More on those secret Federal Reserve loans to banks

The claim that the Federal Reserve extended trillions of dollars in secret loans to banks continues to be spread. Here at Econbrowser we will continue to try to correct some of the misunderstanding that is out there.

Consider for example this item from the Levy Institute blog written by University of Missouri Professor L. Randall Wray, which begins:

It literally took an act of Congress plus a Freedom of Information Act lawsuit by Bloomberg to get [Bernanke] to finally release much of the information surrounding the Fed's actions. Since that release, there have been several reports that tallied up the Fed's largess. Most recently, Bloomberg provided an in-depth analysis of Fed lending to the biggest banks, reporting a sum of $7.77 trillion.

This is a common misunderstanding. As the reporters for the Bloomberg story verified to me personally, their $7.77 trillion figure did not come from the records that Bloomberg obtained under the FOIA. Instead, their figure came from this article published by Bloomberg on March 31, 2009. That original $7.77 trillion estimate in turn was based entirely on publicly available sources which were being quite widely discussed at the time.

Another key fact that seems to be underappreciated by those passing along these numbers is that the vast bulk of this $7.77 trillion figure was never lent at all. For example, $1.8 trillion of the total is attributed to the Commercial Paper Funding Facility. But the fact is that the maximum quantity of loans ever outstanding under this program only came to $351 B. The $7.77 trillion also includes $900 B for the Term Asset-Backed Securities Loan Facility, whose maximum outstanding balance was only $49 B, and includes $540 B for the Money Market Investor Funding Facility
which never lent so much as a single dime.
The table below breaks down the individual components of the $7.77 trillion and compares them with the actual maximum amount lent under each program. Numbers in the first column come from Bloomberg (2009) and numbers in the second column come from the Fed's weekly H.4.1 releases.

Theoretical and actual Federal Reserve lending by category (billions of dollars)

Category Theoreticalpotential
Actual maximumDate of maximum

1Net Portfolio CP Funding1,800351Jan 21 2009
2Term Auction Facility900493Mar 4 2009
3Term Asset-Backed Loan Facility90049Mar 17 2010
4Currency Swaps/Other Assets606 583Dec 17 2008
5MMIFF5400
Sum of 1 through 54,7461,350Dec 17 2008
6GSE Mortgage-Backed Securities1,0001,129Jun 23 2010
7GSE Debt Purchases600169Mar 10 2010
8Commitment to Buy Treasuries300?
9Other1,120
Sum of 1 through 97,766

The $7.77 trillion also includes $1 trillion for the Fed's purchases of mortgage-backed securities. These MBS were guaranteed by Fannie Mae and Freddie Mac. Since the U.S. Treasury had already taken over Fannie and Freddie before the Fed made any of these purchases, essentially the guarantor was the U.S. Treasury. It makes no sense to claim that the Fed's purchases of these securities in any way increased the net liabilities or commitments of the U.S. government. Moreover, at the date when the first 5 categories in the table reached their combined maximum (Dec 17, 2008), the Fed was not holding any MBS. The Fed's MBS holdings would not reach a trillion dollars until Feb 17, 2010, at which date the combined outstanding balance on the first 5 categories was down to $71 B.

The $7.77 trillion also includes $600 B for outright purchases of Fannie and Freddie debt (which turned out to be only $169 B), and another $300 B for Treasury debt. Buying Treasury debt is what the Federal Reserve has always been doing, and certainly the Fed is holding much more today than it did before the crisis. I have made no effort in the table above to quantify what the "actual" purchases of Treasury securities turned out to be, but wish to point out that, if it is the intention of anyone to talk about the magnitude of "Fed lending to the biggest banks", it makes no sense to include purchases of Treasury bonds or either of the previous two categories in the total.

The remaining 14% of the $7.77 trillion comes from 12 other separate items whose details I have not tried to track down.

Now, as for the relevance of the observation that the sum of the "potential" amounts for the first 5 categories is almost 4 times the size of their actual combined amounts at any date, I return to Professor Wray's discussion:

the Fed quibbles about the differences among lending, guarantees, and spending. For the purposes of this blog I will accept these differences and call the sum across the three "commitments." In spite of what Bernanke claims, these do commit "Uncle Sam" since Fed losses will be absorbed by the Treasury.

But the discrepancies between the "potential" and "actual" columns in the table above cannot reasonably be described as "guarantees." For example, the basis for the $1.8 trillion figure for CPFF is the following statement issued publicly by the Fed on Oct 14, 2008:

Limits per issuer
The maximum amount of a single issuer’s commercial paper the SPV may own at any time will be the greatest amount of U.S. dollar-denominated commercial paper the issuer had outstanding on any day between January 1 and August 31, 2008.

The $1.8 trillion was calculated by assuming that every single institution that qualified would in fact issue new commercial paper that would then be purchased by the Fed up to this maximum amount. Notwithstanding, the same Fed statement continued:

The Federal Reserve reserves the right to review and make adjustments to these terms and conditions, including pricing and eligibility requirements.

I do not believe that "guarantees" is the correct term to use for this or most of the other items included in the Bloomberg tally.

A separate issue concerns whether it is appropriate to add together outstanding loan balances as of different points in time in order to generate a total, for example, adding MBS to items 1-5. Here Professor Wray reasons by analogy:

Think about it this way. A half dozen drunken sailors are at the bar, and the bartender refills their shot glasses with whiskey each time a drink is taken. At any instant, the bar-keep has committed only six ounces of booze. That is a useful measure of whiskey outstanding. But it is not useful for telling us how much the drunks drank. Bernanke would like us to believe that if the Fed newly lent a trillion bucks every day for 3 years to all our drunken bankers that we should total that as only a trillion greenbacks committed.

But here's what's wrong with that analogy. The omitted detail is that after each sailor takes a glass of whiskey, he returns the glass, filled with more whiskey than it originally held, to the bartender. In other words, the loans were always paid back with interest before a new loan was extended. If an individual sailor ended up consuming more than one glass, somebody other than the bartender must have paid for it, if the maximum tab with the bartender that any individual sailor ever had at a single point in time was a single glass.

If you take the position that each new loan should be added as a running contribution to some total, then you are led to maintain that when the Fed loans $1 B to Bank A in the form of a 30-day loan, and loans $1 B to Bank B in the form of an overnight loan that is repaid and renewed each day, then the Fed has 30 times the exposure to Bank B as it does to Bank A. You are further led to infer that the Fed could have lost $1 B in lending to Bank A but somehow could have lost $30 B lending to Bank B. And you are led to infer that it is 30 times safer to make a 1-month loan than it is to make a series of overnight loans in the same amount. Good luck managing your or anybody else's finances, if that's your way of thinking.

But Professor Wray goes on to speak admirably about an analysis by his student James Felkerson that does exactly that, and concludes that the Fed lent not $7.77 trillion but instead $29 trillion. For example, Felkerson takes the gross new lending under the Term Auction Facility each week from 2007 to 2010 and adds these numbers together to arrive at a cumulative total that comes to $3.8 trillion. To make the number sound big, of course you want to count only the money going out and pay no attention to the rate at which it is coming back in. If instead you were to take the net new lending under the TAF each week over this period-- that is, subtract each week's loan repayment from that week's new loan issue-- and add those net loan amounts together across all weeks, you would arrive at a cumulative total that equals exactly zero. The number is zero because every loan was repaid, and there are no loans currently outstanding under this program.

But zero isn't quite as fun a number with which to try to rouse the rabble.