Thanks this website
Thanks this website

So this happened:

  1. Two weeks ago, Citigroup announced profits of $3.4 billion for the months of July, August and September.
  2. This week, Citigroup was all “Totes awkward – it’s actually $2.8 billion.”
  3. People-not-in-the-know asked why.
  4. People-in-the-know said it’s because of “rapidly-evolving regulatory inquiries and investigations” in which Citigroup has, perhaps, manipulated foreign currencies.
  5. Meaning that, suddenly, October’s events are driving down 3rd quarter profits.
  6. Which has a few people going “$600 million of profit just disappeared from the third quarter? Finance is magic”.

Many of the accountants in the room will be responding with something along the lines of:

  1. IAS 37.
  2. Subsequent events accounting.
  3. Yeah, we agree that the rules are strange. But they are what they are. You should speak to the FASB. Or the IASB. Perhaps they can tell you why.

Not the most satisfying of answers.

But the thing is, as surprising as it might seem, this accounting treatment – this changing of a 3rd quarter’s results for some events that happened in the 4th quarter – is an unavoidable consequence of having an accounting system that is mostly useful and sufficiently sensible.

A Step Back In Time

Back when the Egyptians (or whoever) first started keeping records of the Nile movements, or sheep, or taxes, you didn’t need a number system. A small sheep on a clay tablet meant one sheep. A large sheep meant a flock of sheep. And it could all be fairly approximate. Especially when everyone was bartering, and money wasn’t really a currency so much as it was bushels of wheat, or dried fish, etc.

Then we got more evolved, and started trading in earnest, and numbers were discovered.

Fast forward to the Venetians, and we suddenly found ourselves using double-entry accounting to keep track of what went in and what went out and what happened in between.

At that point, accounting could probably still have been a set of rules. Such as “When you buy a load of silk for one ingot of gold, you put a 1 load of silk in the lefthand column of the silk ledger, and you put 1 ingot of gold in the righthand column of the gold ledger.”

Then it got complicated.

Somewhere along the line, presumably because someone started trying to buy on credit (those usurious merchants!), the question became “Well, when do I make those entries? Because I’ve got a load of silk, but I’ve still got the gold because I haven’t paid them yet. What do I do?”

And the accounting world went “Oh” and invented the accrual basis.

The Accrual Basis of Accounting

Which basically says: there is no such thing as a single transaction.

To take the silk/gold example, the accrual basis says that in this economic event there are actually two transactions taking place:

  1. There is the buying of the silk (the purchase).
  2. And there is the paying for it (the settlement).

And this idea would be a monumental philosophical shift.

On the “Cash Basis” (which is what came before), there was only one transaction. But under the “Accrual Basis” – paying for something outright was just one of those curious anomalies whereby the purchase and the settlement coincided.

In many ways, this new understanding gave us the ability to develop a whole range of financial structures that eventually led to futures markets and derivatives and the modern banking system. Of course, it might also just have coincided: the development of credit required a way of describing it, etc. But I think that it’s probably a mix of both: the accounting-finance language was used to describe new financial ideas, leading to better financial conversations, leading to more ideas, leading to more development in the language, and so on.

That said, like almost any general principle, when you push it outwards to a very specific logical conclusion – it can quickly get illogical. And often, self-contradictory.

An analogy:

  1. General principle: “All human beings are worthy of respect and therefore deserve to be treated kindly.”
  2. Specific circumstance: Mass-murderers.
  3. Illogical conclusion: “You have to be cruel to be kind” (by locking them up). Or “The rights of the whole over the rights of the few” (by executing them).

The question is: does self-contradiction at the edges make the general principle a bad one?

Bringing It Back To Citigroup

Before I get to the issue of court cases and doing bad stuff that gets you fined (it’s a thorny accounting problem), let me pose a hypothetical:

  1. Let’s say that Citigroup had bought a new head-office building for $10 billion, but the financial people made a finger-error and recorded the building at only $1 billion.
  2. Financial results are released, showing the building at $1 billion.
  3. Citigroup realises this.
  4. Would you rather that they:
    1. Went back and restated the financial results to record the building at $10 billion?
    2. Or that they add $9 billion to the building in the next round of results, and record the difference as a fair value gain?
  5. Well, the better answer has to be “A”. Because otherwise, you’ll have two completely useless sets of financial results on your hands.
  6. But also because it makes sense – a mistake was made, you correct it. The building was always worth $10 billion. You want to get the financials right at the time.
  7. Happily – this is what the accrual basis demands from you.
  8. It doesn’t say “Oh, well, there was a buying event of $1 billion, and a discovering-the-mistake event that led to a $9 billion gain”.
  9. It says “There is only one relevant event related to the building – the $10 billion purchase. Any subsequent events relate back to that purchase – and therefore, you have to go back and adjust your results.”

Another example:

  1. Let’s say that Citigroup closes a major deal and bills a giant $250 million fee.
  2. But the accounting people don’t get the memo, and so they don’t record it.
  3. Citigroup releases their results.
  4. They realise that the $250 million fee is missing from revenue.
  5. Would you rather that they:
    1. Record it in the next period?
    2. Or that they go back and restate?
  6. Again, the answer has to be “go back and restate”.
  7. And this time, it’s not just a question of principle.
  8. It’s because you know that if you allow forgetful mistakes to be made without correction, you’ll end up with all kinds of “mistaken” revenue transactions being recorded twice, or not recorded at all, especially in and around bonus time.

The Awkwardness of Accounting For the Bad Stuff

Up to this point, everyone is usually quite happy. You need to maintain a consistent principle in the preparation of financial results because:

  1. It makes them meaningful
  2. It makes them comparable to history (you can compare last year to this year, etc).
  3. It makes them comparable to other companies (if all banks work off the same principle, then you can compare Goldman to Citigroup, or Goldman to JP Morgan, etc).

Also, you have to insist on correction of mistakes – because the alternatives include:

  1. Misleading financial information; and
  2. Deliberately misleading financial information.

Now, finally, the Citigroup manipulation of financial currencies

The thorny question:

Why is Citigroup recording a fine?

Two possible answers:

  1. Because Citigroup did something wrong; or
  2. Because Citigroup got caught.

If you try to justify the second option on its own, you run into problems. Because if you want to be logically consistent with the two earlier examples, then the second option sounds a lot like:

  • “the only reason the building is worth $10 billion is because they realised their mistake” and
  • “the fee only came into existence when they realised that they hadn’t recorded it.”

Both of those are nonsensical statements. The fee came from closing the deal. The building was worth $10 billion because that’s what it was bought for.

So you’re almost forced to conclude that the fine came about because Citigroup did something wrong.

Then it gets even trickier

Because once you hit that point of saying “Well, actually, the main thing here is that Citigroup did something wrong” – the question then becomes “Well then should we go back and correct everything historically because now we’re caught and we’ve realised our mistake in not raising it?”

So, fortunately, the consensus answer is “no”.

Here is why:

  1. A year ago, Citigroup wasn’t expecting to pay a fine for doing something wrong.
  2. The financial reality at the time was that this event had no monetary consequence (ie. the expected value of the obligation was worth about zero). So there was nothing to record.
  3. But once investigations began, it became clear that the event would have almost certainly have a monetary consequence (ie. the expected value was no longer zero).

So this is a grey area. Unlike the purchase of the building, you are now looking at three key events:

  1. The originating event (doing bad things);
  2. The confirming event (which makes it clear that the doing of the bad things is going to result in some punishment); and
  3. The payment of the fine.

From an accrual perspective, you need both the first and the second events to have occurred before you can record anything: the first is the cause, and the second gives the cause a probable economic value.

If you’re still following the argument, what had happened by the end of the third quarter:

  1. Citigroup had done some bad things with currencies (the originating event); and
  2. Citigroup was being investigated, meaning that it was already highly probable that they would have to pay a fine (the confirming event).

At that point in time, under the accrual concept, the fine had already happened. Sure – it hadn’t been paid. The final amount hadn’t quite been determined. But that’s a question of payment, not a question of economic reality.

And the most likely economic reality was: Citigroup already had an obligation to pay a fine at the end of the third quarter.

So Then Why Change The Value In October?

Once Citigroup decided that it had to accrue for the fine – it then made a call about how much they thought the fine would cost them.

In October, they realised that they had made a total mistake about how high that fine would be. Not because anything had necessarily changed, but because they realised that they had done it wrong. Basically, the equivalent of recording the purchase price of the building at the wrong value.

So in order to be consistent – they restated their results.

Does it sound slippery?


But what the accrual basis really does is separate events in time. And sometimes, you get these absurd moments that look a whole lot like time travel.

But the alternatives are:

  1. A Cash Basis – which is already thrown in for free (the Statement of Cash Flows); or
  2. Unique Accounting for each time and circumstance – which would render the whole thing inarticulate and opaque.

In any case, if you, as a reader, decide that this particular treatment is nonsense and unhelpful – then you are usually given enough information to reverse it out and reconstruct the financials as you think they should be.

Rolling Alpha posts opinions on finance, economics, and the corporate life in general. Follow me on Twitter @RollingAlpha, and on Facebook at