Chapter 6 Quiz — Risk Management

(c). It is an uncertain event, it has an observable moment, and it carries a cost and a duration. (a) and (d) are anxieties, not events. (b) is certain, so it is scope, not risk — it belongs in the estimate at full value.

(c) Mitigate. The drilling does not stop bad soil from existing and it does not move the exposure to anyone else. It converts an unbounded unknown into a bounded, priceable quantity — reducing both probability and impact of the risk by turning part of it into scope.

(b). Weather is typically excusable but non-compensable: the contractor gets a time extension so liquidated damages do not run, and eats its own extended costs. This is why "adverse weather beyond the anticipated days" belongs on a cost register even though the schedule relief is automatic.

(b). $1,320,000 ÷ $43,800,000 = 3.01%. A top-down percentage is fast and defensible in a boardroom and completely uninformative: because no dollar traces to a named risk, it can never tell you whether the amount remaining is enough for the work remaining.

(c). X-1 EMV = $81,000; X-2 EMV = $80,000. Financially identical on paper, and entirely different in life. EMV is a mean and means hide tails. X-2 is a transfer candidate — a bond or a policy — precisely because you cannot absorb it.

(b) 37.0 WD. (30 + 4 × 36 + 48) ÷ 6 = (30 + 144 + 48) ÷ 6 = 222 ÷ 6 = 37.0. Note that the expected duration is a full work day longer than the most-likely duration, because the bad tail is longer than the good one.

(c) about 51%. 0.80 × 0.80 × 0.80 = 0.512. This is merge bias: every feeder looked safe and the merge point is a coin flip. It is the reason a schedule dense with parallel paths is riskier than a schedule with the same number of activities in a line — and the critical-path calculation alone will not show it to you.

(c). Because the GMP is a maximum, unspent money does not convert to profit. On Northgate, $970,000 of total savings produced $727,500 to Meridian and $242,500 to Kestrel — a 13.4% bump on Kestrel's $1,804,800 fee.

(b). 63% ÷ 45% = 1.40, above the 1.3 threshold where the job is eating itself. At this pace the reserve is gone around 71% complete, well before commissioning and closeout risks land. Waiting for the next cycle costs you a reporting period you cannot recover.

(a) Avoid and mitigate. You cannot transfer an injury. Insurance pays medical costs and workers' compensation pays indemnity, and both are necessary — neither makes the injury not have happened. And you cannot accept a safety risk, because acceptance means deciding to let it occur and funding the consequence. The bond does not bring anyone back.

False. Its purpose is to convert arguments you would otherwise have later, expensively and verbally, into decisions made early, cheaply, and in writing — with an owner and a trigger attached to each. A forecast is a by-product, not the point.

False, and in an open-book contract it is worse than false. Contingency is a funded liability against identified risk. In a GMP it is inside the guaranteed maximum, disclosed to the owner, and subject to a savings split — so money carried there with no risk behind it and no intention of spending it is a misrepresentation, not conservatism.

False. It multiplies rank labels rather than quantities, compresses ranges, and is blind above its top band. On the Northgate register it ranked R-11 above R-07 while the expected values ranked them the other way. Use it to communicate; use arithmetic to decide.

False, and the exception is the whole point of allocation. When the transferee genuinely controls the risk it is cheap to them, so they price it below your expected value — a truss supplier accepting a delivery-date liquidated-damages provision for about $9,300 against $21,000 of your expected exposure. Transfer costs more than EMV when you hand a risk to someone who cannot control it.

False. Distributed padding is consumed silently (a crew given 80 days for 72 days of work takes 80), it destroys the honest critical path so you cannot see what matters, and when a real problem hits one activity the protection is sitting inside all the others. A single visible buffer, drawn down like money, is worth far more than the same days scattered.

Risk: you know the event and can put a probability and impact on it — price it, register it, respond to it. Uncertainty: you know the event but cannot honestly put a probability on it — bound it with a range and buy information (a mockup, a trial week, a boring program) to convert it into risk. Ignorance: you do not know the event exists — you cannot register it, so you manage it structurally with checklists built from other people's failures, an outside reviewer, and a contingency loading that assumes the list is incomplete. Northgate's most expensive event, the eleven-day submittal sitting in Kestrel's own office, lived in the third category.

Estimating contingency covers scope you know is coming because the drawings are not finished, and it should shrink as design completes. Construction contingency covers identified execution risk within the contracted scope, and every dollar traces to a register row. Owner's contingency covers scope the owner will add and sits outside the construction contract. Escalation allowance covers price movement in named commodities or trades between GMP and buyout, and nothing else. Management reserve covers unknown-unknowns at the company level and belongs to executives, not to the job.

Before the GMP, information changes the price: whatever you learn becomes scope, priced at unit rates in a competitive market with several bidders. After the GMP, information changes only the loss: whatever you learn becomes a change order priced in a market with one bidder — you — negotiating with a counterparty who has every reason to disbelieve you. The same fact is worth different money depending on when you learn it, which is the whole reason the influence-versus-cost-of-change curve matters.

$5,150/CD of extended general conditions (project staff, trailers, temporary facilities, cleanup, safety) plus $5,500/CD of liquidated damages to Meridian. It is used to convert a register row's schedule impact into dollars — R-04's 18 CD is also $191,700 — and to value the project-level schedule buffer: Wei Chen's 25 calendar days between the internal target of August 24, Year 2 and the contract date of September 18, Year 2 are worth $266,250 of protected exposure.

"Could you show the owner the rows?" Not would you — the answer to that is often no, for legitimate competitive reasons. Could you. It works in both directions because it is about traceability rather than disclosure: if every dollar traces to a named event with a probability and an impact you actually believe, you are on the right side of the line even if nobody ever asks; if it traces to nothing, you are on the wrong side even if nobody ever asks.

Burn ratio = ($364,000 ÷ $520,000) ÷ 50% = 70.0% ÷ 50% = 1.40 — the "job is eating itself" band. At this pace the reserve is exhausted around 71% complete.

What you cannot conclude: whether the draws were legitimate. Every one of them may have been a real, unavoidable cost. The ratio tells you the rate, not the cause.

What you can conclude: the job has no forward visibility. Without row IDs, nobody can say what the remaining $156,000 is protecting against, and without a register nobody can price what is left.

First week, in order: (1) Rebuild the register from the ground up against the remaining scope — not the original one, a forward-looking one, with a named owner and trigger on every row. (2) Reconstruct the log by categorizing all nineteen-plus draws into what they would have been rows for; the proportion with no plausible row tells you how bad the identification was. (3) Build a cost-to-complete forecast, not a cost-to-date report — a job 50% billed and 70% through its reserve is telling you something. (4) Take all three to your operations executive this week with a number and a recommendation, before the next monthly cycle. Burn ratios do not improve on their own.

Top-down: 4% × $6,100,000 = $244,000. Ratio to bottom-up: $244,000 ÷ $118,000 = 2.07×.

What you carry. Not $118,000. That is a mean, and funding a six-row portfolio at its mean means running out roughly half the time; the 80th percentile on a small portfolio where one event can dominate commonly runs around 1.5–2.0 times the mean, so call it $190,000 for the identified rows, then load for what is not on the list. With only six rows on a $7.2 million building, a loading in the range of half again is defensible — which lands you close to the company's $244,000 and tells you the standard is roughly right.

What you transfer. Whichever two rows have the longest schedule tails and a counterparty who genuinely controls them — a supplier for a lead-time row, an installer for a workmanship row — because transfer to the controlling party costs less than the expected value it removes.

What you say. "The register is six rows on a $7.2 million building; the number is defensible and the list is thin. I am carrying $244,000, I have transferred two rows for about $15,000 of hard cost, and if we win I want two weeks to find the rows we missed before I build the budget." The honest answer at hour five is not a different number. It is a number plus a stated plan to fix the thing that is actually wrong.