The growth model, and the one input that controls it
Four inputs, four steps, and only one of them is a fact. Walking the model forwards is easy; the skill is in knowing how far the answer moves when you are wrong about the growth rate.
What you'll take away
- Current earnings per share is the only input you can look up. The other three are decisions.
- Six percentage points of growth rate move the answer by about 70%. That sensitivity is the whole reason for the final halving.
- The model needs a predictable earnings series. On a cyclical or a newly listed company, it does not just lose accuracy — it misleads.
Four inputs, one fact
The model is short enough to hold in your head, which makes it dangerously easy to run without noticing how much of it you made up.
- Current earnings per share. A fact, from the filings. The only one.
- Ten-year growth rate. An estimate, and the input that dominates the output.
- Future price-to-earnings ratio. A convention: twice the growth rate, capped by the company’s own ten-year range of multiples.
- Required annual return. A choice. 15% is the usual figure, and it is the rate you discount at.
The cap on the future multiple is worth dwelling on, because it is the one piece of discipline inside the model. Twice the growth rate would give a company compounding at 12% a future multiple of 24. If the market has never once in ten years paid more than 18 for this company, 24 is not a valuation input — it is an assumption that the market will start behaving differently. Take the lower of the two.
The four steps, worked
Current EPS of 5, a ten-year growth rate of 12%, a ten-year multiple range topping out at 18, and a required return of 15%.
- Grow the earnings out ten years. 5 × 1.12¹⁰ = 15.5 of earnings per share in year ten.
- Price those earnings. Twice the growth rate is 24, the historical ceiling is 18, so use 18: 15.5 × 18 = 279 as the share price in year ten.
- Discount it back at your required return. 279 ÷ 1.15¹⁰ = about 69. That is the most you could pay today and still earn 15% a year, if every assumption holds.
- Halve it. 69 ÷ 2 ≈ 35. That is the entry price, and the halving is not timidity — it is the budget for the assumptions in steps one and two being wrong.
Where the growth rate comes from
One number decides most of the answer, so it deserves more care than the other three combined. There are three defensible places to get it, and they will not agree.
- The historical rates. The four growth metrics over ten, five, three and one year — eight to sixteen figures, from which you take the lower end rather than the headline.
- The equity growth rate. Return on equity multiplied by the retention ratio: the speed at which the business can fund its own growth without issuing shares or borrowing more.
- Analyst consensus. Useful mainly as a sanity check on the other two, and systematically optimistic. If it is the highest of the three, that tells you something about the consensus, not about the company.
Take the most conservative of the three. Not the average — the lowest.
There is also a practical ceiling, somewhere around 15%. Very few businesses compound earnings faster than that for ten consecutive years, and the ones that do are not identifiable in advance from a spreadsheet. Enter a 25% ten-year growth rate and you are not valuing a company, you are making a wish.
How much the answer moves
Same company, same multiple, same required return. Only the growth rate changes.
| Growth rate | EPS in yr 10 | Value today | Entry price |
|---|---|---|---|
| 9% | 11.8 | 53 | 26 |
| 12% | 15.5 | 69 | 35 |
| 15% | 20.2 | 90 | 45 |
Six percentage points — the distance between two perfectly reasonable readings of the same ten years of history — moves the entry price by about 70%. No other input in the model comes close to that leverage.
This is the honest reason for the final halving. It is not a safety blanket bolted onto a precise calculation; it is an acknowledgement that the calculation has a 70% range inside it, and that you would rather sit at the bottom of that range than in the middle of it.
The consequence is that an entry price computed this way will be below the market price most of the time, for most companies. Being unable to buy nearly everything, nearly always, is the function of the method, not a side effect of it.
Three times not to use it
The model assumes a smooth earnings series extended ten years. Where that assumption is absent, the output is not approximately right — it is confidently wrong, which is worse.
- Unpredictable earnings. If EPS has swung between losses and record profits over ten years, there is no growth rate to estimate. Any figure you enter is arbitrary, and the four steps will launder it into something that looks calculated.
- Cyclical businesses. Near the top of a cycle, earnings are at their peak and the multiple is at its lowest, so the model is fed its most flattering inputs at the worst possible moment. On a cyclical, the lowest price-to-earnings ratio is often the most expensive moment.
- Newly listed companies. Fewer than ten years of history means there is nothing to compute a conservative rate from, and no observed multiple range to cap the future multiple with. Both of the model’s disciplines are missing at once.
In all three cases the right response is to drop this model rather than to force it, and to see what the present-tense and cumulative models say instead. That is what running three of them is for.
Worked examples use invented figures to make the arithmetic visible. Moatrics covers US-listed companies.
All articles