The pot everybody raids.

Start with what it is. Contingency is money for things you know will happen, but cannot yet name. Not if — which. Somewhere in this job there is rock, or a late drawing, or a supplier who folds. You do not know which one. You know that on a project of this size, something from that list turns up.

That is a known-unknown, and it is what contingency is for.

Five pots, and you only control one

Track 1 handed you a BAC of $1,000,000 for twenty-seven weeks and never once told you what was inside it. Here it is.

Read that from the bottom.

$827,008 is the distributed budget — the measured works, the preliminaries, the escalation. Every dollar of it is allocated to a control account with a name on it. This is the only money you can plan, schedule, and be held responsible for.

$47,553 is contingency. It sits inside the cost baseline, because it is expected to be spent — but it is not yet allocated to anybody. It is held centrally, and it is released against named risks.

$124,051 is head office overhead and profit. It never lands on a site cost code. If you are spending it, you are not making it.

And $50,000 is management reserve, which sits above the baseline, held by the business, for things nobody imagined at all. Unknown-unknowns. A project manager who can quietly access management reserve does not have a baseline; he has an overdraft.

Where the number should come from

Five percent. Somebody typed five percent.

Ask them why five and you will get one of three answers: it is what we always use, it is what the client expects, or the tender was too high and five was what was left after we took out ten.

Now do it properly. List the risks. Price them. Put a probability on each.

The expected value of that register is $51,400 — and the five percent rule produced $47,553. Close enough that the estimator feels vindicated.

He should not. Look at what the register tells him that the percentage never could.

If every one of those risks lands, the bill is $138,000 — nearly three times the money set aside. And the probability that $47,553 actually covers whatever happens is 49.5%. A coin flip. That is the honest description of the position, and no percentage rule has ever produced it.

Then the deeper problem, and it is the one that catches good engineers. $51,400 is a number that cannot occur. If there is rock in the pile bores it costs $60,000. If there is no rock it costs nothing. It never, on any project, in any universe, costs $24,000.

Expected value is a portfolio number. It is correct across a hundred projects. You are running one.

The raid

Now the part that ruins projects, and it happens in a meeting so ordinary that nobody in the room notices it happening.

It is month five. The reinforcement package from last week is running at a CPI of 0.87 — the gang is slow, the detailing was late, and $190,000 of cost has bought $164,951 of steel in the ground. The cost report is red, and the project manager has to sit in front of a client on Thursday.

So he moves $30,000 of contingency into the reinforcement control account.

Watch what happens. The package budget goes from $289,388 to $319,388. The steel in the ground has not moved — it is still 57% of the work. But 57% of a bigger budget is a bigger earned value: $182,051. Divide by the same $190,000 of real invoices, and the CPI is now 0.96.

Nothing has been fixed. Not one bar has been tied faster. The gang is exactly as slow as it was on Wednesday. But the report is amber instead of red, and the meeting is short.

Here is what he actually did.

He converted a performance problem — which has a cause, an owner, and a fix — into a budget top-up, which has none of those things. The variance is now invisible. The forecast is now wrong in a way nobody can see. And the money that was standing by for the rock is 63% gone.

In month seven the rig hits rock. It was always a forty percent chance. It costs $60,000, and there is $17,553 left in the pot.

The contingency balance is now minus $42,447.

A negative contingency is not a contingency. It is a loss, sitting in your cost report wearing a different hat, and in Week 23 you are going to find out exactly what the accounting rules force you to do the day you first see one. It is worse than you think.

The rules that stop it

Contingency is not the project manager's slush fund and it is not the planner's smoothing tool. Three rules, and every functioning project controls system in the world has them in some form.

Released only against a named risk that has occurred. Not against a variance. Not against a bad month. The rock appeared, the risk register said rock, the drawdown is against that line. If you cannot point at the risk, you cannot have the money.

Every drawdown is a record. Date, risk, amount, approver, and the register is updated. The contingency balance is reported every month next to the cost report, and it is a KPI in its own right. A project with a healthy CPI and an empty contingency is not a healthy project.

Scope change is not contingency. If the client asks for a thicker slab, that is a variation and it goes through change control — new scope, new money, new baseline. The day you start paying for the client's changes out of your contingency is the day you started paying for the client's changes.

Practical insight

Find out two numbers this week. First: what is your contingency balance today? Second: what percentage of the work is complete?

If you have burned 60% of the contingency and completed 30% of the work, you do not have a cost problem next year. You have one now, and you can prove it with two numbers on one line of a report.

Then go and read the drawdown records. If any of them say something like “productivity shortfall” or “budget adjustment” instead of naming a risk from the register, you have just found a raid — and every forecast produced since that date is fiction.

Key takeaways

✔ Contingency is for known-unknowns and sits inside the baseline. Management reserve is for unknown-unknowns and sits above it, with the business.
✔ Of the $1,000,000, you control $827,008. The rest is contingency, margin, or not yours.
✔ A percentage tells you the size of the bet. Only a risk register tells you what the bet is on.
✔ Five risks worth $138,000 at worst, $51,400 expected — and a 49.5% chance the money is enough. That is the real position.
✔ Expected value is a portfolio number. Rock costs $60,000 or nothing. It never costs $24,000.
✔ Raiding contingency to cover a variance took CPI from 0.87 to 0.96 and fixed nothing.
✔ Contingency balance against percent complete is a KPI. Report it every month, next to the CPI.
✔ A negative contingency is a loss. Week 23 explains what that costs you.

What is coming next

That is the estimate finished. You know where a rate comes from, how accurate it is allowed to be, what is not on the drawing, how it maps to the schedule, and what the contingency is really doing.

Now it has to become a budget — a structure with names on it, where every dollar has an owner who can be asked a question and expected to answer.

Next week: the cost breakdown structure, and why an estimate and a budget are not the same document even when they contain the same number.