Chapter 28 Quiz — Cost Control

D. Budget is a controlled document. Committed is a stack of signed agreements. Cost to date is invoices and payroll, with a small judgment component in the accruals. Forecast at completion is arithmetic — cost to date plus cost to complete. Cost to complete is somebody's opinion, and it is the only column that can still be changed by a decision. A cost report that can be produced without a human being making a judgment is not a cost report; it is a spending statement with extra columns.

C. Six packages bought below the estimate for $340,000 (curtain wall $118,000, HVAC $86,000, finishes $64,000, electrical $42,000, fire suppression $18,000, elevators $12,000); four bought above it for $130,000 (earthwork $46,000, miscellaneous metals $38,000, site paving $24,000, masonry $22,000). `$340,000 − $130,000 = $210,000.` The number was a report about procurement, presented where readers expect a report about the job. It contained no information about production, no information about accruals, and no information about the $1,620,000 of scope nobody had bought.

B. Cost to complete = forecast − cost to date = $412,000 − $478,000 = ($66,000). That line is telling you, in the plainest language a spreadsheet has, that nobody re-forecast the code — the forecast was left at budget and the cost-to-complete column was computed as the difference. The correct forecast for a physically complete code is its cost to date, full stop: $478,000, variance ($66,000). A completed code is the easiest forecast on the report and the one most often skipped, because "done" feels like it needs no attention.

C. Forecast = executed subcontract value + executed changes + pending changes you expect to owe. Unless scope changes, you direct extra work, or the subcontractor fails, the number will not move. This is a large part of why general contractors subcontract so much work: you give up margin and you buy certainty. Every dollar you self-perform is a dollar whose cost you must forecast with judgment instead of read off a contract.

C. Both conditions, not either. Twenty to twenty-five percent, because early rates are contaminated by mobilization (a fixed cost over a tiny denominator), by a real learning curve, and by the noisiness of small denominators. Three periods, because one period is an event, two is a coincidence, and three is a system — and a system will keep producing that rate until somebody changes it. Note that the investigation is never premature. Only the forecast can be.

B. This is a Type 2 accrual — an unrecorded commitment. It is real, new, unbudgeted cost that appeared out of a conversation, so both columns move together. Contrast a Type 1 timing accrual — ready-mix delivered but not invoiced — where cost to date rises and the forecast does not, because the work was already in the forecast and you have merely recognized consuming some of it. A Type 2 accrual is, in substance, a change order that has not been written yet.

D. Level of effort measures elapsed duration ÷ total duration, so it reports progress whether or not anything happened. By construction it always says you are exactly on plan. It is the right method for supervision, temporary facilities, and cleanup — costs that genuinely accrue with time — but you must never read it as evidence about production. Cost-to-cost is circular (a wrong forecast produces an equally wrong percentage, and spending inefficiently makes you look further along); milestone credits money moved to a vendor as progress; units installed understates early effort.

C. The test is one question: did the work move, or did only the overrun move? The masonry work will still cost what it costs and the elevator work will still cost what it costs, so the transfer changes no dollar on the job. What it does is delete two pieces of true information — that you bought masonry badly and elevators well — that your estimating department needs for the next bid and your project executive needs this month. The tell is timing: legitimate transfers happen when the work happens; concealment transfers happen in a batch in the last three days of the month.

B. This is red flag #1, and it is the fastest read on any report. A report where nothing moved is suspicious, not reassuring. Ask for the method column; any line whose method is "budget" is unforecast. Then check whether the cost-to-complete figures on those lines are exactly budget minus cost to date — if they are, they are plugs, not forecasts, and the variance column is a column of zeroes because nobody computed anything.

C.

Actual unit rate   = $169,740 ÷ 20,700 SF = $8.20/SF   (budget rate: $322,000 ÷ 46,000 = $7.00/SF)
Percent installed  = 20,700 ÷ 46,000 = 45.0%
Quantity remaining = 46,000 − 20,700 = 25,300 SF
Cost to complete   = 25,300 SF × $8.20 = $207,460
Forecast           = $169,740 + $207,460 = $377,200
Variance           = $322,000 − $377,200 = ($55,200), 17.1% over

The distractors are the three most common errors: A is budget minus cost to date — the plug. B is the remaining quantity at the estimated rate, which quietly asserts the crew is about to become 17 percent more productive tomorrow for no stated reason. D is the forecast at completion, not the cost to complete.

B. $190,000 ÷ $1,824,800 = 10.4 percent. One badly-run self-perform package worth $1.2 million ate a tenth of the profit on a $48 million building. That is the arithmetic of general contracting, and it is why a project manager who thinks $190,000 is a small number on a big job has the scale exactly backwards.

C. Contingency is consumed when identified risks materialize; work is consumed as the job progresses. If the first is outrunning the second, the risk register was optimistic and the remaining reserve will not cover the remaining risk. On Northgate the gap was defensible — soils, dewatering, steel acceleration, an embed misplacement, rock in a trench, winter protection, every dollar nameable and front-loaded — but defensible is not comfortable. The dangerous version is the drawdown nobody can itemize, because that is contingency being used as a slush fund for overruns.

False. At month nine the structure is done and the entire expensive interior has barely started, so a favorable variance is a report about work that has not happened yet. It usually means one of two things: you bought the job well, or the report has not caught up with reality. On Northgate the $340,000 of buyout savings sat on top of $15,818,000 of unperformed work in those same six packages — a cushion of 2.1 percent, on the packages where two-percent adverse moves live.

False, and the error runs in the expensive direction. On a unit-rate forecast, forecast at completion = cost to date ÷ percent complete, so a missed cost understates the forecast by the missed amount ÷ percent complete. A $10,500 accrual missed on a code that is 42 percent complete hides $25,000. At 25 percent complete the same $10,500 hides $42,000. The accrual does not merely change the cost; it corrupts the evidence the forecast is built from.

False — and the distinction is worth being precise about. Budget × (1 − percent complete) can show an overrun once cost to date exceeds budget × percent complete. On the Northgate elevated deck it would have produced $264,000 × 0.42 = $110,880, a forecast of $298,880 against a $264,000 budget — an overrun, though only about half the real one. The method that is genuinely incapable is the one the field engineer actually used: carrying the forecast at budget and letting cost to complete be the plug. A forecast structurally incapable of delivering bad news is not a forecast.

False. It misstates two contracts' cost of the work, it misstates the company's work-in-progress schedule, and on a cost-reimbursable or guaranteed-maximum-price contract it takes money from an owner who is paying the cost of their work. It also never stays small, because the month you would move it back is always a month in which another job needs the room. The damage is rarely the dollars; it is what discovery does to a surety's confidence in every number the company has ever submitted.

False. On a fixed price it is not your direct cost, which is exactly why the committed method works. But a subcontractor losing money is a schedule risk and a forecast risk: pending changes you have not carried, backcharges that shrink in negotiation, and — the one that hurts — a distressed subcontractor whose default replaces a contract price with an emergency price. Watch pay applications against production, crew size trends, supplier calls, and lien notices, and carry the risk in writing even when you do not yet carry it in dollars.

True, with one condition that is not optional: the accrual must be documented, sourced, labeled as an accrual, and reversed when the invoice lands. Yards placed times the purchase-order unit price is a defensible basis; a plug is not. Accrue from the field's record of what was performed, not from accounting's record of what was invoiced — and never guess in secret.

Billing 60 percent of a $40,000,000 contract earns $24,000,000 of revenue. Spending 75 percent of a $36,000,000 cost budget spends $27,000,000. You are $3,000,000 underwater right now.

The trap is that no line has to show a variance for that to be true. Every code can be forecast at budget; the budget column and the forecast column can be identical on all four hundred lines; the report foots; the report is accurate. It is also useless, because it is reporting the plan back to you.

The relationship between money spent and work produced is the entire signal. Cost to date on its own is noise. Cost to date divided by work produced is a unit rate, and that rate applied to the work remaining is a forecast. That chain — actual cost, over actual quantity, times remaining quantity — is the whole of cost control, and it runs toward the future.

Because the actual rate is a measurement of the entire production system as it really is — this crew's composition, this superintendent's sequencing, this site's material haul, this design's detailing complexity, this subcontractor's cooperation, this month's weather. The estimate priced a typical version of the work, by someone who had not met any of those things. When a prediction and a measurement disagree, the measurement wins.

Forecasting the remaining work at the estimated rate implicitly asserts that all of those factors are about to change at once — which they occasionally do, but only when somebody deliberately changes them. Hence the asymmetry: to forecast at the actual rate you owe nothing; to forecast better than actual you owe a written, specific, mechanical reason with a date. "Jamal split the level-4 topping into two placements and added a dedicated finishing crew effective November 8" is a reason. "We expect to improve" is a hope with a number attached.

Reaction one: "How did you let this happen?" Difficult forty minutes, a note in the file, a reputation as the project manager who has problems.

Reaction two: "What are we changing, and when do we know if it worked?" — with the forensics saved for the lessons-learned session at closeout, where they teach something.

The first reaction is more expensive, and the arithmetic says why. An overrun caught at 25 percent complete is mostly preventable; an overrun caught at 75 percent complete is mostly history. A company that makes bad news costly to deliver teaches every project manager to carry the number at budget and hope — which does not stop fades, it only delays them until they are four times bigger and nothing can be done. The pressure to hide bad news comes almost entirely from how organizations react to it. You get the reports your reaction deserves.

Direct cost of work, forecast      $41,240,000 + $88,000 = $41,328,000
General conditions, forecast                              $2,948,000
Insurance and bonds                                         $907,000
Subtotal before contingency, forecast                    $45,183,000
Budget for the same three lines                          $45,007,000
Variance                                                   ($176,000)

Projected contingency at completion = $608,000 − $176,000 = $432,000
   Meridian's 75%  = $324,000
   Kestrel's 25%   = $108,000

Kestrel's projected share falls from $130,000 to $108,000. Kestrel loses $22,000 of the $88,000 — 25 cents on the dollar — and Meridian loses $66,000.

Once contingency reaches zero, that ratio goes to 1:1. Kestrel eats overruns dollar for dollar, because that is what "guaranteed" means. That is the real reason an overrun matters this early: not the $22,000 it costs this month, but that it consumes the cushion protecting you from every remaining risk on the job — and the marginal cost of the next overrun is four times higher on the far side of zero.

Forecast at completion = committed + pending
                       = $4,224,000 + $61,000 = $4,285,000
Cost to complete       = $4,285,000 − $2,956,800 = $1,328,200
Variance               = $4,310,000 − $4,285,000 = $25,000 FAVORABLE

Money spent  = $2,956,800 ÷ $4,224,000 = 70.0% of the committed value
Work complete = 62%
Gap = 8 points

Two separate things are true and you must not confuse them. The forecast is $4,285,000 — that is your cost, because it is a fixed-price subcontract, and the buyout saving has fallen from $86,000 to $25,000 because you directed $61,000 of work.

The 8-point gap and the crew count are a different signal entirely. Billing ahead of production plus a crew halving over six weeks is the classic distress pattern. This week: verify installed quantities independently rather than accepting the schedule of values; check whether the two suppliers who called have been paid and whether any preliminary lien notices have arrived; require a manpower plan and a six-week look-ahead in writing by Friday; and price what it would cost to complete the balance with somebody else, so that if you need that number you are not computing it under duress. Carry the $4,285,000 in the forecast and write the risk note. A distressed subcontractor is a schedule risk and a forecast risk before it is ever a cost variance.