Specimen paper

Halvington Transport's staff cost schedule is built in thousands of pounds, so overheads appear as 850. The group summary sheet works in whole pounds and adds the 850 straight into figures like 1,240,000, understating group costs. Which discipline would have prevented this fault?

• A

  Apply currency formatting with a £000s suffix to every cell, so the scale of each figure is visible on its face.

  This confuses what a cell displays with what it stores. A custom format changes the appearance of the cell while the value handed to the summing formula is still 850, so the consolidation stays wrong by a factor of a thousand. It would be the answer only if number formatting altered the value passed into a calculation, which it never does; formatting can support a units convention but cannot enforce one.

• B

  Label the units on every sheet and block heading, and convert once at a defined point before figures are combined.

  Correct: units are declared in writing, held constant within each calculation area, and changed at a single visible step, which is the only control that acts on the stored value rather than on its appearance.

• C

  Round every schedule to the nearest thousand, so both sheets are carrying the same scale before consolidation.

  This mistakes precision for scale. Rounding leaves 850 as 850 and 1,240,000 as 1,240,000, so the thousand-fold mismatch survives the operation entirely. It would be right only if the two sheets were denominated in the same unit and differed merely in the number of decimals they carried.

• D

  Show two decimal places throughout, so a figure such as 850.00 stands apart from a whole pound total of 1,240,000.00.

  The same confusion in a weaker form: it makes the figures longer and relies on a reviewer noticing the oddity. That treats the fault as a failure to spot rather than a failure to define, when 850 is a perfectly plausible entry in a cost row. It would be right only if human inspection were an acceptable substitute for a stated units convention.

Meltham Utilities prices electricity in usage bands: 0, 500, 1,000 and 2,000 kWh thresholds each map to a tariff. The old model used =VLOOKUP(usage,bands,2,TRUE) on the sorted threshold column. An analyst modernises it to =XLOOKUP(usage,thresholds,tariffs) with no further arguments. A customer using 1,340 kWh now shows an error. Which diagnosis is correct?

• A

  XLOOKUP matches exactly by default, so 1,340 finds no threshold; the match mode must be set to next smaller item.

  Correct: 1,340 sits between the 1,000 and 2,000 thresholds, so an exact-match default returns #N/A until the match mode argument is set to next smaller item.

• B

  XLOOKUP inherits the approximate matching of the formula it replaced, so the fault is an unsorted threshold column.

  Two errors are stacked here. The first is assuming that a replacement function inherits the defaults of the one it replaced, which is the exact misconception this migration exposes. The second is then blaming sort order, a requirement of approximate matching rather than a cause of this error, when the stem tells you the threshold column was sorted. It would be right only if XLOOKUP defaulted to an approximate mode and someone had shuffled the thresholds.

• C

  XLOOKUP requires its lookup and return arrays to be the same size, and the tariff list is one row short of the thresholds.

  This transplants a genuine XLOOKUP rule onto a case that does not have it. A size mismatch raises #VALUE! rather than the failure described, and the stem pairs each of the four thresholds with a tariff. It would be right only if the return range had been truncated, and you would then see a different error code.

• D

  XLOOKUP supports exact matching only, so banded tariffs have to stay on VLOOKUP with its final argument set to TRUE.

  This misremembers the function's capability rather than its default. XLOOKUP offers next smaller item, next larger item and wildcard modes, so banded pricing does not have to remain on VLOOKUP. It would be right only if the match mode argument did not exist, in which case the modernisation would indeed have to be reversed.

Ledbury Chemicals is reviewing its model for sources of circularity before handover. Which of the following drivers creates a circular reference?

• A

  Interest income calculated on average cash balances

  Correct: the income changes profit, profit changes cash generated, cash changes the closing balance, and the closing balance feeds the average on which the income was struck, closing the loop.

• B

  Interest expense calculated on the opening debt balance

  The opening balance is inherited from the prior period and is fixed before anything in the current period is calculated, so no output of the period can feed back into it. Candidates choose this by reasoning that interest is inherently circular without asking which balance the rate is applied to. It would be circular only if the rate were applied to the closing or the average balance.

• C

  Receivables calculated from debtor days applied to revenue

  This is a one-way driver: revenue feeds the receivables balance, which feeds the balance sheet and the working capital movement, and nothing returns to revenue. The error is treating a driver that touches several statements as one that feeds itself. It would close a loop only if revenue depended on the receivables balance, for instance if the model throttled sales when the debtor book grew.

• D

  Depreciation calculated on the opening net book value

  Opening net book value is a fixed inherited figure, so the charge depends on a number the period cannot alter. The misconception is that any calculation appearing in both the income statement and the balance sheet must be circular. It would be circular only if depreciation were struck on the closing net book value, which is itself arrived at after depreciation.

Hartwell Foods forecasts cost of goods sold of £3,650,000 and revenue of £5,110,000 for the year. Inventory is modelled at 60 days of cost of goods sold, and opening inventory is £540,000. What is the cash effect of the inventory movement for the year?

• A

  An outflow of £60,000

  Correct: closing inventory is £3,650,000 divided by 365 and multiplied by 60, or £600,000, and the £60,000 build absorbs cash.

• B

  An outflow of £300,000

  This applies the 60 days to revenue, giving £5,110,000 divided by 365 times 60, or £840,000, and a movement of £300,000. The error is using a selling-price base for a balance carried at cost, which embeds the gross margin in the stock figure. It would be right only if inventory were held at selling prices, which no accounting basis permits; revenue is the correct base for receivables, not for stock.

• C

  An outflow of £68,333

  This uses a 360-day divisor, so daily cost of sales is overstated and closing inventory comes out at £608,333. The error is importing a convention from elsewhere rather than reading the one the model uses. It would be right only if the model's stated day count were 360, and a days assumption should always name its divisor so the reader is not left to guess.

• D

  An inflow of £60,000

  The arithmetic is right and the sign is reversed. This comes from reading a rising asset as a source of cash, or from applying the balance sheet movement in the direction it would take for a liability. It would be right only if inventory had fallen from £600,000 to £540,000, since releasing stock generates cash while building it consumes cash.

Tarn Logistics sells a delivery vehicle with a net book value of £60,000 for £75,000 in cash, recognising a £15,000 gain on the income statement. How should this disposal appear in the indirect cash flow statement?

• A

  Add the £15,000 gain back in operating and show £75,000 in investing

  The investing figure is right and the operating adjustment points the wrong way. Candidates reach this by generalising from depreciation, where non-cash items are added back, without noticing that the reversal must take the opposite sign to the income statement entry. Adding rather than deducting overstates cash by £30,000. It would be right only if the disposal had produced a £15,000 loss.

• B

  Deduct the £15,000 gain in operating and show the £60,000 book value in investing

  The operating adjustment is right and the investing figure substitutes net book value for proceeds. This comes from thinking of the asset as leaving the business at its carrying amount, when the cash flow statement reports what the bank actually received. It would be right only if the vehicle had sold for exactly its book value, in which case there would be no gain to reverse at all.

• C

  Deduct the £15,000 gain in operating and show £75,000 in investing

  Correct: the £75,000 received is the investing inflow, and the gain already sitting inside net profit is removed in operating so the same money is not counted twice.

• D

  Make no adjustment in operating and show the £75,000 in investing

  This leaves the gain inside profit at the top of the statement and adds the full proceeds below, so £15,000 is counted twice and cash is overstated. The misconception is that a gain on disposal is itself an operating cash inflow. It would be right only if the gain had never been recognised in the income statement, for example if proceeds equalled book value.

Bramley Kitchens Ltd reported revenue of £4.00m three years ago. Annual growth in the three years since was 44%, 20% and 0%, taking revenue to £6.912m. For a covenant schedule the analyst must state the compound annual growth rate over the three-year period. What is it?

• A

  20.0 per cent

  Correct: 6.912 divided by 4.00 is 1.728, and the cube root of 1.728 is 1.20, so the constant rate that links the two endpoints over three periods is 20.0 per cent.

• B

  21.3 per cent

  This is the arithmetic mean of 44, 20 and 0. Averaging period rates ignores that each year's growth compounds on a different base, and the arithmetic mean of a set of growth rates always exceeds the compound rate unless every rate is identical. It would be right only if the question asked for the average annual growth rate, which is a different and much less useful statistic for a covenant schedule.

• C

  24.3 per cent

  This divides total growth of 72.8 per cent by three, annualising by simple division and dropping compounding entirely. It would be right only if revenue had grown by a constant amount each year rather than at a constant rate, that is if the base for the percentage never changed.

• D

  14.6 per cent

  The method is right and the exponent is wrong: this takes the fourth root because four revenue figures are involved. The exponent is the number of intervals, not the number of observations, and three growth years separate the two endpoints given. It would be right only if four full years lay between the opening and closing revenue.

Fernbeck Distribution operates two depots, each able to process up to 100,000 orders a year at an annual running cost of £240,000 per depot. It handled 180,000 orders this year at a total depot cost of £480,000. Next year's forecast is 250,000 orders. What depot cost should the forecast carry?

• A

  £667,000, scaling this year's £480,000 depot cost in proportion to the order increase

  This treats a stepped cost as purely variable, deriving a cost per order from this year's total and extrapolating it. The unit cost of £2.67 is an artefact of running two depots at 90 per cent of capacity and says nothing about the cost of the next 70,000 orders. It would be right only if capacity were bought in arbitrarily small increments, for instance if fulfilment were outsourced at a price per parcel.

• B

  £600,000, charging 2.5 depots at £240,000 for 250,000 orders at 100,000 each

  This recognises that capacity comes in units and then forgets that you cannot open half a depot, omitting the round-up a step function requires. It is the near miss that shows a candidate had the right model of the cost and stopped one operation short. It would be right only if a half depot could be run at half the annual cost, which the stem's description of a whole site rules out.

• C

  £480,000, holding depot cost flat because it is fixed within the existing estate

  Depot cost is indeed fixed, but only inside its capacity band, and this answer applies the label without testing the constraint. It would be right only if forecast volume stayed at or below the 200,000 orders the two existing depots can process.

• D

  £720,000, opening a third depot because volume breaches the 200,000-order capacity

  Correct: 250,000 orders exceed the 200,000 the existing estate can handle, so a third depot is required and the cost is three times £240,000.

A junior analyst valuing Saltmarsh Brewing multiplies the company's EBITDA of £60m by the peer group's median P/E of 15x and reports £900m as the equity value. What is the fundamental flaw in this calculation?

• A

  The £900m is an enterprise value, so deducting net debt from it produces the correct equity value.

  This is tempting because the bridge from enterprise value to equity is the right procedure, applied in the wrong place. The £900m was not produced by an enterprise value multiple, so relabelling it and running a valid bridge on an invalid input simply produces a second wrong number. It would be right only if the 15x had been a peer EV/EBITDA multiple.

• B

  EBITDA must first be reduced by depreciation, since P/E rests on a post-depreciation earnings figure.

  This is a partial repair that stops one step short. Deducting depreciation converts EBITDA to EBIT, which is still struck before interest and tax and therefore still belongs to debt and equity together, so the mismatch survives. It would be right only if P/E were defined on operating profit, whereas it requires earnings after interest and tax.

• C

  P/E pairs equity value with post-interest, post-tax earnings, so EBITDA is the wrong denominator.

  Correct: the multiple and the earnings measure belong to different claimholders, so their product measures nothing at all.

• D

  The peer median is the wrong statistic here, and a size-adjusted mean multiple should be used instead.

  This argues about which statistic to draw from the peer set while accepting the mismatched pairing underneath it, and it comes from assuming that peer selection is always where comparables work goes wrong. It would be right only if the multiple and the earnings measure already agreed, and even then the median is usually preferred precisely because it resists the outliers a mean would absorb.

Dunraven Minerals is appraising a quarry with cash flows of -£100m today, +£250m at the end of year 1 and -£156m of restoration costs at the end of year 2. The spreadsheet IRR function returns 20% with its default guess, while a colleague using a guess of 0.35 obtains 30%. The WACC is 12%. How should the team interpret these results?

• A

  The 20% figure is the true IRR, since the default guess converges on the economically valid root.

  This grants authority to a numerical default. The guess is only the starting point of an iterative search and carries no economic meaning, so the function reports whichever root it happens to reach first. It would be right only if the cash flows had a single sign change and therefore a single root, in which case any starting guess would converge on the same answer.

• B

  Take 25%, the mean of the two roots, as the effective return and compare it with the 12% WACC.

  This treats the two figures as competing estimates of one underlying quantity, the way you might average two measurements. An IRR is a root of a polynomial, and the mean of two roots is not itself a root: discounting these flows at 25 per cent does not give an NPV of zero, so the figure describes nothing in the model. Averaging is defensible for estimates of a single true value and never for distinct solutions of an equation.

• C

  Both 20% and 30% solve the equation, so the decision should rest on the NPV at the 12% WACC.

  Correct: with two roots the rate is ambiguous, so the decision moves to NPV at the firm's actual cost of capital, which here is about minus £1.1m and points to rejection.

• D

  Accept the quarry, since both roots comfortably exceed the 12% cost of capital on either reading.

  This is the most dangerous option because it reaches a decision, and the wrong one. The rule that an IRR above the hurdle means value creation depends on a single crossing of zero; with a non-conventional pattern the NPV curve turns back, and here NPV at 12 per cent is negative. It would be right only if the flows were conventional, with one sign change and a single IRR above the cost of capital.

A newly qualified analyst at Fenwick & Doyle, a transport group, is deciding which parts of a modelling assignment to hand to an AI assistant. Which task is the most appropriate one to delegate to the assistant?

• A

  Selecting the discount rate for the impairment review of the haulage fleet, from published sector data

  This sounds mechanical because it involves gathering published figures, but the rate drives the impairment conclusion and is a judgement the accountable analyst must make and evidence. The error is mistaking a technical-sounding input for a clerical one. It would become an acceptable delegation only if the task were reframed as assembling the published evidence for you to weigh, with the selection and its justification remaining yours.

• B

  Drafting first-pass documentation of an inherited schedule's calculation flow, for the analyst to correct

  Correct: the work is laborious, carries no accounting judgement, and can be checked line by line against the live model, which is exactly the profile of a task worth accelerating.

• C

  Writing the fleet utilisation commentary that goes straight into the board pack without further review

  This confuses permission to draft with permission to publish. Drafting commentary is a reasonable use of an assistant; removing the review step is what makes it unsafe, because the analyst signs the pack and inherits every unchecked assertion in it. It would be acceptable only if the words were reviewed and adopted by the person accountable for them, which the option explicitly rules out.

• D

  Deciding whether the group should retain or dispose of its loss-making parcels subsidiary this year

  This hands over a management decision rather than a modelling task, and it is chosen by candidates who read the assistant's fluency about strategy as competence in it. It would be a legitimate request only if reframed as setting out the considerations and the evidence on each side, leaving the decision with management.

An analyst at Silverholt Utilities wants to send a customer-level receivables extract, including domestic customers' names, addresses and balances, to an external AI service to help design an ageing analysis. Which consideration governs this plan?

• A

  The extract is personal data, so minimise it: design on anonymised records, send identifiers only by approved tools.

  Correct: an ageing analysis needs invoice dates and amounts rather than identities, so the design work can be done without the personal data ever leaving the organisation.

• B

  Uploading transfers control to the provider, whose processing terms then carry the organisation's duties.

  This rests on the belief that your obligations travel with the file. A supplier processing data on your behalf takes on duties of its own, but yours do not lapse, and the act of disclosure remains yours. It would be right only if the provider had become a controller in its own right, which an engaged processor does not, and even then your decision to disclose would still need a lawful basis.

• C

  Balances are financial records rather than personal data, so approval from line management is sufficient.

  This draws a false boundary between financial and personal information. The categories overlap, and identifiability is what triggers the duties, so a ledger keyed by a named domestic customer at a home address is plainly personal data. It would be right only if the extract had already been stripped of identifiers, in which case ordinary line management approval would be the appropriate control.

• D

  Customer consent cures the disclosure, so obtain it and then send the full extract to the AI service.

  This treats consent as a universal permission slip. Consent is one lawful basis for processing; it does not displace minimisation, and it cannot authorise a route the organisation's own tooling policy does not permit. Even with consent in hand you would still be sending far more data than designing an ageing analysis requires.

A monthly cash flow model at Kestrelbrook Capital carries a loan balance of £500,000. The analyst prompts an AI assistant: 'Interest accrues at 9% per annum on the outstanding balance; the model is monthly.' The assistant returns '=D8*9%' for each month, explaining that it 'applies the stated rate to the balance', which charges £45,000 of interest every month. Which review comment identifies the defect?

• A

  The base is wrong rather than the rate: each month should charge 9% on the opening balance, not the closing one

  This raises a real convention question, and on another schedule it might be a fair review point, but it leaves the charge twelve times too large. It is chosen by candidates who reach for the familiar debt schedule fault instead of testing the size of the number produced. It would be the right comment only if the periodicity had already been fixed and the sole remaining issue were which balance the monthly rate is applied to.

• B

  An annual rate has been used for monthly periods: each month needs 9%/12, or the compound monthly equivalent

  Correct: 9 per cent on £500,000 is roughly £45,000 for a full year, so charging that amount in every month overstates interest about twelvefold.

• C

  The timing is wrong rather than the rate: the £45,000 annual charge belongs in December of each forecast year

  This makes the right observation, that £45,000 is an annual figure, and then repairs it by abandoning monthly accrual altogether. In a model built to show cash by month it misstates eleven months as nil and one month twelvefold, distorting exactly the profile the model exists to reveal. It would be right only if the question concerned when interest is paid, and even then the accrual would still be earned monthly while the payment fell in one month.

• D

  The rate belongs in an input cell rather than inside the formula, so the charge flexes with the assumption

  This cites a genuine modelling standard that the drafted formula does breach, but it corrects the housekeeping and leaves the number wrong; moving 9 per cent into a cell simply produces the same twelvefold overstatement from a tidier source. It is what a reviewer says when reading the form of a formula rather than testing its output. It would be the right answer to a question about model standards, not to one about the defect in the charge.