Net Net: Promoting innovation and managing change
Net Net: Promoting innovation and managing change

Is Stability Creating Fragility?


Early this week,JPMorgan Chase announced that it had achieved "another major milestone" in the attempt to transform one of the most important markets in the financial system.

Almost no one noticed.

The reason why no one noticed is because the market in which JP Morgan's milestone was crossed—the tri-party repo market—is almost as obscure as it is important.

The tri-party repo market is at the center of our financial markets. It's where cash gets converted into credit that gets used by securities dealers to fund their operations. But its operations are little studied by specialists and academics, much less publicly discussed.

The basic mechanics of the tri-party repo market are relatively straightforward. There are, as you can probably guess, three legs that prop up this market.

The first leg are the primary dealers of Wall Street, the banks and securities broker-dealers (as well as some hedge funds) that require cash from investors to fund their securities portfolios.

The second leg are the investors with tons of cash on hand—big banks, mutual funds and money market funds, Fannie and Freddie, and Japan's ministry of finance—who lend it out to the securities dealers.

The third leg are the "clearing banks" that facilitate the deals between the first two. There are two of these: JP Morgan Chase and Bank of New York Mellon .

There is, of course, some overlap between these groups, which makes things confusing. JP Morgan Chase, for instance, is both one of the largest dealers that uses the tri-party market for funding, one of the largest cash investors in the tri-party market, and one of the two clearing banks.

Eight of the top 25 dealers that rely on the tri-party market are also investors in the market: Bank of America, Barclays, Deutsche Bank, Goldman Sachs, HSBC, Morgan Stanley, UBS , Wells Fargo .

But this overlap can be a bit deceptive. While there is a high degree in concentration among the collateral providing dealers who borrow on the market—the top 10 dealers account for 88 percent of the borrowing—the cash investor side is larger and less concentrated—the 10 ten accounting for less than 60 percent of the cash.

Until very recently, almost all tri-party repos worked like this. At the end of the day a cash investors would choose whether or not to lend its cash out to a dealer. In exchange, it would receive a right to securities that were offered as collateral for the deal.

One of the two clearing banks would hold the collateral, do the bookkeeping for the deal, and evaluate the collateral. Contrary to what a lot of people think, the clearing banks do not directly set up the deals between investors and dealers or play the role of broker in these deals.

The next morning, the tri-party repos were generally unwound, with the cash investor receiving its cash back and the dealer-borrower getting back the securities. At the end of the day, the deals would need to be set up all over again. (For a detailed description of the operations of tri-party repos, see this Fed study.)

One thing this market did is create enormous exposure for the clearing banks during the day, between the morning unwind and the afternoon rewind. During this period, it was generally the clearing banks that bore the credit risk of the dealers. They were funding the portfolios and holding the collateral until the new tri-party repos could be arranged.

The regulators did not like the way this market operated. They worry that even though it functioned well during the financial crisis—with far less strain than other funding markets—it is susceptible to sudden and catastrophic runs that can cause big dealer firms to fail.

A good outline of this regulatory fear of tri-party repo failure can be seen in Antoine Martin's post on Liberty Street Economics, the blog of the New York Fed. Martin explains how the market can fall fear to a "coordination problem" caused by a fear among cash investors that other investors might not invest.

Let's say it's a Thursday afternoon. You're at the tri-party desk of a large cash investor, maybe a money market mutual fund. You get the call that LynchStanleyBrothers wants to borrow $500 million overnight from you.

You know that LSB needs funding not just from you but from at least two other big funds in order to avoid default. You've been reading about troubles at LSB and you worry that even if you provide the funding, others might not.

In that case, you get stuck with the collateral—which you might or might not be able to sell before your own investors panic and demand their funds back—while your competitors kept their cash. So you might well decide not to fund because you fear the defection of other funders.

And, of course, if you are a clearing bank, you worry about this happening and your bank getting stuck holding the collateral when the securities dealer defaults on its obligations to you because it couldn't get overnight funding.

This can be a major headache for the clearing bank. The collateral comes on balance sheet, creating capital adequacy problems. If the collateral value declines, losses can occur. In order to avoid this, a clearing bank would likely refuse to unwind in the morning—forcing the dealer to default on its obligations to the cash investors.

Cash investors know that the clearing bank could do this, so this moves up their plans for refusing morning funding. It becomes a game of hot potato.

Obviously, this is a recipe for reflexivity gone wild. You not only know that you don't want to fund if other funding sources balk, you know that other funders are worried about you balking. Everyone's worries about everyone else's worries threaten to put LSB out of business.

Under the reforms initially proposed by Task Force on Tri-Party Repo Infrastructure—an industry group sponsored by the New York Fed—the clearing banks would not unwind repos every morning. Instead, new and maturing repos would be settled simultaneously, in the afternoon.

This kind of market reform faced some serious challenges. In the first place, one of the reasons the repo market unwound was that securities dealers needed access to their collateral for trading purposes. If the repos didn't unwind in the beginning of the day, how would the dealers trade the securities they put up as collateral?

What's more, putting the unwinds and rewinds together at one point in the afternoon seemed a recipe for chaos. When would anyone have time to count the beans?

While not all the goals of the task force have been accomplished, a lot of progress has been made. The announcement by JP Morgan Chase reflected some of this progress. The firm has cut its intraday credit related to daily unwinds exposure by more than one-third. It has put in place technology allowing borrowers to substitute collateral during the day, which it calls "collateral reoptimization."

What's more, JP Morgan says it will move the unwind back to late afternoon—short of a simultaneous unwind-rewind but creating a far shorter period of risk for the clearing banks. (In February the Task Force released a report detailing its progress on a number of items here.)

The Fed and the clearing banks present these changes as reducing the fragility of the financial system. And, indeed, the reforms to reduce stress points in the tri-party repo system. The regulators, however, see to hope for the gradual elimination of the daily unwind altogether—which may be going a step too far.

Recall that picture we described of reflexive fear in the tri-party market. While the Fed describes this as a source of "coordination error," another way of thinking about it is as a risk-detection process.

The daily unwind is a test of market confidence. It reveals when market participants do or do not have confidence in the credit-worthiness of borrowers and the quality of collateral. What's more, it forces market participants to be keenly attuned not just to their own models but to the likely actions of other similarly situated investors.

Eliminating the daily unwind might make the financial system more stable—but it would allow the build up of undetected fragility. It may have, for example, allowed Bear Stearns or Lehman Brothers have continued to operate for longer—which could have made the eventual detection of their weakness even more painful.

It might be more profitable to think about what kind of reforms could be put in place to make the system truly antifragile, to borrow a term from Nassim Taleb. An antifragile system would need to gain from disorder, rather than simply conceal the stress.

Part of the problem, here, is the belief that the major problem for the market is "coordination error" rather than simply the absence of knowledge. In a reflexive repo marketplace, you face a coordination problem when multiple investors choose to pull out because they wrongly fear the retreat of other investors. If only they could have coordinated their knowledge, they would have remained open to funding the rewind.

But the opposite problem—call it the "herd problem"— exists as well. In a reflexive marketplace, multiple investors will continue to fund so long as they believe others will fund. Their profit-motives prevent them from withdrawing funding while competitors continue to provide funding. A few more conservative or somehow wise sources of funding may withdraw, but the managers of these businesses will face pressure for underperforming so long as the problems with do not become manifest.

This, in turn, means that the managers of dealers will not receive market signals telling them that they are not being prudent, have invested in garbage, and are in danger of insolvency. The continued availability of funding will, to the contrary, send the signal that they are moving in the right direction.

Does any of this sound familiar? It should because this is how fragility gets built up within the system. It's a large part of what went wrong before 2008. And now the Fed and their friends at JP Morgan and Bank of New York want to make the system even more unbalanced, eliminating the coordnation problem"while leaving the herd problem in place.

In none of the literature from the Fed that I've read on the subject have I seen any awareness at all of this problem. That's likely because from the Fed's point of view, this isn't a problem at all. The market will continue to function, the Fed will have achieved stability, despite the growing fragility.

I'd be much more reassured about the attempt to reform the tri-party repo market if I saw a little more attention paid to risk detection and genuine antifragility.

Follow John on Twitter. (Market and financial news, adventures in New York City, plus whatever is on his mind.) You can email him at

Happy Thanksgiving!