Sensitivity and scenario tables

The last two modules delivered an uncomfortable verdict: most of a DCF's value hides in its softest assumption, and its exact-looking output is false comfort. That could leave you feeling the whole exercise is hopeless. It is not. There is a simple, honest fix, and this module is it.

Instead of reporting one number, you report the range — and you show, on a single grid, exactly how the value moves as your two or three most important assumptions change. This is a sensitivity table, and paired with a handful of scenarios, it turns a fragile point into an honest picture. You stop pretending to know the value and start showing what you actually know: a band, and which assumptions decide where inside it the truth lies.

A single figure — ₹166 — makes three quiet lies. It hides how wide the plausible answer is. It hides which assumptions the answer depends on. And it hides where the market price sits relative to your thinking. A sensitivity grid tells all three truths at once.

The idea is plain. Take the two inputs that move the value most — for a typical DCF, the discount rate and the growth rate — and lay one across the top and one down the side. In each cell, put the value the DCF produces for that combination. What was a single number becomes a small map of possibilities, and the map shows you the terrain: how steep the drop as the discount rate rises, how much the value climbs with growth, and how far apart the corners are.

This is being roughly right on purpose — refusing the false comfort of a point in favour of a band you can actually defend. The grid does not make the uncertainty go away. It makes it visible, which is the most honest thing a valuation can do.

Take our composite once more. illustrative ₹100 crore of cash, growing 10% for five years, then settling to a terminal rate. We vary the two inputs that matter most: the discount rate (down the side) and the terminal growth rate (across the top). Each cell is the whole-business value in ₹ crore.

Now read it the way it is meant to be read — not by hunting for "the answer", but by taking in the whole shape.

• The range is the output. The grid runs from ₹1,362 crore in the cautious corner (high discount rate, low growth) to ₹2,159 crore in the optimistic one. At 10 crore shares, that is roughly ₹136 to ₹216 per share. That band is your valuation — not the ₹166 middle cell.
• The tilt tells you what matters. Moving one column right (growth up 1%) adds roughly ₹150-250 crore. Moving one row down (discount up 1%) subtracts a similar amount. Both inputs have a strong grip — which is why both belong on the grid.
• The corners are least likely. The extreme cells stack two favourable or two unfavourable assumptions at once, and, as the next section shows, those combinations are less probable than any single input suggests.

A sensitivity grid varies inputs mechanically, one at a time. A scenario does something more thoughtful: it tells three coherent stories about the business and prices each. This is where the grid meets judgement.

Build three cases for our composite. illustrative

• Bear. Competition bites, growth disappoints, and you demand more for the risk: discount 13%, terminal growth 3%. Value ≈ ₹1,362 crore (₹136/share).
• Base. Things go roughly as expected: discount 12%, terminal growth 4%. Value ≈ ₹1,662 crore (₹166/share).
• Bull. The moat holds, growth stays strong, and the market stays calm about risk: discount 11%, terminal growth 5%. Value ≈ ₹2,159 crore (₹216/share).

Now lay the market price beside them. At ₹150 a share, the market is valuing the business near ₹1,500 crore — below the base case, and only a little above the bear case. That is a precise, useful statement: the market is currently pricing something close to the cautious story. You have learned far more than "the stock is worth ₹166". You have learned which of your three futures the price is betting on — which is exactly the bridge into Part Three's reverse-DCF, where you read the price to find the assumptions it already contains.

One discipline makes scenarios honest: keep the inputs inside each case consistent with each other. The bull case is not "every dial turned to best". Fast growth and low risk rarely coexist — the market charges a premium for safe growth precisely because it is rare. A coherent bull case pairs strong growth with a discount rate that reflects the risk of chasing it, not a fantasy where everything is favourable at once.

A range is more honest than a point, but it is not omniscient, and it brings its own subtler traps.

It cannot assign the odds for you. The grid shows nine values; it does not say which is most likely. That judgement — how probable each future is — is yours, and it is where the real analysis lives. Probability-weighting is a whole later module for exactly this reason.

Its corners can mislead. The best and worst cells stack two assumptions at once and look reachable cell-by-cell, but as combinations they are far less likely than either input alone. Reading value off a corner reintroduces the false confidence the grid was meant to cure.

A wide range is not a licence to pick your favourite cell. The temptation, once you see a band from ₹136 to ₹216, is to quietly settle on whichever end suits your prior view. The grid's honesty depends on you reading the whole of it, especially the cautious end. The margin of safety is measured against the pessimistic end of the range, not the comfortable middle — that is the discipline that keeps a range from becoming an excuse.

1. Reporting the middle cell as 'the value'. The moment you quote ₹166 and drop the grid, every gain of honesty is thrown away. The range is the answer; carry it, not its centre.

2. Stacking a fantasy bull case. High growth, low discount rate and a rich exit multiple all at once is a corner that compounds improbabilities. Keep the inputs within a scenario consistent with one another.

3. Measuring margin of safety against the base case. Safety is judged against the cautious end. Price above the bear case is a warning, however comfortable the base case looks.

4. Varying trivial inputs. A grid is only useful if it flexes the two or three inputs that actually drive the answer. Sensitivity on a line item that barely moves the value is busywork dressed as rigour.

5. Confusing a range with a probability. Showing nine values is not the same as knowing which is likely. The grid frames the judgement; it does not make it for you.