The Fed has re-entered the debate on how to accurately measure Fed Funds volumes. This time the authors of a new staff report, “Identifying Term Interbank Loans from Fedwire Payments Data,” have taken on a bite-sized part of the challenge and come up with some valuable results.
The core issue is whether the Fed’s algorithm for figuring out Fed Funds payments is accurate. It had been thought that the Furfine method had it right, but a more recent paper reviewed by us (“How to track Fed Funds? Even the Fed doesn’t know, and that’s a problem”) disputed some of the important findings. This wasn’t good. The current study follows the idea that Furfine might not be right, adapts it and presents some apparently tight data of their results. There is still room for error but this is pretty good, and is important for tracking volumes and avoiding the potential for inaccurate reporting.
Here’s how the current study works:
“The goal of this paper is to develop a methodology to infer term US dollar interbank loans at a transaction level using payments data from the Fedwire® Funds Service, the large-value bank payment system operated by the Federal Reserve. Amongst other functions, the Fedwire Funds Service is used to settle a significant fraction of US dollar interbank loans amongst financial institutions. The basic idea of the algorithm is to identify pairs of payments that match the properties of the sending and return leg of an interbank loan: namely, a sending payment from bank A to bank B, and a matching return payment for a slightly larger amount from bank B to A on a later date, where the sending and return payment meet a set of criteria consistent with interbank market conventions.”
The variables tracked are: “(i) the ABA number of the sending and receiving institutions; (ii) the payment date and time, (iii) the dollar amount transferred; and (iv) a business function code which provides information on the transaction type.” Check out page 11 for an example of how the tracking worked out in practice.
The authors note that they are not trying to track Fed Funds transactions, but rather “overnight or term loans made or intermediated by banks.” This is an important distinction and one that sidesteps how the Fed is measuring Fed Funds today. Basically the authors are taking Furfine, adding in the refinements of the October paper, and presenting something that could be Fed Funds if everyone agreed to it.
The methodology isn’t perfect, and the authors note that the data could capture transactions that aren’t actually overnight or term loans such as repo payments. There is also the problem of agency vs. principal transactions. But the authors find US$90mm to US$220mm (on two spike occasions) in term transactions between 2007 and 2009 as weighted by maturity. We’d like to see what 2011 to 2013 look like with the same process – that could help inform the LIBOR vs. OIS vs. repo futures conversation.
This report is an important step forward in figuring out what the volume are of Fed Funds transactions. We were pretty perturbed when the October report came out, then was retracted, then published again later on. We are pleased to see the Fed committing more resources to accurately tracking these data.