Forecasting and Investment Appraisal
~50 min · WBS13 · 3.3.3
WBS13 · 3.3.3 · 50 min
A doesn't predict the future — it strips the noise out of the past so the underlying trend becomes visible. does something similar to money itself: it strips out the illusion that a pound arriving in three years is worth what a pound in your hand is worth today.
Key terms in this lesson
Before you read on
Two or three questions on exactly what this lesson teaches. Being wrong here is fine — it's the fastest way to find out what to pay attention to next.
Reading a trend through the noise
A firm's raw sales figures bounce around from one period to the next for reasons that have nothing to do with the direction the business is actually heading — an unusually wet month, a one-off promotional push, a competitor's stock-out. Spec point 3.3.3.1 names three techniques built to separate that noise from the genuine underlying direction, and to use it to look forward: the , which smooths a run of past figures into a single trend line; the calculation of — how far a real figure typically sits above or below that trend line in a given period, averaged across every matching period available, so the gap itself becomes a number that can be added back on to a forecast rather than left as a vague sense that 'this quarter is usually quieter'; and the scatter graph with its , which finds the straight-line relationship between two variables and lets a firm read values off it — including, cautiously, values beyond the data it was actually drawn from.
Two versions of the moving average appear on this paper. A 3-year (or 3-period) moving average is the simpler case: average three consecutive periods, and because three is an odd number the resulting average lands exactly on the middle period — nothing further is needed to line it up with a real point in time. A 4-quarter moving average is the version examiners name specifically, because four is an even number: averaging four consecutive quarters produces a value that falls at the midpoint between two quarters, not on any single one. Fixing that — centring — is a genuine extra step, not a footnote, and it's exactly where marks are commonly lost on this technique (see the chain-drill below).
A scatter graph plots two paired variables — say, monthly advertising spend against monthly sales revenue — as individual points, with no line joining them. The is a single straight line drawn through the scatter (by eye, on this paper — no formula is required), positioned so it passes as close as possible to as many points as possible, with roughly as many points above it as below. Reading a value from a point already inside the range of the plotted data is interpolation; extending the line beyond the last real data point and reading a value there is — a genuine forecast, built entirely on the assumption that whatever relationship produced the line keeps holding outside the range it was actually observed in.
That's why the spec pairs all of this with 'limitations' as their own named sub-point (3.3.3.1c), and it isn't a throwaway line. A real mark scheme credited an answer for noting that 'the actual number of monthly premium subscribers was 2 million more than the forecast number' — a case where the projection held up well — while cautioning just as directly, on a different real extract, that 'the increase in pet ownership in the UK in 2020/21 in Extract E does not mean this will continue.' A moving average or a line of best fit describes precisely what already happened; it forecasts what happens next only on the assumption that nothing changes — and a strong Discuss or Assess answer states that assumption out loud rather than treating the projection as settled fact.
Why averaging shrinks the gap between what you see and what's really happening — before the trend-and-noise model
In plain terms
A firm's true, underlying performance across three years, if every random event could be stripped away, is a smooth, steady climb: Year 1 = 200, Year 2 = 210, Year 3 = 220 — exactly 10 a year, nothing more complicated than that. Now add in what actually happens in the real world: unpredictable one-off events layered on top of that steady climb. Year 1 gets a surprise rush order that adds +15. Year 2 loses a shipment to a supplier delay, subtracting −18. Year 3 has nothing special happen, just a small +4 blip. The figures actually reported — the ones a business owner would see on the page — are the smooth climb plus that year's one-off bump: Year 1 = 200+15 = 215, Year 2 = 210−18 = 192, Year 3 = 220+4 = 224. Look at just those three reported numbers — 215, 192, 224 — with nothing else to go on. They tell a story of a business that dropped sharply in Year 2 and then partly recovered in Year 3. That story is completely wrong: the real underlying performance never dropped at all — it climbed steadily the entire time, from 200 to 210 to 220. The reported figures hid that fact behind three unrelated one-off events. Now average the three reported figures together: (215+192+224)÷3 = 631÷3 = 210.33. Compare that to averaging the three TRUE, no-randomness figures: (200+210+220)÷3 = 210. The averaged reported figure, 210.33, lands almost exactly on the averaged true figure, 210 — off by only 0.33 — even though the individual one-off bumps that went into it were as large as +15 and −18. Add those bumps up and see why: 15−18+4 = +1, and ÷3 = 0.33 — they mostly cancelled each other out, because two of them pointed in opposite directions and were roughly the same size. Compare that 0.33 gap to what any single year told you on its own: Year 2's reported figure, 192, was off from ITS OWN true value, 210, by a full 18 — more than fifty times bigger an error than the 3-year average made. Nothing about averaging removed any individual bump — 15, −18 and 4 are still exactly what they were. What changed is that adding several bumps pointing in different directions together, then dividing, lands much closer to zero than any one of them was alone, while the three true, no-randomness figures (200, 210, 220) were already close to each other, so averaging them barely moves anything. That is the entire trick, done with nothing but addition and division.
In time-series forecasting, every part of that split already has a name. The real, no-randomness climb (200, 210, 220) is the trend. The unpredictable one-off addition each year (+15, −18, +4) is the shock, or noise — random, disconnected from one year to the next, and roughly as likely to be positive as negative. What's actually reported each year — trend plus shock — is the observed figure (215, 192, 224). Averaging several consecutive observed figures together is exactly what a moving average is. The averaged observed figure (210.33) landed so close to the averaged trend (210) for two separate reasons working at once: the shocks (+15, −18, +4) partly cancelled each other when summed, because they didn't all point the same way — and that cancellation tends to get more reliable, not less, the more independent shocks are added into the average — while the trend values (200, 210, 220) barely differ from each other in the first place, so averaging them changes almost nothing regardless of how many are included. Averaging doesn't know which part of any figure is trend and which is shock — it just adds and divides everything it's given — but because of how each of those two components behaves once summed, the result lands close to the trend and far from any single year's shock, purely as a side effect of the arithmetic.
Formally
Model any period's observed figure as X_t = Trend_t + Shock_t, where Shock_t is a random disturbance, independent from one period to the next and roughly as likely to run positive as negative, and Trend_t changes only slowly relative to the length of the averaging window. Averaging n consecutive observed figures gives (X₁+X₂+…+Xₙ)/n = (Trend₁+…+Trendₙ)/n + (Shock₁+…+Shockₙ)/n. Because Trend barely differs across the periods inside a short window, the first term is close to the trend value at the window's midpoint. Because the shocks are independent and roughly zero-centred, the second term tends to be small relative to any individual Shock_t, and shrinks further, on average, as more periods are averaged together. An operation that cannot itself distinguish trend from shock — averaging treats every number identically — ends up preserving one and damping the other, purely because of how each component behaves once several of them are added together and divided. The mechanism below sets out this same argument formally, for shocks of any size in any period, rather than checked once against one specific set of numbers the way it was above.
Mechanism
Why averaging cancels the noise, but not the trend
Model any single period's sales as two components added together: an underlying trend value that moves slowly and predictably from one period to the next, plus a random shock — the one-off weather event, the short promotional spike, the input shortage — that is, by definition, roughly as likely to be positive as negative and unconnected from one period to the next. Average three (or four) consecutive periods together and the trend component survives almost unchanged, because a genuinely slow-moving trend barely differs across three or four adjacent periods to begin with. The random shocks don't survive nearly as well: because they point in independent, unpredictable directions, some of the positive ones and some of the negative ones cancel each other out inside the sum, before it's divided down into an average. The averaging arithmetic has no idea which part of any figure is 'trend' and which is 'noise' — it can't tell them apart at all. It just adds up whatever is there and divides. The reason it still works is a property of the noise itself, not of the arithmetic: genuine random shocks partially cancel when summed together, and a genuine trend does not, so an operation that can't distinguish the two ends up preserving one and dampening the other purely as a side effect of what each component actually looks like.
Worked, in full
Deriving — and testing — the 3-year moving average
- 01
Before building an original example, it's worth anchoring the method in a real one Pearson has actually marked: a genuine mark scheme walks through exactly this calculation on UK pet-ownership percentages — (40% + 41% + 60%) ÷ 3 = 47%. Same method; the numbers below are original, computed independently so every step can be checked. Amberview Sportswear Ltd (VERIDIAN-original) reports annual sales of £210,000, £250,000, £225,000, £270,000, £260,000, £310,000 and £295,000 across Years 1–7.
Earns: K (Knowledge) — the method anchored in a confirmed real calculation before being run on original data.
- 02
3-year moving averages, one for each year with a full year on either side (Years 2–6): Year 2 = (£210k+£250k+£225k)÷3 = £228.33k; Year 3 = (£250k+£225k+£270k)÷3 = £248.33k; Year 4 = (£225k+£270k+£260k)÷3 = £251.67k; Year 5 = (£270k+£260k+£310k)÷3 = £280.00k; Year 6 = (£260k+£310k+£295k)÷3 = £288.33k.
Earns: An1 (Analysis, first step) — every value from the derived formula, not read off a smoothed-looking line drawn by eye.
- 03
Now test the mechanism's own prediction: the moving-average series should swing around less than the raw series did. Raw year-to-year changes: +£40k, −£25k, +£45k, −£10k, +£50k, −£15k — an average swing of about £30,800 per year, ignoring direction. Moving-average year-to-year changes: +£20.0k, +£3.3k, +£28.3k, +£8.3k — an average swing of about £15,000. The swing roughly halves, exactly as the mechanism predicts, purely from averaging three numbers at a time.
Earns: An2 (Analysis, second step) — the mechanism's prediction checked against the actual arithmetic, not just asserted.
- 04
The moving-average series also makes the direction unambiguous — a steady climb from £228.33k to £288.33k — in a way the raw figures alone don't (Year 6's raw £310,000 is followed by a Year 7 fall to £295,000, which could look like the trend has turned if read in isolation). But note the cost, visible in the same table: seven raw data points produced only five moving-average values — one period lost at each end, because the first and last years in the series don't have a full year on both sides to average with. That edge effect gets worse, not better, with a longer window like the 4-quarter version below.
Earns: Eval (Evaluation) — both the benefit (a clearer trend) and the genuine limitation (lost edge data) drawn from the same worked table, not stated as separate, unconnected facts.
Source — Mark scheme, Oct 2022
"(40% + 41% + 60%) ÷ 3 = 47%"
x-axis: Monthly advertising spend, £000 · y-axis: Monthly sales revenue, £000
- Scatter points
- Six months plotted as individual points: (2, 12), (4, 18), (6, 20), (8, 25), (10, 30), (12, 33) — no line joining them yet.
- Line of best fit
- A single straight line drawn through the scatter by eye, as close as possible to as many points as possible with roughly equal numbers above and below — here, close to y ≈ 8.4 + 2.09x. The exact equation isn't required on this paper; reading values off a well-drawn line is.
- Interpolated read, x = 8
- Reading a value at x = 8 — inside the range actually plotted — off the line gives roughly £25,000, matching the real data point closely: interpolation, the safer of the two kinds of read.
- Extrapolated read, x = 14
- Extending the same line beyond the last real point (x = 12) out to x = 14 and reading off roughly £37,600 is extrapolation — a genuine forecast, built entirely on the assumption that the same linear relationship keeps holding beyond the range it was actually observed in.
Common error: Treating an extrapolated read the same as an interpolated one — quoting the £37,600 figure with the same confidence as the £25,000 figure.
Correct: Naming the extrapolated figure explicitly as an assumption-dependent forecast, not a calculated fact — the same evaluative move the mark scheme rewards on the moving-average side of this topic.
In your own words
In one sentence: why does extending a line of best fit to a value of x well beyond the last plotted point carry more risk than reading a value of x that sits between two already-plotted points?
Complete it yourself
Complete the chain — 4-quarter centred moving average
- 01
Amberview Sportswear Ltd's quarterly sales (£000) for two years: Year 1 Q1–Q4 = 40, 65, 80, 45; Year 2 Q1–Q4 = 48, 72, 90, 50. The first 4-quarter moving total covers Year 1 Q1–Q4: 40+65+80+45 = 230, so the first (uncentred) 4-quarter moving average = 230÷4 = 57.5.
- 02
The second 4-quarter moving total shifts forward by one quarter, covering Year 1 Q2 to Year 2 Q1: 65+80+45+48 = 238, so the second (uncentred) 4-quarter moving average = 238÷4 = 59.5.
Worked, in full
Deriving the seasonal variation — and using it to forecast one specific quarter
- 01
A moving-average trend line, on its own, only ever says 'sales are rising' — it can't say how a specific upcoming quarter will compare to that general rise, because the trend has already averaged the quarters together. Spec point 3.3.3.1b names the fix: calculate each period's variation from the trend (actual − trend), average the variations for matching periods, and add that average back on top of an extrapolated trend value to forecast one specific future period. Extending Amberview Sportswear Ltd's quarterly sales one more year, Year 3 Q1–Q4 = £55k, £78k, £98k, £58k, and repeating the exact centring method from the chain-drill above across the full three years produces four more centred trend values on top of the two already found there: Year 2 Q3 = £65.875k, Year 2 Q4 = £67.5k, Year 3 Q1 = £69.25k, Year 3 Q2 = £71.25k.
Earns: K — the centring method from the chain-drill above extended to more data, and the new technique it feeds into named from the spec before any new arithmetic is run.
- 02
Eight quarters now have a centred trend value (Year 1 Q3 through Year 3 Q2). Variation = actual − trend for each: Y1Q3 = £80k−£58.5k = +£21.5k; Y1Q4 = £45k−£60.375k = −£15.375k; Y2Q1 = £48k−£62.5k = −£14.5k; Y2Q2 = £72k−£64.375k = +£7.625k; Y2Q3 = £90k−£65.875k = +£24.125k; Y2Q4 = £50k−£67.5k = −£17.5k; Y3Q1 = £55k−£69.25k = −£14.25k; Y3Q2 = £78k−£71.25k = +£6.75k.
Earns: An1 — every variation shown as its own subtraction against the matching trend value, not read off a chart by eye.
- 03
Grouped by quarter position rather than by year, each quarter now has two variation figures to average — Q1: (−£14.5k, −£14.25k) → −£14.375k; Q2: (+£7.625k, +£6.75k) → +£7.1875k; Q3: (+£21.5k, +£24.125k) → +£22.8125k; Q4: (−£15.375k, −£17.5k) → −£16.4375k. This is the exact same noise-cancelling logic as the mechanism block above: any one year's Q1 shortfall is partly a one-off shock, but averaging two years' worth of Q1 shortfalls together leaves a far more stable read of how far below trend Q1 structurally runs for a business like Amberview, whose sporting-goods sales genuinely do peak around Q3.
Earns: An2 — the averaging step performed across every available quarter, and connected explicitly back to why averaging cancels noise rather than left as an unexplained second calculation.
- 04
The eight trend values rise by an average of about £1,821 per quarter (the eight quarter-on-quarter differences: +£1,875, +£2,125, +£1,875, +£1,500, +£1,625, +£1,750, +£2,000, averaged). Extrapolating three quarters past the last known trend point (Year 3 Q2 = £71.25k) gives a Year 4 Q1 trend of £71.25k + 3×£1.821k ≈ £76.71k. Adding Q1's own average variation, −£14.375k, gives a Year 4 Q1 forecast of £76.71k − £14.375k ≈ £62.34k — a forecast for Q1 specifically, not just 'the trend is rising.' Skipping the variation step and quoting the extrapolated trend alone would overstate Amberview's structurally weakest quarter by roughly £14,400, exactly the kind of forecast a business would plan production and staffing around wrongly.
Earns: Eval — the extrapolated trend and the averaged variation combined into one forecast for a named period, with the cost of skipping the variation step stated as a number rather than asserted as an abstract limitation.
Investment appraisal: three lenses on the same decision
Spec point 3.3.3.2 names three ways to judge whether a specific investment — a new machine, a new site, a new product line — is worth the money: simple (how fast does the cash come back?), the , or ARR (how profitable is it, on average, each year?), and discounted cash flow, examined on this paper as or NPV only (what is the whole project worth in today's money?). All three start from the same two numbers — the initial investment and the forecast net cash inflow for each future year — and can disagree, sometimes sharply, about which project looks best, because each one is asking a genuinely different question of the same data. Running example for the rest of this section: Larkspur Furnishings Ltd (VERIDIAN-original) is considering spending £120,000 on new machinery, forecast to generate net cash inflows of £40,000, £45,000, £50,000 and £50,000 in Years 1 to 4.
Simple payback is the most direct of the three: it asks only how long the initial investment takes to be recovered in cash, ignoring everything that happens afterwards. It's a live calculation topic on this paper — an Airbus aircraft purchase appeared as a 4-mark Calculate question in one series, and Brompton Bikes' new-factory investment was assessed through payback logic worth 12 marks in another (this paper's Assess tariff is 12 marks, not the 10 marks Units 1 and 2 use — see the reference card). The strongest response to that Brompton question is instructive about when payback is actually the right tool to reach for: examiners credited it for linking the method to the business itself, noting that 'Brompton Bikes was in a very dynamic market with technology changing rapidly so therefore needed to use an investment appraisal method which focused on speed of return rather than profitability.'
Average rate of return expresses an investment's profitability as a single annual percentage: the total profit it generates over its whole life, averaged across the years, as a percentage of the amount originally invested. It's a live, real calculation question on this paper too — a real mark scheme states the formula precisely as ARR = average profit per year ÷ cost of initial investment × 100, then applies it to a real company, Bramwell Brown, buying £150,000 of new production machinery: six years of forecast "additional profit" totalling £287,550, an average of £47,925 a year, giving an ARR of 31.95%. Derived in full below, on this section's own running example.
Net present value asks a harder but more complete question than either of the other two: what is this entire investment worth, today, once every future pound is adjusted for the fact that a pound arriving in three years' time isn't worth the same as a pound in hand right now? That adjustment — discounting — is derived in full below, rather than handed over as a formula to memorise.
Mechanism
Why a future pound is worth less than a pound today
Suppose a reliable investment pays 10% a year. £100 placed there today is worth £100 × 1.10 = £110 in a year's time, and £100 × 1.10² = £121 in two years, because it earns a return the whole time it's invested. Now run that logic backwards: how much would need to be set aside today to have exactly £100 in a year, at that same 10%? Call it X: X × 1.10 = £100, so X = £100 ÷ 1.10 = £90.91. £90.91 today and £100 in a year are worth exactly the same amount, because £90.91 invested now grows into precisely £100 by the time the year is up. That's the entire mechanism behind discounting: a future sum is converted back to its present value by asking what smaller sum, invested today at the going rate of return, would grow into it by the time it actually arrives. The rate used isn't arbitrary either — it represents the of capital, the return the business is giving up by tying its money up in this specific project rather than the next-best thing it could have done with it (another investment, a bank deposit, paying down debt). The is just this division done once and reused for any sum: 1 ÷ (1 + r)ⁿ for a sum arriving n years from now at rate r. Multiply any future cash flow by it and the result is that cash flow's value in today's money — ready to be added up, on a genuinely like-for-like basis, with every other year's discounted figure, and compared directly against the cash spent today.
Worked, in full
Deriving the payback period from cumulative cash flow
- 01
Larkspur Furnishings Ltd's £120,000 machine, running cumulative net cash inflow one year at a time: Year 1 £40,000 → cumulative £40,000. Year 2 £45,000 → cumulative £85,000. Year 3 £50,000 → cumulative £135,000.
Earns: K — the payback test built as a running total, not looked up as a rule.
- 02
The £120,000 investment is crossed partway through Year 3: cumulative cash rises from £85,000 (still £35,000 short) to £135,000 (£15,000 past it) across that single year. It is neither exactly recovered at the end of Year 2 (£85,000 < £120,000) nor does it need the whole of Year 3's inflow — the crossing happens somewhere inside it.
Earns: An1 — the exact year identified from the cumulative total's own behaviour, not estimated by eye.
- 03
Find exactly where inside Year 3: the amount still needed at the start of that year is £120,000 − £85,000 = £35,000, out of the £50,000 Year 3 is forecast to bring in. £35,000 ÷ £50,000 = 0.7 of the year, and 0.7 × 12 = 8.4 months.
Earns: An2 — the fraction of the year converted into months by the same division logic throughout, not a separate rule to memorise.
- 04
Payback period = 2 full years + 8.4 months ≈ 2 years and 8 months. The precision-and-format trap applies here just as it does to a percentage ratio: an answer of '2.7' with no unit stated, or '8.4 years' from forgetting to convert the fraction into months, both fail even with a correct method. This paper's confirmed calculation-error pattern is specifically about not giving an answer to the precision or form a question asks for (3 of the 6 series mined this pass show marks lost this way) — a time-based figure like payback carries exactly the same risk, even though the mined examples of this trap are all percentage ratios, not payback specifically.
Earns: Eval — the numeric result connected explicitly to a confirmed, real calculation-error pattern from this paper, not left as an isolated answer.
Worked, in full
From payback to ARR: a second lens, anchored in a real mark scheme
- 01
Before running the method on Larkspur, anchor it in a real one Pearson has actually marked: a genuine mark scheme (Jan 2022, Q1a) states the formula precisely — ARR = average profit per year ÷ cost of initial investment × 100 — and applies it to a real company, Bramwell Brown, which spent £150,000 on new production machinery forecast to generate 'additional profit' of £22,500, £29,250, £38,250, £49,500, £64,350 and £83,700 across Years 1–6, a total of £287,550: average profit = £287,550 ÷ 6 = £47,925; ARR = £47,925 ÷ £150,000 × 100 = 31.95%, exactly the mark scheme's own confirmed answer. Same method; the numbers that follow return to Larkspur Furnishings Ltd, the running example already used above for payback.
Earns: K — the formula anchored in a confirmed real calculation before being run on the section's own original data, the same discipline the moving-average chain above used.
- 02
Bramwell Brown's own extract handed candidates annual PROFIT figures directly — real accounting profit, already net of costs like depreciation, needing no further adjustment. Larkspur's figures above are net cash inflow instead (the same figures already used for payback), so ARR's standard simplification applies first: total profit = total net cash inflow − initial investment. Larkspur: £40,000+£45,000+£50,000+£50,000 = £185,000 total net cash inflow; £185,000 − £120,000 = £65,000 total profit.
Earns: An1 — the genuine difference between a question that hands over profit directly and one that hands over cash inflow named explicitly, not glossed over as if the two inputs were interchangeable.
- 03
Average annual profit = £65,000 ÷ 4 = £16,250. ARR = £16,250 ÷ £120,000 × 100 = 13.54% (to two decimal places — the precision the real mark scheme above itself specifies). Set beside the other lens already calculated on this exact £120,000 investment — payback ≈ 2 years 8 months — the same underlying cash flows already have two different-shaped answers, because each method is built to answer a different question, not to approximate the same one.
Earns: An2 — the new figure connected explicitly back to the lens already derived on identical data, not left as an isolated calculation.
- 04
The real Jan 2022 examiner report on this exact question records the actual failure modes: many candidates 'did not know the formula for ARR so could only score 1 mark for correct placement of £150,000 as the denominator,' some 'confused ARR with simple payback and gave a response in years and months,' and a small number 'correctly calculated the ARR but could only be awarded 3 marks for omitting the % sign.' Every one of those three failure modes is avoidable with what's already been derived above: the formula stated before it's applied, ARR's percentage answer kept visibly distinct from payback's years-and-months answer, and the % sign carried through to the final figure.
Earns: Eval — the confirmed, real error pattern on this exact question connected back to the specific steps in the derivation that prevent each one, not left as generic exam advice.
Source — Mark scheme, Jan 2022
"ARR = average profit per year ÷ cost of initial investment × 100; £47 925 ÷ £150 000 × 100 = 31.95%"
Worked, in full
From ARR to NPV: discounting the same cash flows
- 01
Apply the discount factor derived above, 1 ÷ (1+r)ⁿ, at a 10% discount rate to each of Larkspur's four years, rounded to 3 decimal places (the form usually given in a discount factor table): Year 1 = 0.909, Year 2 = 0.826, Year 3 = 0.751, Year 4 = 0.683.
Earns: K — the discount factors computed from the formula, not pulled from an unexplained table.
- 02
Multiply each year's cash inflow by that year's discount factor: Year 1: £40,000 × 0.909 = £36,360. Year 2: £45,000 × 0.826 = £37,170. Year 3: £50,000 × 0.751 = £37,550. Year 4: £50,000 × 0.683 = £34,150.
Earns: An1 — every present value shown as its own multiplication, checkable line by line.
- 03
Total present value = £36,360 + £37,170 + £37,550 + £34,150 = £145,230. Net present value = total present value − initial investment = £145,230 − £120,000 = +£25,230.
Earns: An2 — NPV built explicitly as (sum of discounted inflows) minus (the one cash flow that needs no discounting, because it happens today), not treated as a single opaque formula.
- 04
The decision rule follows directly from what the number means: a positive NPV means the project returns more, in today's money, than the £120,000 it costs today — after already accounting for what that £120,000 could have earned elsewhere at the 10% rate. +£25,230 says accept. Compare this to the payback answer above (2 years 8 months): payback only ever told Larkspur when the cash came back, never how much the whole four-year project is actually worth once timing is priced in — a genuinely different question, not a rougher version of the same one.
Earns: Eval — the NPV decision rule derived from what the arithmetic represents, and explicitly distinguished from what payback answers, rather than the two methods left to look interchangeable.
In your own words
In one sentence: why can two investments have exactly the same total, undiscounted cash inflow and exactly the same ARR, and yet have very different net present values?
Complete it yourself
Complete the chain — same total cash, same ARR, different NPV
- 01
A manufacturer is choosing between two machines, both costing £150,000. Machine A's forecast net cash inflows are £80,000, £60,000, £40,000, £20,000 (Years 1–4) — front-loaded. Machine B's are £20,000, £40,000, £60,000, £80,000 (Years 1–4) — back-loaded. Both total £200,000 over the four years.
Named traps
- forecast-treated-as-fact-not-assumption
- Pearson's own indicative content for evaluating quantitative sales forecasting is built to reward a balanced answer, not a one-sided one. A real mark scheme praises the technique for being reliable — 'numerical data such as time-series analysis is easy to interpret and analyse… data can be objectively interpreted and bias is often not an issue in comparison to qualitative techniques' — while, on a different real extract, cautioning in the same breath that 'the forecast of the pet care market growing to £7bn… may not occur.' Presenting a moving-average trend or an extrapolated line of best fit as a guaranteed outcome, rather than the mark scheme's own both-sides balance, is the one-sided-Discuss trap this topic is built to catch.
- payback-ignores-time-value-and-post-payback-return
- Confirmed directly in a real examiner report on the Brompton Bikes payback question: the standard counter-argument to using payback is that it 'ignored the time value of money or the overall profitability of the investment.' That's two separate weaknesses in one sentence, and a strong answer names both — payback treats every pound the same regardless of when it arrives (no discounting), and it stops looking at the project entirely the moment the cash is recovered, even if the most profitable years are still to come.
- method-choice-not-linked-to-the-business
- The same confirmed examiner report rewarded an answer specifically for connecting the choice of appraisal method to the nature of the business, not just calculating a number: 'Brompton Bikes was in a very dynamic market with technology changing rapidly so therefore needed to use an investment appraisal method which focused on speed of return rather than profitability.' A calculation with no argument for why THAT method suits THIS business is doing only half the work an Assess or Evaluate question on this topic actually asks for.
- arr-formula-unknown-or-confused-with-payback
- Confirmed directly in the real Jan 2022 examiner report on this exact ARR question: many candidates 'did not know the formula for ARR so could only score 1 mark for correct placement of £150,000 as the denominator,' and some 'confused ARR with simple payback and gave a response in years and months.' The two answers look nothing alike once the formula is actually known — ARR is always a percentage of the original investment, payback is always a length of time — but a candidate who hasn't memorised the formula has nothing to stop the two blurring together under exam pressure, which is exactly what this examiner report caught happening in real scripts.
- percent-and-times-100-omission
- Confirmed across this paper's mark schemes: omitting the % sign on a percentage-ratio answer caps the mark at one below full, and omitting ×100 from the formula loses the knowledge mark even if the final figure comes out right — both confirmed in more than one of the six series mined this pass. ARR is a percentage figure; this applies to it exactly as it applies to gearing, ROCE or any other ratio on this paper.
- decimal-places-and-units-not-given-as-specified
- Confirmed across three of the six series mined this pass: not giving an answer to the number of decimal places a question specifies loses marks even when the underlying figure is correct. That confirmed pattern is about decimal places specifically — the mined examiner reports don't contain a payback-specific example — but the same discipline applies by direct extension on this topic: stating payback in the unit actually asked for (years and months, not a bare decimal), and giving ARR and NPV to the precision the question specifies rather than however many digits a calculator happens to display.
- a-shown-wrong-working-beats-a-blank-answer
- Repeated as explicit advice across three of the six series mined this pass: marks can still be awarded even with an incorrect final answer, provided the correct formula and clear workings are shown. On a calculation this dense — payback, ARR and NPV all involve several intermediate steps — leaving a blank because the final figure feels uncertain loses more marks than showing the right method with one arithmetic slip inside it.
The conditional move
Complete: "Simple payback is the most appropriate primary appraisal method for a business only if ___."
Complete: "Net present value gives a genuinely more reliable investment decision than payback or ARR only if ___."
Beyond the spec
The spec asks for the NPV calculation and its interpretation but doesn't ask why discounting is the correct adjustment to make in the first place — treating it as a procedure rather than a conclusion that follows from how capital actually earns a return. Knowing the theory is what stops NPV from being taught as an arbitrary formula, which is exactly what this course's own mechanism-derivation standard exists to prevent. It also settles a genuinely common real-world question none of this lesson's calculations force a student to confront directly: which cash flows are even allowed into an appraisal in the first place, and which — however large, however emotionally hard to write off — are not.
Irving Fisher's theory of interest (most fully set out in The Theory of Interest, 1930) is the underlying reason a future pound is worth less than a pound today: given the choice, people and firms generally prefer consumption or return now over the identical amount later — time preference — and a competitive capital market prices that preference into an observable rate of return, the same rate this lesson has been calling r. That rate is also, from a firm's side, its opportunity cost of capital: the return it gives up by tying money up in this specific project rather than the next-best alternative use of the same funds, which in a real business is usually estimated as its weighted average cost of capital (WACC) — a blend of what it costs to raise money from both debt (interest) and equity (the return shareholders require). NPV, as taught on this paper, picks a single discount rate in advance and asks whether the project clears it. Professional capital budgeting also asks the mirror-image question: the internal rate of return (IRR) — the exact discount rate at which a project's NPV would equal zero — which tells a firm the maximum cost of capital the project could tolerate before becoming unprofitable. Both descend from the same Fisher mechanism; NPV asks 'is this worth it at our actual cost of capital?' and IRR asks 'how much room for error do we have?' — two questions built on one derivation. The same theory settles a second, entirely practical question every appraisal calculation above has been quietly answering already: every cash flow this lesson has discounted or accumulated is one that hasn't happened yet — a decision still open to change. That's the sunk cost principle: a cost already incurred cannot be altered by any decision still to be made, so it carries no information capable of distinguishing between the options still on the table, and a correctly-built appraisal excludes it entirely. Suppose Larkspur had already spent £15,000 developing a prototype for the £120,000 machine appraised above, before ever deciding whether to proceed. That £15,000 is gone either way — cancel the project and it stays spent, proceed and it stays spent — so it cannot possibly help distinguish 'proceed' from 'cancel' as the better choice; only the cash flows still ahead (the remaining £120,000 outlay and the four years of inflows already calculated) can do that. Folding the sunk £15,000 into the NPV calculation, as though the true cost of proceeding were £135,000, doesn't make the appraisal more complete — it corrupts it, by adding a number that is identical under every option and therefore carries zero decision-relevant information. The sunk cost fallacy is this same error made emotionally rather than technically: continuing to pour money into a failing project specifically because so much has already gone into it, when 'how much have we already spent?' was never the right question — only 'what does spending more return, from this point forward?' ever was.
Retrieval — with feedback on every choice
A firm's annual sales (£000) for five years: Year 1 = 180, Year 2 = 210, Year 3 = 195, Year 4 = 240, Year 5 = 225.
What is the 3-year moving average centred on Year 3?
A business plots six months of scatter data (advertising spend vs sales) and extends its line of best fit to predict sales at double last month's spend — well beyond any point actually plotted. What is the single greatest risk specific to this extrapolation?
A firm's centred trend value for its most recent quarter is £71.25k, and the trend has been rising by an average of £1.821k per quarter. That firm's average variation for the quarter three periods ahead (actual − trend, averaged across every matching past quarter) is −£14.375k.
What is the forecast for that quarter, three periods ahead?
A business spends £80,000 on new equipment, forecast to generate net cash inflows of £25,000, £30,000, £35,000 and £20,000 in Years 1–4.
What is the simple payback period?
The same £80,000 investment: net cash inflows of £25,000, £30,000, £35,000 and £20,000 in Years 1–4.
What is the average rate of return (ARR)?
A business spends £60,000 on a project forecast to generate net cash inflows of £25,000 in each of Years 1, 2 and 3. Discount factors at 10%: Year 1 = 0.909, Year 2 = 0.826, Year 3 = 0.751.
What is the net present value?
Two machines both cost £150,000 and both have an identical total forecast cash inflow of £200,000 over four years, giving both an identical average rate of return. Machine A generates most of its cash inflow in the earlier years; Machine B generates most of its cash inflow in the later years.
Explain, using the concept of discounting, why Machine A is the better investment even though both machines' ARR is identical.
Same question, every level
Discuss the extent to which Amberview Sportswear Ltd can rely on quantitative sales forecasting techniques, such as a moving average, when planning next year's production levels. (VERIDIAN-original question, written to this paper's own confirmed 8-mark Discuss tariff — Section A, 'no conclusion required' — modelled on the real forecasting-reliability questions Pearson has set on Pets at Home (Oct 2022) and Spotify (Jan 2025), not a reproduction of either.)
8 marks available
Amberview Sportswear can use a moving average to see its sales trend and a line of best fit to predict future sales, so it should rely on these techniques when planning its production.
A recall-level assertion with no genuine application to Amberview's own figures and no named limitation of either technique — matches the confirmed 8-mark L1 (1-2) descriptor exactly: 'Isolated elements of knowledge and understanding – recall based. Weak or no relevant application to business examples. Generic assertions may be presented.'
Same question, every level
Evaluate the extent to which net present value is a more reliable method of investment appraisal than simple payback period for a business deciding whether to invest in new production equipment. (VERIDIAN-original question, written in the style confirmed across multiple WBS13 series — not a reproduction of any single past-paper question.)
20 marks available
Payback tells you how fast you get your money back. Net present value works out if an investment is worth it using discounting. Some businesses might prefer one method or the other.
Descriptive definitions with no calculation, no named business context, and no genuine comparison beyond restating what each method is. Matches the confirmed 20-mark L1 descriptor: isolated elements of knowledge with weak or no relevant application, and any argument attempted stays generic and fails to connect causes to consequences.
- 3-yr MA: average 3 periods (auto-centred). 4-qtr MA: average 4, then centre by averaging two consecutive MAs (n even).
- Seasonal/cyclical variation: actual − trend for each period; average matching periods (e.g. all Q1s). Forecast = extrapolated trend + that period's average variation.
- Line of best fit: drawn by eye. Interpolate inside the data; extrapolating beyond it is an assumption, not a fact.
- Payback: cumulative net cash inflow crossing the investment; interpolate the month. Ignores timing of money and post-payback profit.
- ARR = avg annual profit ÷ initial investment × 100% (verbatim from a real mark scheme, Jan 2022 Q1a — Bramwell Brown, £150,000 investment, ARR = 31.95%).
- NPV: DF = 1÷(1+r)ⁿ; sum discounted cash flows minus initial investment. Positive NPV = accept. Assess = 12 marks here, not 10.
Not affiliated with or endorsed by Pearson Edexcel. Every quotation and figure attributed to a mark scheme or examiner report in this lesson was independently verified against the primary Pearson document, not carried over from prior course material. This paper's own evidence base is narrower than some others in this course: no local archive existed for WBS13 before this build, every source was fetched fresh from qualifications.pearson.com, and only 6 of the 8 located exam series (Oct 2021, June 2022, Oct 2022, Jan 2023, June 2023, Jan 2025) were actually read and mined for content across the paper as a whole — Oct 2023's mark scheme remains unopened. The average rate of return (ARR) content, flagged in an earlier pass as resting only on an unverified secondary source, has since been checked against Jan 2022's own question paper, mark scheme and examiner report (WBS13_01_2201) — a seventh series, read specifically for its ARR question (Q1a, Bramwell Brown) rather than mined in full: the formula, the £150,000/£287,550/31.95% worked figures, and the three examiner-report error patterns quoted in this lesson are all confirmed directly against the primary PDF pages. Every calculation in this lesson was independently computed and cross-checked before being written in.
A firm's annual sales (£000) for five years: Year 1 = 180, Year 2 = 210, Year 3 = 195, Year 4 = 240, Year 5 = 225.
What is the 3-year moving average centred on Year 3?
- A195.0
This is the average of Years 1–3 (£180k+£210k+£195k)÷3 — the wrong 3-year window for a value centred on Year 3, which needs the year BEFORE and the year AFTER it, not the two years before it.
- 215.0
Correct. Centred on Year 3 means averaging Year 2, Year 3 and Year 4: (£210k+£195k+£240k)÷3 = £645k÷3 = £215.0k.
- C210.0
This is the simple average (mean) of all five years, (180+210+195+240+225)÷5 = 210 — a different calculation entirely, not a 3-year moving average of any kind.
- D220.0
This is the average of Years 3–5 (£195k+£240k+£225k)÷3 — the wrong 3-year window for a value centred on Year 3, which needs one year on each side, not two years after it.
Traps tested: Wrong window shifted early · Computes overall mean not moving average · Wrong window shifted late
A business plots six months of scatter data (advertising spend vs sales) and extends its line of best fit to predict sales at double last month's spend — well beyond any point actually plotted. What is the single greatest risk specific to this extrapolation?
- The relationship observed within the plotted data range is being assumed to hold beyond it, at a spend level where it has never actually been tested
Correct. Extrapolation extends the line based purely on the assumption that the same straight-line relationship continues past the last real data point — there is no data at that higher spend level confirming the relationship actually holds there.
- BSix data points is mathematically too few to draw any line of best fit at all
Six points is a perfectly usable, if small, sample for drawing a line of best fit by eye — the risk in this scenario is specifically about extrapolating beyond the data, not about whether a line can be drawn from six points in the first place.
- CScatter graphs can only ever be read by interpolation, never extended at all
Extrapolation is a real, named technique on this spec (3.3.3.1b) — it's a riskier read than interpolation, not a forbidden one. The issue is treating an extrapolated figure with the same confidence as an interpolated one, not that extrapolation itself can't be done.
- DThe line of best fit must have been drawn incorrectly if the prediction looks large
A large predicted value isn't evidence the line was drawn wrong — it's simply what a correctly-drawn line, extended a long way past the data, produces. The risk is in the extrapolation itself, not in an assumed drawing error.
Traps tested: Wrong concept entirely · Overclaims a restriction · Assumes a drawing error
A firm's centred trend value for its most recent quarter is £71.25k, and the trend has been rising by an average of £1.821k per quarter. That firm's average variation for the quarter three periods ahead (actual − trend, averaged across every matching past quarter) is −£14.375k.
What is the forecast for that quarter, three periods ahead?
- A£76.71k
This is the extrapolated trend value alone (£71.25k + 3×£1.821k) — correct as far as it goes, but it stops before the last step: the quarter's own average variation still needs to be added on to turn a general trend estimate into a forecast for this specific, structurally weaker quarter.
- £62.34k
Correct. Extrapolated trend = £71.25k + (3 × £1.821k) ≈ £76.71k. Forecast = extrapolated trend + average variation = £76.71k + (−£14.375k) ≈ £62.34k.
- C£56.88k
This applies the average variation to the last KNOWN trend value (£71.25k − £14.375k) instead of the extrapolated trend three periods ahead — the variation has to be added after extrapolating the trend forward to the actual period being forecast, not to a trend value from an earlier period.
- D£91.09k
This adds the SIZE of the variation (£14.375k) rather than its actual negative sign (£76.71k + £14.375k) — treating a quarter that structurally runs below trend as if it ran above it, which reverses the whole point of calculating the variation in the first place.
Traps tested: Extrapolates trend but forgets variation · Applies variation to last known trend not extrapolated trend · Sign error on negative variation
A business spends £80,000 on new equipment, forecast to generate net cash inflows of £25,000, £30,000, £35,000 and £20,000 in Years 1–4.
What is the simple payback period?
- A2 years
Cumulative cash inflow after Year 2 is £25,000+£30,000 = £55,000 — still £25,000 short of £80,000. Stopping at the nearest whole year without checking how far into Year 3 the crossing actually falls skips the interpolation step entirely.
- B3 years
Cumulative cash inflow reaches £90,000 by the END of Year 3 — but the £80,000 investment is actually recovered partway THROUGH Year 3, not at its end. Rounding up to the whole year the crossing falls in, rather than finding exactly where inside it, overstates the payback period.
- C2 years 10 months
This comes from dividing the £25,000 shortfall by Year 2's £30,000 inflow instead of Year 3's £35,000 inflow — the shortfall has to be divided by the cash inflow of the year the payback actually falls INSIDE, not an earlier year's figure.
- 2 years 8.6 months
Correct. Cumulative inflow after Year 2 is £55,000, £25,000 short of £80,000. That shortfall is recovered during Year 3, out of its £35,000 inflow: £25,000÷£35,000 = 0.714 of the year, ×12 ≈ 8.6 months. Payback = 2 years + 8.6 months.
Traps tested: No interpolation within year · Rounds up to whole year · Used wrong years inflow to interpolate
The same £80,000 investment: net cash inflows of £25,000, £30,000, £35,000 and £20,000 in Years 1–4.
What is the average rate of return (ARR)?
- A37.5%
This comes from dividing total profit (£110,000−£80,000=£30,000) straight by the investment, without first averaging it across the 4 years — the missing step is dividing the total profit by the number of years before comparing it to the investment.
- 9.38%
Correct. Total net cash inflow = £25,000+£30,000+£35,000+£20,000 = £110,000. Total profit = £110,000−£80,000 = £30,000. Average annual profit = £30,000÷4 = £7,500. ARR = £7,500÷£80,000×100 = 9.375%, rounded to 9.38%.
- C34.38%
This treats the full £110,000 of cash inflow as if it were profit, dividing by 4 without first subtracting the £80,000 investment — profit is what's LEFT after recovering the original outlay, not the total cash inflow itself.
- D6.82%
This divides the average annual profit by the total cash inflow (£110,000) rather than by the original £80,000 investment — ARR is always expressed as a percentage OF THE INVESTMENT, not of the total cash the project brings in.
Traps tested: Forgot to average across years · Forgot to subtract investment · Wrong denominator
A business spends £60,000 on a project forecast to generate net cash inflows of £25,000 in each of Years 1, 2 and 3. Discount factors at 10%: Year 1 = 0.909, Year 2 = 0.826, Year 3 = 0.751.
What is the net present value?
- A£62,150
This is the total present value of the three discounted cash flows on its own (£22,725+£20,650+£18,775) — the initial £60,000 investment still needs to be subtracted from it to get the NET present value.
- B−£16,625
This only discounts Years 1 and 2 (£22,725+£20,650=£43,375) before subtracting the investment, leaving Year 3's £18,775 out of the total entirely — every year's discounted cash flow has to be included, not just the first two.
- C£15,000
This is the UNDISCOUNTED total profit (£75,000−£60,000) — it ignores the discount factors given in the stimulus entirely and treats every year's £25,000 as worth exactly £25,000 today, the assumption discounting exists to correct.
- +£2,150
Correct. Present values: £25,000×0.909=£22,725; £25,000×0.826=£20,650; £25,000×0.751=£18,775. Total present value = £62,150. NPV = £62,150−£60,000 = +£2,150.
Traps tested: Forgot to subtract investment · Omitted a years cash flow · Ignores time value of money
Two machines both cost £150,000 and both have an identical total forecast cash inflow of £200,000 over four years, giving both an identical average rate of return. Machine A generates most of its cash inflow in the earlier years; Machine B generates most of its cash inflow in the later years.
Explain, using the concept of discounting, why Machine A is the better investment even though both machines' ARR is identical.
- Machine A's net present value is higher than Machine B's, because Machine A's cash inflows arrive sooner and are discounted less heavily — ARR treats every pound earned over the project's life as equally valuable regardless of when it arrives, which is precisely the assumption NPV's discounting removes
Correct, and this is the fully-integrated version: it names the mechanism (discounting penalises delay), the specific reason ARR can't see it (no timing information in the ARR formula at all), and connects both back to the actual scenario rather than restating the question.
- BMachine B is actually the better investment, because cash flows compound and grow more valuable the longer they are left invested before being received
This reverses the direction of discounting. A cash flow arriving LATER is worth LESS today, not more — compounding describes what happens to money already in hand growing forward, not what happens to a future receipt being converted back to today's value.
- CBoth machines are genuinely equally good investments, since their total cash inflow and their ARR are both identical
This ignores discounting entirely — the very concept the question asks the answer to use. Identical totals and identical ARR say nothing about NPV, because ARR by construction contains no information about which year any of the cash arrives in.
- DThere's no way to say which machine is better without first being told the exact discount rate that will be used
A specific discount rate is needed to calculate the exact size of the NPV gap, but the DIRECTION of the result — Machine A ahead of Machine B — holds for any positive discount rate, because Machine A is strictly more front-loaded than Machine B. Falling back on 'not enough information' overstates the genuine uncertainty here.
Traps tested: Direction reversed · Ignores time value of money · Overclaims uncertainty
Practice this for real
This site teaches the mechanism; the exam is sat on Pearson's own real questions. Go find and attempt these yourself — nothing here substitutes for actually sitting a timed paper.
- Mark scheme
- Oct 2022 · Q1b — cited directly in this lesson
- Mark scheme
- Jan 2022 · Q1a — cited directly in this lesson
Select International Advanced Level → Business → any series, then look for WBS13.
Up next
Decision Trees, Critical Path Analysis and Contribution
Three techniques, one shared job: turning a decision that feels like a judgement call into numbers a business can actually check. A decision tree weighs an uncertain choice, critical path analysis finds the shortest a project can possibly take, and contribution tells a firm whether one more unit — even at a strange price — adds to profit or not.
55 min