Reading someone else's accounts

A set of accounts is the most reliable description of a company available to somebody who is not inside it. Somebody spent months assembling it, somebody independent tested it, and it follows a shape you can learn once and then read anywhere. That is a great deal more than you get about most things you have to make decisions about.

What it is not is a complete account of the business. It reports a period that has closed and a day that has passed, at a level of detail chosen for a general reader, using measurement choices made by the people who prepared it. Reading it well means keeping both halves of that in mind: taking it seriously, and knowing precisely which of your questions it was never built to answer.

• The face of the statements. The three statements from session 1, each a page or two of totals. This is what gets quoted and it is the smallest part of the document.
• The notes. Everything behind those totals: the measurement choices, the breakdowns, what is committed but not yet spent, what happened after the period ended. Usually the bulk of the pages.
• The auditor's report. A short statement of what was checked, to what standard, and with what conclusion. Worth reading in full because it is short and because the exceptions live in it.
• The narrative at the front. Written by the company about itself. Useful for understanding what management thinks matters, and not a source of verified numbers.
• Comparatives. The previous period printed alongside the current one, which is where most of the information in a single set of accounts actually comes from.

A standard settles how things are measured and how they are presented, so that the revenue line in one company means roughly what the revenue line in another company means, and so that this year sits next to last year honestly. That consistency is the whole value: without it, comparison is impossible and every set of accounts would need its own translator.

What a standard does not do is remove judgement. It narrows the range of defensible answers and then leaves the choice inside that range to the people who know the business, which is the only place the choice can sensibly sit. Two well run companies in the same trade can therefore report different profit on identical activity, with nobody at fault.

An audit is an independent test of whether the statements, taken as a whole, are free from error large enough to change the conclusions of a reasonable reader. That threshold is the crux. Something can be too small to matter to the accounts as a whole and be precisely the thing your decision turns on: a single customer balance, one contract, the line you happen to be exposed to.

An opinion is also a conclusion about the statements, not a verdict on the business. A company can be audited, clean, and in serious difficulty, and a company can be unaudited and perfectly healthy. Read the auditor's report anyway, in full: it is short, it says what the standard was, and anything the auditor felt the need to draw attention to is in there.

• A statement that the business will survive. The accounts describe a period that has ended. Survival is a forecast, and forecasts are not in the document except where the preparers chose to discuss them.
• A guarantee that every figure is right. The test is applied to the whole. Small things are not individually certified, and small to the accounts may be large to you.
• Enough detail to answer a specific question. A single line can contain the answer and hide it. The notes are where the line is broken apart, and if the breakdown you need is not there, the document does not have it.

A single line on the face of a statement is an aggregate, and an aggregate can be built in several defensible ways. The notes say which way was chosen and what went into it. They are where a revenue figure becomes revenue by type, where a receivables figure becomes receivables by age, and where a cost line splits into the parts that recur and the parts that will not.

They also carry things that have no home on the face at all: money the company has committed to spend but has not spent, obligations it has guaranteed for someone else, dealings with parties it is connected to, and events after the period ended that a reader ought to know about. None of that changes a total. All of it changes what the totals mean.

Session 2 made the point that profit is an opinion. In your own accounts you can see the opinion being formed. In somebody else's, the same opinions have already been formed and then compressed into a total, and the only trace is a paragraph in the policies note.

The recurring ones are worth knowing by name. When revenue is recognised, which decides whether a contract lands in this period or the next. What is treated as a cost now and what is treated as an asset to be spread over later years. How long an asset is spread over: Marchmont writes its equipment off over 60 months, and a company spreading identical equipment over a longer life reports more profit each year with no difference whatever in the cash. How much of what customers owe is judged uncollectable. What unsold stock is judged to be worth. Each of these is a legitimate choice inside the standard, each moves profit, and none of them is visible unless you go and look.

How much judgement should a standard leave in the hands of preparers? This is an old argument between two coherent positions, and it has not been settled.

One position: narrow it. Judgement is where comparability dies and where pressure gets applied, and a reader outside the company cannot audit a state of mind. A tighter rule produces numbers that may fit some businesses badly but that mean the same thing everywhere, which is what an outside reader needs most.

The other position: keep it. A rule tight enough to remove judgement will describe some businesses wrongly, and a wrong number that is comparable is worse than a right number that needs a note. Management knows which contracts are real and which stock will sell, and a standard that refuses to use that knowledge throws away the best information in the building.

Both sides are right about what the other side costs. The practical consequence for a reader is the same either way: the policies note is not boilerplate, it is the terms on which every figure in the document was produced.

Age. The balance sheet describes one specific day, and that day is behind you before the document exists. The books have to be closed, the figures assembled, an audit performed where there is one, and the document approved and filed. By the time a copy reaches you the gap is measured in months, and the company has been trading for all of them. Nothing in the document tells you what it did with that time.

Size. How much you are given depends on how large the company is. Above a size threshold you can expect a full set with extensive notes and an audit. Below it, in many places, a company may file a reduced set: fewer notes, less breakdown, sometimes no cash flow statement and sometimes no audit at all. The thresholds and the exact reliefs differ by jurisdiction and change over time, so check what applies where the company is registered rather than assuming. The practical point is that a small supplier may legitimately give you two of the three questions from session 1 and nothing behind them, and that the missing statement is most often the one on cash.