Project stakeholder management. Project Management Journal 17 (4) 36- 44 Cleland, DI, King, WR 1968. Systems Analysis and Project Management. McGraw-Hill, NY. Coleman, PT, Ferguson, R 2014. Making conflict work: Harnessing the power of disagreement. New York: Houghton Mifflin Harcourt. Creed, WED, Scully, MA, Austin, JR. 2002. Clothes make the person? The tailoring of legitimating accounts and the social construction of identity. Organization Science, 13: 475496. Cyert, MD, March, JG, 1963. A Behavioral Theory of the Firm, NJ. Prentice-Hall Davies, A, Brady T, 2000, Organisational capabilities and learning in complex product systems: towards repeatable solutions, Research Policy, 29, 931-953, Dietz, T., Ostrom, E. Stern, P.C. 2003. The Struggle to Govern the Commons. Science 302:1907-1912. Dvir, D. Lechler, T. 2004. Plans are nothing, changing plans is everything: the impact of changes on project success. Research Policy, 33, 1-15. Eisenhardt, KM 1989. Building theories from case study research. Academy of management review 14(4) 532-550. Eisenhardt K, Graebner M 2007. Theory building from cases: Opportunities and challenges. Academy Management Journal 50(1) 25–32. Engwall, M. 2003. No project is an island: linking projects to history and context. Research Policy, 32, 789-808. Fjeldstad, OD, Snow, CC, Miles, RE, Lettl, C 2012. The architecture of collaboration. Strategic Management Journal, 33: 734-750. Floricel, S., Miller, R. (2001). Strategizing for anticipated risks and turbulence in large-scale engineering projects . International Journal of Project Management, 19(8) 445–455 Flyvbjerg, B., Bruzelius, N., and Rothengatter, W. 2003. Megaprojects and Risk: An anatomy of Ambition. Cambridge University Press. Galbraith, J. R. Designing complex organizations. Addison-Wesley Longman Publishing Co., Inc., 1973 Gann, D., Salter, A. 2000. Innovation in project-based, service-enhanced firms: the 37 construction of complex products and systems. Research Policy 29, 955-972. Gersick, C.J. 1994. Pacing strategic change: The case of a new venture. Academy of Management Journal 37(1) 9-45. Geyer, A., Davies, A. 2000. Managing project-system interfaces: case studies of railway projects in restructured UK and German markets. Research Policy 29, 991-1013. Gibbons, R., Henderson, R. 2011 Relational Contracts and Organizational Capabilities. Organization Science 23: 1350-64. Gil, N. 2007. On the Value of Project Safeguards: Embedding Real Options in Complex Products and Systems. Research Policy, 36 (7) 980-999 Gil, N., Tether, B. 2011. Project Risk Management and Design Flexibility: Analysing a Case and Conditions of Complementarity. Research Policy, 40, 415-428. Gil, N, Baldwin, C. 2013. Sharing Design Rights: A Commons Approach for Developing Infrastructure. Harvard Business School working paper, 14-025, January Gray, B. 1989 Collaborating: Finding common ground for multiparty problems. San Francisco, CA: Jossey-Bass. Gray B 1989a Conditions facilitating interorganizational collaboration. Human Relations 38 (10) 911-936. Gray, B., D. Clyman 2003 Difficulties in fostering cooperation in multiparty negotiations. In M. West, et al. (eds.), International Handbook of Organizational Teamwork and Cooperative Working: 401-422. Chicester, UK: Wiley. Greiman, VA 2013, Megaproject management: Lessons on Risks and Project Managemen from the Big Dig Wiley, Hoboken, NJ. Hall, P 1972. Great Planning Disasters. Berkeley, CA. University of California Press. Hardin, G. 1968. The Tragedy of the Commons. Science 162: 1243–1248 Hargrave, TJ, Van de Ven, AH 2006. A collective action model of institutional innovation. Academy of Management Review, 31: 864-888. HM Treasury (2013). Infrastructure Cost Review: Annual report 2012 to 2013. Hobday, M. 1998. Product Complexity, Innovation and Industrial Organization. Research Policy 26, 689–710. Hobday, M. 2000. The project-based organisation: an ideal form for managing complex products and systems? Research Policy 29,871–893 Hughes, T.P., 1987. The evolution of large technological systems. In: Bijker, W., Hughes, T.P., Pinch, T. (Eds.), The Social Construction of Technological Systems. Cambridge, 51–82 Jessop, B. 1997. The governance of complexity and the Complexity of Governance. In Ash Amin and Jerzy Hausner, eds. Beyond Market and Hierarchy, 95-128. Edward Elgar. Jick, T.D. 1979. Mixing qualitative and quantitative methods: triangulation in action, Administrative Science Quarterly 24 (4) 602-611. Ketokivi, M., Mantere, S. 2010. Two strategies for inductive reasoning in organizational research. Academy of Management Review, 35: 315-333. King, A., Lenox, M. 2000. Industry self-regulation without sanctions: The chemical industry’s responsible care program. Academy of Management Journal 43 (4) 698-717. Knudsen, T., D. A. Levinthal. 2007. Two faces of search: Alternative generation and alternative evaluation Organ. Sci. 18(1) 39–54. Lakhani, KR, Jeppesen LB, Lohse PA, Panetta JA 2007. The Value of Openness in Scientific Problem Solving. HBS Working Paper No. 07-050. Harvard University, Cambridge, MA Langley, A 1999. Strategies for theorizing from process data. Academy of Management Review, 24: 691-710. Lawrence, TB, Hardy, C, Phillips, N 2002. Institutional effects of interorganizational collaboration: The emergence of proto-institutions. Academy of Management Journal, 45: 281–290. Latour, B. 1986. The powers of association. In J. Law (ed.), Power, action and belief: 264-280. London: Routledge. Lincoln, YS, Guba, EG 1985 But is it rigorous? Trustworthiness and authenticity in naturalistic evaluation. In D. D. Williams (ed.), Naturalistic evaluation. San Francisco: Jossey-Bass. Lundrigan, C., Gil, N., Puranam, P. 2015. The (Under) Performance of Mega-projects: A meta38 organizational Approach. Proc. 75th Academy of Management conference. March, JG, Simon, HA 1958. Organizations, New York, John Wiley & Sons, Inc. March, JG, Sutton, RI 1997. Crossroads-Organizational Performance as a Dependent Variable. Organization science 8(6) 698-706. Merrow, EW, McDonwell, L.M., Arguden, R.Y. 1988. Understanding the Outcome of Megaprojects. Rand Corporation. Santa Monica. Mihm, J., Loch, CH, Wilkinson, D. Huberman, BA 2010. Hierarchical Structure and Search in Complex Organizations. Management Science 56(5) 831-848. Miles, M.B., Huberman, A.M. 1994. Qualitative Data Analysis: an expanded sourcebook (2nd edition) Sage Publications, Thousand Oaks. California Miller, R., Hobday, M., Lewroux-Demer, T., Olleros, X., 1995. Innovation in complex systems industries. The case of flight simulation. Industrial and Corporate Change 4 (2), 363–400. Miller, R, Lessard, D 2000. Public Goods and Private Strategies: Making Sense of Project Performance in the The Strategic Management of Large Engineering Projects. Roger Miller and Donald Lessard (edts). Cambridge, MA, MIT Press. Miller R. and Ollero, X.2000. Project Shaping as a Competitve Advantage. in the The Strategic Management of Large Engineering Projects. Roger Miller and Donald Lessard (edts). Cambridge, MA, MIT Press. Morris, P.W. 1994. The management of projects. Thomas Telford Morris, PW, Hough GH 1987. The Anatomy of Major Projects: A Study of the Reality of Project Management. Wiley. Chichester. National Audit Office (2007). Preparations for the London 2012 Olympic and Paralympic Games, Report by the comptroller and auditor general, HC 252 Session 2006-2007 National Audit Office (2014). Crossrail. Report by the Comptroller and Auditor General. Department for Transport. 24 January Norris, E., Rutler, J. Medland, J. 2013. Making the Games. What Government can Learn from London 2012 Institute for Government, January. O'Mahony, S, Ferraro, F 2007. The emergence of governance in an open source community. Academy of Management Journal, 50: 1079-1106. O'Mahony, S, Bechky, BA 2008. Boundary Organizations: Enabling Collaboration among Unexpected Allies. Administrative Science Quarterly 53(3) 422-459. Ostrom, E. 1990. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press, Cambridge, UK. Ostrom, V. 1972. Polycentricity. Presented at 1972 Annual Meeting of the American Political Science Association, Washington, DC, September 1972. Ostrom, E 2005.Understanding Institutional Diversity. Princeton NJ Princeton University Press. Ouchi, W 1980. Markets, bureaucracies, and clans. Administrative Science Quarterly, 25: 125– 141. Pfeffer J, Salancik, GR 1978. The external control of organizations. New York, Harper & Row Pinto, JK, Slevin, DP 1987. Critical success factors in successful project implementation, IEEE Trans Engineering Management, 22-27. Pinto, JK., Winch G. (2015). The Unsettling of “Settled Science:” The Past and Future of the Management of Projects. In press. Reilly 2001. Collaboration in action: An uncertain process. Administration in Social Work 25(1):53-73. Ring, P.S.,Van De Ven, AH. 1992. Structuring Cooperative Relationships between Organizations. Strategic Management Journal, 13 (7) 483-498. Ross, J., Staw, B.M. 1986. “Expo 86: An Escalation Prototype,” Administrative Science Quarterly, 31 (2): 274-297. Rothschild, J., Russell, R. 1986. Alternatives to bureaucracy: Democratic participation in the economy. In J. F. Short (Ed.), Annual review of sociology, 12: 307–328. Palo Alto, CA. Rittel, HWJ, Webber, M 1973. Dilemmas in a General Theory of Planning. Policy Sciences 4, 155-169. 39 Rivkin, JW, Siggelkow, N. 2003. Balancing search and stability: Interdependencies among elements of organizational design. Management Science 49(3) 290–311. Scott, WR 1987 Organizations Rational, Natural, and Open Systems (2nd ed.)Prentice-Hall, Englewood Cliffs, NJ. Shapiro, A., Lorenz, C. 2000. Large-Scale Projects as Complex Systems: Managing Scope Creep. The Systems Thinker 11(1) 1-5. Siggelkow, N. 2007. Persuasion with case studies. Acad. Management J. 50(1) 20–24. Simon, HA 1962. The Architecture of Complexity. Proc. American Philosophical Society, 156: 467-482. Simon H.A. 1969 The Sciences of the Artificial. Cambridge. MIT Press. Simon, HA 1976. From Substantive to Procedural Rationality.In Method and Appraisal in Spiro J. Latsis, ed.Method and Appraisal in Economics, 129-48. Cambridge University Press. Sosa, M., Eppinger, S. Rowles, C. 2004. The misalignment of product architecture and organizational structure in complex product development. Management Science, 50(12) 1674-1689. Steward, D 1981. The design structure matrix: A method for managing the design of complex systems. IEEE Transactions Engineering Management 28(3): 71-74. Stinchcombe, A.L., C.A. Heimer. 1985. Organization theory and project management: Administering uncertainty in Norwegian offshore oil. Scandinavian University Press. Stone, MM, Brush, CG1996. Planning in ambiguous contexts: the dilemma of meeting needs for commitment and demands for legitimacy. Strategic Management Journal, 17 (8) 633-652. Strauss, A., Corbin, JM. 1990. Basics of qualitative research: Grounded theory procedures and techniques.Thousand Oaks, CA, US: Sage Publications, Inc. (270 pp. Susskind L., Cruikshank J. 1987. Breaking the impasse: Consensual approaches to resolving public disputes New York: Basic Books Suchman, MC1995. Managing legitimacy: strategic and institutional approaches Academy of Management Review, 20, 571–610 Szyliowicz, JS, Goetz. AR 1995. Getting realistic about megaproject planning: The case of the new Denver International Airport. Policy Sciences 28 (4) 347-367. Thomson AM, Perry JL 2006. Collaboration processes: Inside the black box. Public Administration Review 66(S1): 20-32 Tuertscher, P., Garud, R., Kumaraswamy, A. 2014. Justification and Interlaced Knowledge at ATLAS, CERN. Organization Science 25(6):1579-1608 Van de Ven AH 1976. On the nature, formation, and maintenance of relations among organizations. Academy Management Review, 1(4):24–36 Von Hippel E Von Krogh G 2006. Open source software and the “private-collective” innovation model: Issues for Organization Science, Organization Science, 14:209-223. Wachs, M. 1989. When Planners Lie with Numbers. Journal of the American Planning Association, 55 (4): 476-79. Weick, KE, and Roberts, KH 1993. Collective mind in organizations: Heedful interrelating on flight decks. Administrative Science Quarterly, 38, 357–381 Williamson OE 1996. The Mechanisms of Governance. Oxford University Press, Oxford. Winch, G., C. Millar and N. Clifton 1997. Culture and Organization: The Case of Transmanche-Link', British Journal of Management, 8, 237-249. World Economic Forum (2015). The Global Competitiveness Report 2014-2015. Klaus Schwab and cavier Sala-i-Martin (editors). Geneva. Yin, R.K. (1984). Case Study Research. Design and Methods (3rd Edition 2003). Applied Social Research Methods Series, Vol. 5. 40 Table 1 - Description of the Sample of Cases, Interviewees, and Archival Database Cases System-level goal London 2012 Heathrow T2 London Crossrail UK High-speed 2 Build an Olympic Build a new terminal Connect the East and Build a cross-country park to host Games campus to create a West outer London railway to increase and catalyse urban dual-hub airport with a new highnational transport regeneration capacity train system capacity Organizational Coalition: UK and Solo: Coalition: UK and Solo: structure of the London governments; BAA (regulated airport London Governments; UK Government promoter BOA(§); LOCOG(§) owner) City of London Outcome of the Formal agreement Legal contract Legal contract Legal contract project frontend Yellow book (2007) 5-year capital Parliamentary bill Parliamentary bill Blue book (2009) investment plan Interviews 36 (11 controversies) 19 (5 controversies) 33 (9 controversies) 35 (12 controversies) Number and 8: 5: 8: 11: description of London2012 (bid STAR Alliance, Air CLRL (promoters’ HS2 Ltd (promoter’s organizations company) ODA Canada, BAA, planning agent); Manchester City interviewed (promoters’ agent); HETCo and Balfour agent);Crossrail, Council (MCC); Greater LOCOG (games Beatty (private design (promoters’ delivery London Authority operator); OPLC and build companies) agent); Network Rail; (GLA);Transport for (park operator); UK Treasury; London (TfL); Borough Transport for London Transport for London of Camden; Transport for (TfL); CLM (TfL);Canary Wharf Greater Manchester (programme (private funder); (TfGM); Network Rail; manager); Land Bechtel, Transcend UK Treasury; Lease (private (consultants) Manchester Airport; developer); Network CH2MHill, AECOM Rail (owner of UK (consultants) rail infrastructure) Archival data: Total Number of Total Number of Total Number of Total Number of Total number Documents: 469 Documents: 114 Documents: 122 Documents: 66 of documents (except news Strategy and Strategy and planning Strategy and planning Strategy and planning articles) planning documents: documents: 74 documents: 74 documents:22 organised by 260 Financial reports: 6 Financial reports: 2 Financial reports: 2 categories Financial reports: 6 Formal Formal Formal communication: Formal communication: 19 communication: 6 20 communication: 5 Newsletters and PR Newsletters and PR Newsletters and PR Newsletters and PR documents: 8 documents: 23 documents:12 documents: 111 Design documents: 4 Design documents: 9 Design documents: 8 Design documents: Meeting minutes: 3 Meeting minutes: 8 Meeting minutes: 2 16 Meeting minutes: 71 (§) LOCOG, London Organising Committee of the Olympic and Paralympic Games and International Olympic Committee (IOC)’s watchdog; BOA - British Olympic Association (*) Inflation in construction prices was nil between 2008-2013 (¥) £15.9.bn (final prices for a completion date around 2016) is about £11.1bn at 02 prices (assuming 3.5% discount factor) 41 Figure 2- Representative excerpts of the Design Structure and Companion Organizational Matrices for the four Projects 42 Table 2- Excerpt of Protocol to Uncover Interdependences between Strategic Choices and Distributed Ownership of Strategic Choices 43 44 Table 3- Excerpt of Evidence for Source of Front-end Dispute, Outcome, and Mechanisms to Resolve Dispute 45 46 47
© Copyright 2026 Paperzz