Intrinsic value and discounted cash flow, made simple
The answer
A discounted cash flow model estimates what a business is worth by adding up the cash it will produce in future, with each future rupee or dollar reduced to what it is worth today. It is arithmetic wrapped around guesses, and most of the answer comes from a guess about the distant future.
So do not use it to produce a target price. Use it backwards: take today's price, and solve for the growth rate the market is already assuming. That sentence is something you can agree or disagree with. A target price is not.
Why this costs you money
Somebody builds a spreadsheet. It has 10 years of projected cash flows, a weighted average cost of capital calculated to 2 decimal places, and a terminal value. It produces a number: this share is worth ₹1,840. The market price is ₹1,310. The conclusion appears obvious. Buy.
Three things are wrong here, and each of them costs real money.
The precision is fake. Change the growth assumption in years 6 to 10 by 2 percentage points, and change the discount rate by 1 percentage point, and the same model will produce ₹1,150 or ₹2,600. Nothing about the business changed. Two small edits to 2 numbers nobody can know moved the answer by more than 100%. A model that swings that far on inputs that cannot be verified is not a measurement. It is an opinion with decimal places.
The terminal value is most of the answer. In a standard 10-year model, the value assigned to everything after year 10 is commonly 60% to 80% of the total . That single number is calculated from a perpetual growth rate the analyst chose. So the model's output is mostly a restatement of what the analyst assumed about a period nobody can forecast. The 10 years of careful projection in the front are, in valuation terms, the smaller half.
The direction of the error is predictable. People build models to justify conclusions they already hold. It is very hard to build a model of a company you admire and arrive at a low number, because you will adjust the growth rate until the answer feels right. The model then becomes a document that makes a hunch look like research. That is worse than a hunch, because it removes your doubt.
The loss is not usually a single bad trade. It is a decade of overconfidence. A reader with a spreadsheet holds a falling position longer than a reader without one, because the spreadsheet says the shares are worth more than the price.
How it works
Three ideas, and then the honest part.
1. Money now is worth more than money later. ₹100 today, invested at 8%, is ₹108 in a year. So ₹108 due in a year is worth ₹100 today. To convert a future amount into today's money you divide by (1 + r) raised to the number of years, where r is the discount rate. This is called discounting.
2. The cash flow you discount is free cash flow. That means cash from operations minus the capital spending needed to keep the business running and growing. It is the cash genuinely available to the people who funded the business. It is not profit. Profit contains non-cash items and ignores the cost of new machines.
3. The discount rate is your required return, and it carries the risk. In professional practice it is the weighted average cost of capital, blending the cost of equity and the after-tax cost of debt. For a private investor, a defensible shortcut is: the yield on a 10-year government bond, plus an equity risk premium, plus more for a riskier business. A higher discount rate produces a lower value.
Now put them together.
Value = the discounted sum of forecast cash flows for N years, plus the discounted terminal value.
Terminal value is the value of everything after the forecast period. The usual method is the perpetuity growth formula:
Terminal value = cash flow in the final year × (1 + g) ÷ (r − g)
where g is the growth rate assumed to continue forever and r is the discount rate.
Look at that denominator. If r is 10% and g is 3%, the denominator is 7%. If you raise g to 5%, the denominator becomes 5%, and the terminal value rises by 40%. A 2 percentage point change in a number describing the year 2065 moves most of the answer.
There is 1 iron constraint on g. No company can grow faster than the whole economy forever, because it would eventually become the whole economy. So the perpetual growth rate cannot honestly exceed long-run nominal GDP growth. Any model using a perpetual growth rate above that is arithmetically claiming something impossible.
Margin of safety, the idea Benjamin Graham built on, exists precisely because of all this. You do not buy at your estimated value. You buy well below it, so that being wrong by a normal amount still leaves you unharmed. If you need your estimate to be accurate for the investment to work, the estimate is doing too much work.
What it tells you, and what it does not
A DCF tells you what has to be true for a price to make sense. That is a genuinely valuable thing and it is the only thing the model does reliably.
It does not tell you what a company is worth. It cannot, because the inputs that matter most cannot be known. It does not tell you what a share will do. And it is close to useless for certain businesses: early-stage companies with no cash flow, cyclical companies where "normal" cash flow does not exist, and banks, whose cash flow statement does not mean what it means elsewhere.
The important limitation is subtler. A DCF is not a bad model because it is imprecise. It is a risky tool because its output is a single confident number, and single confident numbers change behaviour. A range would be honest. A point estimate is not.
The decision rule
Never run a DCF forwards to get a price. Run it backwards to get a sentence.
Here is the reverse question, and it is the whole method:
Take today's market price as correct. Now solve for the growth rate that would justify it. Then say out loud whether you believe that growth rate.
The simplest version needs 3 numbers and 1 line of arithmetic.
Start from the perpetuity formula, rearranged:
Value = free cash flow ÷ (r − g)
so
g = r − (free cash flow ÷ enterprise value)
Free cash flow divided by enterprise value is called the free cash flow yield. So:
Implied growth = your required return − the free cash flow yield.
If a business has an enterprise value of ₹50,000 crore, produces ₹1,500 crore of free cash flow, and you require 11% a year, then the free cash flow yield is 3%, and the implied perpetual growth rate is 8%.
Now you have a sentence: at this price, the market is assuming this company grows its free cash flow at about 8% a year, forever.
That is crude. It assumes cash flow grows smoothly from today, which no business does. It will overstate implied growth for a company at a cyclical trough and understate it for one at a peak. Use it as a first screen, not a conclusion. But it is honest about being crude, which the 10-tab spreadsheet is not.
Then run the number through 4 comparisons.
- Against nominal GDP growth. India's long-run nominal GDP growth has been in the region of 10% to 11%; the United States around 4% to 5%
. A perpetual growth rate above the economy's is not conservative or aggressive. It is impossible.
- Against the company's own last 10 years. Pull revenue and operating cash flow from 10 years ago and today. Compute the actual compound growth rate. If the implied rate is far above what the business has ever achieved, the price requires the company to become better than it has been.
- Against the industry. A company cannot grow far faster than its market forever without taking almost all of it. Ask what share it would need.
- Against the size problem. A company with revenue of ₹500 crore growing 25% a year is plausible. A company with revenue of ₹5,00,000 crore growing 25% a year for a decade would need to add the equivalent of several large listed companies. Growth gets harder as the base gets bigger, always.
If the implied growth survives all 4, the price is defensible. If it fails 2 of them, you have learned something specific, and you have learned it without pretending to know the future.
Try this now
Five minutes, 1 holding you already own. A calculator is enough.
- Find your holding's enterprise value. Market capitalisation, plus total borrowings including lease liabilities, minus cash and current investments. All 3 numbers are in the latest results filing.
- Find free cash flow for the most recent full year. Open the cash flow statement. Take net cash from operating activities and subtract purchase of property, plant and equipment — the capital expenditure line inside investing activities. Do this for 3 years and take the average, because 1 year of capital spending is lumpy.
- Divide average free cash flow by enterprise value. Write it down as a percentage. That is the free cash flow yield.
- Choose your required return. If you have no view, use 11% for an Indian company and 8% for a US company as a starting point. Subtract the free cash flow yield from it.
- Read the result out loud as a sentence: "At today's price, the market is assuming this company grows free cash flow by roughly ___% a year, forever." Then answer 1 question: do you believe it?
What you should see. One of 3 outcomes, and all 3 are useful.
The implied rate may come out low — 2% to 4% — meaning the market expects this business to barely grow. If you think the business will do better than that, you have found a disagreement worth investigating, and you found it without forecasting anything.
The implied rate may come out close to the economy's nominal growth. That is the market saying "an ordinary good business". Fair enough.
Or the implied rate may come out above nominal GDP growth, which means the price requires the company to keep taking share from everybody else indefinitely. Sometimes that is a reasonable bet on an exceptional business. It is never a cheap one, and you should now be able to say what would have to go right.
If free cash flow is negative, the arithmetic breaks. That is also information: this business cannot currently be valued on the cash it produces, so its price rests entirely on cash it has not yet produced.
Three real cases
1. Cisco Systems, March 2000 (United States) — the growth rate nobody said out loud At its peak in late March 2000, Cisco was briefly the most valuable listed company in the world, with a market capitalisation of roughly $500 billion to $550 billion and a share price near $80. Cisco was a real business with real profits and a genuine technology lead. That was never the question. The question a reverse calculation would have asked is what growth rate that price required — and the answer was a rate of profit growth that, sustained for a decade, would have made Cisco larger than plausible estimates of its entire addressable market. The company continued to grow revenue for years afterwards. The share price did not return to that level for more than 20 years. Nothing about the business had to fail. Only the assumption had to be unattainable, and it was.
2. Reliance Power, January and February 2008 (India) — a valuation made entirely of terminal value Reliance Power's initial public offering opened in January 2008 and raised approximately ₹11,563 crore, the largest in Indian history at the time. The company's power projects were largely not built. Almost all of the value being paid for sat in cash flows from plants that did not yet exist, which is another way of saying the entire valuation was terminal value. The shares listed on 11 February 2008 above the issue price and closed the first day sharply below it. When 100% of a valuation depends on the far future, the model is not measuring a business. It is measuring an intention.
3. Nokia, 2007 to 2013 (Finland, listed in the United States) — the terminal value assumption that was the whole answer At the end of 2007 Nokia held roughly 40% of the global mobile phone market. Any reasonable discounted cash flow model built then would have shown a large, safe terminal value, because the assumption behind terminal value is that the business survives in recognisable form. Revenue fell from about €51 billion in 2007 to about €12.7 billion in 2013. The forecast years were not the problem. The terminal value was, and the terminal value was 1 assumption: that the franchise endures. Every DCF you ever build contains that same assumption, usually unexamined.
The question that resolves it
A novice runs a model and asks: what is this company worth?
An expert runs the same model and asks: what does this price require to be true, and would I bet on that?
The first question has no reliable answer, and pretending otherwise produces false confidence. The second question always has an answer, the answer is checkable against history, and it is where the useful part of valuation lives.
What would make this wrong
If reverse-solving for implied growth reliably identified mispricing, then buying every share whose implied growth looked too low would beat the market. It does not. The market frequently assumes low growth for a business that then delivers even less. A low implied growth rate is often correct.
The honest limits of the method in this article are specific. The single-stage formula assumes smooth growth from today, which no real business has; it is wrong at cycle extremes in both directions. It requires positive and reasonably stable free cash flow, which excludes early-stage companies, most banks, and anything mid-restructuring. It uses your required return, which is a personal choice, so 2 honest readers will get different implied growth rates for the same company. And the free cash flow yield can be distorted for a year by a large one-off capital project, which is why you average 3 years.
The strongest argument against the whole approach: implied growth tells you what the market assumes, not whether the market is wrong. Most of the time the market is not wrong. The value of the exercise is that it converts a vague feeling about expensiveness into a specific claim you can test — and it tells you loudly when a price requires something arithmetically impossible.
In India
Where the inputs come from. Free cash flow comes from the consolidated cash flow statement in the annual report, which listed companies file with the NSE and BSE and which is also available on the Ministry of Corporate Affairs portal. Indian Accounting Standard 7 governs the cash flow statement. Use consolidated figures. For a group holding company the standalone statement will show dividends received, not the operations that produced them.
The discount rate has a higher floor. India's 10-year government bond yield has generally run well above the equivalent US Treasury yield. Since the risk-free rate is the base of any discount rate, the same business modelled in India carries a higher discount rate and therefore a lower present value than the same business modelled in the United States. Importing a US discount rate into an Indian model will overvalue everything you look at.
Nominal growth is higher, which raises the ceiling on g. Because India's nominal GDP growth has been substantially higher than that of the United States, a perpetual growth rate that would be impossible in a US model can be defensible in an Indian one. Both halves must move together. A model with a US discount rate and an Indian growth rate is not conservative. It is incoherent.
Capital expenditure is often lumpy and disclosed thinly. Many Indian companies build capacity in large steps rather than continuously. Three-year averaging of free cash flow is not optional here. And check the notes for capital work in progress, which is spending that has occurred but not yet become a productive asset.
Related-party dealings distort the cash flow you are discounting. Where a listed company lends to, buys from or sells to unlisted entities controlled by the same promoter family, part of the cash economics sits outside the entity you are valuing. Read that note before you trust the number.
In the United States
The inputs are easier to find and better standardised. Form 10-K carries the consolidated cash flow statement, and the SEC's EDGAR database provides the full filing history free. Many US companies also state free cash flow directly in their earnings release, though their definition of it may differ from cash from operations minus capital expenditure. Check the definition rather than assuming it.
Stock-based compensation is the trap in US free cash flow. Shares issued to employees as pay are added back in the operating section of the cash flow statement, because no cash left the company. But the cost is real: existing shareholders own less. Many technology companies show strong free cash flow mainly because a large part of their wage bill was paid in shares. A defensible correction is to subtract stock-based compensation from free cash flow before computing the yield, or to check the share count over 5 years and see how much was quietly transferred.
Buybacks change the per-share arithmetic. A company can hold total free cash flow flat and still grow free cash flow per share every year by shrinking the share count. When you compute historical growth to compare against your implied rate, decide which one you are measuring — total or per share — and use the same one on both sides.
Analyst consensus is published, which makes it easy to see what professional forecasters expect. Use it as a comparison against your implied rate, not as a replacement for it. Consensus forecasts are least accurate exactly at the turning points where the answer matters most.
Where they differ, and what that tells you
The same business is worth less in an Indian model than in a US model, and more in an Indian model, at the same time. That is not a contradiction. The discount rate is higher in India, which lowers value. The sustainable growth ceiling is higher, which raises it. The 2 effects partly offset, and how much they offset is the entire debate about whether Indian shares deserve their historically higher multiples.
The practical instruction: keep the discount rate and the growth ceiling in the same country. The most common error in Indian valuation work is a model built with an American textbook's 8% discount rate and an Indian company's double-digit growth expectation. That combination produces enormous values and it is not a view about the company at all. It is an arithmetic mistake.
Free cash flow is more reliable as a starting point in the United States, for 2 reasons. Segment disclosure is deeper, so you can see which part of the business produces the cash. And the reconciliation requirements for non-standard measures are stricter, so the gap between the company's claimed figure and the statutory one is disclosed. In India the reconstruction is more manual, and related-party complexity means the boundary of the business is less clean.
The consequence for the reverse question is asymmetric. In the United States, a reverse calculation is mostly a test of whether the growth assumption is achievable. In India it is that, plus a prior test: is the cash flow you are discounting actually the cash flow the shareholder will receive? Answer the second question first.
Carry this
- A DCF is arithmetic wrapped around guesses, and the terminal value is usually most of the answer.
- Do not solve for value. Solve for the growth rate today's price assumes: implied growth = your required return − the free cash flow yield.
- No company grows faster than its economy forever. If the implied rate is above long-run nominal GDP growth, the price is claiming something impossible.