The frailty of the tri-party repo system has been on the regulator’s minds for some time before the financial crisis. They knew the system worked but had several potentially fatal flaws.
1.) The near guaranteed unwind of deals in the morning before the clearing banks knew if there would be cash committed for a rewind in the afternoon created gigantic exposure to the clearing banks during the day. The crisis brought this into sharp focus and the Fed took action with the reform program. But the result is that instead of the cash lenders having exposure from 4pm to 8am and the clearing bank taking the rest, the cash lenders take the vast amount of exposure. But are they prepared? The recent FSB report suggested that (in order to prevent fire sales) cash lenders may want to limit the paper they take to stuff they can own in their portfolios should there be a default. We’ll be watching that carefully.
2.) There clearing bank exposure still isn’t the “operational moment in time” that the Fed would like, but it is getting better.
3.) Collateral schedules were weak. For asset-backed securities, there was a box to check if you took investment grade, another for non-IG. But that was fine when ABS were credit card and auto loans. But RMBS was considered asset-backed paper too and provided cheap funding for all sorts of sub-prime coming from both dealer books and shadow bank clients. This was a prime example bank/shadow bank inter-connectedness. Shadow banks obtained financing from the banks, who in turn recycled the paper to mutual funds.
4.) Haircut schedules were incredibly sticky for tri-party. Reasons include the difficulty in negotiating new agreement and getting the scheduled haircuts updated to cash investor’s benign neglect of the collateral they received. So while bilateral repo trading haircuts could adjust higher, in tri-party the decision was a simple yes or no. And when it was no, the system shuddered. It made the tri-party susceptible to runs and extreme credit crunches.
5.) Tri-party asset pricing was assumed to be accurate. For liquid assets like Treasuries and Agencies it was. But when the paper was credit intensive and the underlying cash market ground to a halt, the prices were no better than guesses.
6.) The infrastructure was sloppy. Trades didn’t actually get confirmed and acknowledged. Term trades were unwound and rewound just like overnight trades. The Fed tri-party reforms addressed these and other issues.
Why didn’t the bomb go off? The Fed stepped in and created the PDCF and other programs to insure that broker/dealers could source money from the Fed if they had to. This gave confidence to the cash lenders that there was a way out for them…so they kept on lending. This worked brilliantly.
What would that bomb have looked like if it did explode? It depended on the time of day: either the agent banks would be liquidating upwards of $2 trillion of paper or the cash lenders would have to do it. It would have been the mother of all fire sales, an issue the Fed is still grappling with.