Collective Development of Flexible Capital Designs: The case of Future-proofing Infrastructure
This mixed-methods study investigates a dilemma that inter-organizational collectives formed to develop long-lived capital assets face at the project front-end: either invest in flexible designs that cope economically with change in requirements, this is design to evolve—at risk the extra costs upfront will not pay off if the uncertainties fail to resolve favourably later on. Or endorse cheaper but more rigid designs—at risk of higher adaptation costs if the uncertainties materialise in the future. Through an empirical study grounded in the British railway sector, we reveal that the collectives regularly engage in informal future-proofing discussions to address this dilemma. But faced with tight budgets, conflicting interests, and mutual ignorance, the different parties struggle to design in flexibility. Through lab experiments, and taking a flat governance structure as a given, we then unexpectedly find that a formal device to facilitate future-proofing discussions has limited impact on the outcomes. Hence, we argue, a collective action problem is central to the inter-organizational development of flexible capital designs. We conclude by discussing how inter-organizational collectives can better exploit the value of design flexibility at the project front-end.
Managerial relevance
Designing long-lived capital assets to cope with evolution in design requirements when capital resources are scarce is a fundamental problem facing practitioners at project planning, and one we still know little about. These are assets developed not by unitary organizations, but by collectives formed by independent organizations. Our study shows that the collectives invariably engage in future-proofing discussions. But when the capital resources are scarce, the collectives struggle to converge on the value of flexible designs. Through a mixed-methods study, we illuminate the collective action nature of the problem. We demonstrate that with a flat governance structure, formal devices to make the process more efficient have limited impact. We conclude by discussing changes in the formal structure governing the collectives so as to encourage more investment in design flexibility.
Keywords
Collective development, design flexibility, infrastructure, projects, governance, commons
Introduction
Exploring the value of design flexibility to manage the tension between efficiency and effectiveness is an enduring goal in the projects and development literatures, and thus in the development of both new commercial products (Iansiti 1995, Ulrich 1995, Thomke 1997, Baldwin and Clark 2000) and large engineering and infrastructure systems, the so-called capital projects (Klein and Meckling 1957, Morris 1994, Dvir and Lechler 2004, Miller and Lessard 2001, Gil and Tether 2011, Neufville and Sholtes 2011). Efficiency pertains to delivering projects on time and within budget whereas effectiveness to ensuring that the development process copes economically with change in the design requirements.
Flexible designs enable to reduce adaptation costs and avoid premature lock-in into early commitments when requirements are uncertain, and thus mitigate the downside risks of schedule and budget overruns. Flexible designs can be achieved by modularizing design structures (Thomke 1997, Baldwin and Clark 2000) or by safeguarding integral designs (Gil 2007). But physical constraints and technology make it hard to modularize capital designs notably for infrastructure (e.g., transport and power systems, schools, hospitals). Hence Gil and Tether (2011) argue that in capital projects design flexibility and risk management practices, e.g., planning contingencies, project controls, risk registers, are complements.
Complementarities notwithstanding, empirical regularities suggest that risk management practices prevail in capital projects (Morris 1994, Miller and Lessard 2001, Gil and Tether 2011). Noting this, Lenfle and Loch (2010) argue that project management practice puts too much emphasis on control and risk management to the detriment of flexibility.
But capital projects are collective action arenas crowded by many stakeholders who claim rights to directly influence high-order decisions (Gil and Baldwin 2013). Governing these arenas are not unitary actors, but complicated coalitions of promoters with their agent, the project manager. To succeed when planning a new scheme, the coalition engages in informal conversations to craft a mutually acceptable scope and budget (Lundrigan, Gil, and Puranam 2014). But as group studies show lack of familiarity (Newcomb 1956) and formal group-level plans and rules (Leavitt 1958, Sherif 1958, Seiler 1963, Pinto and Pinto 1993) complicate cooperation. And complicated governance structures for allocating roles and controlling and integrating work (Child 1972) will also ‘constrain action’ (Davis, Eisenhardt, and Bingham 2009). This conflation of informal processes and complicated formal governance structures over the planning of a new capital project suggests that attributing the lack of investment in flexible designs to managerial ignorance and/or incompetence is excessively reductionist.
Hence the core question motivating this study is: can we trace the difficulties to invest in flexible capital designs to inefficiencies in the development process and to the governance structure of the inter-organizational collectives that design and develop the new assets?
To address these questions, we combine inductive case reasoning with lab experiments. First, we use an empirical study grounded in Britain’s railway sector to explore the motivations and obstacles to develop flexible designs. Our findings reveal that the new schemes are planned by collectives of autonomous parties which operate under no formal hierarchy, binding contracts or authority of a systems integrator (Brusoni, Prencipe and Pavitt 2001). To address the trade-offs between rigid versus flexible designs, the collectives engage in ad hoc ‘future-proofing’ talks, but they invariably struggle to resolve their differences.
Taking the observed flat governance structure of the collectives as a given, we then use a lab experiment to investigate the impact of facilitating the future-proofing discussions with an administrative device. Our results reveal no push back on efforts to formalise future-proofing talks. But unexpectedly do not reveal statistically significant impacts. These findings suggest a link between flat governance and the difficulties that collectives of heterogeneous, cash-strapped parties face to develop flexible designs. Modifying the governance we thus argue is a pre-requisite to enable the collectives to exploit better the value of capital design flexibility.
We organize the remainder of this article as follows. After reviewing literature on design flexibility in capital projects, we explain our methods. In the analysis, we first empirically examine capital design practices in Britain’s railway sector, and then analyse the results of the lab experiments. After arguing how a collective action problem is central to the inter-organizational development of flexible capital designs, we conclude by discussing whether polycentric governance (Ostrom 1990), an approach laying between flat and centralised governance, can do a better job in facilitating the exploitation of capital design flexibility.
Background
Flexible designs can be produced through near-decomposable (Simon 1962) or modular structures (Baldwin and Clark 2000). Design structures allocate the functions of a product to its components (Ulrich 1995). Modular designs break apart the interdependencies between functional components, and thus have design options built-in that allow the design as a whole to evolve economically. Limited flexibility can also be designed in more integral structures using design safeguards, this is built-in redundancies and spare capacity (Gil 2007).
In commercial product development, flexible designs enable manufacturers to reuse the designs across projects and thus extend design longevity (Thomke 1997). In capital projects, flexible designs also enable design reuse—Intel, for example, reuses designs for semiconductor fabrication facilities (fabs) to speed up new fab development and ramp-up (Gil and Beckman 2007). But more importantly, flexible capital designs reduce the adaptation costs over the project time and then over the asset’s operating life (Gil and Tether 2011).
Crucially, investments in design flexibility increase capital costs (Baldwin and Clark 2000, Gil 2007, Suh, Weck, and Chang 2007). Neufville and Scholtes (2011) contend that capital projects with flexible designs are cheaper than those with integral designs because the capital investment can be staggered as the uncertainties resolve favourably. But if the uncertainties never resolve favourably, investing in a rigid design commensurate with the upfront requirements would be more economical even if riskier; furthermore, investing in options to switch operating regimes requires extra expenditure. Hence design flexibility is like buying an insurance policy (Neufville and Scholtes 2011). There is no free lunch, and mitigating the risks of high adaptation costs through design flexibility involves an extra cost upfront.
The tension between short-term savings associated with rigid designs versus the potential long-term gains of built-in flexibility makes it crucial to evaluate the pay offs of flexbile designs. To this purpose, scholars have used real options theory (Trigeorgis 1996, Neufville and Scholtes 2011, Cardin et al. 2013, Cardin 2014). A real option gives its ‘buyer’ the ability to postpone an investment whilst leaving open the option to invest under uncertainty (‘exercise’ the option). Likewise, flexible capital designs have built-in options that can be exercised if the future resolves favourably. Hence flexible designs enable to benefit from upside scenarios under uncertainty while limiting the losses on the downside.
The uptake of the real options approach has, however, been very slow due to difficulties in ensuring the tractability of real options models, in making reliable numeric assumptions, and in eliminating endogenous processes (Lander and Pinches 1998). An alternative research vein, real options reasoning, derives qualitative statements from real options theory and asks managers to specify their level of agreement before setting priorities and allocating resources (McGrath and MacMillan 2000). This approach has been explored by organizations to inform investment on new technologies and R&D, but less so to evaluate capital design flexibility.
Extraordinarily, extant literature rarely discusses who actually pays for the upfront cost of flexible designs. Real options theory downplays this issue as it assumes that this cost is marginal (Trigeoris 1996). Likewise, in new product development, the pay-offs of modular designs largely outweigh the extra costs incurred upfront (Baldwin and Clark 2000).
But in capital projects, flexible designs require significant investment upfront (Gil 2007). Building a tunnel to safeguard an airport expansion or designing a dual-purpose stadium (to host athletics and football) is not cheap. The extra investments may be the right decisions considering the life-cycle costs, but will not happen if they are deemed unaffordable. This problem is amplified in infrastructure projects where development decisions are not up to a unitary organisation (Klein and Meckling 1958), but to a coalition formed by autonomous parties each with its own interests, beliefs, and planning horizons (Lundrigan et al. 2014).
All in all, our understanding is scarce as to how collectives of autonomous parties decide on design flexibility. Gil and Tether (2011) argue that uncertainty, which makes flexibility more valuable, paradoxically puts off the collectives from investing in flexibility because it precludes the development of a common vision. Since options logic is intuitive, we ask: can the difficulties in developing flexible designs be traced instead to process inefficiencies and the governance structure? These are the questions motivating our study.
Research Methods
We adopted a mixed-methods study to further our understanding of how investments in design flexibility play out in capital projects. Through an empirical study, we first reveal how inter-organizational collectives use options logic intuitively to frame ‘future-proofing’ discussions. We then use a two-group controlled experiment to test whether introducing an administrative device to facilitate future-proofing discussions leads to more efficient discussions, increases participants’ satisfaction with the process, and changes the outcomes.
Lab experiments complement observational studies as they allow for investigating how changes in specific variables influence the participants’ behaviours and outcomes (Ostrom 2006, Croson, Anand, and Agatwa 2007). Lab experiments also allow to control for confounders that would compromise internal validity such as contextual variables, and to create lab conditions that are replicable to test empirical propositions. Studies of the correspondence between experimental and observational findings have refuted claims that lab experiments lack external validity (Anderson, Lindsay, and Bushman 1999), and show that lab studies can simulate effectively critical features of group decision-making (Gersick 1989).
For this mixed-methods study, lab experiments were important to investigate if the observed empirical regularities—the tussles over flexible capital designs—could be attributed to the informal nature of the future-proofing talks. The lab experiments allowed to control for the observed flat governance of the collectives, which we took as a given.
We grounded our field study in the British railway sector. The tension between designing a flexible scheme versus progressing with a rigid one is central to railway projects, which involve capital investments to produce assets that are expected to operate over 100 years. Railways are also an empirically relevant sector attending to the sheer scale of rail work underway around the world.
The UK case is particularly interesting because 70% of its railway infrastructure is around 100 years old (CIRIA 2009). Whilst the Victorian railway infrastructure was designed to evolve, it now operates at full capacity in critical parts, and thus Network Rail (NR)—the regulated public monopoly that owns the UK’s railway infrastructure—spends £2bn yearly in capital projects. To complicate things, the regulator is challenging NR to deliver more for less. ‘Hard times. Great expectations,’ said the NR’s chief executive in 2012 about the mismatch between the government’s plans to cut their budget and the public expectations.
To explore how project collectives negotiate between short-term affordability and long-term flexibility, we formed a diverse and polarised sample as recommended for process-focused inductive studies (Siggelkow 2007). Our sample varies in the role of NR and in the project size. Specifically, Warrington is a £15m scheme to develop a new rail chord in which a local council plays the promoter role and NR the supplier role. Reading is a very large £850m scheme to revitalise a national railway junction funded by the UK government, and thus NR acts both as the government’s agent and project supplier. And Salford is a £12m station redevelopment also funded by the UK government. Table 1 summarises our data.
Table 1: Summary of the sampled projects
Project / Warrington / Reading / SalfordScope / Build a new rail bypass (chord) to connect two existing freight lines / Build new station platforms and concourses; improve the layout / Modernize the station platforms and station building
Goal / Release land to facilitate access to city waterfront / Increase the capacity of a critical railway station / Eliminate overcrowding on the platform
Anticipated final cost / ~£15m / ~£150m (station building); overall project costs ~ £850m / ~£12m
Front-end planning / ~2 years (Oct 2008 - Dec 2010) / ~2 years (Feb 2006 - Jan 2008) / ~1.5 years (May 2009 - Sep 2010)
Estimated project duration / 6 years (2008-2014) / 9 years (2006-2015) / 5 years (2009-2014)
Key participants in front-end planning / NR, local council, freight operating companies, public regeneration agency / Central government (DfT), local council, NR, property developers, train operating companies / University, NR, train operating companies, local council, DfT, regeneration and transport agencies
Sense of urgency / Debatable
Low for NR, high for the Local Council / High across the board
Urgent need to resolve major capacity bottleneck / High across the board
Closure of station imminent due to risks
Data collection involved semi-structured interviews, two workshops, analysis of archival documents, and attendance of eight project meetings. Our key informant was a senior NR programme manager who provided the second author access to the NR Infrastructure Division, restricted access to the intranet, and introductions to key staff members. Between February 2010 and August 2011, we conducted 34 one-on-one meetings with staff from NR (e.g., project manager, engineer, risk manager, commercial sponsor[1]), project clients, and other design participants identified through a snowball effect (Biernacki and Waldorf 1981). We also organised two, two-hour long workshops focused on design flexibility in capital projects, each attended by 8 employees of NR. We recorded and transcribed the interviews and the workshops, and complemented this data with notes taken from informal chats entertained by the second author during a placement at the NR Infrastructure Division.