FINM3005 · Corporate Valuation
Valuation Recommendation
This is the capstone, where the whole course lands. Every method so far — enterprise DCF, reorganised ROIC, forecasting, WACC, growth, multiples, sum-of-the-parts, M&A — converges into one deliverable: a per-share value, compared against the market price, expressed as a Buy / Hold / Sell call over a 1–3 year horizon. No single method is 'the' answer, so you triangulate a range across three lenses (DCF, multiples, SOTP) and argue where in it the stock sits; because multiples are an implied DCF, the lenses should roughly agree, and a sharp divergence tells you which input disagrees with the market. The chapter adds the analyst's toolkit for that judgement: the TRS decomposition (a return splits into earnings growth, P/E change and dividend yield — only the first is real value creation), value-vs-price and the forms of market efficiency, sensitivity (one driver at a time → the tornado) and scenario analysis (a coherent bundle of drivers → worst/base/best, probability-weighted), holdings and minority-interest consistency in multiples, and what a high-marks recommendation always carries. It doubles as the spine of the 35% valuation-report assignment, whose grade is model-building integrity, not a single 'correct' number.
What this chapter covers
- 01Triangulation: reconciling DCF, multiples and SOTP into one range
- 02Value vs price and the forms of market efficiency
- 03The TRS decomposition: earnings growth, P/E change, dividend yield
- 04Sensitivity analysis and the tornado
- 05Scenario analysis: worst/base/best, probability-weighted
- 06Holdings and minority-interest consistency in multiples
- 07The Buy / Hold / Sell call and what a strong recommendation carries
Worked example: probability-weighted value and the call
- +1(a) Probability-weighted value. E[V] = 0.25(18) + 0.50(27) + 0.25(34) = 4.5 + 13.5 + 8.5 = $26.5.
- +1Compare to price. Expected value $26.5 vs price $22 → ~20% upside to the probability-weighted estimate.
- +1(b) The call. Value materially above price (upside beyond a sensible margin of safety) → Buy, anchored on intrinsic value, with a target price and 1–3 year horizon.
- +1State what would change it. The thesis turns to Hold if base growth proves optimistic; to Sell if margins fade toward the worst case — name the contingency, not a false-precision number.
- +1(c) The spread. The $18–$34 range is what feeds the recommendation: a call near the worst case with a wide spread is far riskier than the same call with a tight one. View the valuation as a distribution, not a point.
Key terms
- Triangulation
- Reconciling the three valuation lenses — enterprise DCF, multiples and sum-of-the-parts — into one defensible range rather than a single number. Because every multiple is an implied DCF, the lenses should roughly agree; where they diverge sharply, the divergence is information that one input (growth, ROIC, WACC) disagrees with the market's.
- TRS decomposition
- Total return to shareholders = (price change + dividends) / opening price, which splits into the change in earnings (revenue × margin, the only piece backed by fundamentals), the change in the P/E (a shift in market expectations, a re-rating that can reverse), and the dividend yield. A high TRS driven mostly by P/E expansion is hollow and vulnerable to a de-rating.
- Market efficiency
- Weak form (prices reflect past trading data — technical analysis cannot beat the market), semi-strong (all public information — fundamental analysis cannot earn abnormal returns on listed firms), and strong (public + private). Even under semi-strong efficiency, valuation is indispensable for private firms, divisions and acquisition targets, where there is no market price to lean on.
- Sensitivity vs scenario analysis
- Sensitivity moves one driver at a time (WACC, terminal g, margin, growth, turnover, tax), holding the rest constant, and ranks impact — the tornado, usually topped by WACC and g, which sit in the (WACC − g) denominator. A scenario changes a coherent bundle of drivers to describe a state of the world; build worst/base/best and probability-weight them into the headline value.
- Probability-weighted value
- The expected value across scenarios, E[V] = Σ pₛVₛ = p(worst)V(worst) + p(base)V(base) + p(best)V(best). The weighted figure is the headline, but the spread of outcomes feeds the recommendation: a fair price should already embed the upside and downside surprises, so a wide spread signals more risk around the same call.
Valuation Recommendation FAQ
How do I reconcile a DCF and a comps value that disagree?
Treat the divergence as information, not noise. Because every multiple embeds the same fundamentals a DCF makes explicit (P/E and EV/EBITA are both functions of growth, ROIC and the discount rate), a sharp DCF-vs-comps gap means one of your inputs disagrees with the market's. Find out which — usually long-run growth or margin — decide who is right, and present a triangulated range with a reasoned point inside it rather than forcing a single number.
What does a high-marks recommendation always include?
A clear Buy/Hold/Sell with a target price and horizon; it is tethered to intrinsic value, not price momentum; it states what would change the call (the contingencies and catalysts); it may differ by horizon (short-term trading vs long-term fundamental); and it carries a short peer comparison and a market-efficiency caveat. Present a range, then argue a point inside it — 'fair value $24–$30, we sit at $27, price is $22 → Buy' is more defensible than a single decimal.
What is the TRS decomposition for, and why does it matter?
It tells you what actually drove a past return so you do not extrapolate the wrong thing. A return splits into earnings growth (real operating performance), the change in the P/E (a re-rating, i.e. shifting expectations, which can reverse) and the dividend yield. If most of a year's return came from P/E expansion rather than earnings, it rode rising expectations, not rising fundamentals — vulnerable to a de-rating. Always ask which term did the work.
Why use scenarios rather than just sensitivity tables?
A one-variable sensitivity table pretends drivers are independent, but they are not — raising the operating margin often lowers capital turnover, and a price rise cuts volume. A scenario changes a coherent bundle of drivers to describe a believable state of the world, so worst/base/best valuations, probability-weighted, give a more honest headline value and, crucially, a spread. The spread is what frames the risk around the recommendation.
Exam move
Practise assembling the whole course into one answer: a DCF value, a comps cross-check, an SOTP where relevant, triangulated into a range, then a call. Always present a range and argue a point inside it — a single decimal is less defensible than 'fair value $24–$30, we sit at $27'. Use the TRS decomposition to test whether a past return is real (earnings) or hollow (P/E re-rating), and lean on scenarios over one-variable tables because drivers move together; probability-weight worst/base/best and report the spread, not just the mean. Keep the four recommendation traps in view: a call untethered from intrinsic value, one number instead of a range, inconsistent holdings/minority treatment in multiples, and confusing price momentum with value. This worked answer is also the spine of the 35% assignment, whose grade is model-building integrity — consistency, defensible assumptions and ROIC sanity checks — not a single 'correct' number.
