A ratio is a question, not an answer

Someone opens a screener and sees, next to a company's name, a single line: return on capital employed, 22%. They stop reading. Twenty-two percent beats a bank deposit and beats most businesses, so they mark the company high-quality and move on. They asked "is this a good business?", the screen said "22%", and they heard "yes." illustrative

But the ratio did not say yes. It could not. All it did was take two numbers from the accounts - operating profit on top, capital employed underneath - and divide one by the other. That figure is not a verdict; it is a prompt. It says: the profit this business earns on the money tied up in it is 22% - now go and ask why, and whether that will last. The reader who stops at "22%" has taken delivery of the question and mistaken it for the answer. That is the most common error in the whole of ratio analysis, and this module installs the habit that prevents it: reading every ratio as the question it really is.

A ratio exists to compress. The accounts hold hundreds of numbers; no one can hold hundreds of numbers at once. So a ratio takes two that belong together - profit and sales, debt and equity, price and earnings - and turns their relationship into one figure you can hold. That is genuinely useful: it lets you compare a giant with a minnow, this year with last, one firm with its rival, on a single scale. The danger lives in the same compression. Turning a relationship into one clean number makes it look like a conclusion, because conclusions are also clean and single, so the mind - which likes to be finished - treats the number as the end of the enquiry when it is only the start.

Every ratio has a question folded inside it. ROCE asks how hard the capital is working. A margin asks how much of each rupee of sales survives to this line. A leverage ratio asks whether the business can carry its debt if times turn bad. A valuation multiple asks what the market is paying for each rupee of earnings, and what must come true to justify it. To turn the number back into that question, you ask three things of it, always in the same order.

Compared with what? A ratio on its own has no meaning. Twenty-two percent is high only against something - the same company three years ago, its direct competitors, the cost of the money the business uses. A ratio is a relative instrument pretending to be an absolute one; a single value with nothing beside it is a temperature with no scale, where 22 could be freezing or boiling.

Trend or level? A level is a snapshot; a trend is a story. Two companies can post the identical ratio this year, one having climbed to it and one having fallen to it, and they are not the same business at all. The direction of travel usually tells you more than the height reached, because it points at what is happening - a moat widening or narrowing, discipline tightening or slipping. When the level and the trend disagree, believe the trend. The single year's level is the noisy thing to anchor on; the direction of travel is where the signal lives.

Which two numbers went in? A ratio is only as honest as its numerator and denominator. A high ROE can come from a rising numerator (more profit) or a shrinking denominator (less equity, through buybacks or write-offs) - opposite meanings for the business. A falling ROCE can be a worsening firm, or a new factory sitting in the accounts earning nothing yet. Take the ratio apart before you trust it; half the time the interesting fact is in one of the inputs, not the quotient. Do those three in order and the single number unfolds back into the question it was hiding. Skip them and you hold an answer-shaped object that answers nothing.

A ratio does not even ask the same question in every industry. Here is inventory days - how long stock sits before it sells - carried across four businesses. In some it is a sharp, central question. In others it is barely a question, and reading it as if it mattered will mislead you.

So a ratio is a question in two senses at once. Even where it applies, the number it returns is only the opening of an enquiry. And whether it applies at all depends on the business - the same formula is a keen question for a consumer firm, an inverting one for a cyclical, a triviality for a services firm, and simply the wrong question for a lender. That is the seed of an idea a later session grows into a full skill: every ratio is built on an assumption about how the business is shaped, and the moment a company breaks that assumption, the ratio goes quietly dark while still printing a number. For now, hold the smaller version - never read a ratio without first asking what it is trying to ask of this business, and whether that question makes sense here.

Take a composite mid-size manufacturer. The screener line reads ROCE 19%, operating margin 16%, debt-to-equity 0.6 - every figure respectable, and the temptation is to nod and conclude "solid business." Instead, run the three questions and watch each number unfold. illustrative

Start with the 19% ROCE. Compared with what? Three years ago it was 26%, and the two closest listed peers earn 23% and 25% today - so the same 19% now reads as a business both declining against its own past and trailing its rivals. Which inputs moved? Operating profit is roughly flat, but capital employed has jumped: the company built a large new plant that is only part-loaded, so the denominator swelled while the numerator has not caught up. That single fact reframes the 19% completely. It is not clearly a worsening business; it may be a business mid-investment whose returns are temporarily diluted, and the honest read is to recompute the ratio on the assets actually running and watch whether the new plant fills. The number told you none of that. The question behind it did.

Now the 16% operating margin. Trend or level? It has slipped from 20% over two years. Which input? Revenue is growing, but raw-material cost has risen faster and the company has not passed it through in price - a sign either of weak pricing power or of deliberate share-buying at the expense of margin. Either way, "16%" was never the point; the two-year slide and its cause are. And the 0.6 debt-to-equity is comfortable for a steady manufacturer, but this one's end-markets are cyclical, so the real question is whether 0.6 stays comfortable when the cycle turns and cash flow thins. Three respectable numbers, and every one of them was a door, not a room. The habit to build is to walk through each door rather than read the label and move on.

When management leads with a flattering headline ratio, a sharp analyst pushes past the number to the question underneath. It sounds like this: "You've highlighted the 22% ROCE - can you bridge it? How much is the core business improving, and how much is the new capacity not yet in the capital base? And where does it sit versus three years ago?" The analyst is refusing to take the ratio as an answer and is prising it back open into its question.

A good answer takes its own ratio apart:

"Fair to unpack it. Of the 22%, the underlying business is running around 18%, up from 15% two years ago on better pricing and mix. The reported number is higher partly because we wrote down an old asset, which shrank the capital base - so I'd point you to the 18% as the real trend, not the 22%. On a like-for-like basis versus three years ago, returns are genuinely up, and here's the walk."

It separates trend from level, names the input that flattered the headline, and hands you the number that actually answers the question. An evasive answer does the opposite:

"We're very proud of our 22% ROCE - it's among the best in the sector and reflects the quality of our franchise and our relentless focus on execution. We're confident we can sustain these industry-leading returns going forward."

It repeats the level, claims sector leadership without the peer numbers, and offers "quality" and "focus" in place of a decomposition. It is selling the ratio as an answer - precisely the move this module trains you to distrust. The follow-up that settles it is the one nobody asks: "Strip out the write-down and the under-utilised plant - what is ROCE on the assets actually operating, and what was it three years ago on the same basis?" Watch what happens if it is never asked. If "22%, best in the sector" is allowed to stand unbridged, either the underlying trend is worse than the headline and the silence is convenient, or the analysts who would have decomposed it have stopped doing the work. A room that accepts a ratio as an answer is a room that has forgotten it was ever a question.

A ratio, read as a question, points you at where to look. It never tells you what you will find when you get there. It cannot tell you why a margin fell - input costs rose, or the firm cut prices to win share, or the mix shifted toward a cheaper product. It only tells you that it fell, and sends you to the notes and the management commentary to find the cause. The number is a signpost; the explanation is a separate search the signpost cannot make for you.

Nor can a ratio tell you whether its own inputs are honest. If the profit on top is flattered by an accounting choice, or the capital underneath has been quietly shrunk by a write-off, the ratio computes cleanly and misleads completely. It inherits every distortion in its inputs and adds a coat of respectability, because one tidy percentage looks more trustworthy than the messy figures behind it - and that tidiness is not evidence the inputs were clean. Whether they can be trusted is the subject of a later part; the ratio itself is silent on it.

And a ratio cannot supply the comparison it needs. It cannot know who this company's genuine peers are, or where in its cycle it sits, or what its own history looks like - you bring those. This is where the last trap waits: a screen drops a bank, a cyclical and a software firm into the same ROCE or P/E column as if the number meant the same in each, and it does not. Choosing the true peer set, the mid-cycle level, the relevant span of history is a judgement the number cannot make for you. Give it the wrong comparison and it will answer the wrong question with the same calm confidence - which is why the point of a ratio is to tell you where to look next, never to save you from looking.