Releasing Contingency as Risks Retire
The constraint is rarely the arithmetic. It is who is allowed to decide.
Most capital programmes set a contingency figure at approval and hold it, largely untouched, until the end. Nobody decides to do that. It is what happens when the amount is calculable but the authority to change it was never assigned.
Published
Key Takeaways
- Contingency covers uncertainty, so as uncertainty resolves the required amount falls. A reserve that never moves across a multi-year programme is not prudence, it is an unmanaged asset.
- The mechanism is well established in agency practice: FTA requires contingency drawdown curves in the Risk and Contingency Management Plan. But the framing is often misreported, because FTA's curves define minimum levels to be preserved, not a schedule for handing money back.
- The binding constraint is authority, not analysis. Research on contingency release during execution found the decisions are driven substantially by managerial pressure rather than objective project metrics, which is what happens when no one owns the rule.
Research foundation
Two peer-reviewed sources address contingency release directly: Ford (2002) models the effect of aggressive and conservative release policies, and Ayub, Thaheem and Ullah (2019) examine the release decision during execution and find managerial pressure a substantial driver. Agency practice is documented in FTA Oversight Procedure 40 and WSDOT Policy Statement P 2047.00, and method guidance in AACE recommended practices. One honest limitation is carried through the article: no peer-reviewed study validates the contingency drawdown curve as a technique. The curve rests on agency practice and paywalled conference papers, not on published evidence. The Four A's are the executive lens applied to that material.
Ask a programme director what happens to contingency when a major risk closes out early and you usually get a pause, then a version of: it stays where it is, and we look at it at the next gate. Ask who could release it and the pause gets longer.
That is not carelessness. Holding is the only behaviour with no personal downside. Release it early and you own the decision if something later goes wrong. Hold it and nobody is ever blamed for the money that sat idle, because idle money does not generate an incident.
Why should contingency move at all?
Because of what it is. Contingency is the distance between your target and the confidence level you have chosen to fund, and that distance is a function of the uncertainty remaining in the programme. Retire a high variance item and its spread leaves the model. The distribution narrows. The amount required to reach the same confidence level falls, arithmetically, whether or not anyone acts on it.
A reserve that has not changed in two years on a live programme is therefore making one of two claims. Either no material uncertainty has resolved in two years, which is rarely true, or the number was never connected to the analysis in the first place. Both are worth knowing. The full mechanism is set out in What P80 Means, and the trigger conditions in How Often Should You Re-Run a Risk Analysis.
What is a contingency drawdown curve, exactly?
This is the part most often described loosely, and the precision matters because it changes what the tool actually does for you.
A drawdown curve plots contingency against project advancement. FTA requires them, and its wording is specific: preserved contingency curves “define minimum levels of cost or schedule contingency to be preserved as a function of project advancement,” prepared by the project sponsor and included in the project Risk and Contingency Management Plan, with the oversight contractor reviewing and commenting on their development.
Read that again, because it is the opposite of how the technique is usually pitched. FTA's curve is a floor. It states the least you should still be holding at a given point, so that a project cannot quietly consume its reserve early and arrive at completion exposed. It is a guard against premature depletion, not an instruction to hand money back.
That distinction is worth keeping because both things are legitimate and they are not the same. The floor protects against spending contingency too fast. Release protects against holding capital hostage that the analysis says is no longer needed. A programme wanting the second cannot simply point at FTA and claim it is already required.
An honest caveat belongs here too. I went looking for peer-reviewed validation of drawdown curves as a technique and did not find any. The curve is documented in agency practice and in AACE conference papers that are paywalled and not peer reviewed. If you assert that drawdown curves improve outcomes, that assertion currently rests on institutional practice rather than published evidence, and it is better to say so than to imply a literature that does not exist.
What does the research say about releasing it?
Two studies address the release decision directly, and together they make an uncomfortable point.
Ford modelled contingency management as a policy choice, comparing aggressive and conservative release strategies and their effects on project objectives. The useful finding for a practitioner is that release policy is a lever with consequences in both directions, not a neutral administrative step.
Ayub, Thaheem and Ullah then examined what actually drives release decisions during execution, and found managerial pressure to be a substantial driver, operating largely independently of objective project metrics. In other words, the decision is being made, routinely, on grounds other than the analysis. That is the finding worth putting in front of a board, because it reframes the question. The issue is not whether a programme can calculate a defensible release. It is whether the calculation is what governs when the moment comes.
What does a release rule need to contain?
Three things, all in writing, all decided before the first release is contemplated.
The trigger
What evidence permits a release. The defensible answer is a re-run of the model showing the required amount has fallen because specific uncertainty has resolved, with the resolved items named. Not elapsed time, not percentage complete, and not the absence of bad news.
The authority
Who may authorize it, at what threshold, and who is informed. This is the part that is almost always missing, and its absence is precisely why holding wins by default. A named owner without the standing to act produces the illusion of accountability rather than the substance, which is the argument in Risk Ownership Without Authority applied to money.
The floor
How much must be preserved regardless, expressed against project advancement. This is where FTA's curve belongs, and pairing it with a release rule is what makes the pair coherent: the floor stops early depletion, the rule stops indefinite hoarding.
WSDOT's policy shows one way to structure the whole thing institutionally. It sets a legislative budget value and a lower operational value, at the 60th and 40th percentiles for projects under 100 million dollars and the 50th and 30th above it, and treats the gap between the two explicitly as the project risk reserve. The reserve is visible, bounded, and separated from the working target, which makes drawing on it or returning it an event with a name rather than an adjustment inside an estimate.
A caution on vocabulary
Before writing any of this into a governance document, define the terms locally. There is no shared cross-agency meaning to import.
A 2026 review in Risks documents that GAO and the Department of Energy use contingency and management reserve in effectively opposite senses, with GAO associating contingency with unknown unknowns and management reserve with known unknowns, and DOE's usage inverting that. DOE's own training material adds a further distinction, treating management reserve as budget held by the contractor and contingency as funds held by the department outside the negotiated project scope, schedule and budget. AACE approaches the subject from the estimating side in its general principles of contingency estimating and its principles of schedule contingency management.
None of these are wrong. They are answering different institutional questions. The failure mode is a programme that inherits language from one framework and governance from another and then cannot explain, under scrutiny, what its own reserve is for.
Why is this an Authority problem?
Because every part of this that can be calculated already is, and it changes nothing.
The model can tell you the required contingency has fallen by four million dollars. It cannot tell you who may say so out loud, whether that person will be thanked or exposed, or what happens at the next review if the number moves back. Those are conditions, not computations, and they are what determine whether a quantified reserve behaves like an instrument or like a vault.
The diagnostic is quick. Ask three people who could authorize returning contingency to the organization. If you get three different answers, or the same hesitant answer three times, the number in your risk register is not being managed. It is being stored. And the cost of storing it is real: capital committed to a programme that no longer needs it is capital unavailable to a programme that does, which is a portfolio decision being made by default rather than on purpose.
Evidence matrix
| Claim | Evidence tier | Source |
|---|---|---|
| Release decisions are driven substantially by managerial pressure, not project metrics | Peer reviewed | Ayub, Thaheem & Ullah (2019), Project Management Journal 50(6) |
| Contingency release policy materially affects project objectives | Peer reviewed | Ford (2002), Journal of Construction Engineering and Management 128(1) |
| Drawdown curves are required, and define minimum preserved contingency | Government requirement | FTA Oversight Procedure 40 (October 2023), Appendix N |
| Reserve can be structured as an explicit gap between two funded percentiles | Government policy | WSDOT Policy Statement P 2047.00 (2017) |
| Contingency and management reserve are defined inversely across agencies | Peer reviewed review | Rothwell (2026), Risks 14(5) |
| No published validation of drawdown curves as a technique | Absence of evidence, stated as such | No peer-reviewed source located; AACE conference papers paywalled |
| Holding wins by default when no one owns the release decision | Four A's interpretation | Builders Build, Authority |
What to do with this
Write the release rule before you need it: trigger, authority, floor, in four sentences, in the governance document rather than the slide deck. Then run the diagnostic. Ask three people who can authorize a return of contingency. The variance in their answers is a more reliable measure of how this programme will behave than any curve you can draw.
References
- Ayub, Bilal, Muhammad Jamaluddin Thaheem, and Fahim Ullah. “Contingency Release During Project Execution: The Contractor's Decision-Making Dilemma.” Project Management Journal, vol. 50, no. 6, 2019, pp. 734–748. doi.org/10.1177/8756972819848250. Finds managerial pressure a substantial driver of release decisions, operating largely independently of objective project metrics.
- Ford, David N. “Achieving Multiple Project Objectives Through Contingency Management.” Journal of Construction Engineering and Management, vol. 128, no. 1, 2002, pp. 30–39. doi.org/10.1061/(ASCE)0733-9364(2002)128:1(30). Models aggressive and conservative contingency release policies and their effects.
- Rothwell, Geoffrey. “A Review of Management Reserves in U.S. Government Construction Cost Estimation.” Risks, vol. 14, no. 5, 2026, article 118. doi.org/10.3390/risks14050118. Documents that GAO and DOE use contingency and management reserve in opposite senses. Recent publication; confirm final status before relying on it in a deliverable.
- Federal Transit Administration. Oversight Procedure 40: Risk and Contingency Review. U.S. Department of Transportation, October 2023. transit.dot.gov. Source of the preserved contingency curve requirement and of the P65 cost contingency confidence level adopted since July 2018.
- Washington State Department of Transportation. Policy Statement P 2047.00: Estimating Project Budget and Uncertainty. Effective 20 March 2017. wsdot.wa.gov. Establishes paired legislative and operational budget percentiles with the difference held as the project risk reserve.
- AACE International. Recommended Practice No. 70R-12: Principles of Schedule Contingency Management, As Applied in Engineering, Procurement, and Construction. Rev. 16 October 2013. Cited for scope and applicability; available to AACE members.
- AACE International. Recommended Practice No. 40R-08: Contingency Estimating, General Principles. Rev. 25 June 2008. Cited for scope and applicability; available to AACE members.
- U.S. Department of Energy, Office of Project Management. EVMS Training Snippet 4.3: Management Reserve Versus Contingency and Budget Versus Funds. Posted 9 December 2015. energy.gov. Training material rather than a directive; the governing documents are DOE O 413.3B and DOE G 413.3-7A.
- U.S. Government Accountability Office. Cost Estimating and Assessment Guide. GAO-20-195G, March 2020. gao.gov/products/gao-20-195g.
- International Organization for Standardization. Risk Management: Guidelines. ISO 31000:2018, Clause 4(e) Dynamic. iso.org/standard/65694.html.
About the Author
Dan Flynn
Creator of The Four A's of Organizational Readiness™ · Enterprise Transformation Executive · Author, Builders Build
Dan Flynn has spent thirty years inside federal, defense, and commercial organizations: diagnosing the invisible conditions that determine whether capable people produce extraordinary results. He is the creator of The Four A's of Organizational Readiness™ framework, has reached more than 11,000 professionals across corporate, civic, and national security contexts, and took a federal data platform from one release every six months to seventy-two every two weeks by changing organizational conditions: not people.
His book, Builders Build: The Four A’s of Organizational Readiness™, is forthcoming.