The Federal Reserve’s balance sheet is drawing renewed attention this year. Chairman Kevin Warsh, Treasury Secretary Scott Bessent, and other policymakers and commentators have said that contracting it would be a constructive way to reduce the distortionary effects of the Fed’s financial market “footprint.”¹Indeed, one of the five task forces announced by Chairman Warsh in June is devoted to “Balance Sheet Policy” and has been asked to “examine the costs, benefits, and institutional implications of the Federal Reserve’s current balance sheet regime.”²Many proponents of balance sheet reduction have suggested various measures that would make the process easier.³Some have proposed allowing banking organizations to count their unused discount window capacity toward various regulatory and supervisory liquidity requirements.⁴That proposal is unwise, because it would undermine the avowed purpose of those requirements and likely increase the moral hazard associated with the Fed’s lending backstop. A better way to facilitate balance sheet reduction is for the Fed to reconsider its recently acquired antipathy to high-frequency fluctuations in money-market spreads. More fundamentally, the financial market distortions attributable to the Fed’s “footprint” are more closely related to the composition of the Fed’s balance sheet than to its total size. A more effective approach to reducing those distortions would be to shift toward a bills-only portfolio by shedding agency mortgage-backed securities and longer-term Treasury securities.

An array of regulatory and supervisory rules requires banks to hold enough liquid assets to meet various stipulated minimums. The Liquidity Coverage Ratio requires large banks to hold eligible “High Quality Liquid Assets” sufficient to cover total net cash outflows over 30 days in a “stress scenario.”⁵The Net Stable Funding Ratio is a similar requirement applicable to large banks and utilizing a one-year stress horizon.⁶A somewhat different set of large banks is required to conduct Internal Liquidity Stress Tests and must hold “Highly Liquid Assets” sufficient to cover projected net stressed cash outflows over 30 days.⁷Liquidity standards associated with resolution planning (“living wills”) can also effectively constrain liquidity holdings at large banking organizations.⁸For convenience, I will refer to these collectively as “liquidity constraints.”⁹Proponents differ regarding which of these requirements discount window capacity should count toward. It should be noted that High Quality Liquid Assets (HQLA) and Highly Liquid Assets that have been pledged to a Reserve Bank’s discount window but are not securing outstanding borrowing do count toward their respective liquidity constraints. What is in question is whether unused discount window capacity supported by pledged assets that do not qualify as HQLA should count toward liquidity constraints.

Why Are There Minimum Liquidity Requirements?

The liquidity constraints described above were created after the Great Financial Crisis under authority of the 2010 Dodd–Frank Act,¹⁰and the record shows that they were aimed at mitigating the moral hazard effects created by the expectation that Federal Reserve lending would be forthcoming in the event of financial distress. Their plain purpose was to counteract banks’ incentive to hold too little by way of liquid assets. In the congressional hearings leading up to the drafting of the Act, Secretary of the Treasury Timothy Geithner noted that capital and liquidity standards had been “too low.”¹¹In the Federal Reserve’s congressional testimony, Governor Daniel Tarullo likewise pointed to the moral hazard associated with the “‘too big to fail’ problem” as a motivation for stiffer capital and liquidity requirements.¹²

Unless countervailing steps are taken, the belief by market participants that a particular firm is too big to fail, and that shareholders and creditors of the firm may be partially or fully protected from the consequences of a failure, has many undesirable effects. It materially weakens the incentive of shareholders and creditors of the firm to restrain the firm’s risk-taking, provides incentives for financial firms to become very large in order to be perceived as too big to fail, and creates an unlevel competitive playing field with smaller firms that may not be regarded as having implicit government support. Creation of a mechanism for the orderly resolution of systemically important nonbank financial firms, which I will discuss later, should help remediate this problem. In addition, capital, liquidity, and risk-management requirements for systemically important firms will need to be strengthened to help counteract moral hazard effects, as well as the greater potential risks these institutions pose to the financial system and to the economy.¹³ (emphasis added)

When the bill emerged in mid-2010, the authors pledged that “taxpayers must never again hear that a company is too big to fail,” a message frequently invoked to justify Fed lending to forestall closure. To that end, they created a new body, the Financial Stability Oversight Council, “specifically charged with making sure our large, more complex institutions have more stringent capital requirements, leverage ratios, liquidity requirements, and risk management tools.”¹⁴

When US banking regulators, after passage of the Act, issued a notice of proposed rulemaking on liquidity standards, they tied the need for the liquidity coverage ratio (LCR) directly to the moral hazard engendered by Federal Reserve lending.

A maturity mismatch in a bank’s balance sheet creates liquidity risk. Banks will typically manage this liquidity risk by holding enough liquid assets to meet their usual net outflow demands. The presence of a central bank that can serve as a lender of last resort provides an element of liquidity insurance, which, as is often the case with insurance, creates moral hazard. Because of the presence of a lender of last resort, banks may not hold socially optimal levels of liquid assets. The LCR buffer established by the proposed rule offsets the moral hazard to a degree, and lowers the probability of a liquidity crisis and may limit the severity of liquidity crises when they do occur.¹⁵(emphasis added)

When the agencies issued the final rule in October 2014, they again connected the need for the liquidity coverage ratio to expectations of lender-of-last-resort support.

The agencies believe that the most recent financial crisis demonstrated that large, internationally active banking organizations were exposed to substantial wholesale market funding risks, as well as contingent liquidity risks, that were not well mitigated by the then-prevailing liquidity risk management practices and liquidity portfolio compositions. For a number of large financial institutions, this led to failure, bankruptcy, restructuring, merger, or only maintaining operations with financial support from the Federal government. The agencies believe that covered companies should not overly rely on wholesale market funding that may be elusive in a time of stress, not rely on expectations of government support, and not rely on asset classes that have a significant liquidity discount if sold during a period of stress. . . . Covered companies are expected to maintain appropriate levels of liquidity without reliance on central banks acting in the capacity of lenders of last resort.¹⁶(emphasis added)

In a speech shortly thereafter laying out a broad analysis of the emerging liquidity regulation regime, Governor Tarullo also tied the constraints to the incentive effects of the lender of last resort (LOLR).

The classic responses to this classic problem [of bank runs] have been a combination of deposit insurance and discount window access. . . . Regulatory requirements were imposed to guard against the moral hazard that both programs could create. . . . The US banking agencies have worked with other regulators in the Basel Committee on Banking Supervision to develop two quantitative liquidity standards–the LCR and the NSFR. These standards complement the bank capital framework and resolution mechanisms that have been agreed internationally. They will also help central banks limit the use of LOLR.¹⁷(emphasis added)

Governor Michael Barr, who served in the US Treasury from 2009 to 2011, when the Dodd–Frank Act was being crafted, provided a more recent portrayal of the original motivation for liquidity requirements:

After the Global Financial Crisis (GFC), bank regulators designed and adopted rules that require the largest banks to hold a portfolio of HQLA in amounts calibrated to what their needs would be in stress scenarios. This approach, along with capital requirements, was designed to help the largest banks survive stress events. Rather than allowing these firms to threaten the viability of the financial system, necessitating their bailouts, we instead said they have to basically self-insure by holding an appropriate amount of HQLA.¹⁸

Official commentary makes clear, then, that a central motivation for implementing liquidity constraints in the 2010s was that without them, banks would hold insufficient liquid assets in anticipation of being able to draw on central bank lending or other government support.

Abundant commentary from outside the regulatory sphere made the same point. To cite just one example, three economists at the Board of Governors, Mark Carlson, Burcu Duygan-Bump, and William Nelson, posted a paper in early 2015 directly addressing the relationship between liquidity regulations and the lender of last resort titled “Why Do We Need Both Liquidity Regulations and a Lender of Last Resort?” They argued that liquidity regulations “serve as a tax on liquidity risk by requiring banks to hold low-yielding assets in rough proportion to the amount of liquidity risk they take. As such, they provide an incentive for banks to internalize the externalities associated with liquidity crises, at least to some extent, and accordingly minimize their occurrence.”¹⁹One of the externalities they identified is “LOLR lending” that is “rife with moral hazard and therefore best avoided if at all possible by having robust liquidity and capital regulations.”²⁰They also observed that liquidity buffers provide authorities with time to respond to a potential default, assess the condition of the institution, and arrange an orderly resolution.

The Economics of Liquidity Constraints in Lending Arrangements

The most natural way to think about the connection between the discount window and liquidity constraints is to recognize that the discount window is a secured line of credit, closely analogous to a private line of credit that a bank might provide to a borrowing firm.²¹Through the discount window, a Federal Reserve Bank stands ready, under specified conditions, to provide credit at the initiative of the borrowing bank. Similarly, a private line of credit provider stands ready, under specified conditions, to lend at the borrower’s request. A private line of credit commits the lender to extending credit at a pre-agreed spread over a reference interest rate and thereby provides some insurance against deterioration in the creditworthiness of the borrower. Similarly, the Fed’s discount window provides credit at a posted “discount rate” that is set at a specific relationship to the Federal Open Market Committee’s target range for the federal funds rate; since March 2020 the discount rate has been set at the top of the target range.

Private line of credit contracts generally include loan covenants, conditions that borrowers are required to meet in order to qualify for automatic credit extension. Loan covenants often include periodic financial reporting and maintenance of specified financial conditions such as capital ratios and minimum holdings of liquid assets.²²Similarly, access to “primary credit” at a Federal Reserve Bank is limited to banks that are “in generally sound financial condition,” a qualitative assessment linked to supervisory ratings and capital and liquidity measures.²³In practice this is a fairly broad category. Banks that do not qualify for primary credit may qualify for “secondary credit,” which carries a higher interest rate, currently 50 basis points above the rate on primary credit.

Liquidity covenants can tilt a borrower’s incentives toward more conservative liquidity management practices. Once a line of credit agreement is in place, the borrower may benefit from taking on risks that increase the lender’s exposure. A liquidity covenant mandates an amount of self-insurance and limits risky behavior that is, ex post, disadvantageous to the lender.²⁴Liquidity constraints play a similar role with banks that have access to the discount window, as noted in the liquidity regulation rationales quoted above. They require banks to bear more of the cost of insuring against liquidity stress and thereby offset, at least in part, the incentive to rely on central bank credit.

In competitive markets for private line of credit arrangements, borrower and lenders presumably choose contractual provisions that are efficient for them, ex ante, given the informational constraints and agency problems they face. The stringency of liquidity constraints and other private line of credit covenants presumably balances the benefits of improved alignment of borrower incentives against the costs associated with more prudent liquidity management. Similarly, the implementation of a set of formalized liquidity constraints for banking organizations following the Great Financial Crisis presumably reflects a post-Crisis assessment that the benefits of offsetting the moral hazard of access to central bank credit outweighed the cost of the additional restraints on bank liquidity management. Indeed, documentation surrounding various liquidity regulations published by the Basel Committee on Banking Supervision and by US banking regulators explicitly describes the requirements in terms of such balancing, including their role in constraining the incentive effects of central bank credit.

From this perspective, counting unused discount window capacity toward a liquidity constraint would require an offsetting upward adjustment to the constraint if policymakers wished to preserve its original effectiveness. Consider a business with a $200 million private line of credit with a covenant requiring the borrower to hold $100 million in cash. If the borrower were permitted to count the unused line toward the covenant, the lender would have to raise the covenant to $300 million to preserve the original requirement that the borrower hold $100 million in actual cash. Without that adjustment, the change would weaken the agreement’s original protections and reduce the self-insurance required of the borrower.

The same reasoning applies to the liquidity constraints implemented following the Great Financial Crisis. Allowing unused discount window capacity to satisfy liquidity constraints without an offsetting increase would reduce the amount of liquid assets banks must hold. This would amount to a substantive loosening of liquidity requirements, not an innocuous technical clarification. I am not aware of any proposals to count unused discount window capacity toward liquidity constraints that include such an offsetting increase in liquidity requirements. Indeed, the explicit purpose of the proposals is to reduce required holdings of reserve account balances, in order to reduce aggregate demand for reserve balances and thereby facilitate reducing the overall size of the Federal Reserve’s balance sheet.

The case for counting unused discount window capacity toward liquidity constraints without an offsetting adjustment is therefore equivalent to simply claiming that existing liquidity constraints are currently set too high.²⁵That may or may not be true, but proponents have not, to my knowledge, provided an analytical case for that proposition. If such a case can be made, regulators should simply alter the liquidity regulations directly and transparently. Using discount window capacity to reduce liquidity constraints would obscure the actual policy shift while weakening the effects of those requirements.

These proposals would undermine liquidity regulations that were explicitly designed to discourage discount window borrowing. As noted above, a major lesson legislators and regulators took from the financial crisis was that capital and liquidity standards had been too low. They implemented liquidity constraints specifically to counteract the tendency of banks, particularly large banks, to hold too few liquid assets and to rely too heavily on short-term wholesale funding that can disappear rapidly. Capital requirements were strengthened as well because run risk ultimately comes back to capital, which is what ultimately ensures that adverse liquidity events can be absorbed without undue strain. The post-Crisis regulatory framework was thus specifically designed to encourage self-insurance and reduce reliance on the discount window during episodes of financial distress. Allowing unused discount window capacity to reduce required liquidity would weaken that framework without evidence.

A Historical Perspective

Proposals to count discount window access toward liquidity requirements should be viewed in the broader context of the history of Fed lending doctrine. It is true that the discount window was at the center of Federal Reserve Bank operations envisioned by the founders, but it is clear from the historical record that lending was intended as a means of conducting monetary policy, that is, a means of engineering variations in the supply of high-powered money (reserve balances and paper note supply) in response to fluctuations in demand, something the preceding regime had failed to do reliably due to cumbersome legal restrictions on national bank note issue.²⁶Open market operations soon began to eclipse the discount window as the principal monetary policy instrument, and after the Treasury–Fed Accord in 1951 the window took a subservient role in monetary policy operations. In the 1960s, the Fed began lending to rescue creditors of failing institutions at the behest of other bank regulators that were reluctant to shut them down, a practice that expanded over subsequent decades to create “Too Big To Fail,” the expectation that the Fed would intervene to mitigate ex post costs, both real and political, of resolving failing financial institutions.²⁷Apparently, the Fed has found it difficult to limit ex post support to align with an ex ante optimal lending policy, a conclusion that is implicit in the 2009 testimony of Tarullo and Bernanke. The resulting sequence of expanding precedents incented fragile funding strategies at large financial institutions in the decades leading up to the Great Financial Crisis. Reflecting the influence of the “New Credit View” of the 1990s, Fed credit extension expanded dramatically during the crisis. The Dodd–Frank liquidity constraints were part of a broader effort to tighten ex ante regulations to prevent the occurrence of the events that tended to elicit Fed lending. Efforts to limit Fed lending were less robust.²⁸

Sixteen years on from the DFA, the predictable cyclical backlash to regulatory tightening keeps gathering steam. Proposals to count discount window capacity toward liquidity constraints are part of a broader movement to “normalize” the discount window and encourage tolerance of Fed credit allocation as a routine component of bank liquidity management. That effort gained traction after the failure of Silicon Valley Bank, when regulators cited impediments to SVB borrowing at the discount window, due to “stigma” and a lack of collateral pre-positioning, as contributing to the panicked nature of the resolution, downplaying the decision not to invoke the Dodd–Frank Act Orderly Liquidation Authority, SVB’s fundamental solvency challenges, and bank management failures.²⁹Highlighting discount window readiness deflects attention from more fundamental supervisory failings and safety net incentive effects.

A More Useful Approach Would Be to Reassess “Ample Reserves”

Two further observations concerning the Fed’s balance sheet are worth noting here. First, a direct and promising way to ease the path toward balance sheet reduction would be for the Fed to reconsider its analysis of the level and volatility of money market interest rate spreads. Twice in recent years the Fed has stopped reducing its balance sheet: once in response to an episode of repo rate volatility, and once as the spread between repo rates and the rate of interest on reserves widened. This aversion to money market volatility is a break from past practice. The Fed traditionally has “looked through” transitory high-frequency movements in money market rates, such as quarter- and year-end rate spikes, because longer-term rates (beyond one month) were more essential to monetary conditions and remained anchored by expectations that the Fed would achieve its funds rate target, on average.³⁰Longer-term rates were anchored by the interest rate smoothing that characterized FOMC rate setting.³¹The same logic should apply now when the interest rate on reserves anchors market rates. In the September 2019 repo rate episode, for example, longer-term yields were virtually unchanged.³²If temporary money-market volatility has little effect on broader financial conditions, tolerating more of it could permit a smaller balance sheet without weakening liquidity regulation.

Similarly, the Fed has traditionally allowed market forces to determine average spreads between the policy rate and various money market rates, and the economic case for tightly managing those spreads is not obvious.³³The Fed has been operating under the assumption that a stable demand curve relates the spread of the funds rate over the interest rate on reserves to the aggregate quantity of bank reserves, and that bank reserves must remain large enough to avoid the “steep” part of the demand curve. Econometric estimates of the position of the demand curve are uncertain, however. Moreover, the demand curve appears to have gradually shifted outward during quantitative easing as market participants adapted to a dramatically larger aggregate supply.³⁴This suggests that sustained reductions in supply would gradually induce analogous inward shifts over time as market participants adapted to emerging market-making opportunities, in which case spreads would tighten. Former Fed Chair William McChesney Martin’s insight remains relevant here: The “natural strength and resilience” of market-making institutions and arrangements are endogenous with respect to monetary policy operations and can be “inhibited by official ‘mothering.’”³⁵Chairman Warsh has similarly said that he wants “the architecture to drive the plumbing, not the plumbing to drive the architecture.”³⁶A critical reexamination of the economic implications of both secular and high-frequency movements in money market spreads should therefore ask whether suppressing every sign of scarcity prevents the very adaptation that would support a smaller balance sheet.

But Why Reduce the Balance Sheet?

The ultimate purpose of reducing the size of the Fed’s balance sheet is to reduce the Fed’s “footprint” in financial markets in order to minimize the distortionary effects of the Federal Reserve’s financial market presence.³⁷The composition of the Fed’s portfolio, however, is arguably much more distortionary than the size of the balance sheet per se—especially its holdings of agency mortgage-backed securities and longer-duration Treasury securities.³⁸Because reserve account balances are virtually perfect substitutes for Treasury bills in bank portfolios, market distortions could be reduced for any given-sized balance sheet by shifting assets into bills.³⁹A Fed interested in minimizing financial market distortions caused by its footprint should follow the path laid out by William McChesney Martin toward a bills-only portfolio.

About the Author

Jeffrey M. Lacker is a member of the Shadow Open Market Committee, a senior affiliated scholar at the Mercatus Center at George Mason University, and the former president and CEO of the Federal Reserve Bank of Richmond. Email: [email protected].

Acknowledgments

This policy brief was prepared for presentation at the Shadow Open Market Committee on October 30, 2026, in New York, New York. The author is grateful to Robert Hetzel, Peter Ireland, and Bill Nelson for comments on an earlier draft.