Compliance, Capacity, Confidence: A Better Test of Usable Liquidity
The subject of this series is the Liquidity Coverage Ratio and the gap between regulatory compliance and liquidity that can actually be deployed. The final question is therefore precise: is a compliant LCR the same thing as usable liquidity? The evidence supports a clear answer. No. The LCR is important evidence of short-term resilience and a signal that can improve market funding, but usable liquidity depends on three separate conditions: room above the applicable floor, a contractual and operational path to the borrower or counterparty, and creditors' willingness to keep funding the bank.
That answer is not a rejection of the LCR. It is a better description of what the ratio does. A regulatory measure can be valuable without being complete. The two studies examined in this series show why: one finds that higher buffers supported more corporate credit-line utilization during the acute March 2020 shock, but selectively and temporarily; the other finds that stronger exposure to the LCR was followed by lower wholesale funding costs, longer maturities, and higher borrowing from money-market funds. Regulation can improve a bank's resilience and its access to funding while leaving the final deployment of liquidity constrained.
The same rule changes two different relationships
The Federal Reserve and BIS studies are complementary because they observe different links in the liquidity chain. The Federal Reserve paper asks what happens inside the bank-borrower relationship when a synchronized drawdown shock arrives. Its treatment is a post-regulation LCR buffer, its setting is the COVID-19 credit-line rush, and its outcome is the allocation of credit-line utilization across existing bank relationships. The BIS paper asks what happens in the bank-creditor relationship when the LCR is introduced. Its treatment is a pre-regulation LCR gap, its setting is the 2013 implementation window, and its outcomes are MMF funding prices, maturities, and quantities.
| Question | Federal Reserve evidence | BIS evidence |
|---|---|---|
| Which relationship is observed? | Bank and corporate borrower | Bank and wholesale creditor |
| What is the stress or treatment lens? | Buffer above the applicable floor during COVID Q1 | Pre-regulation gap around LCR implementation |
| What changes? | Credit-line utilization across a firm's bank relationships | Funding cost, maturity, and borrowing quantity |
| Main result | Higher-buffer banks supplied more to prime, high-undrawn borrowers in March 2020; the differential disappeared in Q2 | More-exposed banks later paid about 8% less, roughly 2.5–2.6 basis points at the sample mean, with stronger effects at longer maturities |
| What the result does not show | Universal or permanent lending support | Unlimited freedom to deploy HQLA or effects on every liability |
The table is a comparison of questions, not a pooled estimate. The Federal Reserve sample covers 29 bank holding companies over 2017Q1–2024Q4 and studies a particular crisis episode. The BIS sample covers 28 banks, 223 funds, and 72 fund families from February 2011 through December 2015, before the 2016 MMF reform. The Federal Reserve study classifies banks by a 2019Q4 buffer; the BIS study uses the bank's average pre-regulation LCR gap. A buffer is room after the rule is in place. A gap is exposure to the rule before it takes effect. Their relationship to the regulatory floor is shared, but their causal meanings are not interchangeable.
There is also a useful qualification within the Federal Reserve evidence. The horse-race result supports a buffer-specific stress allocation: the buffer interaction remains significant while the high-LCR-level interaction is effectively zero in the reported specification. Yet a continuous LCR ratio enters significantly in another specification. These are different tests, not necessarily a contradiction. Overall liquidity can contain information about a bank's condition, while the differential response to acute credit-line demand can be concentrated in the room above the applicable minimum. The defensible conclusion is not that the level never matters. It is that a headline level alone does not explain the buffer-related allocation effect.