What FTA Oversight Procedure 40 Actually Requires

Two versions of this procedure are still on FTA's site, along with an abbreviated OP 40b. The current one is dated October 2023 and supersedes the March 2022 revision. If you are working from a summary written before that, check which one it describes.

What confidence level does FTA hold you to?

Sixty-five percent, for cost. The procedure states it plainly:

Since July 2018, FTA has adopted the P65 confidence level (i.e., 65th percentile) as the confidence level to determine cost contingency to allocate to its funded projects. Project sponsors are required to provide cash funding at the 65th percentile (P65) level per FTA's cost risk model for their projects.

And it defines how the contingency figure falls out of that: the oversight contractor recommends cost adjustments to conform to its own stripped and adjusted base cost estimate, and a cost contingency equal to the difference between that estimate and the modelled P65 cost.

This matters because P80 has become an industry shorthand for the responsible confidence level, and a sponsor who funds to P80 on an FTA project is not wrong so much as answering a question nobody asked. The reporting requirement is broader than the funding rule: the oversight contractor is directed to report the P40, P50, P65 and P80 confidence levels, so reviewers see the shape of the curve rather than one point on it. That is good practice worth copying whether or not FTA money is involved, and the reasoning is in What P80 Means.

Why is the schedule rule different?

Because a schedule percentile alone can be gamed by a thin network, so FTA pairs it with a percentage floor. The procedure sets out a top-down expectation:

FTA guidance generally recommends a risk-adjusted schedule of 125 percent of the SABS remaining duration of the critical path (i.e., a top-down schedule contingency of at least 25 percent of the SABS remaining duration).

And then the operative combination:

Generally, the schedule contingency recommended by the PMOC is the larger of 125 percent of the SABS or the 65th percentile, as determined from the bottom-up histogram chart analysis.

So it is a larger-of rule, not a percentile rule. A programme whose bottom-up simulation produces a tight distribution does not get to carry less than the 25 percent floor. If you have seen this summarized as “FTA requires P65 for cost and schedule,” that summary is wrong in a way that will cost you at review.

What are the stripped and adjusted baselines, and why do they exist?

The Stripped and Adjusted Base Cost Estimate and the Stripped and Adjusted Base Schedule are the sponsor's cost estimate and schedule with contingency and allowances removed and with the oversight contractor's adjustments applied.

The purpose is to prevent double counting. An estimate that already contains buried contingency, and is then run through a simulation that adds modelled uncertainty on top, produces a number with no defensible meaning. Stripping forces the padding into the open, where it can be argued about, and it is usually the step that surfaces how much unacknowledged reserve a programme was already carrying.

This is also the step that most often goes badly, because it is not really a technical exercise. Asking a team to declare the conservatism inside its own estimate is an organizational request before it is an analytical one.

What does the drawdown curve actually require?

The requirement is specific, and it is a floor rather than a release schedule:

To assist in the management of contingency, preserved contingency (or ‘contingency drawdown’) curves that define minimum levels of cost or schedule contingency to be preserved as a function of project advancement are prepared by the project sponsor and included in the project RCMP.

The oversight contractor reviews and comments on their development, and guidelines for establishing them sit in Appendix N.

Read that carefully. The curve says how much you must still be holding at a given point. It does not say how much you may hand back, and it is not the mechanism for returning contingency as risks retire. Both things are legitimate and they are different, which is the argument in Releasing Contingency as Risks Retire. A sponsor who wants a release regime needs to build one; pointing at OP 40 will not supply it.

What is the oversight contractor actually looking for?

Less than people fear, and more precisely than they expect. The procedure directs the PMOC to fully identify, describe and analyze the adequacy of the sponsor's cost contingencies, and to make recommendations as to minimum amounts for inclusion in the project management plan, the RCMP and supporting documents. It distinguishes primary from secondary mitigation, expects a risk register, and expects both a bottom-up schedule risk model and a top-down cost risk model.

In practice the findings that hurt are rarely about the simulation. They are about whether the register is real, whether the baseline was genuinely stripped, and whether anyone can explain what changed since the last review. The GAO cost estimating and schedule assessment guides set the same expectations in a broader federal context, and the ten scheduling practices in the schedule guide are a reasonable self-check before an FTA review rather than after one.

Why is this an Authority problem?

Because compliance with OP 40 is achievable by any competent analyst, and nothing about it changes how a programme behaves.

The procedure produces a P65 number, a drawdown curve, and a plan. Whether those artifacts govern anything depends on questions the procedure does not address: who may draw on contingency, who must be told, what happens when the re-forecast moves, and whether the person who owns the risk has the standing to act on it. A programme can satisfy every requirement here and still manage risk as paperwork, which is the failure described in The Risk Register Nobody Reads.

The useful way to read OP 40 is therefore as a floor on rigor rather than a definition of it. It tells you what you must be able to show a reviewer. It does not tell you how to run the programme, and treating the two as the same thing is how organizations end up audit-ready and surprised.

Evidence matrix

What to do with this

Check three things before your next review. That you are funding to P65 and not to an assumed P80. That your schedule contingency satisfies the larger-of rule rather than just the percentile. And that your baseline was genuinely stripped, which you can test by asking the estimator where the conservatism went.