Governability: Holding a Project Coalition Together When Surprise Arrives
Governability: Holding a Project Coalition Together When Surprise Arrives
On this page
Governability: Holding a Project Coalition Together When Surprise Arrives
Across both mature and emerging economies, investment in large capital projects — energy and process plants, transport networks, mining complexes, water systems and industrial infrastructure — continues to bind capital, talent and public attention for a decade or more. Sponsors routinely prepare contracts, contingencies and stage-gate checklists before first concrete is poured. Yet when a turbulent event arrives after commitment hardens, those instruments often prove thinner than expected. Parties who seemed aligned begin to exploit ambiguity. Ordinary contingency is exhausted by the second or third stacked shock. Leadership exhortation does not by itself contain the clash. What then holds the coalition of owners, contractors, financiers, governments and communities together long enough to invent a response?
That quality has a precise name in the research tradition that studied sixty large engineering projects through comparative cases: governability — the property of a project coalition under surprise. It is not the same thing as an owner's gate rules, not the same thing as a lump-sum clause, and not the same thing as a better forecast. It is built from relationships and devices chosen for how they will behave in a crisis, written while commitments are still cheap.
This comprehensive blog will explore governability as a continuous thread across the sponsorship clock of large capital projects. We will delve into the difference between identifiable risk and genuine surprise, the search–sprint–ramp-up journey in which governability is built or neglected, and the four families of devices — reserves, triggers, flexibility and generativity — that keep coalitions from flying apart. Furthermore, we will examine strategic systems and the scope of protection, the layering of risk and who should bear it, the institutional forms that each induce their own failure modes, shaping episodes that produce a committable package, and the neighbouring practices of owner gate governance and honesty incentives that must be stacked rather than merged. Whether you are a project sponsor, owner's representative, government official, investor, lender or project development professional, understanding how to hold a coalition together when surprise arrives is key to delivering large capital projects that remain steerable after the first plan meets the world.
1. When the Coalition Meets What No Clause Foresaw
Large capital projects are one-off, customized and politically exposed. In the comparative sample that anchors this essay, average investment approached a billion dollars, the front-end search averaged well over six years, and the engineering–procurement–construction sprint averaged about four years. Community opposition, international pressure, multiple uncoordinated agencies and political influence on initiation were common, not rare. Failure, in that reading, was more often bad handling of externalities, a thin read of the institutional milieu, or coalition dynamics than a pure failure of estimating technique.
Against that backdrop, the familiar reflex is to write a denser contract or a thicker contingency. Both matter. Neither is sufficient when the event that arrives was never on the risk register that shaped the base case. Identifiable risks can be insured, hedged or contracted. Indeterminacy and surprise cannot be given a specific pre-written script. When those events hit, parties who still have exit options, asymmetric information or political leverage will use them unless devices already exist that trigger joint action, hold deeper reserves, preserve flexibility and allow the coalition to invent a response rather than retreat into enforcement.
Governability is that set of relationships and devices. It is a property of the coalition under turbulence, not a paragraph in the EPC agreement and not a steering-committee calendar. Leadership is necessary and not enough: parties exploit ambiguity. Ordinary contingency funds are not enough when several bad events stack. Building governability is described, in that tradition, as judgment about crisis behaviour — each device chosen for how it will behave when the base-case model no longer holds.
2. Roadmap
The argument that follows is one continuous thread, not a gallery of independent stories. It opens by separating identifiable risk from surprise, then places governability on the sponsorship clock of search, sprint and ramp-up. It unpacks the device families and the broader strategic system of which they form part, follows risk layering into finance and renegotiation, situates projects inside institutional forms that each induce characteristic failures, and closes with shaping discipline and a practice checklist. Neighbouring ideas — owner gate rules that allow or deny a project, honesty incentives in behavioural governance, contract strategy as a governance choice, and scenario work under ambiguity — appear only as bridges. They are not retold here, and their footnotes are not merged with the governability construct.
3. Brief Bridges — What This Post Is Not
Several companion ProTrain pieces sit beside this one. Leading Through Ambiguity addresses contested meaning and turbulent fields with scenario planning as a service to decision readiness — ideas drawn from the Oxford Scenario Planning Approach, applied as transfer to capital projects. The Outside View addresses reference-class forecasting and the planning fallacy. How You Contract Is How You Govern treats contract form as a governance decision and ends with a short governability coda that this essay expands into the main subject. Search, Sprint, Ramp Up traces the decision journey clock that this essay uses as frame without retelling every statistic. Owner gate governance — the set of rules by which an industrial owner allows or denies projects at stage-gates, forcing Basic Data, shaping and front-end loading to be finished — is a neighbouring construct that shares the English word “governance” and a different mechanism. Behavioural work on forecasting, organizing and executing large projects adds honesty incentives and modularity; those belong in the stack as adjacency, not as a substitute for coalition devices under surprise.
Readers who need those companions should read them on their own terms. What follows stays on governability: holding the coalition together when surprise arrives.
4. Identifiable Risk Versus Surprise
A useful first cut separates what can be written into a contract from what cannot. Technical completion risk, construction risk, demand and supply risk, financial-catastrophe risk, and social or institutional risk all appear in the landscape of large engineering projects — and they differ by asset. Oil platforms can be technically hard and institutionally easier; thermal power can push market risk into a purchase agreement or take merchant price risk; hydro is often socially explosive. Risks also move over the life of the project. Managerial responses observed across the sample include selecting projects that can earn revenue, investing in information while commitments are still low, shifting volume and revenue risk with contracts, and embedding shock absorption — shadow revenues, sponsor balance sheets — where a catastrophe would otherwise be fatal.
Layering is the complementary principle: face risks with a stack of moves, not a single hedge. Relative advantage in bearing risk says each piece should sit with the party best able to control or carry it. Completion risk is reduced by early search, not by pretending geology is known. Complete allocation is often impossible. Some risks are large, late and not controllable by contract; those need equity, guarantees or an explicit renegotiation path agreed before the sprint.
None of that layering replaces governability. Layering allocates what can be named. Governability is what remains when the name was never on the list, or when several named risks arrive together and exhaust the ordinary cushion. A complete contract that assumes no surprise is not governability. The renegotiation and reserve rules have to be written while commitments are still cheap — during search, not after the first stacked shock in sprint or ramp-up.
Figure 1: Identifiable risks can be insured, hedged or contracted; surprise requires governability devices that keep the coalition able to act together.
5. The Sponsorship Clock — Search, Sprint, Ramp-Up
Governability is not built in a vacuum. It is built on a sponsorship clock that the same research tradition describes as search, sprint and ramp-up. Search is the long front end: issues, coalition, permits, institutional fit and shaping. In the sample, average front-end search was on the order of seventy-nine months. Sponsors still have freedom to change the configuration. Sprint is engineering, procurement and construction once a committable package exists — coalition formed, rights and permits secured, risk allocation settled. Average EPC sprint in the sample was on the order of forty-nine months. Ramp-up, lasting roughly six months to two years, tests markets, technical function and social acceptability. The project then performs, limps, or is financially restructured. The journey to revenue generation is about ten years.
Competent sponsors treat front-end funds as required: on the order of three to five percent of cost for standard projects and up to thirty-five percent for complex opportunities. A foreword claim in the same tradition puts front-end exploration of coalition, governance, institutions, the state's role, and social and ecological impact as high as a quarter of total project cost in demanding cases. Many executives still treat community, legal and negotiation spend as optional. That underfunding is where governability dies before it is born: the devices that would later hold the coalition are never designed because search was treated as delay.
Two further distinctions keep this clock honest. First, efficiency (time, cost, technical function) is not the same test as effectiveness (social, environmental and developmental fit, and whether the asset is actually wanted). A project can be built “successfully” on efficiency metrics and still fail because effectiveness issues were not designed in early. Second, search–sprint–ramp-up is a sponsorship clock. Owner stage-gate phases used in industrial capital projects — the familiar front-end loading sequence — address a related problem with a different sample, different units and different success tests. Both say the front end is where value is created or lost. They are not interchangeable phase names, and this essay does not merge them.
Figure 2: Search, sprint and ramp-up on the sponsorship clock — governability devices are designed in search, before the sprint begins.
6. Devices That Hold the Coalition — Reserves, Triggers, Flexibility, Generativity
When turbulent events hit, the question is whether the coalition can act together. The devices that support that property fall into recurring families.
Reserves go beyond ordinary contingency. One bad event is not the design case. When several shocks stack, a thin contingency is already spoken for, and the next negotiation becomes a fight over who absorbs the residual. Deeper, deliberately placed reserves — financial, schedule and scope buffers matched to crisis behaviour — are part of governability, not a rounding error in the estimate.
Triggers are pre-agreed conditions that force joint action rather than unilateral retreat. They convert ambiguity into a meeting, a renegotiation window or a shared decision right before positions harden. Without triggers, each party waits for the other to blink, and ambiguity becomes a weapon.
Flexibility preserves options: modular design choices, phased commitments, alternate suppliers, shadow revenues, or contractual paths that allow reconfiguration without treating every change as a breach. Flexibility is costly in the base case; its value appears when the base case is wrong.
Generativity is the ability of the coalition to analyze, debate and invent a response rather than only to enforce a clause. Pure enforcement does not create value in a crisis that the clause never contemplated. Deliberation does. Generativity is therefore not soft culture talk; it is a designed capacity — forums, information rights, facilitation and authority to invent — that must exist before the crisis, because by construction the formal rules are mostly set once the sprint is under way.
Choosing among these devices is judgment. Each is selected for how it will behave when several parties still have exit options and the schedule is already tight. An index of governability in the sixty-project work was built from relationships and devices of this kind. The practical implication is blunt: write the renegotiation path, the reserve rules, the triggers and the generative forums during search, while the cost of designing them is still low relative to the cost of improvising them under fire.
Figure 3: Four families of governability devices — reserves, triggers, flexibility and generativity — chosen for crisis behaviour, not for the base-case model.
7. Strategic Systems — Scope of Protection, Not Scope of Facilities
Individual devices sit inside a broader strategic system: the set of distinct protections a sponsor actually uses to face risk — contractual shifts, design additions, reserves, exchange strategies and similar. Recurring bundles of devices are templates. Every device has a cost; past some point more devices do not create value, and sponsors judge that margin themselves.
The comparative work coded which devices were actually used on each of the sixty projects and built an index of scope — how many different kinds of protection were deliberately built, not the physical scope of facilities. Broader deliberate scope was associated with better project performance. The sample is explicitly too small for statistical generalization; the method is comparative cases plus grounded theorizing. What the association supports as a leadership claim is narrower and still important: a thin register matched only to the base case is a failure mode, because the risks sponsors anticipate and the events that arrive are not the same list.
Templates in the sample are recurring bundles — not checklists to photocopy. Sponsors who inherit a template still must judge which devices fit this institutional form and risk landscape. A thermal merchant plant, a regulated network expansion and a hydro concession with contested land rights do not need the same strategic system. The failure mode is a thin or accidental set of protections matched to a base case that will not be the case that arrives.
Reviewing a risk response set, therefore, means looking for missing layers — social, institutional, financial — not only for a tighter cost contingency. Scope of protection is a design choice in search. It cannot be retrofitted cheaply once the sprint has consumed the coalition's goodwill.
8. Layering Risk and Who Should Bear It
Governability does not abolish allocation; it completes it. The principle that each risk should sit with the party best able to control or bear it remains. Owners typically control scope, market access and regulator relationships. Contractors control means and methods on site. Financiers, insurers and governments have balance sheets and powers that suit particular exposures. Successful sponsors do not rely on a single hedge. They use layered devices so that each risk is matched to an appropriate holder.
Finance is itself a risk-allocation design, not a treasury afterthought. Ownership vehicle, debt versus equity, how macroeconomic and demand shocks will be absorbed, and whether the state guarantees lenders all determine who bears which losses. A guarantee or tariff promise that looked necessary at financial close can later become a political problem once cheaper capacity exists. Debt's contractual payments look cheap until distress. Leverage is limited by tax, risk-bearing ability and regulation as much as by appetite.
Where complete allocation is impossible — risks that are large, late and not controllable by contract — equity, guarantees or a pre-agreed renegotiation path are the remaining instruments. That pre-agreed path is a governability device. Without it, renegotiation becomes opportunism under duress. With it, the coalition has a script for joint action when the unscriptable arrives.
Public–private arrangements deserve particular care. They are more exposed to opportunism on all sides, and contracts alone do not prevent arbitrary actions by governments, whose powers extend beyond any agreement they sign. Governability in those settings depends on institutional arrangements, relationships and credible renegotiation mechanisms as much as on contractual terms.
Figure 4: Layer risks to the party best able to bear them; incomplete allocation needs equity, guarantees or a pre-agreed renegotiation path — the bridge into governability.
9. Institutional Forms — Each Induces Its Own Failure
Well-performing projects are the joint product of institutional arrangements and private strategies. Where institutions are taken for granted, performance is over-credited to strategy. Three historical forms recur.
Entrepreneurial coalitions assembled complementary parties around a local need. Repeated failures and political reaction pushed this form aside as a default. Rational systems — large network operators designing in-house and buying execution by competitive bid — showed fewer failures in the sample than governance arrangements, and later faced a legitimacy crisis from opposition and doubts about monopoly efficiency. Governance arrangements — private sponsors inside public rules of competition, consent and partnership — accounted for about forty percent of the sample. They answer some rational-system failures and are vulnerable to opportunism, because contracts do not stop arbitrary public action.
No form is optimal. Each induces a characteristic failure. The ongoing design problem is how governments and private parties split responsibility and risk, with shock absorbers and renegotiation clauses agreed in advance. Projects also change institutions: a deal can leave new protocols, tariff rules or ownership rights behind. Uncertain rights to use or extract value deter private investment. Early one-off bargains later faced legitimacy challenges. Before copying a contracting recipe from another arrangement, sponsors should ask which arrangement the project is actually in.
The practical implication for sponsors is diagnostic before it is prescriptive. Ask which arrangement the project is actually in, what failure that arrangement characteristically induces, and which shock absorbers — portfolios, shadow revenue, guarantees, pre-agreed renegotiation — are already on the table. Copying an alliance playbook from a rational-system utility into a fragile public–private bargain, or copying a hard arm’s-length bid into a setting that still needs relational design, is how “best practice” becomes the wrong practice. Projects also leave institutional residue: protocols among government, utility and financiers; tariff rules; ownership rights. Sponsors who push for rights at low cost while regulators push for control and legitimacy are playing an unstable game; governability devices are part of how that game stays playable when the next shock hits.
10. Shaping Episodes and the Committable Package
In this tradition, shaping is the sponsor's long work of building a coalition, reducing risk while commitments are still reversible, and creating momentum across many episodes. It is a competitive capability: difficult projects create value for sponsors who can shape them, not for sponsors who only bid. Efficient shaping practices named in the work include bringing bankers and partners in early, integrating suppliers early, and refusing to treat the front end as delay. Issues that hit the political agenda late produce long stalls.
The slow front end closes when there is a committable package. Only then should the sprint start. Starting execution because a schedule milestone arrived, while coalition, rights or risk allocation remain open, does not save time; it moves unresolved questions into the most expensive phase of the project. By construction, once the sprint is under way the formal rules are mostly set, so governability has to have been built earlier.
A neighbouring industrial tradition uses “shaping” for value allocation among stakeholders inside an owner's front-end loading calendar so that scope aligns before authorization. That is a related problem with a different research program, different sample and different success tests. Both are about front-end coalition work. Their footnotes stay separate. This essay's shaping language is the multi-episode sponsorship sense: search measured in years, committable package before sprint, effectiveness as well as efficiency.
Arm’s-length, relational and mixed patterns of who shapes and who executes sit on the same continuous thread. Arm’s-length isolates designers from builders after the contract. Relational patterns keep builders inside the sponsoring coalition or tied by frame agreements. Mixed patterns let a coalition shape the front end and then let EPC at arm’s length. Relational approaches pay off chiefly where many feasible designs exist and some beat current practice; they are not a slogan over an unallocated deal. Choose the pattern for crisis and option value, not for a tidy base-case organogram.
11. Neighbours — Gate Rules and Honesty Incentives (Do Not Merge)
Owner governance in the megaproject sense is a set of basic rules for which projects may proceed and which are denied: fund the best risk-weighted returns; stop projects that will destroy money or reputation. Gates ask whether Basic Data is enough to start scope, whether the business case can stand development, whether killing shaping issues have been examined, and later whether data, shaping and the case remain materially unchanged and the project is actually executable. A technical authority high in the firm must be able to stop a poorly prepared project. A gate that cannot say no is not governance in that sense, even if it has a steering committee. Reported success rates in that tradition have improved and still fall short of what the practices can do — figures that belong to that school's tables, not to the sixty-project qualitative sample.
Behavioural governance work on large projects adds a three-stage stack — forecasting, organizing and executing — with honesty incentives and modularity among its organizing and executing moves. Those incentives matter. They are not a substitute for reserves, triggers, flexibility and generativity when surprise hits a multi-party coalition.
Stack the layers. Do not cite one footnote for the others. Gate discipline decides whether the project should go ahead. Governability decides whether the coalition can survive what happens once it does. Contract strategy shapes who holds information and who carries consequences. Outside-view forecasting disciplines the numbers used to commit. Scenario work under ambiguity prepares decision readiness when meaning is contested. Each is necessary; none absorbs the others.
12. Practice on Large Capital Projects
Before a large capital project leaves search for sprint, a governability review can ask a short set of questions that cut across the thread of this essay.
Has the institutional arrangement been named — entrepreneurial coalition residue, rational-system operator, or public–private governance form — so that recipes are not copied from the wrong failure mode? Is there a committable package: coalition, rights and permits, and risk allocation settled enough that the sprint will not reopen the deal every month? Have identifiable risks been layered to parties with relative advantage, with equity, guarantees or renegotiation paths where allocation is incomplete? Does the strategic system show deliberate scope of protection across social, institutional and financial layers, or only a thin base-case register? Are reserves deeper than ordinary contingency for stacked shocks? Are triggers for joint action written? Is flexibility designed into configuration, not hoped for later? Are generative forums — analysis, debate, invented response — authorized before the crisis? Have renegotiation and reserve rules been written while commitments are still cheap?
On the stack beside these questions sit the neighbours without merging them: gate rules that can still say no; honesty incentives that make deception costly; contract forms chosen as governance rather than paperwork; an outside view beside the bottom-up number; and scenario readiness where ambiguity, not only uncertainty, is the diagnosis.
Principles at the close of the sixty-project work reinforce the checklist. Rehearse collaboration and risk allocation before the sprint. Keep exposures from rising without a decision. Debate jointly in crisis — deliberation creates value; pure enforcement does not. Acknowledge losses to parties who can block cheaply. Continuous shaping is required as the future shows up; intuition is not a substitute. Sponsors used seminars, scenarios, simulations and impact assessments more than a one-time Monte Carlo after scope froze. Successful projects are often the hard ones, because front-end spend that lines parties up creates the value. The coalition that can still invent a response was designed in search, not discovered in ramp-up.
Figure 5: Stack the layers — sponsorship clock and governability devices at the centre; owner gate rules, contract strategy and honesty incentives as neighbours — do not merge the schools.
13. Closing — Build Governability While Commitments Are Still Cheap
Large capital projects will continue to meet events that no base-case clause foresaw. The coalitions that survive those events are not the ones with the densest contracts alone, nor the ones with the most optimistic leadership speeches. They are the ones that treated governability as a design problem in search: reserves beyond ordinary contingency, triggers for joint action, flexibility that preserves options, and generativity that invents a response when enforcement is empty.
Write those devices while commitments are still cheap. Start the sprint only when the package is committable. Name the institutional form so its characteristic failure is not a surprise. Layer what can be allocated; pre-agree the path for what cannot. Keep owner gate governance, forecasting discipline, contract strategy and scenario readiness in the stack — and keep their mechanisms distinct. Governability is the property that holds a project coalition together when surprise arrives. It is built early, or it is improvised at the worst possible time.
References
-
Roger Miller and Donald Lessard, with Serghei Floricel and the IMEC Research Group, The Strategic Management of Large Engineering Projects: Shaping Institutions, Risks, and Governance (MIT Press, copyright 2000).
-
Edward W. Merrow, Industrial Megaprojects: Concepts, Strategies, and Practices for Success, 2nd ed. (Wiley, copyright 2024).
-
Edward W. Merrow and Neeraj S. Nandurdikar, Leading Complex Projects (Wiley, 2018).
-
Dan Lovallo, Matteo Cristofaro, and Bent Flyvbjerg, “Governing Large Projects: A Three-Stage Process to Get It Right,” Academy of Management Perspectives (2023).