The U.S. Treasury announced a comprehensive review of bank liquidity regulations Tuesday, proposing changes it says could unlock hundreds of billions of dollars in new lending capacity targeted at artificial intelligence infrastructure, reshoring of supply chains, and defense manufacturing.
Under Secretary for Domestic Finance Jonathan McKernan delivered the remarks, which were prepared by Treasury Secretary Scott Bessent, and framed the current regulatory framework - developed after the 2008 financial crisis - as overly constraining. McKernan said the regime has "excessively and unnecessarily limited banks' ability to do what they are supposed to do - lend."
According to the Treasury, large banks now hold about 25% of their balance sheets in so-called safe assets, up from roughly 10% before the crisis. The department argues that this shift has reduced the availability of capital for mortgages, small business loans, and financing for critical infrastructure projects.
McKernan criticized the post-crisis approach to liquidity regulation as a form of overcorrection, characterizing it as "novel and even muddled guesswork" created under the pressure of crisis conditions. He noted that, historically, liquidity requirements were enforced through supervisory judgment rather than strict numerical formulas, and said architects of the current rules relied on historical experience and intuition rather than an established theoretical framework.
At the center of the Treasury's reform agenda is the liquidity coverage ratio. The department proposes modifying how that ratio treats a bank's borrowing capacity tied to collateral prepositioned at the Federal Reserve's discount window. The proposed change would allow some recognition of discount window borrowing capacity, but that recognition would be subject to a cap.
McKernan observed that, in practice, banks have been reluctant to draw down liquidity buffers in times of stress even though regulators designed those buffers to be used. He said both regulators and market participants have treated these buffers as inviolable minimums, which has had the unintended effect of amplifying liquidity stress rather than absorbing it. He also criticized the current rules for reinforcing stigma around discount window use by effectively requiring banks to exhaust regulatory buffers before accessing the facility, even though the lender of last resort function is intended for use in severe episodes of stress.
The Treasury highlighted the March 2023 banking turmoil involving Silicon Valley Bank, Signature Bank, and First Republic Bank to illustrate its point. McKernan noted that each institution held sizable positions in Treasury securities and agency mortgage-backed securities, but that liquidity was largely only on paper because collateral was not prepositioned and discount window access had not been tested.
On the mechanics of implementation, the Treasury said regulators could consider sizing the cap on recognized discount window borrowing capacity according to each bank's historical use of the facility. One option mentioned is to limit recognition to the smaller of an absolute ceiling and some multiple of a bank's borrowing over a defined past period.
Beyond liquidity accounting, McKernan said the administration will keep advocating for expanded deposit insurance coverage for noninterest-bearing transaction accounts. He also said the Financial Crimes Enforcement Network and bank regulators will soon propose a rule to shift anti-money laundering supervision toward assessing program effectiveness rather than other measures.
The Treasury indicated it will also begin reassessing which activities are appropriate for banking organizations, with an emphasis on enabling responsible adoption of new technologies.
Impacted sectors and considerations:
- Banks and financial institutions - potential changes in balance sheet allocations and liquidity reporting.
- Housing and small business finance - possible increased availability of mortgage and small business lending.
- Technology and manufacturing - targeted lending capacity for AI infrastructure and defense manufacturing.