The CAP turned a moat into a duration. So why learn another method? Because Greenwald's three-tier stack turns the moat into a number you can compute from a single year of statements, and it forecasts nothing at all. That makes it the cleanest barrier test in the phase.
The construction is Greenwald's, set out with Judd Kahn, Paul Sonkin and Michael van Biema in Value Investing: From Graham to Buffett and Beyond (2001; second edition 2020). You get three valuations of the same company, each answering a different question, in increasing order of how much judgement they demand.
How do you calculate asset reproduction value?
Tier 1, asset reproduction value. What would a competent entrant spend to stand where this company stands today, with the same capacity, the same shelf presence and the same customer list? It isn't book value, and it isn't liquidation value. Work down the asset side and restate each line at what recreating it would cost. Plant and equipment go in at the current cost of equivalent capacity, which is usually below gross block for a technology that has become cheaper and above net block for one that hasn't. Receivables enter at face less what's already known to be uncollectible, and inventory at current cost. Only the cash the business needs to operate belongs here, because excess cash is a separate asset an entrant doesn't have to reproduce. Then subtract the spontaneous liabilities, trade payables above all, since the entrant gets those free as well.
Two lines matter most, and they appear on no balance sheet. One is the accumulated selling, brand and product-development spend an entrant would have to repeat to arrive with a comparable position. Capitalise it over a period matched to the industry's customer-acquisition cycle, and don't pull that period from the air. The other is an allowance for the operating losses of the ramp, because nobody reaches scale profitably from day one. Leave both out and reproduction value comes out too low, which manufactures a franchise that isn't there. It's the most common error in the method.
Advertisement
How do you calculate earnings power and franchise value?
Tier 2, earnings power value. What are today's earnings worth if the company never grows and never improves? Take operating profit on a normalised margin, averaged across a full cycle instead of read off last year. Add back the excess of depreciation over the capital expenditure really needed to hold capacity. Strip out the one-offs and tax what's left. Then capitalise it at the cost of capital. Earnings power value for the operations is adjusted net operating profit after tax divided by the weighted average cost of capital. Add excess cash and subtract debt to reach equity. There's no growth term anywhere in it, and that's deliberate. Nobody can inflate this valuation with a story.
Notice what that expression already is. It's the steady-state term of the Miller-Modigliani skeleton, NOPAT / WACC, arrived at from the opposite direction. Greenwald's earnings power value and the value of a no-barrier firm are the same quantity.
Tier 3, franchise value, as a residual. Franchise value is earnings power value minus reproduction value. Nobody estimates it. It falls out of the other two, and it has one meaning only, which is what makes the stack a barrier test and not a valuation exercise.
| Reading | What it says | What to do |
|---|---|---|
| EPV well above reproduction | Something stops an entrant from competing the return away. Barriers exist, and the gap is their price. | Name the barrier from the three real advantages, then grant a CAP using the competitive advantage period |
| EPV at reproduction | No barriers. Everything earned is the return on assets anyone can buy. | Value at steady state; CAP 0 to 3 years |
| EPV below reproduction | The industry holds more capacity than demand will pay for, so no rational entrant builds. | Value between EPV and liquidation; you are in M4.06's stage 5 |
One identity makes the residual rigorous. Franchise value also equals (adjusted NOPAT − WACC × reproduction value) / WACC, which is the excess return on the reproducible asset base, capitalised. Compute it both ways every time. If the two disagree, you've made an arithmetic error, and your judgement has nothing to do with it.
Worked example: Satara Abrasives
Worked reconciliation: Satara Abrasives (synthetic, India, ₹ crore, hurdle 12%, tax 25%). Revenue is ₹2,400 crore at a through-cycle EBIT margin of 15.5%. Depreciation is ₹95 crore against ₹72 crore of maintenance capital expenditure, and there's ₹180 crore of excess cash and ₹260 crore of debt.
| Reproduction line | ₹ cr |
|---|---|
| Plant and equipment at current replacement cost | 780 |
| Receivables | 380 |
| Inventory | 460 |
| Operating cash | 60 |
| Trade payables | (300) |
| Three years of channel and brand spend at ₹90 cr | 270 |
| Start-up operating-loss allowance | 70 |
| Asset reproduction value | 1,720 |
Now earnings power. Normalised EBIT = 2,400 × 15.5% = ₹372 crore. Add the ₹23 crore by which depreciation exceeds maintenance capex and you get adjusted EBIT of ₹395 crore. After 25% tax, adjusted NOPAT is ₹296.25 crore, and capitalised at 12%, EPV of the operations is ₹2,468.75 crore. Franchise value is therefore 2,468.75 − 1,720 = ₹748.75 crore. Cross-check it with the identity. Excess return is 296.25 − 0.12 × 1,720 = ₹89.85 crore, and 89.85 ÷ 0.12 = ₹748.75 crore. Both routes give the same answer. Put another way, the reproducible asset base earns 17.22% against a 12% hurdle, a 5.22-point spread, and 30.3% of the company's earnings power is franchise and not assets.
Now run the CAP skeleton on the same company and watch a third number appear. Suppose Satara can reinvest 35% of NOPAT (₹103.69 crore a year) at a 20% incremental return, so growth is 0.35 × 0.20 = 7%. Suppose too that the evidence supports a ten-year CAP. Each CAP year is worth 103.69 × (0.20 − 0.12) ÷ (0.12 × 1.12) = ₹61.72 crore, so ten years add ₹617.19 crore and enterprise value is ₹3,085.94 crore.
| Number | ₹ cr | The question it answers |
|---|---|---|
| Asset reproduction value | 1,720.00 | What an entrant pays to be here |
| Earnings power value | 2,468.75 | What today's earnings are worth with no growth |
| CAP-adjusted enterprise value | 3,085.94 | Today's earnings plus ten years of profitable reinvestment |
What do the three numbers tell you?
These three disagree because they answer three different questions, and the two gaps are different kinds of thing. The ₹748.75 crore between reproduction and earnings power is the existing franchise. It protects assets already in the ground, and it's evidence, because it was computed from statements without forecasting anything. The ₹617.19 crore between earnings power and the CAP valuation is the future franchise. It's the spread on money not yet spent, and it's a grant, because it exists only for as long as the CAP you awarded. Reproduction and earnings power test the moat. The CAP prices it.
A practical rule follows straight away. If the market pays ₹3,400 crore for Satara, the implied CAP is (3,400 − 2,468.75) ÷ 61.72 = 15.1 years. Implied total franchise is 3,400 − 1,720 = ₹1,680 crore, and the statements alone support only ₹748.75 crore of it. The remaining ₹931 crore is the part the four-lens protocol has to earn with share tables and barrier evidence. Quote both numbers in a dossier. If you report only the DCF, you've hidden which half of the value was measured and which half was granted.
What are the limits of earnings power value?
The method has limits, and they're real. Reproduction value is the softest of the three numbers. Replacement cost of plant is a judgement, and the number of years of brand spend to capitalise is a bigger one, so run reproduction value as a band and see whether the barrier verdict survives the low end. Normalised margin is a cycle judgement, and a deep cyclical at the peak will show a spurious franchise. The method also says nothing useful about a company whose assets are almost entirely intangible. There, reproduction value is nearly all capitalised spend, and the answer is whatever you assumed. And a franchise measured today is a fact about today. Watching it move is a job for the erosion lens.