What.

Banks Keep Forgetting How Much Money They Made Last Quarter

Matt Levine is a Bloomberg View columnist. He was an editor of Dealbreaker, an investment banker at Goldman Sachs, a mergers and acquisitions lawyer at Wachtell, Lipton, Rosen & Katz and a clerk for the U.S. Court of Appeals for the Third Circuit.
Read More.
a | A

A few weeks ago I did some hand-waving about how banks are mysterious creatures, and how a bank's "income" is just its best guess at the midpoint of a wide probability distribution of how much money it might actually have made, filtered through one of several semi-arbitrary accounting conventions. I used the specific example of Bank of America, which had just announced net income, under U.S. generally accepted accounting principles, of $168 million. That number represented roughly 0.008 percent of its assets, or 0.009 percent of its liabilities, meaning that if its measurements of its assets or liabilities were off by just a teensy little bit, its income for the quarter would vanish.

Guess what happened!

Subsequent to the company’s earnings announcement on October 15, and prior to the filing of the company’s Quarterly Report on Form 10-Q for the quarter ended September 30, 2014, the company has been engaged in separate advanced discussions with certain U.S. banking regulatory agencies to resolve matters related to its foreign exchange business. As a result of those discussions, the company recorded a $400 million non-deductible charge and adjusted its third-quarter 2014 financial results to a net loss of $232 million or $(0.04) per share.

Bank of America's third-quarter profit lasted for just three weeks, and then vanished back into the depths whence it came. For myself, I will always remember it fondly, and I suspect many people at Bank of America will join me in mourning its passing. Here is the 10-Q, scrubbed of that fleeting profit.

What is going on? Bank of America is the second bank to revise its earnings for last quarter to reflect increased legal expenses; Citigroup did the same thing last week, knocking $600 million off its net income. When that happened I mostly boggled at how weird it was. The foreign-exchange matters that BofA and Citi blamed for their problems have not yet been made entirely public, but they do seem to have ended in 2013 at the latest. The nine-or-more-digit fines and settlements that the banks will pay to resolve those problems will occur, at the earliest, tomorrow, and that is frankly pushing it. Nothing relevant to these charges happened in the third quarter of 2014. And yet Bank of America and Citi combined to lose a billion dollars in the third quarter, based on those non-events.

What seems to have happened is that U.S. regulators sat all the big banks down in late October and told them how much they'd be paying. For some banks, this was more than they were planning to pay. (For others, not: JPMorgan and Goldman seem to have filed 10-Qs without incident. ) These conversations occurred after the third quarter, but their effects nonetheless bled back into the third quarter. With bank earnings it's not only the earnings that are probabilistic. Time itself is blurry. Something that happened in late October might as well have occurred in late September. There's no need to get hung up on details like that. These are bank earnings we're talking about.

The way the rules work, when a loss becomes probable, and you can reasonably estimate its size, you need to reserve for it. (And by "reserve for it," I mean basically account as though you have already lost it.) Here's the rule:

An estimated loss from a loss contingency shall be accrued by a charge to income if both of the following conditions are met:

a. Information available before the financial statements are issued or are available to be issued ... indicates that it is probable that an asset had been impaired or a liability had been incurred at the date of the financial statements. Date of the financial statements means the end of the most recent accounting period for which financial statements are being presented. It is implicit in this condition that it must be probable that one or more future events will occur confirming the fact of the loss.

b. The amount of loss can be reasonably estimated.

It goes on:

If a loss cannot be accrued in the period when it is probable that an asset had been impaired or a liability had been incurred because the amount of loss cannot be reasonably estimated, the loss shall be charged to the income of the period in which the loss can be reasonably estimated and shall not be charged retroactively to an earlier period.

So if you did bad stuff in 2012, and found out in late 2013 that it was going to come back to bite you, but only found out in the third quarter of 2014 how big that bite would be, then that reduces your income for the third quarter of 2014, even though the bad stuff happened in 2012 and the actual bite will occur in the fourth quarter or later. You can't travel back in time to reduce income for the periods when the bad stuff happened. That would be silly.

Of course BofA and Citi didn't find out how big the bite would be in the third quarter. They found it out sometime between Oct. 15 (BofA's earnings statement) and Oct. 30 (Citi's revision). But this too is dealt with by the accounting:

An entity shall recognize in the financial statements the effects of all subsequent events that provide additional evidence about conditions that existed at the date of the balance sheet, including the estimates inherent in the process of preparing financial statements.

For instance:

If the events that gave rise to litigation had taken place before the balance sheet date and that litigation is settled after the balance sheet date but before the financial statements are issued or are available to be issued, for an amount different form the liability recorded in the accounts, then the settlement amount should be considered in estimating the amount of liability recognized in the financial statements at the balance sheet date.

BofA and Citi don't have a settlement, but they have something like it. They have a regulator-influenced estimate of their future liability for past misconduct, which they obtained after the quarter ended but before the financials were issued. And that, according to accounting, means that they have to take a loss for that liability in the last quarter. You can travel back in time to last quarter before you file the 10-Q, but not after. It's all very simple!

  1. Since the global probes were well underway by 2013. Even if you think that very similar badness occurred in 2014, or even the third quarter of 2014, it seems unlikely that that particular badness has already been caught and wrapped up in a settlement process. Regulators take longer than that.

  2. I just mean, they haven't occurred yet. I have no reason to think they'd be tomorrow.

  3. Actually Goldman sort of lowered its legal bill, though not in an accounting way, or in a way that adjusted its earnings release. But it's imaginable that that might be related to being pleasantly surprised about the FX thing. Disclosure: I used to work at Goldman.

  4. This is from the Accounting Standards Codification, ASC 450-20-25-2. Here's a link, but you'll need a (free) log-in.

  5. ASC 450-20-25-7.

  6. Obviously it's possible that they found out at different times. I imagine a big sit-down, but I have no reason for that.

  7. ASC 855-10-25-1.

  8. ASC 855-10-55-1-a.

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.

To contact the author on this story:
Matthew S Levine at mlevine51@bloomberg.net

To contact the editor on this story:
Zara Kessler at zkessler@bloomberg.net