Ample Reserves, Uneven Liquidity

01 · RESERVES AND MARKET FUNDING

Ample is a relationship between supply and demand

A reserve total is a stock of balances. Whether it is ample depends on how that supply compares with demand and how sensitive overnight rates become when the supply changes.

The New York Fed's reserve-demand-elasticity research estimates that sensitivity. Its October 15 update reported an estimate very small and statistically indistinguishable from zero, consistent with abundant reserves. The feature is a research estimate, rather than an official determination by the FOMC or the Federal Reserve System.^[1]

That is useful evidence about the aggregate market. It does not describe every bank's cash position, every dealer's financing capacity, or every institution's ability to obtain settlement funds at a particular hour.

In his March discussion of reserve management, Roberto Perli described an ample system in which individual banks can still borrow to smooth reserve distribution and manage intraday positions.^[12] Ordinary intermediation and modest rate variation are compatible with an ample framework. The analytical task is to identify when that variation becomes persistent, costly, or difficult to accommodate.

My starting point is therefore two questions. Is the total supply sufficient for effective rate control? And can the institutions that need funds obtain them on workable terms? The first question gives the system-wide context. The second helps locate a possible constraint within it.

Those questions require different evidence. A falling aggregate balance is worth monitoring, but it is not enough to establish scarcity. A higher rate paid by one institution may reflect its own circumstances, rather than a shortage across the entire banking system.

Back to top

02 · RESERVES AND MARKET FUNDING

The latest overnight rates show a range of prices

The October 28 implementation note maintained interest on reserve balances, or IORB, at 3.90%, the standing repo rate at 4.00%, and the federal funds target range at 3.75%–4.00%, effective October 29.^[4]

The latest overnight observations retrieved from the New York Fed's API are for October 29. EFFR was 3.88%, two basis points below IORB. SOFR was 3.94%, four basis points above it. The published 99th-percentile rates were 3.94% for fed funds and 4.02% for SOFR.^[5]

The range is part of the information. A median tells us where the middle of transaction volume trades. Percentiles show more of the price distribution, although they still do not identify which institutions face which terms.

EFFR and SOFR also cover different transactions. EFFR measures overnight unsecured federal funds borrowing. SOFR measures the cost of overnight borrowing secured by Treasury securities.^[6][7] Their relative prices can reflect collateral supply, intermediation, balance-sheet costs, and different participants. A positive SOFR spread to IORB is not, by itself, a finding of banking-system scarcity.

The 99th percentile is neither the maximum rate nor a statement that every borrower in a particular category paid that price. A SOFR percentile above the standing repo rate also does not identify the reason for that trade or prove that an eligible counterparty was unable to use the Fed operation.

The observation date matters. Overnight reference rates are published for the prior business day. The API's SOFR averages and index can carry a later date than the underlying overnight-rate observations. Mixing those records would give the impression of a newer transaction print than the data actually contain.

Back to top

03 · RESERVES AND MARKET FUNDING

Who trades can change what the headline tells us

The October 16 FEDS Note, by Sriya Anbil, Sebastian Infante, and Zeynep Senyuz, separates fed funds activity by lender and borrower type. It describes a market dominated by Federal Home Loan Bank lending and borrowing by U.S. branches of foreign banks, alongside a smaller domestic interbank segment.^[2]

FHLBs cannot earn IORB. Lending below that rate can still provide them a return on balances that would otherwise be unremunerated. An eligible bank borrowing below IORB can earn a spread, subject to its own costs and constraints. That transaction can be attractive without being driven by an urgent shortage of liquidity at the borrower.

Domestic bank borrowing can have a different purpose. It may be used to manage a cash need or a change in reserve distribution. The research finds that lending between banks is more sensitive to reserve positions than the nonbank-lender segment. Its empirical figures include a sample running from October 2015 through January 2024.^[2]

This is a historical research result published in October 2026, rather than a new daily measure proving that small banks are currently under stress. Its contribution is diagnostic: the same headline market contains transactions motivated by different incentives.

That limits how much reassurance a single rate can provide. A stable aggregate rate can coexist with a change in a smaller segment. It also limits how much alarm we should infer from an isolated transaction. We need to know the participants, volumes, persistence, and context.

Back to top

04 · RESERVES AND MARKET FUNDING

A median can stay put while a smaller segment gets more expensive

The New York Fed calculates EFFR as a volume-weighted median of eligible reported overnight transactions.^[6] A median is useful for summarizing a market, but it does not have to move whenever a minority of trading volume becomes more expensive.

Consider a PN illustration with $100 billion of volume. Assume $90 billion trades eight basis points below a reference rate and $10 billion trades five basis points above it. The volume-weighted median is eight basis points below the reference rate.

Now keep the $90 billion segment unchanged, but raise the smaller segment's price to fifteen basis points above the reference. The median stays where it was, even though the smaller segment's rate has increased by ten basis points.

These are assumed volumes and spreads. They do not reproduce the October 29 EFFR calculation or estimate the size of a stressed domestic bank segment. The example demonstrates a property of the statistic, rather than an empirical finding about current market conditions.

The practical response is to examine more than the median. Published percentiles and volumes can add context. Institution-level or participant-level data would be more informative still, where available. A public dashboard cannot infer the identity of a borrower from a rate percentile alone.

This is the angle I think deserves closer attention: a smaller segment can provide an early signal without yet moving the central price of the largest segment. To turn that possibility into a current conclusion, the signal needs to persist and be supported by evidence about actual funding demand or constraints.

Back to top

05 · RESERVES AND MARKET FUNDING

The reserve decline needs a balance-sheet explanation

The October 29 H.4.1 release reports Wednesday reserve balances of $2.895 trillion for October 28. Its weekly-average measure was $2.894 trillion, down $68.837 billion from the preceding week's average.^[3] Those are different observations, even though the rounded levels are close.

In the weekly-average comparison, the Treasury General Account rose $47.725 billion. Other nonreserve deposits rose by a net $15.743 billion, calculated from the published deposit and TGA changes. Changes in currency, reverse repos, other absorbing factors, and reserve-supplying factors complete the reconciliation.^[3]

The chart translates those published changes into their contributions to reserve balances. It is an accounting reconciliation of changes in weekly averages, not a record of gross cash flows or an estimate of which individual banks lost reserves.

A reserve decline does not automatically mean quantitative tightening has restarted. A movement between liabilities on the Fed's balance sheet can change reserves while the relevant securities policy remains unchanged. Treasury balances, currency, and deposits held by other entities all matter.

The October 15 reserve-demand estimate predates this October 28 balance-sheet observation. It should not be treated as a real-time test of that later week's conditions. The correct approach is to use the earlier aggregate assessment as context and then monitor how prices and market behavior respond to subsequent changes.

For this week, the balance sheet explains the reserve movement. It does not establish that the movement caused a distributional funding problem. That additional claim would require additional evidence.

Back to top

06 · RESERVES AND MARKET FUNDING

Treasury cash and reverse repo need their own definitions

Treasury cash management can move reserve supply independently of a new monetary-policy decision. When a payment to Treasury is funded from a bank deposit, settlement can reduce the bank's reserve balance and increase the TGA, holding other factors constant. Treasury spending in the other direction can add reserves and customer deposits.

The aggregate result depends on the financing path and the other changes occurring at the same time. If a money fund reallocates from the Fed's overnight reverse repo operation into bills, the reserve effect need not match a bill purchase funded directly from a bank deposit. A one-variable rule cannot describe every combination.

The reverse-repo line also needs a scope check. On October 28, H.4.1 shows $325.837 billion of total reverse repos, of which $325.661 billion belongs to foreign official and international accounts. The remaining “Others” category is $176 million.^[3] The large total should not be described as the domestic money-fund ON RRP buffer.

This is a recurring problem with liquidity summaries. Several different accounts can be compressed into one familiar label, and the resulting number is then assigned a role it does not actually measure.

PN's Liquidity Framework & Market Plumbing follows reserves, Treasury flows, collateral, repo, and intermediary capacity together.^[13] The current data reinforce the need to preserve those separate definitions. A useful aggregate signal still needs a map of the transactions and institutions behind it.

Back to top

07 · RESERVES AND MARKET FUNDING

Plenty in total does not guarantee access at one bank

Reserve balances are held within the banking and settlement system. They are not a pool that every household, company, or investment fund can draw from directly. A customer's bank deposit and its bank's reserve balance are different claims.

When funds move between banks, reserve balances can move as part of settlement. That redistributes the existing aggregate pool. Bank lending also involves the creation of deposits and subsequent funding and settlement requirements; it should not be modeled as a bank handing its reserve account directly to a household borrower.

Now consider a deliberately simple three-bank illustration. The system has $100 billion of reserves: Bank A holds $80 billion, Bank B $15 billion, and Bank C $5 billion. Assume each bank wants an internal $20 billion buffer. Total reserves exceed the combined desired buffers, but two banks are below their own targets.

If Bank A is willing and able to transfer only $10 billion through the modeled lending relationships, with $5 billion reaching each of the other banks, the balances become $70 billion, $20 billion, and $10 billion. Aggregate reserves remain $100 billion. Bank C remains $10 billion below its desired buffer.

The assumed buffer is not a regulatory reserve requirement, and the transfer limit is not estimated from real banks. The example isolates an access constraint. It deliberately leaves out other banks, central bank borrowing, collateral sales, and management responses that could resolve the gap.

In an actual market, the constraint might involve a credit limit, the price a lender requires, collateral, operational readiness, or payment timing. Which explanation applies cannot be read from the aggregate total. Nor does being below an internal target establish insolvency.

The question is whether available channels can redistribute or supply liquidity before obligations become due, and at what cost. That is the link between an aggregate reserve assessment and the experience of a particular institution.

Back to top

08 · RESERVES AND MARKET FUNDING

A standing facility works through eligibility and settlement

The standing repo operation provides an important financing channel, but its operation has specific terms. The New York Fed's current FAQs identify eligible counterparties and securities, two daily operation windows, a 4.00% rate, and same-day settlement through the tri-party platform.^[10]

Counterparty status also requires legal, operational, and technical setup. The eligibility page specifies institutional and balance-sheet criteria, together with arrangements to receive cash in the tri-party market.^[11] The published rate is therefore one part of the transaction.

A firm that is not directly eligible may depend on a bank or dealer to intermediate the funding. That intermediary has to be able and willing to do so. Eligible collateral, available balance sheet, counterparty relationships, and the time remaining before a payment is due can affect the terms.

This helps explain why a standing operation can support rate control without mechanically fixing every private transaction at its rate. It also explains why a trade above that rate does not identify one unique cause. The borrower, collateral, timing, and operational path need to be examined.

Facility usage needs context in both directions. Occasional use can be consistent with the intended functioning of an ample framework. Low use does not establish that every institution has plentiful access to every funding channel. More informative evidence would combine take-up, market rates, timing, collateral, and conditions facing the relevant borrowers.

For the current article, there is no institution-specific claim that an eligible bank was denied funding. The mechanism is an explanation of what must be checked before drawing such a conclusion.

Back to top

09 · RESERVES AND MARKET FUNDING

Reserve support is adjustable, and the current purchase pace is zero

The Fed retains authority to add securities when appropriate to maintain ample reserves. That authority is different from a commitment to purchase a fixed amount every month.

For October 15 through November 13, the New York Fed's operational schedule lists approximately $14.6 billion of reinvestment purchases and no reserve-management purchases.^[8] Reinvestment and an additional reserve-management purchase are separate operations; the planned reinvestment amount should not be relabeled as a fresh $14.6 billion reserve expansion.

In September, Perli explained that reserve-management purchases had been set to zero since mid-August as the Desk revised its supply outlook and assessed market conditions. He also said the amounts could be adjusted again as those conditions evolved.^[9]

The October implementation note continues to authorize appropriate purchases of bills and, if needed, other Treasury securities with short remaining maturities. It also directs Treasury principal rollover and agency principal reinvestment into bills.^[4] That describes the operating framework; it does not override the published monthly schedule.

The important forecast question is what would prompt another adjustment. A changed reserve-supply outlook, a change in reserve demand, and a sustained deterioration in relevant money-market conditions would provide different kinds of evidence.

It would be premature to infer a restart date from one weekly reserve decline. It would be equally misleading to treat the current zero pace as a permanent promise. The documented framework is conditional, and the next schedule and market response should be checked together.

Back to top

10 · RESERVES AND MARKET FUNDING

A June operational event shows a third source of pressure

Reserve supply and distribution are not the only issues. Funds also have to arrive when the settlement system needs them.

In his September speech, Perli described an operational issue at DTCC at the end of June. A faulty trade submission prevented transactions from settling and cash from returning to member banks. Deposits held by DTCC at the Fed stayed higher, producing an unexpected, short-lived reduction in bank reserve balances.^[9]

He also said the ample framework accommodated the surprise without broader disruption and cautioned against drawing sweeping conclusions from the event. The episode is evidence that settlement timing can affect reserve availability. It is not evidence that the same problem is ongoing or that the framework failed.

The analytical implication is broader than that single incident. A balance can exist somewhere in the system while remaining unavailable to the institution expecting it at a particular time. Operational delay can therefore resemble a cash need even when the underlying assets and claims remain intact.

That creates three different monitoring problems: a system-wide quantity problem, a distribution or access problem, and a timing or operational problem. They can interact, but they call for different evidence and potentially different responses.

Adding reserves can help with a quantity shortfall. An access constraint may require an effective financing relationship. An operational delay requires the relevant transactions and settlement processes to work. A diagnosis that treats all three as the same event can miss the mechanism that actually needs attention.

Back to top

11 · RESERVES AND MARKET FUNDING

The first effect on credit could be selective

A bank facing less dependable funding may become more cautious about a new commitment before it stops making payments. A dealer may require different terms to intermediate collateral. Another institution with stronger access may continue operating normally.

That is a conditional transmission channel, not a claim that the October observations already demonstrate broad credit contraction. The mechanism would run from a funding cost or timing problem into a balance-sheet decision, then into the terms offered to the next borrower.

It could appear as a wider financing spread, a smaller line, a greater collateral requirement, or more conservative liquidity management. Each outcome has other possible explanations, including borrower credit quality and ordinary changes in risk appetite. Establishing a reserve-distribution contribution would require evidence connecting the steps.

Adding settlement balances does not instruct a lender to approve a particular corporate loan. The lender still evaluates capital, credit risk, collateral, profitability, and funding.

The potential early signal is therefore unevenness. One funding segment can become more expensive while another remains stable. One type of borrower can face tighter terms while an aggregate benchmark barely changes. Tracking that variation is more useful than declaring all financing conditions easy or tight from a single reserve number.

The thesis would gain support if persistent funding differences were followed by documented changes in lender behavior. Without that sequence, it should remain a mechanism to monitor rather than a conclusion about the current economy.

Back to top

12 · RESERVES AND MARKET FUNDING

Three paths through the next funding cycle

The paths below are conditional scenarios. They carry no assigned probabilities and are not predictions of a particular institution's failure.

The first path deserves real weight. A functioning market can price temporary needs and move funds without a policy change. A rate movement around a reporting date that reverses promptly can be consistent with ordinary balance-sheet management.

The second path is the central research question in this article. It would be supported by concentration: recurring pressure among particular participants or funding channels, accompanied by evidence that access or intermediation is becoming more restrictive. A calm headline median would not rule it out.

The third path requires broader confirmation. A more negative reserve-demand-elasticity estimate, changes in the relationship between rates and reserve supply, persistent funding pressure, and an operational response would provide a stronger case than a weekly balance decline alone.

An operational interruption can cut across all three. Its timing, duration, and resolution need to be identified before it is interpreted as a change in reserve demand or borrower credit quality.

Back to top

13 · RESERVES AND MARKET FUNDING

A dashboard that can test the argument

The dashboard should preserve the definition and observation date of each measure. A monthly estimate, a weekly average, a Wednesday stock, and a daily overnight rate are complementary observations, rather than one synchronized print.

The strongest support for the access thesis would join several observations: persistent pressure in a relevant segment, evidence about the borrowers or intermediaries involved, and a documented financing or liquidity-management response.

The thesis would weaken if the apparent pressure normalized quickly, funding remained readily available, and participant evidence showed ordinary calendar effects. Broader reserve tightening would be a different conclusion requiring broader evidence.

The public data have a limit. Published rates and percentiles do not provide a live map of each domestic bank's reserve position and transactions. The historical research identifies useful channels; a public monitor has to be candid about which of those channels it can observe directly.

Back to top

PATTERN NEXUS VIEW

The Pattern Nexus view

The reserve pool is one part of the liquidity system. Its distribution, the agreements connecting institutions, and the timing of settlement determine how that liquidity can be used.

The October research provides a concrete reason to examine who is trading beneath the headline fed funds rate. The balance sheet adds the fiscal and account-level movement. The operating documents explain where additional funding can enter and which participants can use it directly.

My working view is that the next useful warning could be a persistent difference in access before it becomes an obvious aggregate shortage. That would show up through prices, timing, financing relationships, and subsequent behavior. It is a testable hypothesis, not something established by the word “uneven” in a headline.

The same discipline applies to reassurance. Aggregate reserves can be abundant and the framework can work well, as the available research and operating accounts describe. The task is to keep watching the channels through which a particular institution obtains the funds it needs.

That is the addition to the PN liquidity map: quantity establishes the system's capacity; distribution and access determine where the next constraint might emerge.

Back to top

READER QUESTIONS

Reader questions

Can reserves be ample while one bank needs to borrow?

Yes. Aggregate supply can be sufficient for rate control while individual banks redistribute balances to meet their own needs. Borrowing by itself does not establish a system-wide shortage.

Does a stable EFFR mean every bank has the same funding conditions?

No. EFFR is a volume-weighted median. A smaller segment can change price without moving that median. The hypothetical example illustrates that statistical property.

Does the latest reserve decline prove that QT restarted?

No. Changes in the TGA, currency, reverse repos, and other Fed accounts can move reserve balances. The current securities directives and purchase schedules must be checked separately.

Is SOFR above the standing repo rate proof that the backstop failed?

A transaction's eligibility, collateral, counterparty, timing, and settlement path need examination. A percentile above the posted rate does not identify one cause or establish that a particular eligible institution could not obtain funds.

Are reserve-management purchases continuing at a fixed monthly pace?

The October 15–November 13 schedule lists no reserve-management purchases and approximately $14.6 billion of reinvestment purchases. Future amounts depend on the Desk's assessment and published schedules.^[8]

Are the three-bank and median charts real market data?

They are PN illustrations with explicitly assumed inputs. The October 29 rate distributions and the H.4.1 weekly-average reconciliation are the empirical figures.

What would make the distributional concern more convincing?

Repeated pressure in a relevant funding segment, evidence about the participants' constraints, and a documented response in financing or liquidity management. A one-day rate movement or one aggregate reserve decline is insufficient.

Back to top

METHOD AND EVIDENCE

Method and evidence boundaries

Research cutoff: October 31, 2026. Overnight-rate data use October 29 effective dates from the New York Fed API snapshot. H.4.1 was released October 29 and reports October 28 Wednesday stocks and weekly averages. The reserve-demand update was released October 15. The empirical FEDS Note was published October 16 and draws on historical research samples.

Figure 1 displays published 1st and 99th percentiles with the volume-weighted median, expressed relative to the 3.90% IORB rate. These are percentile ranges, not minimum-to-maximum trading ranges. EFFR and SOFR measure different markets. Spread arithmetic uses 100 basis points per percentage point.

Figure 3 uses changes in weekly-average balances throughout. Its non-TGA deposit contribution is the total nonreserve-deposit change less the TGA change. “Other absorbing factors” combines Treasury cash holdings and other liabilities and capital. The contributions reconcile to the published $68.837 billion reserve decline.

Figures 2 and 4 use assumed inputs. Neither estimates an actual constrained bank, the share of stressed trading, or a regulatory buffer. The data audit preserves the observed inputs, calculation rules, dates, and illustrative assumptions. The New York Fed's API documentation supplies the retrieval framework.^[14]

Distributional pressure, downstream credit effects, and the three future paths are PN analysis. No hidden institution, impending failure, fixed reserve-scarcity threshold, or probability of intervention is inferred. Historical research results and a dated aggregate assessment are kept separate from current observations.

Back to top