Chapter 18: Plan and Control Cost, Funding, and Cash

Four numbers, one project

It is month twenty at BlueLine, and the steering committee has four numbers for one project, and all four are true. Marta Reyes, the transport authority’s finance officer, reads the commitment ledger: 1,330 million units committed against a plan of 1,290, which sounds like trouble, the project signing ahead of itself. Lena Voss, the project director, reads the cost variance: 1,190 million units incurred against a plan of 1,260, seventy million units under budget, the word the mayor’s office will like. The accountant reads the cash ledger: 1,010 million units actually paid against 1,150 planned, which sounds like the corridor is ahead of its own spend. And Daniel Osei, the consortium’s finance director, has the fourth number, the one nobody quoted first: the forecast at completion, about 2,550 million units against the 2,540 the second-half authorization carried in month seventeen. Ten million units. Rounding noise, on a corridor of that size, except that it sits inside a cash curve that spikes in month twenty-five and a funding staircase sized to an earlier plan. The ten million is not the story. The cash curve is the story.

Four people, four numbers, one project. The committee’s first question decides the meeting: which number is the project? The honest answer is the one that makes everyone uncomfortable: all of them, and none of them alone. Commitment says what the project owes. Incurred says what the work has cost. Cash says what has left the bank. Forecast says what the end will look like if the current evidence holds. They disagree in ordinary, predictable ways, and the disagreement is not a defect in the reports. It is the information. A project leader who cannot say which ledger each number comes from, and why the ledgers diverge, will be told the project is under budget on the same morning it cannot pay its contractors.

This chapter’s promise is deliberately modest: not to make a financial specialist of you, but to give a nonfinancial project leader enough financial fluency to make responsible decisions. The specialist builds the model, knows the tax treatment, negotiates the facility. The leader decides what the money means, which story is being told, and what to do when the ledgers disagree. Chapter 7 built the business case; chapter 15 built the estimate. This chapter turns the estimate into the machine that runs alongside the schedule of chapter 17: the cost model, the time-phased budget, the funding profile, and the cash curve, read together, because the corridor is about to discover that the schedule decision of month nineteen has a money shape, and nobody has drawn it yet.

Four ledgers, four stories

Every project runs on four ledgers, and the first skill of financial fluency is naming them, because the words are used loosely and the looseness costs money. The commitment ledger records money the project has legally obligated: contracts signed, purchase orders issued. Commitment is the moment the project loses its freedom, not the moment money leaves, and most projects discover their real cost profile when they discover their commitment curve, because commitments run ahead of everything else. The accrual ledger records work actually done and certified, whether or not it is paid; it is the ledger of progress, the closest of the four to the schedule of chapter 17. The cash ledger records what has actually been paid: invoices settled, retention released, the mobilization fee transferred. It is the ledger of survival, because a project can be fully accrued and still die for lack of cash. The forecast ledger records what the project expects to cost by completion, the estimate at completion. It is the ledger of judgment, the only one that looks forward.

The discipline that makes the four ledgers useful is the reconciliation, one equation every project leader should be able to write without looking it up: paid, plus unpaid accrued, plus committed-not-yet-incurred, always equals committed. Every unit committed is in exactly one of three states: paid, owed because the work is done and certified, or owed because the contract is signed and the work is not done. Nothing else exists. The reconciliation is boring arithmetic, and it is the most valuable hour in the month, because it is where the favorite-number error gets caught.

The month-20 ledger at BlueLine makes the equation concrete.

The equation holds: 1,010 paid, plus 180 unpaid accrued, plus 140 committed-not-yet-incurred, is 1,330 committed. And the three variances tell three different stories. The commitment variance, 40 over plan, is the acceleration: the eastern package and the fast-tracked integration were signed early, exactly what the schedule decision of month nineteen required. The incurred variance, 70 under plan, is the delay: the eastern works could not start until the drainage approval landed, so the month’s valuations ran below a plan that assumed they would run. The cash variance, 140 under plan, is both: less certified, and the corridor pays certified claims on sixty-day terms, so cash lags both. Three variances, three causes, and none of the three is the project’s health. The health is in the fourth ledger, the forecast, and it says the corridor will finish about ten million units above the month-seventeen forecast, inside the width of the estimate. That is the real headline, and not the headline anyone quotes.

The favorite-number error is the failure pattern of the ledger, and it deserves its name because it is so human. Every room has a favorite number, the one that tells the story the room wants to hear, and the room reports it and stops. The project that is behind schedule loves the incurred variance, because under budget sounds like health. The project accused of waste loves the commitment ledger, because committed sounds like discipline. The project that wants the gate to pass loves the cash ledger, because spent sounds like progress. And the project in denial loves the forecast, because a forecast can always be adjusted. The tell is the question the room cannot answer: “and what do the other three ledgers say?” A project leader who answers that in one breath, without opening the pack, cannot be misled, because the four ledgers only tell the truth when they are read together.

The minimum viable instrument of the whole discipline is therefore not a budget and not a dashboard. It is the reconciliation: one line of arithmetic, run monthly, signed by a named owner, presented before the variance narrative. A project that reconciles its ledgers monthly cannot be surprised by its own money. A project that reports a single favorite number has already chosen which truth to carry.

What the money buys

The cost model is the map from the work to the money, and it is where the cost conversation starts, because a budget not attached to the work is a number with nowhere to live. Chapter 14 built the decomposition, the work breakdown structure that turned the corridor into segments, packages, and work items; the cost model is that decomposition in money terms, every work package carrying a cost line, an estimate, and an owner. The model’s first virtue is that it can be wrong in a useful place: when a package overruns, the model says which one, and the conversation moves from “the budget is over” to “the eastern culverts are over,” which are different conversations with different fixes.

The categories of cost are few and routinely confused, so they are worth naming once. Direct costs attach to a work package: the civils gangs, the ticketing hardware, the integration team. Indirect costs attach to the project without attaching to a package: the site office, the insurance, the shared traffic-management staff. They are where the invisible money lives: a project that tracks only direct costs discovers its overhead in the closing accounts. Fixed costs do not change with the amount of work, the depot lease, the licensing fee; variable costs do, the concrete, the cloud usage. The two historical categories are the ones leaders misuse most. Sunk costs are the money already committed or spent that no future decision can recover: the 1,330 million units on the corridor’s commitment ledger, of which, as chapter 7 said, a thousand million are gone, and no forward decision should treat them as reclaimable. Opportunity cost is the value of the alternative forgone: the do-minimum row from chapter 7, the 2,250 million units the city does not spend, and the programs it funds instead. Sunk costs get quoted as reasons to continue; opportunity costs get forgotten because they are invisible. A cost model with no line for either has deleted the two decisions that matter.

The full cost of ownership is the second lens: the project’s cost is not what the project spends, it is what the asset costs across its life, capital to build, operating to run, maintenance to keep, renewal to replace, decommissioning to close. The corridor’s capital envelope is 2,400 million units, and its operating subsidy is 90 million units a year, and the second number belongs in the same conversation as the first, because the money is the money whether spent in year zero or year twenty. The model must show the boundary explicitly: what sits inside the project’s budget, the capital and the transition, and what sits outside but on the project’s conscience, the operating subsidy, the maintenance regime the operator inherits, the renewal the city will fund in year fifteen. A leader who has built the full-cost view once never again approves a capital-only comparison.

The third lens separates expense from cash requirement, and it is where the money conversation gets interesting, because not everything the project needs is a cost. The KijaniPay lending pilot is the standing example. The scope statement of chapter 16 commits a working-capital lending pilot for 500 merchants in Lagos by 31 January, and the budget has no line for it; the seam walk named the contradiction. The reason is not that the budget is wrong: lending capital is not an expense. It is money the project must put on the balance sheet, lend out, and get back, with interest and with risk, and the project’s cost is the interest, the defaults, and the operations, not the capital itself. The same distinction appears everywhere, the security deposit, the licensing bond, the prepayment, the settlement float: cash the project must command, none of it cost, and a budget built only of costs will be surprised by the cash its own plan requires.

The cost model’s failure mode is the decomposition’s from chapter 14, applied to money: the model that mirrors the organization chart instead of the work. The check is the seam walk from chapter 16 done in money — take the scope statement and ask what each promise costs, take the risk register and ask which lines the contingency covers, take the schedule of chapter 17 and ask which packages spend when. The moment it stops being walked, it is decoration.

The pockets and the limit

The estimate, from chapter 15, is the honest range of what the work will cost. The budget is the plan’s decision about what the project intends to spend. Between the two sits the structure that decides who absorbs the unexpected, which the book has been calling the pockets, and which deserves a full treatment here because it is where most cost governance lives.

The contingency is the pocket for named risks, held by the project, sized by the risk analysis, and released against evidence. At BlueLine, the month-seventeen authorization carried a forward base estimate of 1,390 million units and a contingency of about 110 million, roughly 8 percent, held by the consortium and drawn against the named risks: the eastern flood-plain drainage approval, the ticketing integration outcome, the final scope of the access standard. By month twenty, about 30 of the 110 has been applied against the drainage conditions that added the culvert works, and the remaining 80 is the corridor’s freedom of action for the risks it has not retired yet. The rule is chapter 15’s: contingency moves within the project against named risks, spent by decision, with the trigger that names the evidence.

The management reserve is the pocket for the unknown unknowns, held above the project, by the organization, and released only through governance. The discipline is the two-pocket structure from chapters 15 and 16: the project manages the risk it can name, the organization prices the risk it cannot, and the two pockets are never merged. The project that merges them spends the unknown-unknowns money on known problems and calls it health. At BlueLine the reserve takes the form the city gave it in month seventeen: beyond the 80-percent upper edge of 1,500, the modeling runs to a 1,570 ceiling, and the band from 1,500 to 1,570 is the city’s, released only through re-authorization with the council’s visibility. The band is a governance fact before a financial one: a tolerance written at the top of the decision-rights map from chapter 8, and its release is a gate decision, not a finance entry.

The funding limit is the term most often confused with the budget. The budget is what the project plans to spend. The funding limit is what the organization has actually committed to provide, and the two should be reconciled separately, because the gap between them is the project’s true constraint. A project can be fully budgeted and underfunded — budget approved, money not appropriated — or over budget and fully funded, with the organization committed to more than the plan requires. At BlueLine the forward funding limit is the 1,500 million units the council appropriated against the second-half authorization, with the band to 1,570 available by re-authorization, and the corridor’s cost forecast is inside it. The interesting question is never “is the project on budget.” It is “what is the funding limit, what is the forecast, and what is the difference,” because the difference is the organization’s actual exposure, and it is the number the sponsor, not the finance director, must own.

The pockets have a fourth dimension that is quietly decisive, and it is time: nominal versus real. The appropriation is in nominal units, the units printed in the year they are spent; the cost of the work moves with the price level; and a project that plans in nominal units against real cost will discover the gap in the closing accounts, unless the gap is owned somewhere.

The cash curve and the staircase

The time-phased budget is the cost model with time added: each package’s cost spread over the calendar, month by month, so the project can say not only what the work will cost but when the money is expected to flow. The schedule of chapter 17 provides the time, the cost model provides the amounts, and the product is the S-curve of expected cumulative spend: shallow at the start, steep through the works, flattening as the corridor approaches acceptance. It is the plan’s answer to “when does the money need to be there,” and it is the single most under-built artifact in project management, because it requires the integration of schedule and cost model that most projects never perform.

The cash curve is not the same shape as the cost curve, and the difference is the discipline’s whole content. The cost curve follows progress: work done, valued and certified. The cash curve follows payment: invoices issued, paid on the contract’s terms after certification, thirty days, sixty, ninety, with retention held back until handover. Every contract is a small machine that converts progress into cash with a delay and a haircut, and the project’s cash curve is the sum of those machines. The corridor pays its civils contractors on sixty-day terms with 5 percent retention, so its cash curve lags its cost curve by about two months and is permanently short of it by the retention pool, which is not an error and is not recoverable until acceptance. The funding staircase is the third shape: the appropriation tranches the organization has committed, arriving on its calendar, month twenty-two, month twenty-six, month twenty-eight, sized against the plan’s spend curve when the plan was approved. And the gap between the cash curve and the staircase is the project’s real financial question: the bridge, the amount the project must command between the moment the money is needed and the moment it arrives.

The month-20 picture at BlueLine is the chapter’s primary figure, because it shows all four shapes disagreeing at once.

Figure 18.1: The money picture at BlueLine, month 20. Four shapes
answer four questions: what the remaining work costs (the cost
baseline), who absorbs the unexpected (the contingency, held by the
project, and the reserve, held by the city and released only by
re-authorization), what the city has appropriated and when (the
funding staircase), and when the invoices must be paid (the cash
curve). The gap between the staircase and the cash curve is not a
cost overrun and not a budget cut. It is a timing decision, and at
month 20 it has a name: the eastern acceleration.

  THE POCKETS (million units, approved forward structure, month 17)
  1,570  reserve ceiling; the city holds the band above 1,500 and
         releases it only through re-authorization
  1,500  the funding limit of the approved forward ask; the
         80-percent upper edge of the estimate from chapter 15
  1,390  forward base estimate, the plan's expectation; by month 20,
         about 30 of the 110 below it have been drawn against the
         eastern drainage conditions
  1,280  the 80-percent lower edge; the estimate's width is a fact
         about the model, not a promise about the world

  THE TIMING (million units, forward, cumulative)
  month       20    21    22    23    24    25    26    27    28
  cash out    120   280   460   660   860   1,050 1,230 1,390 1,540
  funding in    0     0   300   300   650     650 1,050 1,050 1,350
  gap        +120  +280  +160  +360  +210    +400  +180  +340  +190

  The gap peaks at month 25, the phased-opening month, at about
  +400, and even at month 28, with every approved tranche drawn and
  the reserve band fully released, the 70 above 1,350, the corridor
  is still about 120 short of its forward cash curve.

The arithmetic of the picture tells the story the variance narrative missed. The forward cash curve sums to 1,540 million units, the estimate at completion minus what has already been paid. The forward funding, the tranches still to arrive, sums to 1,350, and the reserve band, fully released, adds 70, for 1,420. The corridor is short about 120 million units against its own approved envelope, and the shortfall is not a cost overrun: the cost forecast is inside its limit. It is a timing fact: the eastern package had to be committed early, to beat the wet season that the schedule decision of chapter 17 identified, and the acceleration moved the cash curve left, against a staircase sized to the older, smoother plan. The corridor that is 70 million units under budget on the incurred ledger is simultaneously about 400 million units short of cash at its peak. Both statements are true, the first one is the headline, and the second one is the project.

The instruments for closing a cash gap are few, and the craft is choosing among them with the pockets and the schedule in the room. The re-profiled draw moves the appropriation tranches earlier within the same total, borrowing from month twenty-six to pay month twenty-one; it costs nothing in money and everything in governance, because the city’s appropriation calendar is a commitment the council made. The renegotiated payment term gives the eastern contractor staged mobilization, a smaller advance with more frequent certifications, slowing the spike by the quarter the wet season allows. The bridge is a credit facility drawn against committed funding, with a price, the interest between draw and reimbursement, and a risk, the covenant and the counterparty. The re-authorization releases the reserve band against evidence — the 80-percent forecast, the named risks, the trigger, the whole discipline of the chapter 8 gates — adding money and scrutiny in equal measure. And the fifth instrument exists and must be named because it is the one the room reaches for first: do nothing and call it cash management, which is not an instrument but a forecast that the problem will solve itself, and a cash gap has never solved itself.

The question that decides among the instruments is the one the chapter has been building: which ledger is telling a story the other ledgers do not confirm. If the gap is timing, the re-profiled draw and the payment terms fit, and the reserve stays closed. If the gap is cost, if the forecast has moved outside the limit, the re-authorization fits, and the reserve opens with evidence. The month-20 corridor is a timing story, which is why reading all four ledgers before choosing is what saves it from opening its reserve for a cash-flow problem that re-profiling solves.

Forecasting the end

The estimate at completion is the fourth ledger’s number, and it deserves its own treatment. It is the number every decision ultimately lands on, and the number most often produced by a formula without anyone saying what the formula assumes. The formula does not matter. The assumption does.

The corridor’s own forecast is built from the bottom up: the four delivery partners re-estimate their remaining work against the current evidence, the drainage conditions, the access-standard scope, the integration outcome, and the sum lands at 1,220 million units of remaining work; committed 1,330, forecast at completion 2,550, about ten million above the month-seventeen forecast of 2,540. The movement is the schedule decision priced: the eastern additions add about 45, the resequenced integration and the phased opening save about 25, the fast-track premium adds about 10, and the re-phased mitigation line defers about 20; the four movements sum to ten. The bottom-up forecast is the team’s own claim about the future, and its virtue is that it is owned: every component has a name and an estimate behind it.

The formulas exist because bottom-up forecasting is slow and the world moves monthly, and the standard earned-value formulas give the project a faster instrument that extrapolates from the measured past. The corridor’s systems package, the one whose integration the schedule of chapter 17 resequenced, is the worked example, and the numbers are small enough to reproduce by hand. The package’s budget at completion, BAC, is 120 million units. Through the month-22 review, the planned value is 60, half the package by the calendar; the earned value is 45, the package 37.5 percent done; and the actual cost is 48. Two ratios follow. The cost performance index, CPI, is earned value divided by actual cost, 45 over 48, 0.9375: the project is getting about 94 cents of progress for each unit spent. The schedule performance index, SPI, is earned value divided by planned value, 45 over 60, 0.75: the package is three quarters of the way through its planned progress.

Three estimate-at-completion forms use the two ratios, and the choice among them is an assumption about what persists.

The rate-based form assumes the cost performance continues, and it says the package lands at 128, 8 over budget, because the project is spending more per unit of progress than planned. The schedule-inclusive form assumes both cost and schedule performance persist, and it says about 155, because a package a quarter behind schedule must be finished somehow, and finishing late has a price: the compression of chapter 17, the premium, the overtime, the rework. The bottom-up form says the team’s own re-estimate, 126, because the remaining work is different from the past work, the integration against the frozen design is a new shape, and the past rate is not a prediction about it.

The three forms disagree by 30 million units on a 120 million-unit package, and the disagreement is the teaching. The estimate at completion is not a number. It is a set of assumptions about what persists, and the question to ask of any forecast is not “what is the number” but “what does this number assume persists, and do we believe that?” The corridor’s bottom-up forecast of 2,550 assumes the team’s re-estimates are right and the acceleration holds; the rate-based form applied to the whole remaining work would say something more pessimistic, because the corridor’s past cost performance includes the waste of the delay; and the schedule-inclusive form would say something harsher still, because the phased opening compresses the systems and ticketing work into a shorter window, and compression has a price. The chapter 15 discipline applies whole: the forecast is the current evidence-based expectation, revised monthly, with its assumptions and its width, and the corridor’s 80-percent range on the remaining work runs to about 2,630, still inside the total authorized limit of 2,720, which is the sentence that lets the corridor choose its instruments calmly instead of in crisis.

Chapter 38 builds the full earned-value system, and this chapter uses only its two most useful ratios, because they are enough to make the point that matters here: the past rate is a hypothesis, not a law, and the formula that assumes the past persists must be able to say what it is assuming. The discipline is the re-forecast as learning, the monthly revision that chapter 15 built, with the forecast’s history kept, so the room can see the corridor’s forecast move as evidence landed. A forecast that moves with evidence is the project learning. A forecast that never moves is the project in denial. And a forecast that moves by formula, without anyone saying which formula, is the project delegating its judgment to arithmetic.

Escalation, taxes, and the cost of money

The financial environment adds three layers the model must carry, and each one is small until it is not.

Escalation is the contract’s mechanism for inflation, and it is a risk allocation before it is a calculation. The eastern culvert package, 45 million units at the base price, carries the contractor’s standard clause: the index movement applies to 60 percent of the base, the labor and materials that track the construction-cost index. If the index runs at 6 percent over the package’s life, the escalation is 45 times 0.60 times 0.06, about 1.6 million units, paid on top of the base. The clause is the parties deciding who carries the inflation risk, the corridor through the clause or the contractor through a fixed price that prices the risk in advance, and the decision belongs in the chapter 20 procurement conversation: a project that accepts an escalation clause without estimating its cost has made a procurement decision by default. And the nominal-versus-real discipline applies: a project that plans its reserve in real terms while its contracts escalate in nominal terms will find the difference in the closing accounts.

Tax is the layer the cost model must include because it is a cost and a cash flow, and the layer projects most often discover in the invoicing stage, the worst time to discover it. The value-added tax on the ticketing hardware, the withholding on the contractor’s certificates, the import duties on the fare gates, the recovery of input tax: each is a percentage of a cash flow with a timing, and each sits in a different ledger, because the tax that is a cost and the tax that is a cash-flow timing item are different facts. The discipline is not to become a tax specialist; it is to have the treatment named in the model by the people who own it, amounts estimated and timing stated, so the project is not surprised by the difference between the invoice and the payment.

The cost of money is the layer where the month-20 cash gap gets its price. The corridor’s bridge, the 120 million units it must command between the cash curve and the staircase, carries a financing cost, and the arithmetic changes the decision: at a 7 percent annual facility rate, bridged for an average of six months, the cost is 120 times 0.07 times 0.5, about 4.2 million units, against which the re-profiled draw costs nothing but governance and the renegotiated payment terms nothing but negotiation. The cost of money also speaks in the corridor’s own language, the 5 percent discount rate of the chapter 7 case: a unit spent earlier is worth less than a unit spent later, and the acceleration that moved the cash curve left moved the present value left with it. Time is money, and the project that ignores the time of its money is spending its own case against itself.

The fourth element is currency, for projects that cross borders. KijaniPay settles merchants in Lagos and Nairobi, in two currencies, and its float, the money between collecting from the customer and paying the merchant, sits in both, carrying an exchange risk the launch project’s budget never named. The discipline is the tax discipline again: the exposure is owned, the rates named in the model, the sensitivity run, and the decision about who absorbs the movement made in the light. A project leader does not need to trade currencies; the leader needs a named owner for the exposure, because the exchange movement will arrive whether or not the project named it.

The money rhythm changes with the work

The cost discipline is universal, and its rhythm is not: the delivery styles differ in the way the money breathes, and the difference is the source of half the method arguments in the profession.

The predictive project breathes in gates. Its money lives in a baseline with a change threshold, and the rhythm is the appropriation tranche and the stage gate: the organization authorizes what it can see, the next tranche is earned by evidence, and cost control is variance against the baseline, escalated through the governance of chapter 8 when the variance crosses the tolerance. The BlueLine corridor is the standing case: the second-half authorization, the tranches at months twenty-two, twenty-six, and twenty-eight, the reserve band held by the city, and the monthly reconciliation that tells the committee which ledger is moving. Its strength is that the money is predictable, and its failure mode is the gate that approves on habit, the tranche released because the work is going well rather than because the outcome is still worth it — the sunk-cost gate: if we knew now what we know now, would we authorize this again?

The adaptive project breathes in renewals. Its money lives in a budget that is a hypothesis, renewed on evidence, and the rhythm is the short cycle: the team earns the next quarter by what the last quarter showed, and cost control is the empirical review, spend against outcome, adjusted at the cadence of the learning. The KijaniPay platform is the standing case: its 210 million-unit budget is a spending plan, not a promise, the product council renews funding against demonstrated evidence, the change threshold from chapter 16, a budget movement above 5 million units, escalates, and the lending pilot’s cash requirement is managed as a balance-sheet decision rather than a budget line. Its strength is that the money follows the evidence, and its failure mode is the renewal that becomes a ritual, the quarter approved because the team is busy — the sunk-cost gate dressed in velocity.

The hybrid project breathes in seams, and the seams are where the money story gets honest. The Meridian clinics are the standing case: the grant pays on evidence, the access outcomes of wait times and screening uptake, and it ends at month thirty-six, so the money rhythm is the grant’s reporting calendar, the evidence claims, the quarterly reimbursements, running against the construction gates, which pay on certified progress, and the platform releases, which renew on demonstrated adoption. The three rhythms disagree, and the craft is the seam: the construction certifies, the grant reimburses, the platform renews, and the project leader’s job is to make the three calendars visible in one picture, because the money will follow whichever rhythm the project forgets. The Northstar response is the compressed version, the restricted-funding window where unspent money lapses: the rhythm is the donor’s calendar, and the staged payments tied to verified deliveries are the cash discipline wearing the ethics discipline’s clothes.

The method-neutral rule holds: the difference between the styles is not whether cost is controlled, it is how often the control happens and what evidence earns the next unit. The predictive project controls against the baseline, the adaptive against the hypothesis, the hybrid against the seam. All three reconcile the ledgers monthly, and all three know which ledger their organization is prone to favor, which is the difference between a money rhythm and a money story.

The money story can lie in four ways

The failure patterns of financial management deserve to be named as characters, because each is produced by competent people doing what the room rewarded, and each has a signal that reveals it.

The favorite-number executive reports the ledger that flatters and stops there. Under budget on incurred, so the project is healthy; committed ahead of plan, so the project is disciplined; cash under plan, so the project is efficient. The tell is the sentence the executive cannot finish: “and the other three ledgers say…” The cost is the decision made on one truth while the project lives in another, the month-20 corridor, seventy under budget on incurred and four hundred short of cash at the peak, with the first number quoted and the second one the project.

The contingency cashier spends the risk pocket on ordinary problems. The eastern conditions arrive, and the cashier draws the contingency to cover the culvert works, which is legitimate, and then draws it again for the cash gap, which is not, because the contingency is for the cost of named risks, not the timing of cash, and a project that spends its risk money on its cash flow is a project whose named risks arrive later with nothing behind them. The tell is the contingency balance falling while the risk register is unchanged. The cost is the reserve empty at the moment the risk it was built for actually lands.

The commitment hoarder signs ahead of the money. The acceleration requires the eastern package early, and the hoarder signs everything the schedule wants, and then some, because signing feels like progress and the invoice is someone else’s problem. The tell is the commitment curve rising while the funding staircase stays flat, the gap that no report names. The cost is the cash crisis that arrives as a surprise to everyone except the hoarder, who knew the commitments were there and hoped the money would follow, which is not a plan, it is a bet.

The invoice waltzer reclassifies to look green. The certification that would move a cost from committed to incurred is delayed, so the incurred variance stays favorable; the invoice that would move it from incurred to paid is held, so the cash ledger stays calm; the tax treatment is revised, so the line moves to a pocket that nobody watches. The tell is the reconciliation that no longer sums, or sums only after three adjustments, and the meeting that no longer asks why. The cost is the four ledgers uncoupled, the favorite-number executive’s supply chain, because the waltzer is the one who makes the favorite number available.

The four characters share a root: each one replaced the reconciliation with a performance of it, the favorite-number executive the variance, the contingency cashier the reserve, the commitment hoarder the approval, the invoice waltzer the accounts. The controls that keep each honest are the chapter’s instruments used as designed: the four ledgers reconciled monthly, by a named owner, before the narrative; the contingency drawn against named risks with triggers; the commitment curve reviewed against the funding staircase at the same cadence; and the reconciliation that must sum before anyone speaks. The money story does not need to be comfortable. It needs to be complete, because the decision the room makes on a complete money story is a decision the project can live with, and the decision it makes on a favorite number is a decision the project will pay for.

The ethics of the money story is chapter 4’s forecasting ethics in the financial register, and it has teeth because the victims are concrete: the contractor whose certified work goes unpaid while the project reports under budget, the donor whose restricted funds are spent on the wrong line, the community whose clinic waits while the grant lapses, the taxpayers whose reserve is opened for a cash-flow problem that re-profiling would have solved. The standard for the report is the one chapter 40 will build in full: context, evidence, interpretation, options, recommendation, with the uncertainty named. And the standard for the leader is one sentence: never let the report tell a story the ledgers do not confirm, because the ledgers are the only witnesses, and the room cannot check what the leader will not show.

What the machine can do with the money files

The assistant has a genuine and bounded place in the money work. It can reconcile the four ledgers from the provided files, paid plus unpaid accrued plus committed-not-yet-incurred against committed, and flag the rows that do not sum, exactly the check the invoice waltzer depends on nobody doing. It can compute the earned-value ratios and the estimate-at-completion variants from the provided planned values, earned values, and actual costs, with each formula labeled and each assumption stated, and it can produce the cash curve and the funding staircase from the provided spend and appropriation data, flagging the months where the gap appears. It can draft the variance narrative from the reconciliation — what moved and by how much — labeled as a draft for a named owner to verify. The source data is the approved, non-confidential cost and funding content, redacted before prompting; the outputs are drafts until a named owner verifies them; and the audit record says what was generated, from what, checked by whom, and decided by whom.

Four boundaries matter, because money is where the machine’s confidence is most dangerous. Classification is human: whether a line is an expense, a capital item, or a balance-sheet requirement — the lending capital that is not a cost, the tax that is a timing — is a judgment no model can make from the files. The release of the pockets is human: the machine can compute what the reserve would cover, and the decision to release it is governance, made against the evidence by the people chapter 8 named. Borrowing is human: the machine can price the bridge at the stated rate, and the decision to take on debt belongs to the organization. And the data is sensitive: contract rates, tax positions, donor terms, and merchant settlement data stay on the project’s owned infrastructure. The verification is the room: the leader walks the reconciliation with the finance owner, tests the machine’s flags against the contract calendar, and makes the funding decisions in the light, while the machine accelerates the arithmetic and the flagging, and the classification, the release, and the decision stay human.

Practice

One. A quick classification. For each statement, name the cost category, and say which ledger it would sit in. (a) “The corridor’s traffic-management staff are charged to the project during construction and to operations after opening.” (b) “The consortium’s head-office rent is apportioned to the project each month.” (c) “The 1,330 million units already committed to the corridor cannot be recovered if the program stops.” (d) “The city forgoes the 2,250 million units it would save by not building, and the programs those units would fund.” (e) “A new fare-gate machine that will serve passengers for fifteen years.” (f) “The monthly cloud bill for the KijaniPay test environments, which grows with usage.” (g) “The eastern culvert package, signed on the first of the month, with the mobilization invoice due in thirty days.”

(a) is an indirect cost during construction, a direct operating cost after; the boundary is a full-cost-of-ownership decision, not an accounting habit. (b) is an indirect cost, and its ledger is the accrual, because it is consumed monthly whether or not it is paid. (c) is a sunk cost, its ledger the commitment, and the only honest forward use of it is the chapter 7 question — is the continuation worth it — answered without the sunk amount in the arithmetic. (d) is an opportunity cost, and it lives in the business case of chapter 7, not in the project’s ledgers, which is exactly why it is so often forgotten. (e) is a capital cost, an asset whose life extends past the project, and the model must hand it to the operator with the maintenance and renewal lines visible. (f) is a variable cost, growing with usage, and its ledger is the forecast, because the actual depends on the usage the pilot produces. (g) is a commitment, and the mobilization invoice will move it to unpaid accrued when certified and to paid when settled; the three states of the reconciliation are the sentence.

Two. A numbers drill: the forecast you can reproduce. Verify the systems-package arithmetic from this chapter, then run the sensitivity. (a) Confirm CPI 0.9375, SPI 0.75, and the three estimate-at-completion forms, 128, 154.7, and 126. (b) Suppose the review comes a month later and the earned value is 52 against the same actual cost of 48 and the same planned value of 60. Compute the new ratios and the rate-based and schedule-inclusive estimates, and say what changed. (c) Recompute the eastern culvert escalation if the index runs at 9 percent and the clause applies to 70 percent of the 45 million-unit base.

(a) The arithmetic is reproduced in the table: CPI is 45 over 48, 0.9375, SPI is 45 over 60, 0.75; the rate-based estimate is 120 over 0.9375, 128; the schedule-inclusive is 48 plus 75 over 0.703, about 154.7; the bottom-up is 48 plus 78, 126. The three disagree because they assume different futures: the rate persists, the rate and the schedule persist together, or the remaining work is a new shape. (b) CPI becomes 52 over 48, about 1.08, the project is now getting more than a unit of progress per unit spent; SPI becomes 52 over 60, about 0.87; the rate-based estimate is 120 over 1.08, about 111; the schedule-inclusive is 48 plus 68 over 0.94, about 120. The schedule-inclusive estimate landing near the budget while the rate-based estimate is below it is the whole lesson: the package is still behind plan, and the compressed finish eats the cost efficiency. (c) 45 times 0.70 times 0.09, about 2.8 million units; the escalation doubled with the index and the linked share, which is why the clause’s parameters are a negotiation, not a formality.

Three. A field drill: build the ledger for your own project. Take the project you lead or know best. (a) Write the reconciliation for the current month: committed, and its three states, paid, unpaid accrued, and committed-not-yet-incurred, and check that the equation holds. (b) Draw the cash curve and the funding staircase for the next six months: when the invoices will land, when the money arrives, and where the gap is. (c) Answer three questions: which ledger does the room look at, which ledger would surprise it, and what is the one decision the disagreement forces?

The drill succeeds when the reconciliation sums, because a reconciliation that does not sum is the first signal of the invoice waltzer, and when the cash gap has a name and an owner, because an unnamed gap is a forecast that the problem will solve itself. The most common failure is the project that cannot produce the commitments ledger at all, because the contracts live in procurement and the purchase orders live in finance and nobody owns the map between them; the repair is a named owner for the commitment ledger, a governance decision from chapter 8 before it is a finance decision. The second failure is the cash curve drawn from the budget instead of the payment terms, the S-curve that ignores the sixty-day lag and the retention; the repair is to ask each contract what it pays, and when. The third failure is the drill itself deferred, the leader who knows the reconciliation is valuable and has not run it, which is the favorite-number executive in embryo.

Four. A decision room: the grant that pays in arrears. It is month fourteen at Meridian, and the money rhythm has collided with itself. The grant funds the program and pays quarterly in arrears, on evidence, the access outcomes the grant was written for. Clinic two’s construction certifies monthly, about 6 million units of certified progress a month, payable on thirty-day terms, and the grant claim for clinic one is in review, so the project faces a peak working-capital gap of about 18 million units through month sixteen. The options on the table: draw on Meridian’s operating reserve, which costs about 8 percent a year, so carrying 18 million for an average of six weeks costs roughly 170,000 units; negotiate a milestone draw with the funder, which takes about six weeks and moves the first reimbursement earlier; re-phase clinic two’s fit-out by a month, which slips the adoption corridor and delays the evidence the grant pays on; or absorb the gap as a financing line in the cost model and present it to the board as the true cost of the grant’s rhythm. Decide what Dana and the finance director should do, what they should refuse, and what the board report should show.

The defensible answer combines the milestone draw with a small, named financing line: negotiate the draw because the funder’s rhythm is the problem and the funder has an interest in the evidence arriving, and carry the residual gap explicitly in the cost model, because the 170,000-unit cost of bridging is the true cost of the grant’s payment rhythm and hiding it is how the favorite-number error starts. The re-phase is the weakest move: the grant pays on evidence, the evidence is adoption, and the adoption corridor is the project’s success metric from chapter 2, so a month of fit-out slips the thing the money is actually for, the sunk-cost gate in reverse, saving cash by destroying the value that earns it. The refusal is the informal bridge, the contractor asked to carry the delay informally: the payment terms are a contract, the working-capital problem belongs to the project, and chapter 20 will price the trust that gets spent this way. The board report should show the three rhythms on one page, the grant’s evidence claims, the construction certifications, and the platform’s renewals, with the financing line named and its cost stated, so the board sees the money following the rhythm the project forgot. The evidence that would change the answer: whether the funder’s milestone draw actually lands inside the gap window, and whether the adoption corridor can absorb a month, which is an assumption with a test date and an owner.

Five. The mastery drill: under budget, out of cash. It is month twenty-two at BlueLine, and the steering committee has the report Daniel prepared: incurred variance 70 million units favorable, under budget, and the cash gap from the month-20 picture, peaking at about 400 million in month twenty-five, the phased-opening month, with about 120 million unfunded even after the reserve band is fully released. Marta asks the question the chapter has been building toward: “Explain to the council how a project can be under budget and still face a cash crisis. Then tell me what you want to do about it.” The instruments on the table: re-profile the appropriation tranches within the same total, moving about 150 million from months twenty-six and twenty-eight to months twenty-one and twenty-two; release the reserve band, 70 million, by re-authorization; draw a bridge facility, about 120 million at 7 percent for an average of six months, about 4.2 million in interest; renegotiate the eastern contractor’s payment terms, staged mobilization and more frequent certifications, slowing the cash spike by about a quarter; or some combination. Choose the combination, defend the trade-off, name the evidence that would change your answer, and say what the committee should refuse to do.

The explanation is the chapter’s whole argument, in four sentences. First, the ledgers measure different things: incurred records work done and certified, cash records money paid, commitments record money obligated. Second, the favorable incurred variance is the delay wearing a compliment: the eastern works could not start until the drainage approval landed, so the month’s valuations ran below a plan that assumed they would run. Third, the cash gap is the acceleration wearing a warning: to beat the wet season, the eastern package had to be committed early, and the invoices from that commitment land before the funding staircase, sized to the older plan, arrives. Fourth, the two stories are connected by payment terms, the sixty-day lag that makes the cash curve follow the commitment curve at a distance, and the retention that holds money back until acceptance. A project can be under budget on the ledger that measures progress and out of cash on the ledger that measures survival, because the two ledgers answer different questions, and the favorite-number error is reporting one answer as the whole truth. The defensible decision starts with the re-profiled draw: it adds no money, moves the tranches within the total the council already approved, and it is a governance decision under chapter 8, re-authorization of the timing, made with the evidence on the table. The second move is the staged payment terms, because the cash spike is exactly one quarter long and the contractor’s package can carry the rhythm, negotiated as part of the eastern commitment, not after it. The reserve band is opened only if the 80-percent evidence requires it, because the corridor’s forecast is still inside its limit and the reserve is for the tail, not for the timing. The bridge facility is the residual, priced, 4.2 million, carried only for what re-profiling and payment terms cannot cover. The refusals are the chapter’s failure patterns in drill form: do not draw the remaining contingency, 80 million, against the cash gap, because the contingency is for named risks and the named risks have not retired; do not reclassify the eastern commitment or hold its certification, that is the invoice waltzer and the reconciliation would catch it within the month; and do not freeze the culvert works, because the wet season closes the window and converts a cash problem into the schedule problem chapter 17 already priced. A defensible alternative is the bridge-first combination if the council cannot re-profile quickly, and credit belongs to any answer that names the ledgers, prices the moves, and keeps the pockets intact.

Six. The transfer question. Which ledger does your project’s status report actually show, and which three does it hide? Where does your commitment ledger live, and who reconciles it monthly? What is your cash curve, and what is your funding staircase, and who owns the gap between them? If the answer to any of those questions is “nobody,” you have just found the chapter’s minimum viable instrument for your project, and the cost of building it is one line of arithmetic a month.

The durable principle: a project has four ledgers, commitments, accruals, actuals, and forecast, and they tell four stories that only reconcile into one, and mastery is reading all four before deciding, because the favorite number is a story and the reconciliation is the project. The pockets are the discipline that keeps the story honest: contingency for the risks you can name, reserve for the tail you cannot, funding limit for what the organization has actually promised, and the cash curve for when the money must actually be there. The most common next failure is quieter than the four characters: the ledgers get reconciled, the pockets get built, the cash curve gets drawn, and then the cadence decays, the monthly reconciliation becomes quarterly, the gap gets a name and loses its owner, and the project returns to the favorite number. The money must be carried into execution itself, the monthly reconciliation, the commitment review against the staircase, the forecast revised with the evidence, because the money machine is not finished when it is built; it is finished when the project’s financial decisions are made on it, every month, in the light. And the money machine has one more dimension that this chapter has only pointed at: the cash that must flow depends on the capacity that must exist, the eastern acceleration needs gangs and specialists the corridor has not yet confirmed, and the schedule’s four moves from chapter 17 and the money’s four instruments from this chapter both land in the resource plan, which is the next chapter’s craft.

Notes

• The composite cases remain author-created illustrative material. The BlueLine month-20 cost review, the four-ledger table, the forward cash curve, the funding staircase, the estimate-at-completion arithmetic, and all named characters are the author’s teaching constructions consistent with the facts established in earlier chapters: the 2,400 million-unit capital envelope, the 140 million-unit mitigation line, the operating subsidy of 90 million units a year, and the discount rate of 5 percent from chapter 7; the roughly 1,150 million units committed by month seventeen, the 1,390 million-unit forward base estimate, the 80-percent range of 1,280 to 1,500, the 110 million-unit contingency, and the 1,570 million-unit reserve ceiling from chapter 15; and the month-nineteen schedule decision, the resequenced integration and the phased opening of the central and northern segments at month twenty-five with the eastern segment later, from chapter 17. The month-20 ledger figures are teaching numbers, and they reconcile as stated: paid 1,010, plus unpaid accrued 180, plus committed-not-yet-incurred 140, equals committed 1,330; the forward cash curve sums to 1,540, which equals the estimate at completion of 2,550 minus the 1,010 already paid; the gap peaks at 400 in month twenty-five, cumulative cash 1,050 minus cumulative funding 650, and ends at 190 in month twenty-eight, leaving about 120 above the 1,420 available when the reserve band of 70 is fully released. The systems-package arithmetic is reproducible from the text: CPI is 45 over 48, 0.9375; SPI is 45 over 60, 0.75; the rate-based estimate is 120 over 0.9375, 128; the schedule-inclusive estimate is 48 plus 75 over 0.703, about 154.7; the bottom-up estimate is 48 plus 78, 126; the sensitivity run gives CPI about 1.08, SPI about 0.87, and schedule-inclusive about 120. The escalation arithmetic is 45 times 0.60 times 0.06, about 1.6 million units, and the bridge arithmetic is 120 times 0.07 times 0.5, about 4.2 million units. The Meridian decision-room figures, the 6 million-unit monthly certification, the 18 million-unit peak gap, and the 170,000-unit bridging cost, are author-created teaching numbers consistent with the grant of 250,000 units per clinic-month and the month-thirty-six window from chapters 1 and 8; 18 times 0.08 times 6 over 52 is about 0.17 million units. The KijaniPay budget of 210 million units and the 5 million-unit change threshold are from chapter 16; the lending-pilot capital treatment is the author’s synthesis of the budget-to-scope seam identified there.
• The terminology of commitments, accruals, actuals, and forecasts follows general accounting and project-cost practice, and the distinction between expense and balance-sheet cash requirements is standard financial treatment; see Richard A. Brealey, Stewart C. Myers, and Franklin Allen, Principles of Corporate Finance, 13th edition (McGraw-Hill, 2020), for the corporate-finance treatment of sunk costs, opportunity costs, working capital, and the time value of money, the same source cited in chapter 7. The earned-value vocabulary, planned value, earned value, actual cost, cost performance index, schedule performance index, and the estimate-at-completion formulas, follows the standard earned-value practice codified in ISO 21508:2018, Earned value management in project and programme management, and this chapter presents the formulas as standard formulations whose choice is an assumption statement, not a prediction; chapter 38 builds the full earned-value system. The general project-management treatment of cost, funding, and financial management follows ISO 21502:2020, Project, programme and portfolio management, Guidance on project management, and the related vocabulary of the PMBOK Guide, Eighth Edition (Project Management Institute, November 2025), which treats cost, like schedule, inside its performance domains, per the book’s reference baseline of 1 August 2026; this book describes the ideas in its own words and remains independent of PMI and the standards bodies. The reserve-band and rolling-authorization structure follows the chapter 8 discipline of tolerances, gates, and re-authorization, and the two-pocket contingency and management-reserve structure follows chapter 15. The escalation-clause example, the tax and currency treatment, and the restricted-funding and reimbursement patterns are described from general practice without reproducing any contract, statute, or donor framework. No proprietary certification manual, commercial text, or framework guide is reproduced or paraphrased here.