The principles for designing a safe financial infrastructure are a bit like those for making a safe bridge or building. Safety-minded engineers should design the most critical components as simply and reliably as possible. They should use shock absorbers to reduce the frequency of failures, and establish backup mechanisms to limit their consequences.
One of the most critical components of any financial system is its mechanism for providing liquidity, including both the conversion of assets into the means of exchange (market liquidity) and the financing of an asset inventory for the purpose of making markets (funding liquidity).
In the U.S. financial system, the market for repurchase agreements (repo) is the most prominent source of short-term funding liquidity. A repo is a collateralized loan, usually with an overnight maturity. If the borrower fails to pay, the lender receives the collateral and can sell it to satisfy the claim.
At its peak prior to the financial crisis, U.S. primary dealers in securities reported repo borrowing of about $4.5 trillion each day – nearly one-half the size of the commercial banking system. More recently, that volume has shrunk to about $2.4 trillion, but the repo market remains the principal source of funding liquidity for many banks and securities dealers.
According to the Federal Reserve Bank of New York (FRBNY), the largest portion of the repo market is the “tri-party repo” market in which, as the name indicates, a third party facilitates the credit and collateral transfer between the lender and the borrower. Only two banks – J.P. Morgan Chase (JPMC) and Bank of New York Mellon (BNYM) – play this clearing role, together accounting for about $1½ trillion each day.
In theory, such a market could function quite simply and reliably. Each day, at a specific time, say 9am, the loans from the previous day would unwind and the new one-day loans would begin. The clearing institution would simply transfer the proceeds to the previous-day lenders and return the collateral to the borrowers. The next contracts would begin without any time lapse, so that borrowers in need of rolling over their loan could do so as long as a lender would accept their name and collateral.
At least until recently, however, the tri-party repo market was neither simple nor reliable. Typically, the previous-day contracts ended early in the morning, but the next-day contracts began only late in the afternoon. In the interim, until they could match lenders and borrowers once again for the next evening, the two clearing banks provided intraday or daylight credit to the borrowers – in the trillions of dollars! As a result, the clearing banks were exposed to borrower default, and the repo parties were exposed to clearing bank default.
Unsurprisingly, this design proved quite fragile in the financial crisis. The clearing banks, whose intraday credit provisions far exceeded their capital, had strong incentive to withdraw credit from risky security dealers, potentially contributing to a run on their liquidity. Indeed, according to the report of the Lehman bankruptcy examiner, JPMC required cash collateral from Lehman in September, 2008, in return for maintaining its credit.
Fortunately, the FRBNY has pushed tri-party repo participants to make the system more robust. Over the past two years, automation of the lender-borrower matching process has reduced the need for daylight credit from about 100 percent of tri-party repo volume in 2012 to about 20 percent today, and the FRBNY expects this share to drop to 10 percent this year.
Still, it’s hard to see why any intraday credit should be needed in an efficiently organized repo lending system. And, why should the clearing process for a critical security funding mechanism be performed by a private bank that, were it to fail, could infect the broader payments mechanism? As Darrell Duffie points out, this practice violates principles for financial market infrastructure promulgated by the Committee on Payment and Settlement Systems (CPSS) and the International Organization of Securities Commissions (IOSCO).
Just as the intraday credit in the repo market is not recorded on the balance sheets of the clearing banks, it also can be viewed as an unrecorded contingent asset of the central bank. The Fed can ill afford for the repo market (or a clearing bank) to fail in a crisis, so it would almost certainly have to provide a substitute source of credit in its role as lender of last resort. The result is an obvious moral hazard that encourages over-reliance on repo funding (and, especially, on tri-party repo cleared by one of the two large banks that are systemic as a result). Getting rid of daylight credit in the repo market would reduce, but not eliminate, this moral hazard.
If reforms of the repo mechanism can resolve these thorny issues, will it be safe? Unfortunately, the answer is still no. The big lingering risk that will remain even in the absence of daylight overdrafts is that of a fire sale of the collateral. Many lenders in the repo market – such as money market funds – simply cannot hold the less-liquid securities that often are pledged as repo collateral on their own books. Should their counterparty not be able to deliver the cash to unwind the transaction, they would be forced to sell. As regulators are well aware, such forced sales of collateral could trigger a vicious cycle of margin calls, deleveraging and asset price declines like those that characterized the crisis of 2007-2009. More than five years after the crisis, regulators are still looking for a solution.