The Constitutional Case for the Fed's Treasury Backstops

A budget in chains.

A budget locked in chains. One of the more controversial provisions of the CARES Act consists of the $454 billion it allows the U.S. Treasury to devote to "backstopping"  (that is, to supplying partial funding for) the Federal Reserve's emergency lending.

In a previous post, I argued that these Treasury backstops help to preserve the "boundary line" separating fiscal from monetary policy. I've since engaged in a spirited Twitter exchange with Peter Conti-Brown, Dan Awrey, and Kate Judge—all legal scholars specializing in monetary policy. That exchange inspires me to offer this more complete, but qualified, defense of the Treasury's backstopping of risky Fed lending programs.

Leverage, Smeverage

Before I take up the legal case for Treasury backstopping, I want to clear the air of a distracting pseudo-economic rationale for it: the "leverage" argument.  According to Secretary of the Treasury Mnuchin, with the help of $454 billion in CARES Act funds, the Fed can "lever up" its emergency lending by some $4 trillion. Jay Powell has also appealed to the leveraging rationale, saying that every dollar of additional Fed loss-absorption capacity is worth another $10 in Fed emergency lending.

Were the Fed an ordinary bank, this "leveraging" argument might make sense. Ordinary banks can't operate without capital, which allows them to take risks by absorbing unanticipated losses, from bad loans or otherwise. Both economic incentives and legal requirements compel such banks, not only to stay solvent (that is, to not exhaust their capital), but to limit their lending to a finite multiple of their capital. Once it's fully levered-up, an ordinary bank can't lend more unless more capital comes its way.

But the Fed isn't an ordinary bank. It's a fiat money-issuing central bank acting on the federal government's behalf; and such a bank can technically function with no capital at all, and even with negative capital. Unlike an ordinary bank's liabilities, the Fed's "liabilities" are (in John Exter's immortal words), "IOU nothings." The Fed never has to redeem them, except by replacing worn notes with new ones and that sort of thing. If its earnings fall short of its expenses, it can just "print" the difference. It follows that it might remain a going concern even if most of its assets weren't worth diddley. Nor, as Nathan Tankus points out, is there a "statute, court case or any other binding legal constraint… that requires the Federal Reserve to have a positive net worth."

In short, and speaking generally, the Fed can "lever-up" all it likes without a Treasury backstop. The sole caveat is that, if its losses are so great that the present discounted value of its future net earnings is itself negative, it can lose control of inflation. That could happen were the Fed to disburse "helicopter money" on a grand enough scale, or were it forced by the U.S. Treasury to accept a deposit of several trillion-dollar platinum coins. But it's unlikely to happen otherwise.

Finally, from a consolidated government balance sheet perspective, there is little practical difference between $1 trillion in lending financed entirely by Congress and $1 trillion in lending financed mostly or entirely by the Fed. In one case the Treasury issues more debt to finance the lending; in the other, the Fed's liabilities, including bank reserves on which it pays interest, increase. The necessary amounts are borrowed either way, with the Treasury ultimately bearing the interest burden. The only real difference is that, when the Treasury's own liabilities increase, so does the federal debt, whereas when the Fed's liabilities increase, the federal debt stays the same. In other words, to the extent that it isn't backstopped, Fed-funded lending is "off budget."

Dodd-Frank and Fed Risk Taking

Though no law calls for the Fed to be in the black, or otherwise limits its total lending capacity in any direct fashion, the Fed's lending, and its Section 13(3) lending in particular, is subject to important legal limits. The most obvious of these are to be found in Section 13(3) of the Federal Reserve Act itself, as amended by the Dodd-Frank Act.

In answering the question, "Is Treasury investment in the Fed’s emergency lending necessary?," Peter Conti-Brown observes in a Brookings article  that "Nowhere in any part of the Federal Reserve Act does Fed lending require Treasury participation, nor did Congress prevent the Fed from taking losses in its emergency lending." The first part of this statement is true, so far as it goes. But the second seems to me somewhat misleading. In defending it Peter notes that, although the law calls for the Fed's 13(3) lending to be “indorsed or otherwise secured" to the lending Fed bank's satisfaction, this falls short of dictating that "the Fed avoid taking first losses in an emergency lending facility." But Peter overlooks another Section 13(3) clause—13(3)(b)(i)—according to which the Fed must "establish, by regulation, in consultation with the Secretary of the Treasury, the policies and procedures designed to ensure … that the security for emergency loans is sufficient to protect taxpayers from losses." "It seems to me," I said in our Twitter exchange, "that this can be taken to mean that the Fed isn't supposed to lose money."*

Congress, on the other hand, is mostly free to dispose of taxpayers' money as it sees fit. That includes appropriating funds for the specific purpose of covering losses from insufficiently secured Fed loans. The Fed can, in other words, lose money to the extent that Congress has appropriated money for it to lose, but not otherwise. Treasury backstops sanctioned by Congress are nothing other than such appropriations.

That, I submit, is the real rationale for Treasury backstopping of risky Fed lending. But to defend this view, I must trace that rationale's roots beyond the Federal Reserve Act, to their underpinnings in the U.S. Constitution.

Constitutional Underpinnings

Because those underpinnings are very clearly explained in "Congress' Power of the Purse," a 1988 Yale Law Review  article by Kate Stith, I need only summarize them here. Stith's lengthy article is loaded with supporting evidence and arguments.

By stipulating that "No Money shall be drawn from the Treasury, but in Consequence of Appropriations made by Law," the Constitution, Stith observes, "places the power of the purse in Congress." That assignment of power, she adds, "is at the foundation of our constitutional order," for it's by controlling the things on which public monies are spent that Congress "defines the contours of the federal government."

Congress' power of the purse has as corollaries two "governing principles." The first, which Stith calls the "Principle of the Public Fisc," "assert[s] that all monies received from whatever source by any part of the government are public funds." The other is a "Principle of Appropriations Control" that "prohibit[s] expenditure of any public money without legislative authorization." Together these principles imply that "Agencies and officials of the federal government may not spend monies from any source, private or public, without legislative permission to do so." Congress, on the other hand, has not only the power but the duty to exercise legislative control over federal expenditures. "If Congress permits the Executive access to the public fisc without effective appropriations control [it] abdicates, rather than exercises, its power of the purse."

Furthermore, Stith claims,

it is not enough for Congress to direct federal agencies to produce a better world. …Congress must affirmatively authorize the funds to do the job. …Even where the President believes that Congress has transgressed the Constitution by failing to [fund some] activity, the President has no constitutional authority to draw funds from the Treasury to finance the activity. Spending in the absence of appropriations is ultra vires.

Finally, and crucially,

Where Congress thus denies appropriations, the denial is not merely a determination that the public fisc cannot afford spending any money on that activity. By such appropriations legislation, Congress decides that, under our constitutional scheme, for the duration of the appropriations denial, the specific activity is no longer within the realm of authorized government actions.

Fitting the Fed In

What has this to do with the Federal Reserve System?  Were the Fed just another government agency, the answer would be easy, for in that case it would be obliged to surrender to the Treasury, for disposal by Congress, any funds it acquired from any source, including the Reserve Banks' earnings. It could not fritter away part of those funds unless it had Congress' express permission to do so, in the shape of a corresponding appropriation.

Alas, things aren't quite so simple. Although the Board of Governors is an independent government agency, the Federal Reserve Banks are so many private corporations. Because the System's earnings come entirely from those Reserve Banks, this arrangement raises doubts concerning whether those earnings qualify as "public funds." Until 2016, those doubts were reinforced by the fact Reserve Banks' surplus revenues—meaning their revenues after deducting the System's operating costs, dividends on member bank stock, and funds sufficient to keep the banks' surplus capital equal to their paid-in capital—were remitted to the Treasury, not as a matter of law, but only in accordance with Board policy.

The passage of the 2015 FAST Act has, however, eliminated much of this ambiguity. That measure capped the Federal Reserve System's surplus capital at $10 billion, while requiring that any surplus funds beyond those needed to maintain that amount be surrendered to the Treasury. The 2018 Economic Growth Act subsequently lowered the Fed's surplus capital limit to $6.825 billion. These changes make it impossible to continue to regard the Fed's Treasury remittances as voluntary contributions of "private" funds. Congress has instead asserted its right to such remittances.

By so doing, Congress asserts its right to determine the how the Reserve Banks' earnings are employed, albeit with one important exception. That exception is provided for by Federal Reserve Act's stipulation that any "assessments levied" by the Board upon the Reserve Banks "to defray its [that is, the Board's] estimated expenses and the salaries of its members and employees… shall not be construed to be Government funds or appropriated moneys." Funds so employed are considered private funds. Congress is thus denied the power to deprive the Board of an operating budget, even though that Board is a government agency. This is the (admittedly slim) legal substance of the Fed's vaunted budgetary autonomy.**

While Congress can't deprive the Board of its sustenance, and while it may take a generous view of what that sustenance costs, the law doesn't seem to supply grounds by which the Board can dispose of other Reserve Bank earnings without Congress' express permission. Seen in this light, the Dodd-Frank requirement "that the security for emergency loans is sufficient to protect taxpayers from losses," far from being nugatory, supplies additional clarity, by specifically preventing the Board from treating such losses as a component of its ordinary expenditures and, therefore, as funds to which it's entitled without Congress' express permission.

Not so Novel

It remains for me to address Peter Conti-Brown's claim that, far from being an application of long-standing Constitutional principles, the Treasury's backstopping of the Fed's recent emergency lending initiatives is sui generis. Peter makes this claim both in his previously-linked Brookings article and in a subsequent Macro Musings interview. In the last he says that backstops are something

we've never seen before in the United States, in the history of the relationship between the Fed and Treasury and the Fed and Congress. …$454 billion, not trivial money, is appropriated to one part of the government for the exclusive purpose of investing in another part of the government, this is, these are Treasury funds that can only be used as loans, loan guarantees or investments in Fed programs or facilities.

In his Brookings piece, Peter adds that backstops were not even a feature of "the vast experimentation in emergency lending that we saw in the 2008 financial crisis."

Such claims appear to pose a serious challenge to my understanding of backstops as a Constitutionally-required accompaniment to risky Fed lending. After all, if they're so necessary, why haven't we seen them before?

However, a closer look at history reveals that the challenge isn't as daunting as it seems. This is true in part because until recently the Fed simply didn't make many risky loans. Until 2008, the Fed hardly used its 13(3) lending powers at all. The only exceptions were 123 mostly very small loans it made between 1932 and 1936, all of which were well secured.

The Fed did make potentially risky 13(3) loans during the 2007-8 crisis. But then as now it relied on either Treasury or other backstops to cover the risks it took. The TALF, for instance, involved a $20 billion Treasury backstop. JP Morgan Chase and certain AIG affiliates backstopped the Fed's Maiden Lane loans. The government rescues of Citigroup and Bank of America, to which the Fed contributed, were also structured so that the Fed was unlikely to bear any losses. In those instances both the Treasury and the FDIC stood to lose money before it did. In fact, the Fed lost no money at all from any of its 2008-era 13(3) loans. It did incur losses on its outright agency MBS purchases, but those purchases were authorized not by Section 13(3) but by Section 14(b) of the Federal Reserve Act.

Although the Fed made no risky 13(3) loans before 2008, it made other risky loans before then, using a separate 13(b) authority granted it between 1934 and 1958. 13(b) loans were aimed not at banks or "systemically important" nonbank firms, but at smaller businesses, and as such could be secured by any sort of collateral. In practice, this meant that many were hardly secured at all, and were therefore very risky—so risky, indeed, that by 1940 the Fed's 13(b) lending had yielded it a return of minus 3 percent.

If Peter's view of things were correct, the Fed should have been allowed to lose that money without Congress' express permission, in the shape of a Treasury backstop or otherwise. If, on the other hand, I'm right in thinking that the Constitution requires such permission, Congress would have had to appropriate funds to cover any such losses.

The Fed's 13(b) loans were, in fact, backstopped by $140 million in Treasury-supplied funds, and limited to a total of $280 million, which was equal to the sum of the Treasury's contribution and the Reserve Banks' own surplus capital at the start of the program. The Fed might, in other words, have lent to the full limit of its 13(b) authority, and lost every nickel, without going broke. That's taking Congress' power of the purse very seriously indeed! For comparison, the Fed's recently-unveiled Main Street New Loan Facility has a backstop equal to just one-twelfth of that facility's potential lending capacity. In short, by New Deal standards Congress today is playing relatively fast and loose with its "duty to exercise legislative control over federal expenditures."

Backstop or Back to Basics?

I said that I would offer a "qualified" defense of the Treasury's backstopping of risky Fed lending programs. My argument is that such backstopping is justified in so far as it respects Congress' "power of the purse," whereas any risky lending the Fed does without such backstopping would contravene that power. When Jay Powell said on TODAY that the Fed lending was limited by its ability to take losses, whether he knew it or not, he was appealing to these Constitutional principles, not making stuff up.

BUT, Treasury backstops are only one way for the Fed to abide by both the Constitution and the specific requirements of the Federal Reserve Act. The other, more straightforward way is by renouncing altogether its authority to make loans for which adequate security is lacking. That is the course William McChesney Martin chose to take in the 1950s when, having first opposed legislation that would have strengthened the Fed's 13(b) lending powers, he then encouraged Congress to deprive the Fed of those powers altogether, and to instead leave the task of risky emergency lending to the Reconstruction Finance Corporation and the recently established SBA. The Fed, Martin testified,

would favor neither the financing of such institutions by the Federal Reserve by purchase of stock or otherwise, nor the exercise by the System of any proprietary functions. …Basically, our concern stems from the belief that it is good government as well as good central banking for the Federal Reserve to devote itself primarily to objectives set for it by the Congress, namely, guiding monetary policy and credit policy so as to exert its influence toward maintaining the value of the dollar and fostering orderly economic growth.

Were the Fed to take a similar stand today, Congress would in turn have to establish an alternative, strictly fiscal emergency lending facility—a new RFC, perhaps. It would also have to pluck-up the courage whenever necessary to fund that facility to whatever extent it considered sufficient to avert a crisis.

Compromises between such a strictly fiscal risky lending arrangement and the current set-up might also be considered. One such compromise has been proposed by James Nason and Charles Plosser, who suggest a new Fed-Treasury "accord" by which the Treasury would agree to trade its own securities for any non-Treasury securities the Fed acquires during an emergency after a set period of time. Congress would then have to authorize and fund the Treasury's commitment. "With such an accord," they say, "fiscal policy remains outside the province of the Fed, but policy has the flexibility to respond to a crisis in the short run." A stricter version of this proposal would call for the Treasury to purchase, not only any non-Treasury securities the Fed acquires, but all of its less than fully secured 13(3) loans. As the late Marvin Goodfriend testified in 2013, emergency Fed lending, like emergency Fed asset purchases, "should be authorized before the fact by Congress in  its oversight role, but only as a 'bridge loan' and accompanied by a 'take out' arranged and guaranteed by Congress."

It is of course unfair to blame the Fed for not drawing a line in the sand in the middle of an unprecedented crisis. Back in 2008, Goodfriend says, the Fed found itself

in a no-win situation. Given its wide powers to lend, the Fed could disappoint expectations of accommodation and risk financial collapse or take on expansive, underpriced credit risk… with the implied promise of similar actions in times of future turmoil.

Today it may likewise be said that the Fed had little choice but to take on risks and responsibilities that a pusillanimous and small-minded Congress wouldn't take on itself. Karen Petrou is nevertheless right to fear that it will end up regretting its further entanglement in the nation's fiscal affairs. "Congress and the Administration," she says, "now get to stand back, taking credit if the economy recovers and knowing where 20th and Constitution Avenue is if it doesn’t."

As for the rest of us, we must ask what point there is in giving Congress the power of the purse, if instead of defining "the contours of the federal government" itself, it gives a free hand to a coterie of unelected officials.


*According to April 2020 CRS report issued as I wrote this essay, but only discovered by me much later, "ESF backing of Fed facilities may reflect the significant and uncertain economic risks associated with COVID-19. Use of the ESF may be seen to allow these facilities to meet the Dodd-Frank Act’s (P.L. 111-203) requirement that the Fed’s 13(3) lending is secured 'sufficient[ly] to protect taxpayers from losses'." The report adds that "given ESF losses would ultimately be borne by taxpayers, it is unclear if this requirement is actually being met in a broader sense."  As I explain here, although backstops obviously don't prevent Fed losses from being borne by taxpayers, they do prevent unappropriated funds from being employed to cover them. [Note added 12/20/2020.]

**The provision designating as private funds the Board's assessments covering its operating expenses dates from the Banking Act of 1933. Revealingly, as Peter Conti-Brown notes in his 2015 article "The Institutions of Federal Reserve Independence"  (p. 280n102), a decade earlier the U.S. Comptroller General had instead expressly designated those same assessed amounts as public funds "subject to various restrictions and impositions."

In this same article Peter appears to argue that the provision allowing the Board to fund its expenses by assessments against the Reserve Banks does not allow it to employ the System's interest earnings from its SOMA portfolio for that purpose. Assuming that I've understood him correctly, I confess that I'm unable to grasp his reason for taking this view. The proceeds from the Fed's open-market operations are part of the Reserve Banks' earnings, the rest of which consist mainly of fees they collect from member banks. The Board’s assessments are made, not against Reserve Bank fee income only, but against the fungible sum of all Reserve Bank income; and the appropriations exception appears to hinge, not on the ultimate source of funds employed by the Board, but on the way in which it employs them.