It's Basel, trust me. Photographer: Michele Tantussi/Getty Images

Banks Prefer Losses They Don't Have to Talk About

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

If you own a bond and interest rates go up, your bond will lose value. That's just a fact. But the consequences of that fact, if you're a bank, are all over the place. If you own that bond in your trading portfolio, you have what is colloquially called a trading loss: You owned a thing, it went down in price, you have lost money. Not in a strict literal sense -- you have not turned the bond back into money, so your losses are "on paper" -- but in every relevant accounting and capital and so forth sense. So your net income goes down (or becomes a net loss), and your regulatory capital goes down, by the amount of value that your bond has lost.

If you own that bond for investment purposes, and you don't have any "intent of selling it within hours or days," you have an investment loss on paper, but you get to treat it a bit more gently. (This is called "available for sale," or AFS.) The loss doesn't flow through your net income; instead it flows through a different place called "other comprehensive income," and everyone agrees to treat that as somewhat less important than net income. Everyone except Basel III bank capital regulation: Last year, regulators ungallantly decided to require you to treat those unrealized investment losses as reducing your capital.

If you have the "positive intent and ability" to hold the bond until it matures, then you can just ignore the economic loss until maturity. (This is called "held to maturity," or HTM.) You just keep the bond on your books at the price you paid for it, and at maturity you get back par and your loss vanishes. You ignore the loss in net income, other comprehensive income, your balance sheet, your capital, whatever.

The point is: The economic loss is the economic loss, but the accounting treatment varies from "recognize the economic loss immediately" through "recognize it for some purposes but not for others" all the way to "ignore it in all ways forever." And you get some choice about what accounting treatment you'll use.

Generically speaking, ignoring losses is more attractive than recognizing losses. So you'd expect that, when losses seem likely -- say when you have a giant portfolio of bonds in a low interest rate environment in which everyone expects long-term interest rates to increase in the next few years -- held-to-maturity treatment would be more attractive than available-for-sale. And, lo:

The largest U.S. lenders are moving assets into the "held-to-maturity" column of their books instead of designating them as "available for sale," an accounting method that under post-crisis banking regulations allows paper losses to erode measures of their health. The change pushed the share of securities that the five biggest banks keep in the held-to-maturity category to 8.4 percent, the highest in almost two decades, according to Credit Suisse Group AG analysts. ...

U.S. deposit-takers invest more in the $5.4 trillion of Fannie Mae, Freddie Mac and Ginnie Mae mortgage bonds than in other securities because they carry little default risk, offer higher yields than Treasuries and are easy to trade.

The top 50 banks owned $1.2 trillion at year-end, increasing those designated as held-to-maturity, or HTM, by $38.4 billion in the three months ended Dec. 31 and reducing available-for-sale, or AFS, investments by $23.3 billion, according to data compiled by Morgan Stanley analysts.

There are many delights in the story. But let's quickly pick two.

First, one reaction to a rising interest rate environment might be to reduce one's holdings of long-dated fixed-income securities. One might say "hmm, I should maybe sell my trillions of dollars of very liquid Fannie Mae bonds that I expect to lose value in the next few years." Or not, I mean, I'm not advising anyone to predict the timing of interest rate rises. The point is though that banks seem to be reacting to their expectations of rising interest rates with the opposite of the economically rational strategy: Not "sell bonds to avoid losses later," but rather "reclassify bonds as hold-forever to avoid recognizing losses later." The accounting provides opposite incentives from the economics.

Second is this very deep passage at the end of the article:

At the same time that regulators are putting bank capital at greater risk from bond slumps, they're also working on separate rules stemming from Basel III that will make lenders increase holdings of easy-to-trade assets such as agency mortgage securities so they can survive a financial crisis.

The use of HTM accounting, which so far appears to be getting done in a "thoughtful" and limited way that shouldn't increase banks' risks, is a logical response to the pair of new regulations, said Greg Hertrich, the head of U.S. depository strategies at Nomura Holdings Inc.'s securities arm.

"If you tell banks they have to hold" a certain amount of easy-to-liquidate assets, "it behooves them to find a way to do it in which there isn't a significant amount of capital volatility from quarter to quarter," he said.

Did you get that? New Basel III related regulations require banks to have a "liquidity buffer" of assets they can use to raise cash in an emergency. Banks are willing to do this, but they don't want their liquidity buffer to create "a significant amount of capital volatility." That is, they don't want mark-to-market losses on their liquidity buffers to reduce their regulatory capital. This is part of their broader desire not to have any mark-to-market losses on anything reduce their regulatory capital, but the liquidity buffer is part of that, sure.

So why not classify (some of) your liquidity buffer as held-to-maturity, that is, as bonds that you plan to hold on to forever? You tell your accountants "we never plan to sell these bonds," and you tell your regulators "well we can always sell these bonds in an emergency." It's its own little miniature regulatory arbitrage: You treat the same thing in opposite ways under different regulatory regimes. Normally that takes some significant effort, but here, just pronouncing a few magic words seems to do the trick.

(Matt Levine writes about Wall Street and the financial world for Bloomberg View.)

  1. It's somewhat less of a fact if it's a floating-rate bond, but it's probably not. Also I don't want to hear about credit spreads being inversely correlated to rates or whatever, you know what I mean.

  2. Even in the tax sense, if you're a securities dealer.

  3. When you sell the bond, if you still have that loss, it moves from OCI into net income. Incidentally here and in the surrounding paragraphs I'm quoting from ASC 320-10-25 (classification of investment securities) and 320-10-35 (subsequent measurement). (You need to register to access that page, but it's free, and worth it if you like accounting.)

  4. Amortized cost, but let's not split hairs.

  5. Which makes a sort of sense: If you never sell the bond in the market, you never realize the loss in the form of getting fewer dollars for your bond than you paid for it. Your loss becomes just an opportunity cost: Instead of getting 5 percent interest and your money back, you got 4 percent interest and your money back. But you got your money back. Everything is fine.

  6. I mean, to the company with the losses! Not, like, to Society.

  7. Down here, two bonus delights. One is less of a delight: One rough model you could have of banks' vast Fannie/Freddie portfolios is that the banks are flush with deposits and somewhat light on lending opportunities, but that eventually that will get better, they will sell all their agency bonds, and will start lending money to the real economy again. Reclassifying your agency bonds as held-to-maturity rather cuts against that model, and provides some support for the alternative model of banks as mostly an arbitrage of borrowing cheaply in government-supported short-term markets and lending expensively to government-sponsored agencies in long-term markets.

    Another is just a longstanding complaint I have, which is that JPMorgan got in just so much trouble for losing $6 billion-ish in mark-to-market losses, while other banks have gotten in significantly less trouble for losing similar amounts of money in available-for-sale losses. (That first link was to a post about Bank of America's $5.7 billion loss in the second quarter of 2013, for instance, and lots of banks have similar numbers.) The economic losses are the same, but the accounting treatment really drives public perception: A big loss that hits net income is a scandal, a big loss that hits other comprehensive income is not. Held-to-maturity is the logical next step; a big loss that disappears is no scandal at all.

  8. I guess you're really telling your regulators "we can always raise money against these bonds in an emergency," which is not quite the same as selling them. You could borrow against them in repo markets, for instance. (Though technically a repo is a sale and a repurchase agreement? Best not to think too hard about this.)

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

To contact the editor on this story:
Toby Harshaw at