M4.5 — Contingency & Uncertainty

M4.5 — Contingency & Uncertainty

1. Practical outcome

By the end of this section, the learner should be able to establish, explain, govern and reassess cost and schedule provision for uncertainty without hiding it inside the base plan. They should be able to distinguish base budget or duration from contingency and, where the organisation uses it, management reserve; connect provision to qualitative and quantitative risk evidence; select a confidence or governance basis; decide where contingency should be held; define ownership, drawdown and release rules; track consumption separately from performance; and judge whether the remaining provision is still proportionate to the remaining exposure.

The objective is not to create the largest possible reserve or to force every uncertainty into a precise number. It is to make uncertainty provision transparent enough that management can see what the base position is, what protection exists, what that protection is intended to cover, how it may be used, and whether the balance remains credible as the project changes.

M4.5 completes the Module 4 learning chain. M4.1 identifies uncertainty, M4.2 prioritises it, M4.3 quantifies selected exposure, and M4.4 turns exposure into management action. M4.5 asks what cost and time protection should remain available after those decisions, how that protection is governed, and how it changes as risks materialise, retire or are reduced.

A useful question throughout this Knowledge Area is: “If management asks what protection remains between the current plan and the project’s risk appetite, can we show the amount, the evidence behind it, the rules for using it and the exposure it still has to cover?”

2. Core concepts

Contingency is provision for uncertainty, not spare budget or spare time

Contingency is a controlled allowance for uncertainty within the project’s governance model. It may be expressed as money, time, or another constrained resource. It is not an invitation to spend up to a limit and it is not an automatic cushion for poor performance. Its purpose is to protect objectives against defined uncertainty while keeping the base plan visible.

The same principle applies to schedule protection. An eight-day gap between the deterministic forecast and an opening date is not automatically “eight days of contingency”. The project must understand what uncertainty sits behind the forecast, what confidence that margin represents, and whether the time is deliberately protected or can be consumed by ordinary planning decisions.

Base position, contingency and management reserve should be distinguishable

Terminology varies by organisation, so the Open Standard does not impose one universal accounting or governance definition. However, the underlying distinctions should be explicit:

·       Base cost or base duration — cost or time for the defined scope under stated assumptions, with the treatment of explicit or embedded uncertainty allowances disclosed. Distinguish the base used to calculate QRA allowance from the authorised budget and from a current deterministic forecast that may already contain expected risk costs.

·       Contingency — provision for recognised uncertainty within a defined governance boundary. It may be held at package, control-account, project or another level depending on the organisation’s rules.

·       Management reserve — a separately controlled provision whose definition depends on the adopted system. In formal EVM it sits outside the PMB and addresses risks within authorised scope; see M5.3. An owner’s broader project reserve may follow different rules and should be labelled accordingly.

·       Margin or buffer — a time-based protective interval or decision allowance whose ownership and use should be governed just as deliberately as cost contingency.

Use the definitions required by the adopted contract or control standard. Where terminology is discretionary, define it explicitly. Reconcile the categories to the authorised total and explain what each can fund; similar labels do not guarantee identical boundaries.

Hidden contingency weakens both planning and assurance

Allowances become difficult to govern when they are embedded invisibly in activity durations, unit rates, productivity assumptions, quantities and package values while a separate central reserve is also held. The result can be double counting, misleading performance measures and false confidence. A ten-day activity that contains seven days of expected work and three days of undisclosed “safety” cannot be analysed or challenged in the same way as a seven-day activity with a visible project-level schedule allowance.

Not every estimate or duration range can be stripped to a single uncontested “risk-free” number. Professional judgement is still required. The goal is not theoretical purity; it is enough transparency to prevent the same uncertainty being funded or scheduled several times without anyone knowing.

Confidence levels can inform provision, but they do not choose governance

M4.3 introduced probability distributions and confidence points such as P50 and P80. The difference between a deterministic or base outcome and a selected confidence point can inform contingency. For example, if a credible cost model has a base outcome of 47.0m CU and a P80 outcome of 49.7m CU, the 2.7m CU difference is useful evidence when deciding whether a 3.0m CU provision is proportionate.

The statistical result does not make the governance decision. P80 is not automatically the correct budget, commitment or reserve level. The appropriate basis depends on consequence, risk appetite, public or contractual commitment, ability to respond later, model quality, funding constraints and organisational policy. A project may deliberately choose a higher confidence for a fixed public opening than for an internal intermediate milestone.

Expected value and percentile-based provision answer different questions

Expected monetary value is the probability-weighted mean consequence. Expected values add even when risks are dependent, but the aggregate distribution and its tail depend on that relationship. A 30% chance of a 1.0m CU loss has an expected value of 0.30m CU; if the event occurs its stated loss is 1.0m CU. EMV can support comparison, but it is not automatically sufficient contingency for a project commitment. Use the distribution and the chosen confidence objective where those inform the decision.

Neither technique is universally superior. The practitioner should match the method to the decision and explain what the number represents.

Cost contingency and schedule contingency are related but not interchangeable

A risk can affect both money and time, but the two forms of protection should not be treated as the same control. Paying for acceleration may protect time by consuming cost contingency. Holding schedule margin may reduce the need for acceleration but does not create funding. Response choices therefore change both the exposure and the contingency requirement across objectives.

Where the fixed opening date is the dominant objective, management may accept additional cost to protect schedule confidence. Where the funding cap is immovable, the project may instead preserve cost headroom and accept a wider date range. M4.5 makes those trade-offs visible rather than leaving them implicit.

Where contingency is held affects behaviour and control

Contingency can be held centrally, distributed to packages, or managed through a combination. Each design creates different incentives and decision speeds:

·       Package-held contingency can support rapid local decisions and clear linkage to specific uncertainty, but may become fragmented or treated as entitlement.

·       Project-held contingency supports pooling across uncertainties and stronger central governance, but can become remote from the teams that see early warning signs.

·       A hybrid model can hold defined package uncertainty locally while retaining project-level provision for cross-cutting or aggregated exposure.

The structure should reflect where uncertainty is understood, where decisions need to be made and where authority can be exercised. It should not simply mirror the organisation chart.

Aggregation can reduce or increase required provision

Pooling uncertainties can reduce aggregate provision at a chosen confidence relative to adding individual worst cases. Common causes and positive dependence can increase the adverse tail relative to a model that assumes independence. This does not mean that dependence makes the total exceed the sum of fixed individual maxima. A late building handover can jointly affect installation, training, testing and supplier attendance; model that shared driver without counting it repeatedly.

This is why central contingency should not be allocated by mechanically adding individual risk maxima, and why correlation assumptions in M4.3 matter to M4.5. Provision is an aggregate management decision informed by how uncertainty interacts.

Drawdown is a governance transaction, not a forecast correction

When a qualifying uncertainty materialises, the project may authorise contingency drawdown or transfer. That transaction changes where authorised budget or time protection is held; it does not automatically change what the work is forecast to cost or how long it will take. Forecast and budget remain different control views.

A project can therefore have a forecast increase with no contingency drawdown, a contingency drawdown with no increase to the total authorised budget, or both. The reason, evidence and authority should be traceable in each case.

Contingency should not repair unrelated performance failure

Suppose installation productivity is below plan because the team ignored known access constraints and failed to manage resources. The resulting forecast overrun is not automatically a risk event eligible for contingency. If the governance rule says contingency covers defined uncertainty, routine inefficiency or unmanaged scope should not be re-labelled as risk simply to make the variance disappear.

Conversely, when a recognised uncertain event materialises and the project uses provision exactly as intended, drawdown is not automatically evidence of poor control. The correct question is whether the event was within the approved uncertainty boundary and whether remaining exposure is still covered.

A materialised risk becomes a fact; remaining uncertainty still needs provision

Once a risk event occurs, the realised consequence should be reflected in the current schedule, cost forecast and issue/change processes as appropriate. The uncertainty associated with that event may reduce, but secondary effects can remain. The project should not leave the full original risk allowance untouched while also adding the realised cost to the forecast unless there is evidence that the residual exposure genuinely remains.

Equally, a risk can retire without materialising. That does not mean its original nominal allowance must be released one-for-one if other aggregate exposures have increased. Contingency is managed against the remaining portfolio of uncertainty, not only against isolated line items.

Contingency consumption and remaining exposure must be reviewed together

A reserve-balance report alone is weak control. “We have used 60% of contingency” says little unless management also knows how much uncertainty has retired or materialised. A project halfway through delivery may reasonably have used most of its provision if most uncertainty has also been resolved; the same consumption could be alarming if the highest-risk work is still ahead.

Useful review therefore considers both sides of the equation: remaining provision and remaining exposure. The relationship need not be expressed as one universal ratio, but the project should be able to explain whether the balance is still adequate for the risk profile and chosen confidence objective.

Schedule contingency needs explicit ownership and protection

Time is often consumed invisibly. Teams bring work later, move internal milestones, use float, or allow contingency activities to become normal working space. If schedule protection matters, ownership and drawdown rules should be explicit. The project may define a protected opening buffer, an internal readiness target earlier than the public commitment, or controlled milestone margin that can only be consumed with named authority.

This does not mean padding every activity. It means preserving visibility of the difference between expected delivery and protected commitment, then monitoring whether the probability of success remains acceptable as that difference changes.

Where the project adopts PASEG-style explicit schedule-margin tasks, keep them distinct from productive work and do not resource-load them. In the analysis copy, zero their durations for schedule risk assessment so protection is not counted again as uncertainty is introduced; preserve a reconciliation to the controlled schedule. Still assess the time-dependent cost and staffing consequences of later delivery. Do not remove genuine work or evidenced duration requirements as though they were margin.

Opportunity can change contingency requirements, but benefits should not be banked too early

Credible opportunities can reduce cost or schedule exposure, as illustrated by staged training in M4.3 and M4.4. However, contingency should not be released merely because an opportunity is listed. The opportunity must become sufficiently evidenced, executable and reflected in the forecast or risk model. Banking unproven upside creates the same false confidence as reducing a risk score before a response has worked.

Provision should evolve as project maturity changes

As scope and delivery become better defined, replace arbitrary allowances with provision supported by current uncertainty evidence. Review the remaining exposure before retaining or releasing funds. Late integration, commissioning, acceptance or commercial settlement can still carry substantial uncertainty; elapsed time or apparent percentage completion alone does not justify release.

A contingency profile should therefore change because uncertainty changes, not because time has simply passed.

Contingency, change and funding are connected but distinct

A client-requested scope increase is a change decision, not uncertainty merely because its final cost is uncertain. A funding increase is an authorisation decision, not automatically a new cost baseline. A contingency drawdown may finance a recognised uncertainty without changing the total approved budget. These distinctions protect transparency across M3, M4 and M6.

Where the rules of an owner-held reserve permit funding an approved scope change, identify it as a change and follow the relevant authorisation route. Do not extend that permission to formal EVM management reserve, which cannot fund scope outside the contract. Reserve use never bypasses scope or baseline control.

The model should remain reproducible and auditable

Whether contingency is based on expert judgement, ranges, EMV, QRA percentiles or a combination, another competent practitioner should be able to reconstruct the rationale. That means preserving source risks, assumptions, model version, confidence basis, exclusions, correlation decisions, governance threshold, approvals, drawdowns, releases and subsequent reassessments.

The key M4.5 principle is: keep the base visible, keep the uncertainty provision visible, and keep the reason for movement visible. Hidden allowance may create comfort, but it weakens control.

3. Step-by-step method

The following 18-step method can be scaled from a simple package-level allowance to a project-wide cost and schedule contingency regime. The method deliberately connects risk evidence, quantitative analysis, budget/schedule control and governance rather than treating reserve as a stand-alone finance number.

1. Define terminology and governance. Confirm how the organisation uses the terms contingency, reserve, management reserve, schedule margin, buffer and risk allowance. State who owns each category, what it may fund or protect, and what authority is required to move it.

2. Establish the credible base position. Confirm the base estimate, control budget, deterministic schedule and current forecast that will be protected. Identify hidden allowances in rates, quantities, productivity assumptions, durations or milestones and remove or document them where practical.

3. Define the uncertainty boundary. State which types of uncertainty the provision is intended to cover and which are excluded. Clarify treatment of known scope change, routine performance variance, force majeure, commercial claims, escalation, opportunities and broader management uncertainty.

4. Reconcile the uncertainty sources. Link the provision basis to the current risk register, assumptions, dependencies, estimate/schedule uncertainty and relevant commercial exposure. Remove duplicates and check that the same uncertainty is not included both as a discrete risk and inside a broad range without justification.

5. Decide the analysis depth. Use qualitative judgement for small or early decisions, simple range/EMV methods where suitable, and QRA where the consequence and data justify it. Document why the method is proportionate to the decision.

6. Set the confidence or risk-appetite basis. Define the management question: for example, protect an 80% confidence of opening by the fixed date, or maintain a stated probability of remaining within the 50m CU funding limit. Record any policy-mandated threshold and why it applies.

7. Quantify cost and schedule exposure separately where material. Determine the cost distribution, date distribution or other uncertainty range. Do not assume the same confidence or provision is appropriate for every objective.

8. Test aggregation, dependencies and common cause. Review whether uncertainties are independent, correlated or driven by shared events. Check that central provision reflects portfolio behaviour rather than a simple sum of individual worst cases.

9. Determine the proposed provision. Translate the evidence into a cost amount, time margin or combination. Show the relationship between base, deterministic/current forecast, selected confidence point and proposed contingency. Explain any judgemental overlay.

10. Decide where provision will be held. Allocate or retain contingency at package, control-account, project or management level according to decision rights and pooling benefits. Prevent duplicated ownership and entitlement.

11. Separate contingency from management reserve as required by the adopted system. State each provision’s scope boundary, authority and permitted use. Use the formal EVM distinction in M5.3 when applicable, and prevent package teams from treating centrally held reserve as available budget before authorised release.

12. Define drawdown criteria and evidence. For each material category, state what event, trigger or evidence permits drawdown, who submits the request, who approves it, how quickly the decision must be made and what records are updated.

13. Define transfer, release and increase rules. State how unused provision can be released, moved between packages, rebalanced across risks or increased when exposure changes. Clarify the relationship to funding approval and baseline change.

14. Integrate contingency with schedule, cost and forecast controls. Ensure drawdown can be traced to the affected WBS/CBS, schedule activity or milestone, risk/issue and change record. Preserve the difference between authorised budget movement and forecast movement.

15. Record consumption transparently. Maintain opening balance, approved drawdowns/transfers, releases, closing balance, event linkage, authority and date. For schedule contingency, record consumed and remaining protection against the relevant commitment.

16. Reassess remaining exposure at each material review point. Update risk evidence, responses, realised events, QRA inputs and confidence levels. Compare remaining provision with remaining uncertainty rather than with elapsed project time alone.

17. Release, replenish or escalate through governance. Release provision when exposure has genuinely reduced and the organisation no longer needs the protection. Seek additional provision or a changed commitment when the remaining authorised protection is no longer adequate. Do not hide the shortfall by changing assumptions silently.

18. Preserve the audit trail and communicate the decision. Record model version, assumptions, approved contingency basis, movements and current adequacy. Report management information that answers: what protection remains, what it covers, how much confidence it provides, what has consumed it, and what decision is now required?

4. Worked Reference Project example — Community & Learning Centre

Context — the 50m CU envelope and the fixed opening date

The Community & Learning Centre has fixed authorisation of 50.0m CU. At approval, 47.0m CU is distributed to delivery budgets and 3.0m CU is held centrally. The distributed 47.0m CU already includes the Technology package authorisation of 3.8m CU, comprising 3.711m CU base control budget and 0.089m CU package contingency. Do not describe the whole 47.0m CU as allowance-free base scope: local contingency is inside it and separate from the central 3.0m CU.

At Month 16, M4.3 also showed a deterministic current cost forecast of 49.2m CU against the 50.0m CU approved budget. The cost QRA indicated only an illustrative 57% probability of remaining within 50.0m CU under the then-current uncertainty. On schedule, the deterministic readiness forecast sat eight working days before the fixed opening date, but the initial modelled probability of achieving opening was 72%.

M4.4 then implemented targeted fire-system assurance, secured the specialist supplier support window and controlled staged training. Evidence supported the response assumptions, and the illustrative schedule QRA improved to an 88% probability of meeting the opening date, with P80 three working days before the commitment. The project must now decide what cost and time protection should remain after that response improvement.

Step A — make the base and held provision visible

Reconcile authorisation separately from forecast uncertainty. The 50.0m CU approval includes the 3.0m CU central budget provision. That does not establish whether any allowance is included in the 49.2m CU EAC. Define the EAC boundary, identify embedded risk costs and construct the QRA without adding a duplicate allowance. Never increase the reported funding cap by adding held contingency to it.

The current ledger shows that 1.2m CU of the original central contingency has previously been authorised into delivery budgets for qualifying uncertainty events, leaving 1.8m CU held centrally. The project total remains 50.0m CU: contingency drawdown changed the distribution of authorised budget, not the total approval.

Within the Technology package, 0.060m CU of the original 0.089m CU package contingency has been drawn for a specifically approved technology uncertainty, leaving 0.029m CU locally held. The package-authorised total remains 3.8m CU.

Step B — reassess the cost confidence rather than relying on the reserve balance

Rerun the cost analysis using current response evidence and an explicit forecast boundary. For this illustrative snapshot, assume the 49.2m CU deterministic EAC includes planned response work and known costs, but excludes undrawn central contingency as a lump-sum forecast item. Sample only remaining uncertainty not already priced. The assumed post-response outputs are P50 49.55m CU, P80 50.20m CU and approximately 69% probability of remaining within 50.0m CU. These are teaching assumptions, not reproduced simulation results or a revised budget.

The 1.8m CU central balance is budget authority already inside the 50.0m CU cap; it does not provide another 1.8m CU above that cap. Deterministic forecast headroom is 50.0 − 49.2 = 0.8m CU. The P80 increment above the defined EAC is 50.20 − 49.20 = 1.0m CU, leaving a 0.2m CU gap between P80 and authority. Reserve balance, forecast headroom and QRA allowance are different measures; compare them on their stated boundaries.

Step C — choose the governance basis explicitly

For the public opening date, the Project Board has set a minimum 80% confidence objective because failure would have high public and operational consequence. For cost, the organisation has not mandated that the budget itself equal P80; instead it requires the project to report the probability of remaining within authorised funding and escalate when that probability falls below 70% or when P80 exceeds funding by more than 0.25m CU.

The 69% result breaches the illustrative probability threshold of 70%; the 0.20m CU P80 excess does not breach the separate 0.25m CU threshold. Because the rules use “or”, the probability breach triggers escalation. The Board receives the decision now, considers further cost-tail responses, and explicitly accepts, funds or otherwise resolves the exposure within its authority; it does not treat a near-threshold result as compliant.

Step D — protect schedule contingency without padding activities

The deterministic readiness forecast is eight working days before opening, while post-response P80 readiness is three working days before opening. The allowance from deterministic readiness to P80 is five working days; the additional margin from P80 to the fixed commitment is three working days. These are distinct intervals. Governance protects their combined eight-day space while reporting the five-day modelled allowance and three-day confidence headroom separately. Do not pad productive activities or call the three-day remainder the entire P80 contingency requirement.

If a known testing event moves deterministic readiness two working days later, the deterministic gap becomes six days. Under an unchanged five-day uncertainty allowance, confidence headroom would be one day, but that is only a conditional arithmetic check. Update actual/remaining work, remove the realised event from unrealised sampling, retain residual uncertainty and rerun QRA before reporting a new P80 or success probability. Record authorised use of protection separately from the forecast movement.

Step E — demonstrate a legitimate cost drawdown

Assume the formal fire witness identifies an unexpected but recognised cross-system defect whose approved correction costs 0.35m CU. The event falls within the defined central-contingency boundary and is linked to the fire-system risk already analysed through M4.1-M4.4. The Project Director approves a 0.35m CU drawdown from central contingency into the affected work-package budget.

The transaction reduces centrally held contingency from 1.8m CU to 1.45m CU and increases distributed budget by the same amount. The authorised project total remains 50.0m CU. The issue and current forecast are updated separately to reflect the actual expected cost of the defect. If the defect costs exactly what the forecast already anticipated, the budget transfer can occur with little forecast movement; if the new evidence changes expected cost, the forecast changes too. The two records are related but not interchangeable.

Trace the qualifying event through the control model

For the 0.35m CU fire-defect example, retain one transaction trail: source risk and exposure window; observed defect and issue ID; approved correction scope and schedule work; evidenced forecast consequence; authorised contingency transfer; revised budget distribution; and reassessed residual risk/QRA. The central budget ledger reconciles 3.0 − 1.2 − 0.35 = 1.45m CU. Distributed budgets reconcile 47.0 + 1.2 + 0.35 = 48.55m CU, so total authorisation remains 50.0m CU.

Do not assume that a 0.35m CU transfer produces a 0.35m CU EAC increase. Reconcile the amount already forecast for the consequence, the new known cost and any changed remaining estimate. Remove only the duplicated unrealised event; retain evidenced residual uncertainty. Release or replenish provision after reviewing the aggregate exposure and authority, with a separate approval and retained before/after balance.

Step F — reject an ineligible drawdown

In the same period, a delivery team reports 0.25m CU of additional labour caused by sustained productivity below the approved plan. Investigation shows that the loss arose from poor work-front coordination that had already been identified as a management action but was not implemented. Under the worked governance rules, this is normal performance variance rather than a qualifying uncertainty event. The forecast must show the adverse effect, but contingency is not automatically transferred to erase the variance.

This is an important control behaviour. A reserve that is used whenever a package is overspent stops being uncertainty provision and becomes an opaque budget-balancing mechanism.

Step G — retire risk and reassess the aggregate provision

Later, the specialist supplier support window is completed successfully and no further exposure remains from that event. The project does not mechanically release the full amount once associated with that risk. It first reruns the aggregate cost and schedule uncertainty because fire acceptance, remaining testing and operational-readiness duration are now the dominant exposures. If the new P80 cost reduces to 49.85m CU and the probability of remaining within 50.0m CU rises to 84%, the Board may decide that part of the remaining central contingency can be released or retained for the final high-risk testing window according to policy.

“The 50.0m CU authorisation includes central and distributed provision; 1.8m CU remains centrally held before the later fire-defect drawdown. Deterministic EAC is 49.2m CU, giving 0.8m CU forecast headroom. Assumed post-response QRA gives P80 50.20m CU and 69% probability of meeting the cap, triggering the stated 70% escalation rule. Deterministic readiness is eight working days before opening; P80 is three days before, implying five days of modelled allowance and three days of remaining confidence headroom. Use of protection needs authority and evidence. Update known events and residual exposure before rerunning the model or releasing provision.”

Step H — management statement

“The project remains within the fixed 50.0m CU authorised envelope, which includes centrally governed contingency established at baseline. Following earlier approved drawdowns, 1.8m CU was held centrally at Month 16. Current deterministic EAC is 49.2m CU; post-response QRA indicates approximately 69% confidence of remaining within 50.0m CU and a P80 of 50.20m CU. The opening-date position has improved to 88% confidence, with P80 three working days before the public commitment; those three days are being protected as project-level schedule contingency. Any drawdown must be linked to qualifying uncertainty and approved separately from forecast movement. Management should continue the planned cost-tail responses and reassess after the formal fire witness before releasing or seeking additional provision.”

5. Interpretation and professional judgement

A large reserve can coexist with poor confidence

A project can report a sizeable contingency balance while still having inadequate protection if the base forecast has deteriorated or the remaining risks are highly correlated. Equally, a small balance near completion may be entirely adequate if most uncertainty has retired. The balance is context, not the conclusion.

Using contingency is not automatically failure

If an identified uncertainty materialises and the approved reserve is used exactly as intended, the control system is working. The stronger test is whether the risk was recognised, the drawdown was authorised, the forecast was updated, and the remaining provision was reassessed. Hiding the event would be poor control; transparent drawdown is not.

Unused contingency is not automatically success

A project can finish with all contingency unused because the reserve was excessive, because risk identification was weak, or because costs were hidden elsewhere. Conversely, a project can use most of its contingency and still be well controlled. Success is better judged by forecast quality, transparent decision-making and whether the provision remained proportionate to exposure.

P80 is a decision input, not a universal target

A senior manager may legitimately choose P70, P80, P90 or a non-percentile rule depending on the consequence and decision context. The practitioner should resist presenting one confidence point as professionally “correct” in all circumstances. What matters is that the threshold is intentional, understood and supported by a model fit for the decision.

Do not confuse available budget with forecast headroom

Available budget, forecast headroom and uncertainty allowance must be reported on separate boundaries. Held contingency does not increase the authorised cap. Whether adding allowance to a deterministic forecast double counts uncertainty depends on what the forecast already includes, not simply on whether the budget contains contingency. Reconcile embedded allowances, sampled risk costs and remaining provision explicitly.

Schedule margin can be consumed without anyone calling it a drawdown

Time protection often disappears through small date movements, late decisions and “temporary” resequencing. The experienced practitioner watches internal targets, critical/near-critical paths, QRA confidence and milestone margin together. A protected buffer that is not explicitly governed will often be spent by default.

Release should be evidence-led, not calendar-led

A rule such as “release 25% of contingency at halfway” may be administratively simple but can be irrational. If procurement uncertainty has retired early, release may be justified before the midpoint. If commissioning and regulatory acceptance remain highly uncertain late in the project, significant provision may still be needed. Release should follow exposure, not elapsed time.

Provision can be rebalanced without pretending the risk disappeared

As the project matures, risk can migrate between packages and objectives. A package allowance may be released centrally while aggregate project provision remains unchanged; cost contingency may be used to buy schedule protection; or a schedule buffer may be reduced after an acceleration decision. These are management choices that should remain traceable to the underlying exposure and authority.

Contingency adequacy depends on model quality

A precise P80 derived from missing risks, optimistic ranges or a poor schedule is not stronger evidence than a well-reasoned qualitative allowance. The practitioner should challenge the quality of M4.3 inputs before using the output to justify reserve. False precision in provision can be more dangerous than an explicitly judgemental range.

Management reserve should not become invisible contingency

Where management reserve exists, its higher-level purpose should remain clear. If it is routinely distributed to cure package overruns without explaining the cause, the distinction from contingency has no practical meaning. If management reserve is not used by the organisation, the project should not invent it merely to copy another framework; it should still preserve equivalent clarity over decision rights and uncertainty categories.

Contingency drawdown and baseline change require careful language

Some organisations treat authorised contingency drawdown as a budget transfer within the existing baseline; others formally update control baselines when reserve is allocated. The Open Standard does not prescribe the accounting convention. It does require the project to state which convention it uses and preserve before/after values so performance history remains interpretable.

The most useful question is forward-looking

The weak review question is “How much contingency have we used?” The stronger question is “Given the risk that remains, the responses now in place and the confidence we have committed to, is the remaining protection enough - and what decision follows if it is not?”

6. Practice diagnostic

Use this table to challenge the quality of the approach or deliverable. The rows consolidate the most important recurring red flags and checks; they are prompts for professional judgement rather than a substitute for the method above.

A good M4.5 control position lets management distinguish four things immediately: what we planned, what we now forecast, what uncertainty still surrounds that forecast, and what authorised protection remains. If those four views are blurred together, contingency is not being controlled transparently.

7. Try-it-yourself practical exercise

Use the Month-16 position before the later fire-defect drawdown. Authorisation is 50.0m CU: initially 47.0m CU distributed budgets, including stated package contingency, plus 3.0m CU held centrally. After 1.2m CU approved central drawdown, 48.2m CU is distributed and 1.8m CU is held centrally. Deterministic EAC is 49.2m CU on the boundary defined above. Assumed post-response QRA gives P50 49.55m CU, P80 50.20m CU and 69% chance of meeting the cap. Deterministic readiness is eight working days before opening, P80 is three days before, and assumed opening probability is 88%.

Complete the exercise in eight parts:

·       1. Reconcile the 50.0m CU authorisation, 48.2m CU distributed budget and 1.8m CU central balance. Calculate deterministic headroom and the P80 increment above the defined EAC. Explain why held budget is not additional funding and why a forecast-boundary check is needed before adding uncertainty.

·       2. Recommend a cost-confidence governance position. State whether you would accept the current 69% probability, seek additional responses, seek more funding or change the commitment. Explain the decision without assuming that P80 must automatically become the budget.

·       3. Calculate the interval from deterministic readiness to P80 and the separate interval from P80 to opening. Define ownership and approval rules for each. Then test a two-day known delay without asserting that unchanged uncertainty or the old 88% probability can be assumed.

·       4. Assume a recognised fire-system defect materialises and requires 0.35m CU. Prepare the contingency drawdown entry: risk/issue reference, amount, source reserve, destination, approving authority, forecast effect and post-drawdown reserve balance.

·       5. Assume a separate 0.25m CU labour overrun arises from unmanaged productivity. Decide whether it qualifies for contingency under your rules and explain how the forecast and variance should be reported if no drawdown is approved.

·       6. One supplier risk then retires without cost. Explain why you would not automatically release the exact amount previously associated with that risk, and describe what analysis you would rerun before releasing provision.

·       7. Identify two ways contingency could be double counted in this project and state the control that would prevent each one.

·       8. Write a management statement of no more than 200 words explaining current cost and schedule protection, what has been consumed, remaining confidence, the next drawdown/release decision and any escalation required.

Final challenge question: “If the contingency balance stayed exactly the same next month, what evidence would you need before deciding whether the project had become safer, riskier or simply unchanged?”

8. Bridge to capability

The How-to teaches how uncertainty evidence becomes governed cost and schedule provision. At APPLIED and above, capability assessment should test whether the practitioner can establish or administer realistic contingency, justify drawdown or release and reassess remaining exposure. Reconciled provision records, approvals and current risk evidence should support the decision. Supporting work products are illustrative and should be assessed with the practitioner’s reasoning. Sample learning questions illustrate the judgement at each recognition level; they are discussion prompts rather than validated assessment items.

Learning question: Do you understand how uncertainty becomes controlled provision?

Capability question for APPLIED and above: Can you demonstrate justified contingency decisions without hiding exposure, double counting cost or erasing performance variance?