The short answer: a dividend is safe when the business generates more cash than it pays out, in a bad year, without borrowing. Everything else — the payout ratio, the decades-long increase streak, the reassuring language in the annual report — is a proxy for that one question, and every proxy can be gamed. What follows is the order we check it in, and what each step actually tells you.
It matters because the cut is not the loss. The loss is the price, which falls long before the announcement, and the tax bill on selling afterwards, and the years of compounding that do not happen. Getting this right is mostly about refusing a handful of obvious traps.
Start with cash, not earnings
The payout ratio is a good question asked of the wrong number
Almost every screener shows dividends as a percentage of earnings per share. Earnings are an accounting opinion: they carry depreciation schedules, write-downs, share-based compensation and one-off charges that have nothing to do with whether the cheque clears. Dividends are paid in cash.
So compute the ratio against free cash flow — operating cash flow minus capital expenditure — and you get an answer you can trust. A company paying 55% of earnings but 110% of free cash flow is not paying you out of profits. It is paying you out of the balance sheet, and that has a maturity date.
The one adjustment worth memorising is that different structures report their spare cash differently. Ask for the number the business itself is run on:
| Type of business | Measure the dividend against | Why |
|---|---|---|
| Ordinary company | Free cash flow | Earnings carry non-cash charges; the dividend does not. |
| REIT | AFFO (adjusted funds from operations) | Property depreciation crushes reported earnings without consuming cash, so an EPS payout ratio above 100% is normal and meaningless. |
| Bank or insurer | Earnings, plus the regulatory capital ratio | Cash flow statements are near-useless here; the binding constraint is the regulator, not the cash. |
| Cyclical (energy, mining, autos) | Free cash flow at mid-cycle prices | The payout looks trivially covered at the top of the cycle and impossible at the bottom. Neither is the real number. |
| BDC or fund structure | Net investment income | Distributions are mandated by structure; what matters is whether they are funded by income or by returning your own capital. |
Read the balance sheet before the yield
Debt does not cut the dividend. Debt coming due does.
Dividends are cut when a company needs the cash somewhere else, and the most common somewhere else is a maturing bond. A business with a comfortable payout ratio and a wall of debt refinancing into a higher-rate market is in more danger than one with a stretched ratio and nothing due for a decade.
Three things to look at, in this order: net debt to EBITDA (roughly, how many years of profit it would take to clear the debt), interest cover, and the maturity schedule. The first two are ratios you can compare across an industry. The third is the one that actually sets the date.
A dividend is a promise made with money the lenders have first claim on.
Ask what the payout looks like in a bad year
Not a forecast — a stress test you can do in two minutes
You do not need a model. Take the worst revenue year in the last decade and the margin that came with it, apply both to today's business, and see whether the dividend still fits inside the cash. For most quality companies it does, with room to spare. For the ones that make you nervous, the arithmetic usually stops working somewhere in the second step — which is the answer.
For cyclical businesses this is the whole exercise. The useful question is not "is the dividend covered?" but "at what commodity price does the cover break?" If the answer is a price the market has seen twice in ten years, that is not an income stock. It is a bet with a coupon attached.
Read the history as behaviour, not as a guarantee
What a company did in 2008 and 2020 tells you what it will do next time
A long record of increases is real information — it tells you the board treats the dividend as a commitment rather than a residual. But treat it as evidence about people, not about cash. Managements defend streaks, and the ways they defend them are not all healthy: underinvesting in the business, selling assets, borrowing to pay you. A streak that survives on borrowed money buys a few more years and then costs you the cut anyway, with a weaker company on the other side of it.
The more useful question: what did they do in the last real crisis? A company that held the dividend through 2008 or 2020 while its peers cut has told you something durable about both the business and the board. A company that cut has told you where the dividend sits in its list of priorities — which is worth knowing, and is not automatically disqualifying.
Check who the dividend is for
The most under-rated signal in income investing
Ask who receives the money. A controlling family, a founder, a pension fund whose obligations are paid out of this specific payout — these are owners with a powerful reason to keep the dividend intact, and they usually have the votes. Contrast that with a company where the dividend was introduced to put a floor under the share price after a bad run: that payout is a marketing expense, and it gets cut like one.
This is also the point where the payout ratio stops being a safety metric and becomes a capital-allocation question. A business earning high returns on capital that pays out most of its cash is quietly telling you it has run out of things to do with it. That can be perfectly honest — some great businesses genuinely cannot reinvest — but it caps your future return at roughly the yield plus inflation, and you should price it that way.
If you only have a moment: dividend versus free cash flow, net debt to EBITDA, what happened in the last recession. Two out of three clean is usually enough to keep looking. Two out of three ugly is usually enough to stop.
What this looks like on a real company
Three payouts we have taken apart in full
The checks above are quick to describe and slower to apply, so it helps to see them run end to end. Each of these reports walks the same ground for one company — the cash cover, the balance sheet, the record through the bad years and the verdict that falls out of it.
Business, moat, management, the numbers, valuation and an honest verdict.
What to do with all this on Monday
Pick the highest-yielding position you own and put it through the five-minute version. If it passes, you have bought yourself the right to ignore the next scary headline about it — which is most of what this work is for. If it does not, you have found out now rather than on the morning of the announcement, when the price has already moved and your options have narrowed to one.
The purpose of all this is not to avoid every cut. It is to make sure that when one arrives, it is a surprise to the market and not to you.
Frequently asked
What is a safe payout ratio?
There is no single number, because the right denominator changes with the business. As a rough anchor: under roughly 60% of earnings for an ordinary industrial or consumer company, under roughly 75% of free cash flow for anything capital-hungry, and under roughly 85% of AFFO for a REIT. What matters far more than the level is whether the ratio was built out of cash the business actually generated, and whether it holds up in a bad year rather than a good one.
Does a long streak of dividend increases mean the dividend is safe?
It means management has been unwilling to cut, which is useful information — but it is a statement about their behaviour, not about their cash flow. Streaks are defended long past the point of prudence, sometimes with borrowed money. Treat a long record as a reason to look harder at the cash cover, not as a substitute for looking.
Is a very high yield always a warning sign?
Nearly always, yes — with one honest exception. Some structures (REITs, BDCs, certain tobacco and telecom businesses) legitimately pay out most of what they earn, so a high yield is normal for them and should be compared against their own peer group, not against the wider market. Outside those cases, a yield far above a company's own history usually means the price has fallen because the market expects a cut.
What is the single best early warning of a dividend cut?
A dividend being funded by something other than operating cash: rising debt, asset sales, or a payout that exceeds free cash flow for more than a year or two. The cut is usually announced long after the cash stopped covering it.