Banks Will Miss Libor When It's Gone

(Bloomberg View) -- SOFR vs. Libor.

Libor, the London Interbank Offered Rate, is a lot of things. It is the standard interest rate for floating-rate loans: If a company borrows money at a rate that resets every three months, the rate that resets every three months is normally Libor. (Well, Libor plus a spread: A company might borrow at, say, Libor plus 3 percent or whatever.) It is the reference rate for trillions of dollars of interest-rate derivatives: If you are making some sort of bet on the future of interest rates, there's a decent chance that the interest rate whose future your betting on is Libor. Beyond that, it was -- it's not really anymore, but for a long time it was -- the standard "risk-free rate" that people would use in financial models for pricing other derivatives; if you bought, say, a call option on Microsoft stock, the formula for valuing that call option has a spot to plug in an interest rate, and the interest rate you'd plug in was probably Libor. Almost every part of the financial world was touched by Libor; it was plausibly "the world's most important number"; it was just part of the atmosphere, a basic given of the financial system.

But deep down, at its origins, Libor was a simple risk-management tool for banks. The classic business of banking is borrowing short-term to lend long-term: Banks take deposits (short-term loans), and give out business loans and mortgages and stuff (long-term loans). If you borrow short-term and interest rates rise, you soon have a higher interest bill. If you lend long-term and interest rates rise, you don't get paid any more interest on those long-term loans. But if you instead lend long-term at floating rates -- if you reset the loan rate every three months -- then you do get paid more when interest rates rise; your interest-rate risk is reduced.

Even better, though, if you lend long-term at Libor, then you really reduce your interest-rate risk, because Libor is notionally the short-term rate at which banks themselves can borrow. Libor, in its original form, was meant to represent the banks' cost of getting money to make loans, and Libor-based loans were a simple cost-plus model of lending. If a company borrowed at Libor plus 3 percent, then that meant more or less that the bank would pay Libor to get money, and lend it to the company at Libor plus 3 percent, and so lock in a profit of 3 percent on the loan. 

There is an obvious conflict of interest here, which is that Libor is calculated by surveying the banks to see what their cost of borrowing is, and the banks get paid on their loans based on Libor. So there's a simplistic incentive for the banks to respond to the survey "yeah our cost of borrowing is 30 percent, weird," so that Libor will be really high, so that they'll get paid a lot of money on their loans. It is a curious historical fact that that never really happened. Oh Libor absolutely was manipulated, but the manipulation wasn't really for the dumb simple reason that overstating your costs in cost-plus lending model will make you more money. Instead, as Libor became a vital reference rate for interest-rate derivatives, the banks' derivatives traders pushed their Libor submitters to manipulate Libor, up or down, depending on their positions. And then, as Libor became a proxy for the health of the banking system during the financial crisis, the banks' executives pushed their Libor submitters to understate Libor, to make them look healthier than they were. 

This manipulation was bad enough that Libor fell into disrepute, and regulators and market participants demanded a replacement, and slowly and creakily some replacements are coming online. In the U.S., the preferred replacement is the Secured Overnight Financing Rate, or SOFR, which the Federal Reserve Bank of New York started publishing this month, and which is "a broad measure of the general cost of financing Treasury securities overnight." Libor is notionally the interest rate that banks pay to borrow money unsecured from each other for a night or a month or three months or whatever; SOFR is the interest rate that you pay to borrow money secured by a Treasury bond overnight. The advantage is that banks don't really borrow money unsecured from each other in all of the Libor tenors, so the Libor numbers are unavoidably made up; the repo market -- the market for borrowing secured by Treasuries -- is quite robust, so SOFR is based on real transactions and thus much harder to manipulate.

But change is hard. Here is a fascinating post from Joe Rennison at Alphaville about how hard it will be to replace Libor with SOFR in the floating-rate loan market: Interest-rate derivatives markets are relatively willing to transition to SOFR, but "Libor has some features which work for the $4.1tn syndicated loan industry that are less crucial to the derivatives industry."

There are at least four distinct things going on here. One is that there are a lot of existing loans that reference Libor, and they're not going to magically become SOFR loans overnight, so it will be a long time until SOFR replaces Libor as the most common reference rate for loans. (Or derivatives for that matter.) Another is that there are a lot of people -- bankers, lawyers, corporate treasurers -- who are used to Libor, and who will grumble a bit at having to change their documents and look up a new loan rate. These are both real and important social problems in the adoption of a new reference rate, but they're not particularly financially interesting.

A third issue is that SOFR is an overnight lending rate, and no one wants a floating-rate loan that floats every day. Floating every three months is okay, but you want a little bit of predictability; you want your interest costs locked in for at least a quarter. Rennison:

SOFR is an overnight funding rate that resets every day, whereas Libor measures a bank's cost of borrowing cash over different periods (one month, three months, six months, etc).

Corporate borrowers say that Libor's term structure provides more predictability and helps them manage cashflow. And because no predictable term structure for SOFR exists just yet, treasurers will need to either stick with Libor until one is provided -- or grapple with the uncertainty of paying interest based on a daily rate.

This does not strike me as an especially substantive problem. It's easy enough to build a three-month SOFR. A confession: When I was a banker, I thought for an embarrassingly long time that there was a thing called "five-year Libor." There is not. (The longest Libor is 12 months.) But there are interest-rate swaps that reference Libor, and there's a five-year swap in which I pay you a fixed rate and you pay me six-month Libor every six months, and that swap trades quite liquidly and is a good market reference point, and it is effectively an extension of Libor out to five years: The fixed rate on that swap is the fixed rate that buys you five years' worth of Libor. For many practical purposes -- figuring out what sort of rate a company should pay on a five-year bond, for instance -- that swap rate really works like five-year Libor.

Similarly, if SOFR really becomes the standard reference rate for interest-rate derivatives, then it will not be a problem to find a fixed rate that buys you three months' (or whatever) worth of SOFR. And in fact that rate is coming pretty soon:

The CME's launch of SOFR futures might help, since the contracts measure the compounded cost of interest over a three-month period. But there is still a difference between futures estimating a three-month borrowing rate and an actual three-month borrowing rate.  

Eh, not really; a thing that lets you pay a fixed rate in exchange for three months' worth of overnight rates really does give you a three-month rate. (Here is CME Group's description of the SOFR futures, and the contract specification.) And it seems to me that there's no special trick about writing a floating-rate loan contract where the floating rate is not overnight SOFR but the three-month SOFR rate determined by the futures. A floating-rate loan indexed to SOFR itself would be annoying for a corporate treasurer, but a floating-rate loan indexed to three-month derivatives indexed to SOFR would be fine. And three-month SOFR, like actual SOFR, and unlike three-month Libor, would be computed by real market transactions rather than guesswork. (One hopes -- of course the SOFR futures market might not take off.) Derivatives markets might have made Libor unsustainable, but they could also make SOFR work.

But there is a fourth issue, which is that Libor was fundamentally a risk-management tool for the banks, and SOFR isn't. Rennison hints at that issue here:

The derivatives industry appears intent on moving away from Libor. If it does so before the loan market can figure out a way to create a long-term SOFR rate, it would create an entire new basis market between Libor and SOFR.

Say Company Z gets a loan from a bank and then hedges out its interest-rate risk in the derivatives market by putting on an interest-rate swap (let's say receiving the floating side). The company would have a basis differential between its Libor-based loan and its SOFR-based hedge. But problems could easily arise if the financial system comes under stress. ... Its Libor-based loans would push up its borrowing costs in a banking crisis, while the payments it would receive from its SOFR-based swap would fall, as rising demand for Treasuries would depress overnight rates.

So, yes, if companies borrow at Libor and hedge at SOFR, they will effectively take on the risk that banks' funding costs rise. But that problem doesn't go away if companies borrow at SOFR and hedge at SOFR. Instead it just becomes the banks' problem. Banks ... well, they don't exactly borrow at Libor, do they, but they borrow at something like Libor. They borrow at the rate at which banks borrow. Libor was meant to measure that, and that measurement broke down, but the point is that banks do more or less fund their business, at the margin, by short-term borrowing. If they make floating-rate loans indexed to Libor, and their cost of funds rises because people start worrying about the banking system, then their lending income will automatically rise, and they will be hedged. But if they make floating-rate loans indexed to SOFR, and their cost of funds rises because people start worrying about the banking system, then their lending income won't automatically rise -- it will probably fall -- and the problem will get worse. Libor gave banks a useful safety valve; it's a shame they ruined it.

Don't short to zero!

The way short selling works is that, if you want to bet against a stock, you borrow it from someone who owns it, and then you sell it to someone else. Eventually, you go out into the market, buy the stock back, and deliver it back to the person who owns it. If all has gone well, you will buy it back for less than you originally sold it for, and the difference will be your profit. Well, part of the difference will go to pay a fee to the stock lender, who will generally charge you a percentage for each day that you borrow the stock. But hopefully the stock will go down a lot, and the difference between your selling and buying price will more than cover that fee.

How far do you want it to go down? This seems like simple stuff: If you sell the stock for $20, and it goes to $15, you make $5 (minus the stock-lending fee), if it goes to $10, you make $10; if it goes to $5, you make $15; if it goes to $1, you make $19, etc. etc. etc., the arithmetic is pretty straightforward. And yet there are weird singularities. If the stock goes to zero, I guess you make $20? That's the goal, right? Except, what does it mean for a stock to go to zero? If a stock goes to $1, that means that people are buying and selling it on the stock exchange for $1.00. But if it goes to zero ... who is selling stock for $0.00? "Go to zero" is a conventional expression, especially popular among short sellers, but you have to be careful with it. If a stock "goes to zero" by going from $20 to 2 cents, or if it "goes to zero" through a formal bankruptcy process, then that is a pretty unequivocal win for the short seller. But some other paths to zero are ... actually ... bad?

Gates says he is stuck short China-Biotics, for instance, a stock that was halted in November 2013 when the Securities and Exchange Commission revoked its registration. The shares were delisted, and the company shut down. Gates showed us a January 2018 letter that he says he sent to the stock market record keepers at The Depository Trust Co., in which he laments that a TFS mutual fund has paid hundreds of thousands of dollars in stock-loan fees on China-Biotics and remained subject to the $2.50 a share margin required by Financial Industry Regulatory Authority rules. A handful of other short positions have him in the same predicament.

Gates is Rich Gates of TFS Capital, who "has joined an industry committee that the Securities Industry and Financial Markets Association calls the Worthless Securities Working Group." He shorted some stocks that he thought were frauds, and the SEC agreed that they were frauds and halted them, and then ... things got worse for him. The shares were worthless, but they didn't trade at zero or $0.01 or whatever: They didn't trade at all, so he couldn't buy them back to deliver to his stock lenders. Here is his letter to DTCC, which notes:

DTCC maintains securities in its records indefinitely that are worthless and illiquid; these securities have not been "taken down" by DTCC despite the fact that they have been deregistered by the SEC, delisted by the exchanges, have no existing business location, operations, management or employees and their registration is not in good standing in their state of incorporation. 

Gates blames the brokers who loaned him the stock for this situation (and calls it "a glaring conflict of interest" for them), but that is not the whole story. "Gates’ prime brokers aren’t necessarily pocketing stock-loan fees—they may have to pass them on to other firms that supplied the borrowed shares," notes Bill Alpert at Barron's. The only way to (legally) short stock is to borrow it from someone who owns it, and your stock borrow fee makes its way back to that owner-lender (with brokers taking a cut). The actual conflict is that whoever was long China-Biotics stock is still making money from it in its limbo; if the stock borrow fees are high enough and the limbo goes on long enough, they might make more from the stock borrow fees than they would have if the company had been successful. The incentives in stock lending are always a little weird -- you are making a little extra money on the stock you own by helping someone bet against that stock -- but here they are downright perverse.

Obviously this result is Wrong, and should probably be solved. ("Put them on the blockchain," you might shrug, though of course this problem has nothing to do with tracking ownership; it's about deciding when a company is not a company anymore, which is harder to blockchain.) Obviously the situation is not fun for Gates. I confess that it makes me rather gleeful. It is Wrong, of course, but it is Wrong in such an evocative way. It is, structurally, hard to bet on disaster, because when a disaster happens, that's when bets tend not to get paid off. If you bet on the end of the world, and you are correct, no one will be around to pay you. If you bet on the collapse of the financial system by buying credit-default swaps on banks, then you are probably buying those swaps from banks, and you'll have trouble collecting if your bet comes good. But that is true even in miniature: Even if you bet on a disaster for some penny-stock fraud, and even if that disaster occurs exactly as you predict, you might get swept up in it as well.


I used to work at an investment bank, and part of my compensation was paid in shares of my bank, and a popular rule of thumb among low-level drones like me was that you sold your shares as soon as they were delivered. Senior partners might be doing wild stuff like buying more stock, or selling puts, or whatever, but they were rich. For the rest of us, our careers were already so leveraged to the health of our bank that it seemed silly to hold any more shares than we had to. So every year some of my stock compensation would become unrestricted, and I'd sell it that day.

But I don't recall ever having selling reflexes quite as catlike as this:

A Samsung Securities employee caused the firm to pay out a massive dividend in the form of its own shares to 2,018 employees who were members of a company stock-ownership scheme.

The dividend was supposed to be 1,000 won ($0.94) per share. But the employee mistook the form of measurement, confusing won and shares. The error caused Samsung Securities to issue a dividend that was 1,000 times the value of each share held by the group of employees. ...

Sixteen staff members sold a collective five million shares worth about $186.9 million minutes shortly after receiving them. It took the brokerage 37 minutes to completely block employees from selling the accidental shares, according to South Korea’s Financial Supervisory Services, a financial watchdog.

Some of those employees -- ones who "sold stock despite receiving warnings from the company not to do so" -- have been suspended, and the transactions seem to have been reversed. But really those 16 Samsung Securities employees should be among the great folk heroes of finance. Imagine coming to work one day; expecting a nice, you know, extra $500 of company stock; checking your account the moment the stock is supposed to hit; finding an extra $19 million instead (Samsung Securities shares traded in the neighborhood of 38,000 won at the time, so paying 1,000 shares rather than 1,000 won was a 38,000-fold increase); carefully pondering the likelihood that you were supposed to get a $19 million bonus and the ethical and career implications of cashing it in; and then cooly putting in a sell order and blasting out the entire windfall -- all in the 37 minutes before your employer noticed the problem. That is the sort of careful attention to detail, decisive action, coolness under pressure, sense of humor and ethical flexibility that you want in a stock trader. If Samsung doesn't appreciate these people, I am sure someone else will.

The crypto.

I feel like these days I am constantly reading about hedge funds who had a few years of 30 percent returns and who are now down like 6 percent year-to-date and facing redemptions. So it's pleasing to read these absolutely cartoon numbers:

Pantera Capital, a hedge fund that gained attention for returning 25,000 percent over its lifetime through the end of last year, saw the value of its cryptocurrency fund cut nearly in half in March, according to an investor letter Tuesday.

Actually "the firm's Digital Asset Fund is down 54.9 percent year to date but is still up 10.3 percent life to date"; the 25,000 percent comes from its Bitcoin Fund. I suppose if your fund were up 25,000 percent and then down 50 percent, your investor letter could still be pretty swaggery. "We're up 12,500 percent cumulatively," you would say, "so, you know, whatever." 

Elsewhere in crypto, "The world’s first blockchain toothbrush lets you mine coins by brushing your teeth," and somehow it is not from Dentacoin. And: "In Defense of CryptoKitties." 

People are worried about bond market liquidity.

Bloomberg Intelligence's Eric Balchunas pointed out on Twitter that "junk bond ETFs had biggest quarterly outflow ever" in the first quarter of 2018: "Assets shrunk by 10% pretty quickly w/ no issues." I made a similar point in February, when a junk-bond exchange-traded fund had the biggest daily outflow ever, and bond prices were fine. For a long time -- years? it felt like forever -- we talked in this space about people's worries that the bond market couldn't handle rapid withdrawals from bond ETFs, and that those ETFs were creating a "liquidity illusion" that could crash the market when investors withdrew money. Nope!

Things happen.

Wall Street Faces Higher Capital Demands Under Fed Proposal. BofA Will Stop Lending to Makers of Assault-Style Guns. Senators Tell Facebook CEO the Days of Self-Regulation May End. Instagram Looks Like Facebook’s Best Hope. Mark Zuckerberg has been apologizing for reckless privacy violations since he was a freshman. China Adds Flesh to Bones of Plan for Big Bang Financial Opening. Theranos Lays Off Most of Its Remaining Workforce. Russia Investors Offload Ruble as U.S. Sanctions Spook Market. Deripaska’s Two-Decade Wooing of U.S. Ends in Financial Meltdown. Carl Icahn to Sell Federal-Mogul for $2.44 Billion. Forest City Board Shake-Up Could Be a Harbinger for Other REITs. Tiny Town Shuts Police Program That Gave Robert Mercer a Badge. Koch Bodyguards Got Police Badges From Same Tiny Town as Mercer.

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Thanks! 

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

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.

To contact the author of this story: Matt Levine at

For more columns from Bloomberg View, visit

©2018 Bloomberg L.P.