It feels virtuous every so often to take glance over at the triparty repo market. You get a nice dose of horrified vertigo and then go back to your life and don’t think about it for a while and that always feels better. Now is a good time to get back to it, what with continued worrying about money-market funds – a core player in the market – and two interesting things this week about triparty repo: this testimony from Matthew Eichner of the Fed to a Senate subcommittee, and this report from Fitch.
Here is how I imagine triparty repo:
- A bunch of money market funds and other cash investors keep $1.8 billion of cash at JPMorgan and Bank of New York Mellon, the “clearing banks” in the triparty system.
- A bunch of securities dealers keep a pile of securities – worth, on a good day, more than $1.8bn – to JPM and BoNY Mellon.
- The dealers need money to fund those securities, because what are they going to do, pay for them themselves?
- Every afternoon, the cash investors and the securities dealers frantically negotiate which dealers swap their securities (at negotiated haircuts) for which cash investors’ cash.
- Every night, the cash sleeps in the (notional) arms of the securities dealers, while the securities (and a promise to buy them back in the morning) sleep in the (notional) arms of the cash investors.
- Every morning, the cash wakes up and springs from the dealers’ beds back into the waiting arms of the cash investors, and vice versa etc.
- Which means that the dealers need to borrow cash to be able to give it back to the investors. Where do they get the money?
- Well, from JPMorgan or BoNY.
- Where do JPM and BoNY get the money?
- Well, from deposits.
- Whose deposits?
- Well, the deposits of the cash investors.
More or less, right? From a certain angle, nothing happens during the frantic activity every evening: cash investors are always getting interest on their cash by investing it in a debt claim on a pool of securities, and securities dealers are always paying interest to debt finance their pool of securities. During working hours, the cash investors get their interest from the clearing banks and the dealers pay interest to the clearing banks; at night, the cash investors get their interest directly from the dealers but it’s all still intermediated by the clearing banks.
This is something not far from magic, and you can see why Fitch begins its report by calling repos “a core part of the ‘shadow banking’ system,” because they are literally an after-dark equivalent for activity that during daylight is basic bank financing.
Fitch is particularly concerned about the quality of the collateral that the biggest money market funds are actually financing; the report is frustrating because there’s a lot of cherry-picking* but if you want to feel alarmed by it you have my blessing. Basically big money market funds put a good chunk of their money in repo, and a good chunk of that is funding tiny positions in illiquid deeply discounted junk-rated structured finance securities, and blah blah blah Countrywide!** And as money market funds are not natural holders (or efficient sellers) of illiquid junk-rated structured finance securities, that is worrying: if the market lost confidence in a repo borrower, causing it to default and leave the money market funds to liquidate their collateral, that would not go well for the funds.
The Fed is also worried about what would happen on a loss of confidence, particularly during the magic moments of the “unwind” and “rewind,” when the dealers’ portfolios shift each morning from being funded by cash investors to being funded by clearing banks (and vice versa). Because the clearing banks have no obligation to fund the dealers each morning, and because the cash investors mostly have no obligation to fund the dealers each evening (to the extent repo is overnight), each unwind and rewind point is an opportunity for the dealer to not get funding, default on its debt, and have to hand over its collateral to the clearing bank or, even worse, to the cash investors. And when an investment bank’s cats-and-dogs CCC-rated structured finance portfolio gets handed over to a money market fund that was funding it at a 92% LTV … that money market fund is going to liquidated it quick and get back less than 92 cents on the dollar. Then the world ends.
So the Fed would like to robustify the unwind and rewind a bit. Here is Eichner:
A solution to this fire sale problem likely requires a marketwide collateral liquidation mechanism. The challenges in designing and creating a robust mechanism–which would almost certainly need the capacity to fund a significant volume of collateral for some period of time–are appreciable, and include assuring adequate liquidity resources even under adverse market conditions and developing rules for the allocation of any eventual losses across market participants. Such capabilities typically exist today in the context of clearing organizations that have a formal membership structure, which allows for capital assessments and the sharing, or “mutualization,” of potential losses. How this model might be adapted to a market more loosely organized around clearing banks, particularly in which certain less-liquid collateral types continue to be funded, remains unclear and will surely need to be the focus of much additional study.
Study it up, Fed, study it up.
I feel like there is much, metaphysically, to love about all of this. My bullet points above stylize the process, perhaps to the point of uselessness, but I don’t think so. During the day this looks like a banking system, with all the attendant risks of a banking system: if all the cash investors pull their deposits, that is a “run” and the bank will be left having to liquidate its assets (intraday loans to dealers, or realistically the collateral for those loans), likely at fire sale prices. So we have things to deal with that, like deposit insurance (not relevant in these markets), capital requirements for the banks, prudential regulation, reputational constraints on JPMorgan and BoNY Mellon, etc.
At night it’s the same thing but spookier. “Capital requirements” become repo haircuts negotiated between the money market funds and the dealers, regulation is relaxed, and reputational constraints don’t really bite: it’s not JPMorgan’s fault if a dealer can’t find someone to repo its portfolio overnight, or if a money market fund breaks the buck because its repo counterparty defaults. They’re just the marketplace.
As Eichner points out, though, lots of marketplaces do take the responsibility on themselves, not (entirely) because regulators force them to but because it makes for a good market and thus good marketing: if you know that your trades will actually settle on the NYSE but might not settle on the Lower Broadway Bucket Shop Exchange, you’re more likely to use the NYSE.*** So it’s a little puzzling that JPMorgan and BoNY haven’t taken any steps to fix this themselves: couldn’t they in theory attract more triparty repo business with a more robust structure – say, one that required minimum haircuts and featured a coordinated system in which the clearing bank liquidates collateral of defaulters to maximize recovery rather than handing it to lenders to bash out on their own?
Steve Randy Waldman’s musings about financial complexity and banking at Interfluidity may provide one answer. Everyone on the regulatory side seems to agree that a more coordinated system that explicitly acknowledged the risk of failure would be better, but that would require explicitly acknowledging the risk of failure. Money market funds can’t fail! They can’t go below $1! Offering them a system that handles default better, at the cost of acknowledging the possibility of default, may appeal to regulators (and, for that matter, to the clearing banks), but it’s not exactly good marketing to the lenders. Better to pretend that the risks of the triparty repo system are overblown, and go back to cheerfully ignoring them.
Testimony of Matthew J. Eichner, Deputy Director, Division of Research and Statistics: Tri-party Repo Market [Fed]
Repos: A Deep Dive in the Collateral Pool [Fitch, miserably gated]
* And misleading use of statistics. Like here is a table to show “ooh there are lots of tiny odd-lot securities positions that would be hard to liquidate”:
OMG 53% of the structured finance repo pool is positions of under $1mm! Except … that’s clearly 53% by number of issues, right? Not dollar amount? And the 47% of positions that are over $1mm are clearly a much bigger chunk of the market? Why would you do that? Blargh.
** In more detail: (1) the ten biggest prime money market funds have about $116bn of repo collateral, or about 1/6th of their total assets, (2) over a third of that is in “risky” (non-Treasury/Agency) collateral), with $21bn in structured finance repo (or only about 3% of total assets, so, y’know), and (3) the structured finance stuff has a lot of crap: “structured finance repos are typically collateralized by pools of security that are of lower credit quality (e.g. ‘CCC’ and below), deeply discounted, and small in size.” Also some eyebrow-wagging about the average haircut for the structured finance stuff is 7.6%, meaning that it’s levered 14x, and the #1 issuer of it is Countrywide, which, Countrywide.
*** Joke’s on you.