UPDATE: New evidence on "Is the Fed Financing the FDIC?"

A new submission to the congressional record suggests more FDIC liquidity being sourced from the Fed

I wrote two weeks ago about the interaction of the Fed and the FDIC in resolving the 2023 failed banks—a financing relationship that has continued to today. If you missed it, see the note here:

Is the Fed Financing the FDIC?

That note concluded with the following:

Certainly in the case of First Republic, all or effectively all the collateral that was posted to the Fed was sold to JPMorgan. So there’s no collateral left at the Fed—which means the FDIC […] likely simply replaced the collateral altogether with the FDIC guarantee.

The Fed is likely just playing ball here and letting the FDIC sell the collateral out from under the Fed loans. But, at least in the case of First Republic’s BTFP loan, we know the FDIC isn’t paying down the Fed following the sale of the underlying collateral. Thus, if nothing else, we can say with virtual certainty that the Fed’s BTFP loan to First Republic is now “financing the FDIC.” Presumably there are other loan balances like this from the three banks’ discount window borrowings.

Due to a new submission to the congressional record, we now have some interesting new information with additional evidence on the financing arrangements—confirming the likely existence of those other loan balances financing the FDIC.

On March 29, 2023, the House Financial Services Committee had a hearing called “The Federal Regulators' Response to Recent Bank Failures,” which featured testimony from Fed Vice Chair Michael Barr, FDIC Chair Marty Gruenberg, and Treasury Undersecretary for Domestic Finance Nellie Liang. While the testimonies and hearing are old news, we now have the answers to the questions that the committee’s members submitted for the record.

Onto the new info.

The Rate

As written in the previous note:

You’d think the FDIC guarantee would make the Fed extremely comfortable. Normally, the discount window and BTFP charge a slight premium to the policy/Treasury rate. Yet, per the Fed [emphasis added]:

The outstanding loans accrue interest at 100 basis points above the applicable discount window or BTFP rate until they are repaid in full. The Federal Reserve expects that these loans will be repaid in full before the end of 2023.

So, the FDIC is paying the Fed a 100-basis-point premium relative to what private banks pay—including First Republic in its last days and the bridge banks.

Per an answer submitted for the record by FDIC Chair Gruenberg:

While the bridge banks were open, the rate was 5 percent. Once the bridge banks were closed, the borrowings were transferred to the receiverships and were considered in technical default, resulting in a 100 basis point penalty. The current rate, including the penalty, is 6.25 percent.

Still though: technical default? Why would that matter, as long as there’s an FDIC guarantee? There’s also nothing in the Fed’s discount window statute suggesting there’s any requirement for the upcharge.

Discount Window Activity at the time of Failure

While SVB and Signature both had trouble borrowing materially from the Fed for operational reasons in advance of their failure, their bridge banks were able to scale discount window borrowing substantially.

The Fed’s postmortem of SVB said that pre-failure (and pre-bridge-bank) SVB

had limited collateral pledged to the Federal Reserve’s discount window, had not conducted test transactions, and was not able to move securities collateral quickly from its custody bank or the FHLB to the discount window.

The NYDFS postmortem similarly said of Signature on March 10, its last day of operation [emphasis added]:

Ultimately, to close a cash deficit of $3.9 billion, the FRBNY loaned $5.6 billion to Signature, secured by $6.5 billion of collateral Signature had already posted with the FHLB. The process of pledging that collateral held at the FHLB to FRBNY was significantly challenged because Signature did not have existing arrangements in place to pledge any available collateral directly to the FRBNY. As an accommodation, given the urgency of the situation, FHLB agreed to subordinate its interest in Signature collateral to the FRBNY in light of Signature’s critical liquidity needs and its lack of timely viable alternatives.

Regulators closed SVB on March 10 and Signature on March 12, but both reopened as bridge banks on March 13 (post systemic risk exception). Signature’s resolution came March 19 and SVB’s came March 26 (both Sunday nights), when the deposit franchises were sold and the bridge banks put into resolution. Chair Gruenberg wrote in an answer that SVB had $126.5 billion of discount window borrowing outstanding by the time of failure, and Signature had $53.6 billion—a pretty substantial increase in borrowing in two weeks and one week, respectively.

Notably, he also revealed that the SVB borrowing appeared to have little margin and the Signature borrowing appeared to be substantially underwater.

The SVB borrowings of $126.5 from the discount window were against collateral with a book value of $152.2 billion and fair value of $127.5 billion—a haircut of less than 1%. Prior to the changes effective March 13 (after its failure) to discount window margins, even Treasury bills—which had the lowest possible haircuts—had haircuts of 1% of fair value.

Signature’s outstanding $53.6 billion of discount window borrowing at the time of failure was backed by collateral with a book value of $65.3 and fair value of $42.3 billion—a “haircut” to fair value of about -25%.

While the Fed can get valuation wrong, it wouldn’t get it wrong that badly that fast.¹ That is, it seems extremely unlikely, especially with historical context, that this kind of valuation gap would emerge within a week (not least because the Fed margin calls daily); it seems more likely that the Fed received non-collateral security.²

Thus, the valuation details strongly suggest that even prior to the FDIC selling the failed bank franchises, it was creating liquidity for itself at the discount window by providing FDIC guarantees as collateral. This goes one step further than the previous note, which could only demonstrate that the FDIC in some cases appeared to be creating liquidity by selling the underlying collateral of the Fed loans without immediately paying back the loan. The new information suggests that, irrespective of the failed banks’ collateral in the bridge banks, the FDIC was newly borrowing from the Fed with just its guarantees as collateral.

As the Fed initially wrote in its weekly H.4.1 balance sheet release on March 16, following the creation of “other credit extensions” to cover lending to FDIC-run institutions:

The Federal Reserve Banks' loans to these depository institutions are secured by collateral, and the FDIC provides repayment guarantees.

In the newly released hearing record, Chair Gruenberg simply says:

The FDIC, in its corporate capacity, guaranteed all discount window borrowings, and will repay those loans from proceeds from the sale of collateral and other assets of the receivership, any proceeds from the transaction with the acquiring institution, and the Deposit Insurance Fund (DIF).

The new information suggests that these guarantees were not duplicative with collateral. That is, where liquidity was desired over and above available eligible collateral, the FDIC was simply “posting” its guarantee as collateral for new discount window loans to the bridge banks. This is no problem for the Fed, who gets a rock-solid guarantee from the USG as collateral (and still charges a super penalty rate for some reason).³

As I wrote last time, the two most apparent possible reasons for the FDIC to use this mechanism were simply uncertainty over how much the DIF was going to be stressed by future bank failures and the impending bind of the debt ceiling.

Gruenberg sort of alluded to the former, writing:

Since these bridge banks were open and operating national banks that had access to the Federal Reserve's discount window, it was preferable to use this liquidity option to conserve the funds in the DIF for any potential subsequent resolution activity and/or other needs.

However, this does not explain any liquidity creation via getting loans beyond the value of the collateral or maintaining the loans after the underlying collateral is sold.

As I signed off last time:

Should/do these loans count against the FDIC statutory borrowing limit? And the debt ceiling? […] Should they be classified as borrowing from the Fed, something not provided for in the FDIC’s borrowing statute?

To be clear: Aside from the legal questions, there doesn’t seem to be anything inherently sinister about this setup. It may be notable as a financing tool for the FDIC to use (or repurpose in a more legally conforming way) in the future: To the extent it can sell collateral out from under any failed banks’ loans that it assumes and replace that collateral with an FDIC guarantee, it can create liquidity for itself.

Yet, this goes further if the FDIC can also generate new liquidity where no collateral previously existed by posting its guarantees through bridge banks.

9/24/2023 UPDATE – FDIC partial paydown: https://www.linkedin.com/pulse/fdic-borrows-50-billion-pays-down-some-debt-fed-steven-kelly

1/6/2024 UPDATE - Another congressional record update, from the written record for a May 16 hearing of the House Financial Services Committee called “Oversight of Prudential Regulators”: On p. 187, Chair Gruenberg says, “The Fed loans, which are currently covered by an FDIC guarantee rather than collateral…”

Comments welcome via email (steven.kelly@yale.edu) and Twitter (@StevenKelly49). View Without Warning in browser here.

1. Moreover, the NYDFS report cited above shares that some of Signature’s troubles with respect to getting more last minute discount window financing were that the Fed needed time to appropriately value new collateral Signature was trying to bring. i.e., the Fed wouldn’t just massively overpay—especially in a way it never had in history.

2. The steeply negative haircut looks even too big for the BTFP, which lent against the par value of securities. Moreover, Chair Gruenberg specifically said these were “discount window” totals, AND the Fed said only that the bridge banks borrowed from the discount window and that, of the failed banks, only First Republic (pre failure, no bridge bank) borrowed from the BTFP. Also, as noted in the last post, any BTFP borrowing that was being classified as “other credit extensions” would show up in the BTFP reports to Congress—as the $14 billion to First Republic does; yet, only First Republic shows up in that data. It seems clear the bridge banks did not borrow from the BTFP at all. The bridge banks also likely wouldn’t have been eligible for the BTFP even with the FDIC backstop given that the post-Dodd-Frank Section 13(3), unlike the discount window, explicitly calls for borrowers to be solvent—and both had been determined nonviable by their regulators.

3. Gruenberg also wrote that, “per the FDIC's agreement between the receiverships of the bridge banks and the respective Federal Reserve Banks, all loans will be repaid in full before December 31, 2023.”