Money Creation and Bank Clearing

Fordham Journal of Corporate & Financial Law
Volume 28 Issue 1 Article 2
2023
Money Creation and Bank Clearing

Nadav Orian Peer
Colorado Law
Follow this and additional works at: https://ir.lawnet.fordham.edu/jcfl
Part of the Banking and Finance Law Commons, Business Organizations Law Commons, Corporate
Finance Commons, and the Finance and Financial Management Commons
Recommended Citation
Nadav Orian Peer, Money Creation and Bank Clearing, 28 Fordham J. Corp. & Fin. L. 35 (2023).
This Article is brought to you for free and open access by FLASH: The Fordham Law Archive of Scholarship and
History. It has been accepted for inclusion in Fordham Journal of Corporate & Financial Law by an authorized editor
of FLASH: The Fordham Law Archive of Scholarship and History. For more information, please contact
[email protected].
MONEY CREATION AND BANK CLEARING
Nadav Orian Peer*
Like many other countries, the U.S. money supply consists primarily
of deposits created by private commercial banks. How we understand
bank money creation matters enormously. We are currently
witnessing a debate between two competing understandings. On the
one hand, a long-standing conventional view argues that bank money
creation originates in individual market transactions. Based on this
understanding, the conventional view narrowly limits the scope of
banking regulation to market failure correction. On the other hand,
authors in a new legal literature emphasize the public aspects of bank
money creation, characterizing it as a “public franchise,” a “public-
private partnership,” and part of the “social contract.” This new legal
literature has a broader vision of banking regulation, and has raised
ambitious proposals in areas including financial stability, civil rights,
climate action, and financial technology.
This Article bridges a gap in the new literature that has held it back
from achieving its full potential. While the new literature recognizes
bank money creation as public in important ways, it has dedicated
little attention to the question of how banks are able to engage in
money creation in the first place, thereby leaving key aspects of the
conventional account unchallenged. The Article fills this gap by
focusing on the process of clearing, through which banks pay trillions
of dollars in obligations they owe each other every day. To assess the
conventional account, the Article presents a case study of daily
clearing practice in an environment that seems as market driven as
* Associate Professor, Colorado Law. The author would like to acknowledge helpful
comments and suggestions by Christine Desan, Roy Kreitner, Adam Feibelman, Charles
Kahn, Virginia France, Benjamin Geva, Robert Steigerwald, John Crawford, Gustavo
Ribeiro, Sloan Speck, Harry Surdan, Margot Kaminski, Sharon Jacobs, Scott Skinner-
Thompson, Blake Reid, Ben Levine, and Noam Magor. Research for the Article has
benefited from generous financial support by the Weatherhead Center for International
Affairs (Harvard University). The Fourth Law & Macroeconomics Conference (2021),
and the Annual Conference of the American Society for Legal History (2019) provided
insightful forums for discussion. Larkin Dykstra and Lukas Knudsen provided
outstanding research assistance. My thanks also go to the editors of the Fordham Journal
of Corporate & Financial Law for their excellent work in preparing the manuscript for
print. This Article is dedicated with love to my wife, Virginia.
35
36 FORDHAM JOURNAL [Vol. XXVIII
OF CORPORATE & FINANCIAL LAW
possible: the New York Clearing House Association prior to the

creation of the Federal Reserve system. Building on novel primary
sources, the case study demonstrates that daily clearing presented
NYCHA banks with serious challenges. Addressing these challenges
required governance both at the level of the state, and through bank
cooperation on nonmarket terms. These findings expand our
understanding of how bank money creation occurs and how it should
be regulated.
INTRODUCTION .............................................................................. 36
I. THE KEY QUESTION: HOW DO BANKS CLEAR THEIR DAILY
OBLIGATIONS? ....................................................................... 42
A. Clearing is Essential to Bank Money Creation ................. 42
B. The Conventional View Abstracts from Clearing ............. 49
II. SIDESTEPPING THE QUESTION: ASSUMPTIONS BEHIND
ABSTRACTION FROM CLEARING ............................................ 53
A. Netting is Freely Available to Banks ................................ 54
B. The Law of Large Numbers Prevents Drains .................... 56
C. Interbank Markets Can Solve Drains................................. 59
III. ADDRESSING THE QUESTION: DAILY GOVERNANCE IN THE
NEW YORK CLEARING HOUSE ASSOCIATION ....................... 62
A. Netting Was Not Freely Available .................................... 64
B. Clearing Drains Did Occur ................................................ 71
1. Daily Drains ................................................................. 72
2. Weekly Drains ............................................................. 77
C. Interbank Markets were Absent ......................................... 79
1. The Use of Reserves .................................................... 79
2. The Use of Call Loans ................................................. 82
CONCLUSION .................................................................................. 89
APPENDIX ....................................................................................... 91
INTRODUCTION
Like many other countries, the U.S. money supply consists primarily
of deposits created by private commercial banks. The $18 trillion in these
deposits far exceed the $2 trillion in cash, currency directly issued by the
government.1 Deposits are commonly referred to as “money” because they
1. Deposits, All Commercial Banks, FED. RSRV. BANK ST. LOUIS (Jan. 21, 2022),
https://fred.stlouisfed.org/series/DPSACBW027SBOG [https://perma.cc/G2MD-
WG53] [hereinafter Deposits]; Liabilities and Capital: Liabilities: Federal Reserve
2023] MONEY CREATION AND BANK CLEARING 37
are the most common form of payment in the economy, from buying a cup
of coffee, to payroll, to paying for a house or a securities transaction.2 In
jargon, one speaks of “bank money creation,” to capture commercial banks’
ability to issue deposits and the monetary use of these deposits. Given
money’s elemental role in economic activity, how we understand bank
money creation matters enormously.3
We are currently witnessing a high-stakes debate between two
competing understandings. On the one hand, a long-standing
conventional view holds that bank money creation originates in individual
market transactions. As that account goes, the countless individual
choices of banks, depositors, and borrowers ultimately produce bank
money creation as a generally efficient private arrangement.4 Based on
this understanding, the conventional view has for decades limited the
scope of U.S. banking regulation to the relatively narrow task of market
failure correction. Exceptions to this notion are so narrow they highlight
the norm.5
On the other hand, since the Global Financial Crisis of 2007-2009 (the
“GFC”), an emerging legal literature has recognized bank money creation
as a “public franchise,”6 a “public-private partnership,”7 a “public utility,”8
and part of the “social contract.”9 This approach enables scholars to identify
the ways banking produces, impacts, or exacerbates some of society’s most
Notes, Net of F.R. Bank Holdings: Wednesday Level, FED. RES. BANK ST. LOUIS (Jan. 19,
2022) https://fred.stlouisfed.org/series/WLFN [https://perma.cc/3PYB-QSAY].
2. See infra Section I.A.
3. For money’s role in shaping economic activity, see CHRISTINE DESAN, MAKING
MONEY: COIN, CURRENCY, AND THE COMING OF CAPITALISM (2014); Roy Kreitner, Legal
History of Money, 8 ANN. REV. L. & SOC. SCI. 415 (2012).
4. See infra Section I.B.
5. See discussion of bank runs, public supports, and the ensuing need to address
moral hazard, infra Section I.B.
6. Robert C. Hockett & Saule T. Omarova, The Finance Franchise, 102 CORNELL
L. REV. 1143, 1147 (2017).
7. See MORGAN RICKS, THE MONEY PROBLEM: RETHINKING FINANCIAL
REGULATION 200 (2016).
8. See Morgan Ricks, Money as Infrastructure, 2018 COLUM. BUS. L. REV. 757,
763-64 (2018).
9. Mehrsa Baradaran, Banking and the Social Contract, 89 NOTRE DAME L. REV.
1283, 1313 (2014). Other contributions to this literature include Adam J. Levitin, Safe
Banking: Finance and Democracy, 83 U. CHI. L. REV. 357 (2015); Nadav Orian Peer,
Negotiating the Lender of Last Resort: The 1913 Federal Reserve Act as a Debate over
Credit Distribution, 15 N.Y.U. J.L. & BUS. 367, 369 (2019); Lev Menand, Why Supervise
Banks? The Foundation of the American Monetary Settlement, 74 VAND. L. REV. 951,
967 (2019); Dan Awrey, Bad Money, 106 CORNELL L. REV. 1, 3 (2020).
pressing challenges: financial instability, racial and wealth inequality,

climate change, and financial technology (“fintech”).10 Scholars in the new
literature have made a number of ambitious proposals in all these areas:
proposals for panic proofing the financial system,11 increasing access to
credit and financial services in underserved communities,12 funding the
green economy,13 and creating an inclusive central bank digital currency.14
Ideas and proposals from the new literature are increasingly entering the
public conversation, including in legislative and adjudicative settings.15
This Article bridges a gap in the new literature that, despite its many
achievements, has held it back from achieving its full potential. The new
literature has dedicated little attention to the question of how banks are
able to engage in money creation in the first place, thereby leaving key
aspects of the conventional account unchallenged. The new literature, for
example, emphasizes the dramatic government interventions that stabilize
bank money creation in crisis times like the GFC and the March 2020
financial market disruptions due to COVID-19.16 This is evidence that
bank money creation has a public dimension. But holders of the
conventional view respond that crises are localized failures in market
activity that can only justify narrowly targeted interventions, including
stabilization and central bank lending-of-last-resort in times of crisis, and
prudential regulation in ordinary times, e.g., capital and liquidity
10. For proposals, see notes 11-14 infra.

11. For two different proposals designed to enhance stability, see Ricks, supra note
8; Levitin, supra note 9.
12. See, e.g., Mehrsa Baradaran, It’s Time for Postal Banking: The USPS Should
Help Extend Credit to the Unbanked Population, 127 HARV. L. REV. F. 165, 168 (2014);
John Crawford et al., FedAccounts: Digital Dollars, 89 GEO. WASH. L. REV. 113, 127
(2021).
13. See, e.g., ROBERT. C. HOCKETT, FINANCING THE GREEN NEW DEAL: A PLAN OF
ACTION AND RENEWAL (2020).
14. See generally Saule T. Omarova, The People’s Ledger: How to Democratize
Money and Finance the Economy, 74 VAND. L. REV. 1231 (2021).
15. See, e.g., Nomination Hearing of Dr. Saule T. Omarova to the Comptroller of
the Currency, Before the S. Comm. on Banking, Hous., & Urb. Aff., 117th Cong. (2021);
Brief of Thirty-Three Banking Law Scholars as Amici Curiae, Lacewell v. Off. of
Comptroller of Currency, 999 F.3d 130 (2d Cir. 2021) (No. 50) (opposing the OCC’s
fintech charter); TNB U.S.A. Inc. v. Fed. Rsrv. Bank of N.Y., No. 1:18-CV-7978 (ALC),
2020 WL 1445806 (S.D.N.Y. Mar. 25, 2020) (denying The Narrow Bank’s petition for
relief against the New York Fed’s refusal to open a master account); S. 579, Sen. No.
665, 192d Gen. Ct. (Mass. 2021) (a bill to establish a Massachusetts Public Bank).
16. See, e.g., Baradaran, supra note 9, at 1284-85; Awrey, supra note 9, at 4-5.
requirements. These are far narrower measures than the new literature’s
reform proposals in financial stability, civil rights, climate policy, etc. As
long as the methodological individualism of the conventional approach
prevails, the new literature will keep running into this difficulty. 17 What
is needed is direct engagement with the conventional view’s premise that
bank money creation arises from individual market decisions, and as such,
enjoys a presumption against any but the narrowest interventions.
This Article fills this gap by focusing on bank clearing. Clearing is
the crucial but little discussed process through which banks pay trillions
of dollars in obligations they owe to each other every day. 18 These large
obligations are a direct outcome of the monetary function deposits serve
in our economy. Every time a depositor makes a payment to a recipient
in another bank, that transaction must clear. How banks manage to clear
these vast obligations on an extremely tight timeline, ought to be a central
question for any theory of bank money creation. But despite its ubiquity
in day-to-day practice, clearing is strikingly absent from the conventional
view.19 Instead, its focus is on the relationship between depositors and
their banks, and the conditions under which depositors maintain sufficient
trust in banks’ ability to redeem their deposits. That is the trust that is
undermined in crisis and can lead to bank runs and financial panics.
Meanwhile, the conventional view leaves aside the ways in which banks
manage to clear the trillions of obligations they owe to each other every
day. This neglect of day-to-day practice is based on an impression that
clearing is an easy process, one that does not raise any particular
17. Another example is the turn to endogenous money theory in some of the new
literature. See, e.g., Hockett & Omarova, supra note 6. Endogenous money theory
highlights that bank deposits are created through the process of bank lending, without
need for preexisting deposits of physical currency (a common assumption in textbook
models; for an extensive discussion of endogenous money theory and its historical
origins, see L. RANDALL WRAY, MONEY AND CREDIT IN CAPITALIST ECONOMIES: THE
ENDOGENOUS MONEY APPROACH (1990)). To be sure, to many, grasping the more
proactive role that banks play in money creation casts them in a more public light. At the
same time, endogenous money views are today ubiquitous in policymaking circles. These
views have been easily assimilated into the conventional understanding that bank money
creation is private, and regulation should be limited to market failure correction.
18. See infra Section I.A. My interest in clearing builds takes its inspiration from the
work of financial economist Perry Mehrling. See, e.g., PERRY MEHRLING, THE NEW
LOMBARD STREET (2010).
19. See infra Section I.B.
difficulties to banks, and is therefore not integral to bank money

creation.20 The new literature has not dispelled that impression.21
Assessing the impression that clearing is easy involves several
methodological difficulties. At the most basic level, modern clearing
practice is fundamentally shaped by the activities of central banks.22 The
central bank’s daily involvement in clearing removes many of the
challenges banks would have otherwise faced if left to their own devices.
Meanwhile, the key role that central banks play in clearing remains
largely invisible: we overlook the governance work that actually makes
clearing possible.23 To truly assess the conventional view’s account, we
must study the workings of clearing in an environment predating the
modern central bank. To this extent, this Article presents a case study of
the New York Clearing House Association (the “NYCHA”) prior to the
creation of the Federal Reserve system in 1913.24 During that time, the
NYCHA stood at the apex of the U.S. payments system, and cleared not
only local New York transactions, but a large portion of the country’s
interregional payments. The case study builds on previously unexplored
primary sources to reconstruct a highly detailed picture of daily practice
at the NYCHA.25 If NYCHA members indeed faced serious challenges in
their daily clearing and those were not solved through private contracting,
such findings would call into question the conventional view’s account of
bank money creation. That is indeed what we find.
Specifically, NYCHA banks faced large and threatening clearing
“drains” in their day-to-day operations.26 Contrary to modern assumptions,
these drains were not managed through contracting between the NYCHA
20. See infra Part II.

21. While the new literature includes a number of discussions of clearing, these
discussions do not provide a comprehensive account of the daily challenges clearing
involves, and how they are met in practice. See, e.g., Hockett & Omarova, supra note 6,
at 1162; RICKS, supra note 7, at 58-61; Awrey, supra note 9, at 20-22. But see discussion,
infra note 23.
22. See infra Part II.
23. Hockett & Omarova take a step towards addressing that governance work when
they discuss the central bank’s accommodation of bank money creation as part of its role
in presiding over the payments system and administration of monetary policy. Hockett &
Omarova, supra note 6, at 1162.
24. See infra Part III.
25. While the NYCHA has been the studied by financial historians, those works
focused on rare crisis dynamics rather than daily routines. See infra discussion and
references in Part III; see infra note 118 and accompanying text.
26. See infra Section III.B.
banks. Instead, these banks were able to successfully clear by relying on

two distinct forms of governance. The first was a very large amount of
reserves of state issued money that enjoyed legal tender status under the
law.27 These reserves were not held merely to satisfy regulatory
requirements or meet withdrawals from depositors, but for actively meeting
clearing drains. The second form of governance was undertaken by
cooperation among the NYCHA banks themselves. To keep their reserve
position in-tact, the NYCHA banks did not trade reserves among
themselves as predicted by economic theory. Instead, these banks created a
system of “secondary reserves” by making overnight loans to stockbrokers
on the New York Stock Exchange, and carefully recycling their
clearinghouse balances through these brokers.28 At a sum total, the case
study demonstrates that bank money creation relied on clearing, and that
clearing relied on governance, some of which was provided by the state
(the issuer of reserves), and some was provided through non-market
cooperation among the banks. In modern times, central banks fulfill similar
daily functions and help commercial banks overcome challenges they
cannot overcome through market activity.
By highlighting the role of governance in bank money creation, the
article provides a stronger foundation for the new literature on the public
dimensions of bank money creation.29 This foundation responds to the
conventional view that bank regulation should be narrowly limited to
market failure correction. The workings of bank money creation—even in
ordinary times—are far more complex than suggested by the image of
simple contracting between banks, depositors, and borrowers. There is no
reason to afford a presumption of market efficiency to a system that relies
on considerable nonmarket components for its operation, not only rarely,
in crisis, but daily, in its most quotidian operations. Recognizing the key
roles of clearing, and governance in clearing, thus opens greater space for
discussion of recent reform proposals raised in the new literature. One can,
of course, disagree with these proposals as a matter of substance. But the
notion the proposals are prima facie flawed because they “tinker” with
market transactions is at odds with the history.
The remainder of the Article proceeds as follows. Part I introduces the
key concepts involved in clearing and explains why clearing is essential
27. See infra Section III.B.

28. See infra Section III.C.
29. My interest in the governance that enables money creation takes its inspiration
from Christine Desan’s CONSTITUTIONAL APPROACH TO MONEY. See DESAN, supra note
3, at 775.
to bank money creation. It then surveys the conventional view’s account

of bank money creation pointing out to its abstraction from clearing. Part
II discusses three specific assumptions authors use to justify their
abstraction from clearing: first, that netting of obligations is freely
available to banks; second, that the probabilistic law of large numbers
prevents banks from experiencing destabilizing clearing drains; and third,
that interbank markets offer a solution to drains even if they do occur.
This Article argues that these assumptions miss the challenges clearing
raises because they take for granted the proactive role that central banks
play in modern clearing. Part III provides the historical case study of daily
routines of banks at the New York Clearing House Association before the
creation of the Federal Reserve in 1913. The case study is organized
around each of the three assumptions underlying the conventional view’s
abstraction from clearing. The findings call into question each of the three
assumptions: netting was not free, but instead involved a considerable risk
exposure between banks in the event of default; large clearing drains did
in fact occur and could lead to default; and default was avoided through
reserves (issued by the state) and call loans (made possible through
collective action), not through trading in interbank markets. These
findings can be used as ingredients for a general theory of clearing as a
public function in future work.
A last note by way of introduction. Given the topic’s relevance for
different areas of law (civil rights, environmental law, fintech) this Article
explains all key concepts with a non-technical reader in mind. Readers
interested in the more technical aspects may consult the Appendix.
I. THE KEY QUESTION: HOW DO BANKS CLEAR THEIR DAILY

OBLIGATIONS?
A. CLEARING IS ESSENTIAL TO BANK MONEY CREATION
Every day, the commercial banking system stands ready to clear

trillions of dollars of obligations banks owe to each other. These
obligations are a product of the monetary function deposits serve in our
legal system, that is, the way deposits settle payments between bank
customers as if they were cash. Without clearing, there can be no deposit
transfers, and no bank money creation. To develop our understanding of

clearing, we must begin by understanding the legal nature of deposits.30
The term “deposit” is notoriously confusing. 31 It evokes the image of
a bank customer providing a bank with actual cash (Federal Reserve
notes) in return for a deposit. In reality, only a small fraction of deposits
is created in this way. More commonly, deposits are created in the process
of commercial bank lending.32 When a bank provides a borrower with a
loan, the bank credits the borrower’s account with a new deposit it creates
on its books. No actual cash needs to be involved in that transaction. The
deposit is simply a new bank liability that the bank creates to fund the
new asset it acquired, that is, the loan (the borrower’s obligation for future
payment). This deposit creation process is presented in Balance Sheet
1.33 This deposit creation process is what leads to the typical balance sheet
structure of a bank and its high degree of “maturity transformation.” The
assets, consisting mainly of long-term loans, are funded through liabilities
consisting mainly of short-term deposits.34
While cash is not required for the creation of a deposit, a deposit is a
legal obligation for a bank to pay and to make that payment on demand.
Under the law, a bank’s payment to a depositor could take place in two
different ways. The simpler, but less common way, is paying cash to a
depositor, e.g., when withdrawal is made at the ATM. Under U.S. Code
§ 5103, federal reserve notes “are legal tender for all debts.”35 A deposit
is a form of debt, so by delivering the notes, the bank lawfully redeems
the deposit. According to recent data, annual ATM withdrawals amount
to around $0.80 trillion.36 The far more common way to redeem a deposit
30. Discussion in the following paragraphs builds on JOHN MAYNARD KEYNES, THE
COLLECTED WRITINGS OF JOHN MAYNARD KEYNES: VOLUME V: A TREATISE ON MONEY
20-29 (Austin Robinson & Donald Moggridge eds., 2d ed. 2012).
31. See Ricks, supra note 8, at 57.
32. This insight is known as “endogenous money.” See, e.g., Michael McLeay et al.,
Money Creation in the Modern Economy, Q. BULL. (Bank of Eng., U.K. 2014); WRAY,
supra note 17; MARC LAVOIE, POST-KEYNESIAN ECONOMICS: NEW FOUNDATIONS 182-
274 (1st ed. 2014). In legal scholarship, see discussion by Hockett & Omarova, supra
note 17.
33. See Appendix, Balance Sheet 1.
34. See Pascal Paul, Banks, Maturity Transformation, and Monetary Policy (Fed.
Res. Bank of S.F. Working Paper 2020-07, 2021).
35. 31 U.S.C. § 5103.
36. GEOFFREY GERDES ET AL., FED. RSRV. SYS., THE 2019 FEDERAL RESERVE
PAYMENTS STUDY 11 (2019), https://www.federalreserve.gov/newsevents/pressreleases
/files/2019-payments-study-20191219.pdf [https://perma.cc/V2R4-58PP]. As discussed
is through a deposit transfer.37 Here, the depositor is not interested in

holding cash, but in making a payment to some third party with a bank
account. Such payments also enjoy the sanction of law. For example,
under Uniform Commercial Code (U.C.C.) § 4A-406,38 a payment by a
funds transfer39 generally discharges an obligation “to the same extent
discharge would result from payment to the beneficiary [the payee] of the
same amount in money . . . .”40 Payments by transfer approximate $100
trillion a year,41 and once financial transactions are incorporated, that
figure grows to several trillions per day.42 In other words, deposit transfers
in a single day exceed (by multiples) amounts withdrawn from ATMs
over an entire year. Banks’ need to clear is a product of these enormous
flows arising from deposit transfers to settle payments between
customers.
The next step is to understand how transfers between depositors
translate into clearing obligations between their banks. This process is
below, that amount is small in comparison to the trillions of dollars per day (rather than
annually) required for clearing.
37. In this Section, the term “deposit transfer” is used as an umbrella term bringing
together the distinct legal frameworks that allow payment through direct use of a deposit.
Those include, for example, funds transfers, checks, debit and credit cards and other
forms of transfers. For a general discussion of the laws and regulations governing each
of these, see COMMITTEE ON PAYMENTS AND MARKET INFRASTRUCTURES, Payment,
Clearing and Settlement Systems in the United States, 2012 RED BOOK 471 [hereinafter
BIS RED BOOK].
38. U.C.C. § 4A-406 (AM. L. INST. & UNIF. L. COMM’N 2020),
39. See id. at 477-78. “Funds transfers” are a legal term-of-art governed by U.C.C.
§ 4A-406. They are a subset of my general use of the term “deposit transfers” explained
in note 37, supra. The examples that follow make reference to funds transfers because by
their volume they are by far the most significant legal framework governing payments.
40. U.C.C. § 4A-406 (AM. L. INST. & UNIF. L. COMM’N 2020), subject to narrow
exceptions in subsection (b)(i)-(iii). The notion of a discharge of the payor’s obligation
by means of a deposit transfer is often referred to as “payment finality.” Within the broad
category of deposit transfers, different legal frameworks have different requirements for
payment finality. See discussion in Benjamin Geva, Payment Finality and Discharge in
Funds Transfers, 83 CHI.-KENT L. REV. 633 (2008).
41. See Gerdes et al., supra note 36, at 13 (Table B.1) (value of core noncash
payments for 2018 at $97 billion).
42. See About Payments and Financial Market Infrastructures Statistics, BANK FOR
INTERNATIONAL SETTLEMENTS, https://www.bis.org/statistics/payment_stats.htm?m=6
_36 (last visited Jan. 22, 2022) (U.S., T9: Value of transaction processed by selected
payment systems). 2020 “total gross value” for Fedwire and CHIPS were $840 trillion
and $382 trillion respectively. Id.
described in Balance Sheet 2.43 Consider a Buyer of widgets (the payor)

interested in paying $1 million to a Seller of widgets (the payee). The
Buyer instructs its bank (the Buyer’s Bank) to make a $1 million funds
transfer to the Seller’s account in some different bank (the Seller’s
Bank).44 The Buyer’s payment to the Seller would be complete when the
Seller’s bank account is credited with the $1 million amount.45 From the
point of view of the Buyer and Seller, the deposit transfer allows the debt
to be repaid, just as if the Buyer had provided the Seller with federal
reserve notes. Not so from the point of view of their banks, who now need
to work out their internal financial relationship.46
The world would have been simple if all buyers and sellers held
accounts in a single large bank. That large bank could have simply debited
the Buyer’s account and credited the Seller’s account by the same amount
(Balance Sheet 3).47 In such a world, clearing would not have been
necessary.48 Our world has multiple banks, so the obligation between
Buyer and Seller transforms into an obligation between their respective
banks. Note that the funds transfer requires the Seller’s Bank to credit the
Seller’s account with a new deposit, meaning that the Seller’s Bank needs
to increase its liabilities. The Seller’s Bank cannot agree to increase its
liabilities for free, or it will suffer a loss.49 For that reason, when it credits
the Seller’s account by $1 million, the Seller’s Bank will acquire a $1

44. “Funds transfer” is a technical term governed by U.C.C. § 4A (AM. L. INST. &
UNIF. L. COMM’N 2020).
45. U.C.C. § 4A-406. § 4A includes detailed rules determining when a beneficiary’s
bank (like the Seller’s Bank in the example) is considered to have accepted a payment
order. These rules are beyond the scope of the current Article. See, e.g., U.C.C. § 4A-
209; Geva, supra note 40.
46. For expositional convenience, this Section presents the settlement of obligations
between clients (through the crediting of the beneficiary’s bank account) as preceding
interbank clearing. In modern practice, with the rise of Real Time Gross Settlement and
the New CHIPS (see discussion infra Section II.A.), the order is often reversed (or the
two are made simultaneously). See Geva, supra note 40, at 646. Historical practice was
more diverse, as reflected by U.C.C. § 4A-405(c). See id.
48. See Perry Mehrling, The Central Bank as a Clearinghouse, B.U.: MONEY &
BANKING LECTURES (last visited Oct. 16, 2022), https://sites.bu.edu/perry/mb-lectures
/section-1/5-the-central-bank-as-a-clearinghouse/ [https://perma.cc/X3J8-J4AN]. The
proposal advanced by Morgan Ricks in his book, THE MONEY PROBLEM, would in fact
eliminate the need to clear. RICKS, supra note 7, at ch. 9.
49. According to the famous balance sheet equation, Shareholders’ Equity = Assets
– Liabilities. A bank increasing its liabilities without gaining an asset of the same amount
would have its equity reduced by the same amount.
million claim against the Buyer’s Bank (in jargon, such claim is often
booked as “due from banks”).50 This claim is a new asset that balances
the new liability (deposit to Seller). This accounting can be understood
intuitively. By crediting the Seller’s account with a deposit, the Seller’s
Bank is doing the Buyer’s Bank a favor, namely, it is making it possible
for the Buyer’s Bank to “pay” its deposit to the Buyer. The $1 million
claim against the Buyer’s Bank is the price of that favor. This is the
obligation that now needs to be “cleared,” and as noted above, daily
clearing volume measures in the trillions per day.
Clearing can take place in two different ways. Those are known as
“settlement” and “setoff” or “netting.”51 Settlement generally means
payment in public (state issued) money. For banks, this public money
comes in the form of “reserves” with the Federal Reserve System. Like the
cash we use in retail payments (Federal Reserve notes), reserves are issued
directly by the Federal Reserve, but instead of being issued as paper notes,
they are balances (book entries) in accounts commercial banks hold with
the Federal Reserve.52 This is an instance of the Fed’s role as a “bankers’
bank,” meaning that commercial banks maintain accounts with the Fed just
like retail depositors maintain accounts with commercial banks. Reserve
balances have legal characteristics that are similar to the legal tender quality
of cash discussed above. A transfer of reserves constitutes final payment
between banks.53
Fed creation of reserves—at the writing (January 2022), nearly $4
trillion—occurs when the Fed purchases assets (so called “open market
operations”) or when it makes loans.54 Throughout the course of each
50. For a more nuanced discussion of the synchronization of the different steps, see
supra note 46.
51. For definitions of these terms, see COMMITTEE ON PAYMENT AND SETTLEMENT
SYSTEMS, BANK FOR INT’L SETTLEMENTS, A GLOSSARY OF TERMS USED IN PAYMENTS
AND SETTLEMENT SYSTEMS (2003) https://www.bis.org/cpmi/glossary_030301.pdf
[https://perma.cc/8XHY-FBBT] [hereinafter GLOSSARY].
52. Commercial banks also keep accounts with each other, an arrangement known
as “correspondent” banking. For discussion of historical significance, see infra Section
II.
53. The statutory basis of the legal tender quality of Federal Reserve notes and
reserve in a Federal Reserve account is different. While the former derives from U.S.C.
§ 5103, the latter derives from a combination of U.C.C. § 4A and Federal Reserve
Regulation J, 12 C.F.R. 211, Subpart B, § 210.30. See Geva, supra note 40, at 653; 12
C.F.R. § 211.20-211.30 (2022).
54. Data from Reserves of Depository Institutions: Total, FED. RSRV. BANK OF ST.
LOUIS, https://fred.stlouisfed.org/series/TOTRESNS [https://perma.cc/ETA5-LMS4]
business day, banks use the Fed’s payment system—Fedwire—to transfer

reserves to each other in settlement of payments. For 2020, the average
value of daily Fedwire transfers was over $3 trillion.55 The key to
appreciate here is that for the commercial banking system, the Fed is in
fact able to operate as a single large bank, where a payment from the
Buyer’s Bank to the Seller’s Bank simply means a debit of reserves in the
Buyer’s Bank’s Fed account, and a credit of reserves to the Seller’s
Bank’s account.56 Settlement in reserves is so important because it shows
us that bank money creation ultimately rests on the availability of public
money to settle in.57
The second way to clear is called “netting” or “setoff” of obligations.
The idea is simple. A typical bank is sometimes in the position of needing to
make payments to other banks (when its depositors are initiating transfers)
and is other times in the position of receiving payments (when its depositors
are receiving transfers). That is, throughout the course of the day, a bank is
sometimes like the Buyer’s Bank, and other times like the Seller’s Bank.
Netting or setoff is a legal concept that—provided certain conditions are
met—allows a debtor to reduce the amounts it owes by the amounts it is
owed. Assume that a certain bank has $500 million in payment obligations
and $400 million in claims to receive payment. Where netting is permissible,
$400 million can be canceled (or “netted”), and only the remaining balance
of $100 million will need to be settled in public money (reserves transferred
through Fedwire).
Sounds simple enough, but the basic arithmetic understates the legal
and institutional complexity that netting involves. The right of setoff is
governed by state law that typically incorporates two common law
requirements: for obligations to have “matured” and for there to be
(last viewed Jan. 22, 2022) [hereinafter Reserves]. For legal authorities, see Federal
Reserve Act, 12 U.S.C. §§ 226, 14 (open market purchases), 10B (lending to member
banks), and 13(3) (lending during unusual and exigent circumstances).
55. See BIS RED BOOK, supra note 37. Daily volumes are averaged across 260
business days. See also Fedwire Funds Service – Annual Statistics, FED. RSRV.,
https://www.frbservices.org/resources/financial-services/wires/volume-value-
stats/annual-stats.html [https://perma.cc/KFU4-CGVK] (last visited Jan. 22, 2022)
[hereinafter BIS Statistics].
56. See Mehrling, supra note 48 at 5.
57. For this hierarchy between deposits and reserves, see generally Perry Mehrling,
The Inherent Hierarchy of Money, in SOCIAL FAIRNESS & ECONOMICS: ECONOMIC
ESSAYS IN THE SPIRIT OF DUNCAN FOLEY (Lance Taylor, Armon Rezai & Thomas Michl
eds., 2015).
“mutuality” between parties.58 Maturity means that both obligations are

already due. Where Bank A owes Bank B now, but Bank B owes Bank A
tomorrow, there can generally be no setoff. 59 Mutuality means that
obligations must be owed between the same parties.60 A situation where
Bank A owes $1 million to Bank B, and Bank B owes $1 million to Bank
C, does not meet the requirement for mutuality, so Bank B would have to
pay Bank C, and be paid by Bank A. To fulfill these rather strict
requirements, banks can operate as members of a clearinghouse.
Clearinghouse rules determine that all obligations are due at the end of
the day (thereby fulfilling the maturity requirement) and all obligations
are owed to and from the clearinghouse (thereby fulfilling the mutuality
requirement).61 The largest payments clearinghouse in the U.S. dollar is
called CHIPS and during 2020 it cleared about $1.5 trillion daily. 62
Because it seems like reserves are not required for netting, 63 netting
appears to represent a kind of independence of bank money creation from
public money. As discussed below, this mindset, emphasizing the
importance of private arrangements, sits comfortably within the
conventional view, and is one of the reasons for its abstraction from
clearing.
To summarize, without clearing there can be no deposit transfers,
and without deposit transfers, deposits cannot serve to settle payments
between bank customers. The volume of clearing is enormous—for 2020,
nearly $5 trillion per business day (about one third of the outstanding
deposits during that time) and some $1,200 trillion over one year
58. See Peter M. Mortimer, The Law of Set-off in New York, INT’L FIN. L. REV., May
1983, at 24, 27 (“In addition to the requirement that the debt against which the set-off is
effected be mature, the law of setoff generally requires that the debts be between the same
parties.”).
59. The requirement for matured obligations has exceptions, most importantly under
the U.S. Bankruptcy Code. Note, however, that in the context of clearing, the restriction
on setoff goes beyond even the requirement for maturity. With clearing, it is not simply
that a later obligation exists, but has not yet matured (B owes A tomorrow) but that the
obligation itself has not yet been created (B may, or may not, owe A at a later point
depending on a flow of payments that is yet unknown).
60. See Benjamin Geva, The Clearing House Arrangement, 19 CANADIAN BUS. L.J.
138, 142 (1991).
61. See id. at 143 (for the mutuality requirement).
62. See BIS Statistics, supra note 55 (CHIPS clearing volume for 2020 was $382
trillion, divided by 260 business days).
63. In Part III below, this Article argues reserves continue to play a crucial role in
netting schemes.
(meaning that every $1 in deposits supports, on average, about $80 in

payments over one year).64 Given the centrality of clearing, one would
expect that an account of bank money creation would begin by explaining
how banks are able to successfully clear their obligations. As we now turn
to discuss, that is not the path taken by the conventional view.
B. THE CONVENTIONAL VIEW ABSTRACTS FROM CLEARING
The conventional view’s account of bank money creation has no

clearing in it. That account focuses instead on the individual decisions
depositors make as to the trustworthiness of their bank. Left out of the
picture are the financial relationships between the banks themselves, and
with them, the need for governance or central authority. This abstraction
relegates governance to the modest role of correcting occasional market
failures (depositor bank runs) rather than allowing money creation from the
onset. The following paragraphs take a closer look at these themes by
breaking the conventional view into three basic steps.65
The first step is locating the moneyness of deposits in private
contractual arrangements. As we have seen, a deposit contract includes a
bank’s promise to pay the depositor money—cash—on demand. As the
argument goes, if the depositor trusts the bank’s ability to redeem, the bank’s
promise—the deposit—can become as good as money itself. It is common,
64. See Deposits, supra note 1 (total deposit figures for 2020 were $13.2 trillion (Jan.
1) and $16.1 trillion (Dec. 30)); and BANK FOR INTERNATIONAL SETTLEMENTS, supra note
42 (2020 clearing figures totaling $1,222 trillion avaeraged across 260 business days).
The payments-per-deposit ratio of about 80 is calculated as $1,222 trillion divided by
$14.5 trillion (rough average of deposits for 2020).
65. My focus in this Article is with narrative descriptions of the conventional
account, but the same basic approach can be seen in the three foundational formal models
of bank deposits. See, e.g., Charles W. Calomiris & Charles M. Kahn, The Role of
Demandable Debt in Structuring Optimal Banking Arrangements, 81 AM. ECON. REV.
497, 508 (1991) (commitment device models); Gary Gorton & George Pennacchi,
Financial Intermediaries and Liquidity Creation, 45 J. FIN. 49 (1990) (information
asymmetry models); Douglas W. Diamond & Philip H. Dybvig, Bank Runs, Deposit
Insurance, and Liquidity, 91 J. POL. ECON. 401 (1983) (consumption insurance models).
In the second and third models (information asymmetry device and consumption
insurance models) the “banks” being models do not make obligations to pay in money,
but rather in a consumption good. With no dollar obligations, there is no question of
clearing. In the first model (commitment device) the bank’s obligation is denominated in
dollars, but the model explicitly assumes away any transactions between depositors, i.e.,
it assumes away the monetary use of a deposit that gives rise to clearing. The model also
assumes a single monopolist bank, which makes clearing unnecessary. For a typology of
these models, see generally Ricks, supra note 8.
for example, to speak of the origins of banking as a warehousing

arrangement.66 According to that account, early depositors found it
inconvenient to use coin (which is subject to theft, erosion, etc.) so they
exchanged it for certificates of deposit with a currency warehouse. Given
their convenience, people gladly accepted the certificates in payment,
knowing they could always redeem them for coin when needed. Over time,
the currency warehouse noticed few of the certificates were ever redeemed,
so its coin was lying idle. The clever warehouse management decided it could
benefit from this fact by offering loans. The warehouse would now issue new
certificates (promises to pay coin on demand) and lend them in return for the
borrower’s own promise for future payment.67 Interest on these loans is a
profitable business for the warehouse. As more loans are made, the amount
of coin at the warehouse would become smaller in proportion to the growing
number of certificates. Gradually, the warehouse’s balance sheet acquires the
typical shape of a bank discussed above, with long-term loans funded
through short-term liabilities (the certificates). In jargon, we also speak of
banks as operating on a “fractional–reserve” basis, meaning that the coin
reserve is a fraction of the certificates.
The details of such accounts vary, but they all attribute a sense of
agency to individual decisions in making deposits function as money.
What drives these accounts is ultimately the convenience a depositor
attributes to the certificate,68 and their confidence in the bank’s ability to
redeem. At the outset, that trust hangs on prudent conduct by the banker.
The loans the banker makes need to be sound; the banker needs to have
adequate capital to absorb losses69 and to maintain an adequate cash
66. See George Selgin, Those Dishonest Goldsmiths, 19 FIN. HIST. REV. 269, 270-
72 (2012) (The Article, which offers a critique of the warehouse origins story, begins by
documenting the ubiquity of that warehouse origins story in leading works in economics);
Ricks, supra note 8, at 55-57; see also The Economist, Could Digital Currencies Put
Banks Out of Business?, YOUTUBE (May 8, 2021), https://www.youtube.com/watch?v=
fpb-qJv6dBs [https://perma.cc/3J36-NU4N].
67. In some accounts, instead of issuing new certificates, the warehouse lends out a
portion of its bullion.
68. This line of thinking resonates with current efforts (both outside and inside the
banking system) to use blockchain technology in creating means of payment that are
faster, more efficient, etc. See, e.g., Our Story, RIPPLE, https://ripple.com/company/
[https://perma.cc/326A-FTZE] (last visited Jan. 22, 2022); J.P. Morgan Creates Digital
Coin for Payments, J.P. MORGAN: ONYX (Feb. 1, 2021), https://www.jpmorgan.com
/solutions/cib/news/digital-coin-payments [https://perma.cc/LE6Z-5GDX].
69. In banking regulation, “capital” is roughly synonymous with “shareholders’
equity,” which is the difference between assets and liabilities. See supra note 49. Capital
reserve for occasional withdrawals. When these conditions are met,

deposits are assumed to acquire “moneyness” through the countless
optimizing decisions of depositors, bankers, and borrowers. This image
soon acquires a normative significance—it is perceived as efficient and
desirable. After all, the depositor enjoys a more convenient medium to
transact in, and the borrower enjoys new credit. These benefits are
possible thanks to the bank’s incentives to expand the money supply and
lend to maximize profit.
Things would be simple if this were all, but the conventional view is
deeply aware that bank money creation is unstable. The second step is an
occasional market failure that disrupts the efficient arrangements
described above. Enter the bank run. The run is commonly (though not
exclusively) understood as a coordination failure among depositors. 70
Recall that a defining feature of banking is the bank’s maturity
transformation and its operation on a fractional reserve. This business
model exposes the bank to liquidity risk. If more than a small fraction of
depositors simultaneously demands cash payment, the bank will exhaust
its small reserve, and will have to suspend further payments. Depositors
are aware of this, and their awareness may drive them to run and redeem
before the cash reserve is depleted. The result is a self-fulfilling prophecy,
where the fear of a run is sufficient to generate a run. The initial reason
for the run may involve doubts over the bank’s performance. At the same
time, the history of bank runs in the United States reveals that even
solvent banks are vulnerable to run dynamics, especially in an
environment of panic and financial contagion. As shown by the Great
Depression, the results of these runs to the real economy can be
devastating.
This is where the third and final step comes in. To avoid disastrous
runs, the state steps in to protect the moneyness of deposits.71 Lender-of-
last-resort support by the central bank provides banks with immediate
thus represents loss absorption capacity that protects a corporation’s creditors. For capital
requirements under U.S. law, see 12 U.S.C. §§ 1831 and 5371.
70. See, e.g., E. GERALD CORRIGAN, FED. RSRV. BANK OF MINNEAPOLIS, MINN., ARE
BANKS SPECIAL?, ANN. REPS. 1982 (1983), http://www.bu.edu/econ/files/2012/01/Corrigan-
Are-Banks-Special_main-text.pdf; Diamond & Dybvig, supra note 65; Ricks, supra note 8, at
62-73; Baradaran, supra note 9, at 1314. For further discussion of runs see Orian Peer, supra
note 9, at 378-81.
71. Daniel K. Tarullo, Member, Bd. of Governors of the Fed. Rsrv., Remarks at The
Clearing House 2014 Annual Conf.: Liquidity Regul. (Nov. 20, 2014) (transcript
available at https://www.federalreserve.gov/newsevents/speech/tarullo20141120a.pdf
[https://perma.cc/WP69-NT52]); see CORRIGAN, supra note 70.
cash against collateral, thereby helping meet the immediate obligation to

redeem.72 Federal deposit insurance goes even further, guaranteeing
depositors would maintain their moneyness even in the event of bank
failure.73 At present, the deposit insurance cap is $250,000 and total
deposits insured are about $9.5 trillion, around one half of total deposits. 74
Because insured depositors are indifferent to a bank’s failure, they do not
run. That was precisely the goal, but it also creates a side-effect in the
form of moral hazard. After all, if insured depositors may rationally
choose to stay with a risky bank, what is preventing that bank from taking
outsized risks? Bank shareholders would enjoy any outsized profits, and
the government would absorb any outsized losses.75
Following this way of thinking leads to the role of modern banking
regulation. As the largest potential creditor of the banking system, the
government steps in to require that banks comply with prudential
regulation.76 In this way, the state’s regulatory role is understood to mimic
the behavior of private depositors in a counterfactual world without public
crisis supports. This leads to the limited role of public policy in banking
discussed in the Introduction. Because the state is merely addressing a market
failure (depositor runs), its interventions must be narrowly tailored to address
that failure, while trying to minimize interference with the underlying private
activity. An important consequence of this approach is limiting public policy
to prudential regulation (adequate levels of risk). Exceptions to this notion
are so narrow they highlight the norm. The Community Reinvestment Act
(“CRA”), to take one example, requires banks to affirmatively promote
the credit needs of low- and moderate-income (“LMI”) communities, that
make 43 percent of the U.S. population.77 While there are certain
72. See Federal Reserve Act, 12 U.S.C. §§ 10B, 13(3); Orian Peer, supra note 9.
73. See 12 U.S.C. § 16 (Fed. Deposit Ins. Corp.).
74. See Deposit Insurance FAQs, FED. DEPOSIT INS. CORP. (Dec. 8, 2021), https:/
/www.fdic.gov/resources/deposit-insurance/faq [https://perma.cc/X4CM-8E75]; Quarterly
Banking Profile: Third Quarter 2021, 15(4) FDIC Q. 1, 24 (2021).
75. See MICHAEL BARR ET AL., FINANCIAL REGULATION: LAW AND POLICY 248-49
(2016).
76. See, e.g., Tarullo, supra note 71, at 3 (“Regulatory requirements were imposed
to guard against the moral hazard that both programs [lender of last resort and deposit
insurance] could create.”).
77. See Low- and Moderate-Income Residences Can Help Modernize the U.S.
Electric Grid, NAT’L RENEWABLE ENERGY LAB’Y (Apr. 25, 2018), https://www.nrel.gov
/news/program/2018/lmi-residences-can-help-modernize-us-electric-grid.html
[https://perma.cc/TW6B-3XPJ].
difficulties quantifying the total proportion of bank resources dedicated

to CRA compliance, the percentage of loans earning CRA credit is
strikingly small, under 3 percent of bank assets funded through deposits. 78
Looking at the conventional view, one is left to wonder how banking
regulation—and banks’ social responsibilities—could have been differently
understood if the question of clearing came to the fore. It is important to
understand that abstraction from clearing is not generally the result of
ignorance. Most authors are familiar with the basic aspects of clearing, and
some have done highly sophisticated work on various aspects of clearing.
The issue is not lack of awareness of clearing per se. It is maintaining the
knowledge of clearing and the account of bank money creation separate
from each other. The next Part turns to examine how the conventional
view fills this gap between its theoretical account and the institutional
reality.
II. SIDESTEPPING THE QUESTION: ASSUMPTIONS BEHIND

ABSTRACTION FROM CLEARING
The conventional view’s abstraction from clearing is an “as if”

proposition. Authors are aware that banks need to clear but believe one
could describe the fundamental causes of bank money creation (market
forces) “as if” clearing was not an important concern. Three assumptions,
often quite subtle and implicit, are behind this “as if” proposition: first,
that netting is freely available to banks; second, that the law of large
numbers prevents clearing drains; and third, that interbank markets can
solve drains if they do arise. This Section elaborates on each of these
assumptions. It explains why they are difficult to test in a modern setting,
and sets the stage to examine them in the historical case study in Part III.
78. The under 3 percent figure reflects the sum of around $334 billion “earning CRA
credit [in 2018]” divided by total commercial bank deposits of about $12 trillion for 2018.
See Laurie Goodman et al., Under the Current CRA Rules, Banks Earn Most of Their CRA
Credit Through Community Development and Single-Family Mortgage Lending, URB.
INST.: URB. WIRE (July 9, 2020), https://www.urban.org/urban-wire/under-current-cra-
rules-banks-earn-most-their-cra-credit-through-community-development-and-single-
family-mortgage-lending [https://perma.cc/6YKR-CWA5]; Deposits, supra note 1. As
Goodman et al. note, due to data limitations, their data does not include CRA investments
(which are distinct from CRA loans). Adding investments to the numerator would likely
result in a figure larger than 3 percent. That said, CRA investments often tap into other
public subsidies, e.g., Low-Income Housing Tax Credits and New Markets Tax Credits.
A. NETTING IS FREELY AVAILABLE TO BANKS
The first assumption is that banks can freely use netting to minimize
their need for reserves. Remember that, over the course of the day, a bank
is sometimes in the position of owing other banks, and is other times in
the position of being owed. The net figure of these payments is
considerably smaller than the “gross” figure (total outgoing payments).
By netting these payments a bank can dramatically reduce its need for
reserves, which is to say, its need for state provided money. 79 This makes
clearing seem far less challenging or worthy of analysis.
While netting is indeed an important concept, speaking of netting as a
mere matter of arithmetic—payments in offsetting payments out—misses
crucial aspects of the legal arrangements that netting requires. This
abstraction from law creates a sense that netting is a free lunch. In reality,
netting creates risks that someone has to bear. At a basic level, netting
involves an extension of credit between banks. Instead of paying
immediately, the hour of settlement is extended until the end of the day,
allowing banks that happen to run morning deficits to recover them over
the course of the day. These deficit banks, however, can also default at any
time during the day, meaning someone would have to bear that risk. As the
historical case study demonstrates, default risk in the clearinghouse
triggered a liability on behalf of surviving banks, and that liability could
have further led to a cascading default.80 The modern image of netting as a
free lunch takes attention away from default risk and the daily governance
that is necessary to manage it.
Modern commentators’ assumption that netting is freely available is
a product of modern central bank practice. Central banks remove much of
the risk involved in clearing from banks by assuming it themselves. The
obvious case in point is Fedwire, that accounts for over two-thirds of daily
clearing volume in the United States. Fedwire is not formally a
clearinghouse. But while Fedwire does not provide banks with netting in
a legal sense, it certainly provides them with the practical benefits of
netting. Here is how it works: Fedwire is a Real Time Gross Settlement
(“RTGS”) system.81 This means that a bank lacking reserves in its Fed
account during the morning cannot simply go ahead with the payment and
net it out against incoming payments in the afternoon. That bank would be
79. For an empirical example from NYCHA practice, see infra Section III.A.
80. See infra Section III.A.
81. See BIS RED BOOK, supra note 37; see also GLOSSARY, supra note 51, at 41.
stuck, and other banks relying on the incoming reserves to make payments
would be stuck as well.82 This potential gridlock poses a problem to the
Fed, which is interested in facilitating smooth payments. That problem was
especially acute in the pre-GFC era, where the supply of reserves was very
low compared to today-only a few tens of billions compared to nearly $4
trillion (January 2022).83 To address this gridlock, the Fed provides banks
short of reserves with so-called “intraday loans” or “daylight overdrafts.”84
These are loans that can be taken during the day and paid back from
incoming payments before the close of the day.85
Note how by coupling RTGS with daylight overdrafts, the Fed is
replicating the benefits of clearinghouse netting to banks, but it does so
without them having to take any risk of default. If a bank defaults during
the day, the default will not affect banks who have already received
payments, because payment in reserves is final, and all risk is borne by the
Fed (in its capacity as intraday lender to the defaulting bank).86 By
removing risk from banks, the Fed is also removing any need for them to
engage in collective action to avoid default in the first place. The result is a
system that seems more privatized than it really is.
But what about CHIPS? Unlike Fedwire, CHIPS is a clearinghouse,
and does operate a netting system. Presumably, CHIPS would involve the
kind of default and cascading default risk mentioned above, as well as the
82. See Morten L. Bech, Intraday Liquidity Management: A Tale of Games Banks
Play, 14 ECON. POL’Y REV. 7, 8 (2008).
83. See Reserves, supra note 54. The rise in reserves was a result of the Federal
Reserve’s Large-Scale Asset Program (LSAP) in the aftermath of the GFC, and similar
policies since March 2020. Rising reserves were a side-effect rather than the goal of these
policies. See Stefania D’Amico et al., The Federal Reserve’s Large-Scale Asset Purchase
Programs: Rationale and Effects, (Fin. and Econ. Discussion Series, Fed. Rsrv. Bd.,
Working Paper, 2012).
84. See generally Douglas Evanoff, Daylight Overdrafts: Rationale and Risks, FED.
RSRV. BANK CHICAGO: ECON. PERSPECTIVES, May-June 1988, at 18.
85. See Federal Reserve Regulation J, 12 C.F.R. § 210.28 (2022); BD. OF
GOVERNORS OF THE FED. RSRV. SYS., FEDERAL RESERVE POLICY ON PAYMENT SYSTEMIC
RISK (2021), https://www.federalreserve.gov/paymentsystems/files/psr_policy.pdf [https
://perma.cc/QQU3-ESSQ] [hereinafter POLICY]. The demand for daylight overdrafts has
fallen substantially since the rise of reserves in the post-GFC era (discussed in Bech,
supra note 82). For data, see Daylight Overdrafts and Fees, BD. OF GOVERNORS OF THE
FED. RSRV. SYS. (Feb. 4, 2020), https://www.federalreserve.gov/paymentsystems/psr_
data.htm [https://perma.cc/RJB2-57A2].
86. See Regulation J, 12 C.F.R. § 210.31 (2022); E.J. Stevens, Risk in Large-Dollar
Transfer Systems, ECON. REV.: FED. RSRV. BANK OF CLEVELAND, Fall 1984, at 2, 4; Geva,
supra note 40, at 652-53.
need for governance to manage them. The complicated answer is that the
modern CHIPS differs in important ways from the traditional netting
systems that many commentators still have in mind. Since 2001, CHIPS
relies on liquidity available through Fedwire, so much so, that it can
virtually eliminate default risk altogether.87 While formally still a netting
system, CHIPS is more akin to a sophisticated algorithm that benefits from
central bank liquidity. Even prior to the 2000s, policymakers highlighted
the ways in which settlement on CHIPS was indirectly benefiting from
central bank liquidity to settle netted positions.88 As it turns out, under
modern conditions, there is no way to disentangle the workings of netting
systems that reduce the need for reserves from central bank practice to
flexibly provide those reserves as needed.
B. THE LAW OF LARGE NUMBERS PREVENTS DRAINS
While the first assumption is that netting is freely available to banks,

the second assumption concerns its enormous power. Specifically,
authors discuss the way in which the law of large numbers (“LLN”)
reduces large clearing flows to a small and predictable net position.
Scholars note that the LLN enables banks to engage in maturity
transformation with only a small reserve.89 Implicit in this position is the
assumption that the workings of the LLN prevent clearing drains—days
and weeks where a bank experiences exceptionally large negative clearing
positions. In this way, reliance on the LLN leads to an underestimation of
87. See BIS RED BOOK, supra note 37, at 490; Stacy Panigay Coleman, The
Evolution of the Federal Reserve’s Intraday Credit Policies, FED. RSRV. BULL., Feb.
2002, at 67, 74; James McAndrews & Samira Rajan, The Timing and Funding of Fedwire
Funds Transfers, 2000 FRBNY ECON. POL’Y REV. 17, 26 (2000). For context to these
reforms, see infra discussion in text accompanying notes 141 and 143.
88. See Panigay Coleman, supra note 87, at 68 (for reliance on intraday liquidity);
see Stevens, supra note 86 (for reliance on lender-of-last-resort).
89. See ERIC A. POSNER, LAST RESORT: THE FINANCIAL CRISIS AND THE FUTURE OF
BAILOUTS 11 (2018) (“The key to maturity transformation is a statistical law—the law of
large numbers.”); Manmohan Singh & Peter Stella, Money and Collateral (International
Monetary Fund, Working Paper 12/95, 2012)(“Owing to the law of large numbers, banks
have—for centuries—been able to safely conduct this business [maturity transformation]
with relatively little liquid reserves, as long as basic confidence in the soundness of the
bank portfolio is maintained.”); Anat R. Admati & Martin F. Hellwig, Bank Leverage,
Welfare, and Regulation 5 (Rock Ctr. for Corp. Governance at Stan. U., Working Paper
No. 235, 2019); RICKS, supra note 7, at 758.
the risk to which banks are subject, and the real-life mechanisms used to
protect against that risk.
The law of large numbers is a foundational theorem in probability. It
states that when an experiment is repeated many times, the average of the
observed results would converge to the expected value. For example, a
fair coin (equal probability of heads and tails) flipped many times should
come close (though not identical) to 50 percent heads.90 The LLN could
be translated into a highly stylized clearing arrangement. Consider two
banks, Bank A and Bank B, each with 500 deposits of $1.91 A coin is
tossed 1,000 times, representing transactions taking place between
depositors over the day. When the coin comes out as heads, a depositor in
Bank A is receiving a transfer from a depositor in Bank B, resulting in a
clearing inflow of +$1 to Bank A. Conversely, when a coin comes out as
tails, a depositor in Bank A is making a transfer to a depositor in Bank B,
resulting in a clearing outflow -$1 to Bank A. The expected clearing
position in this experiment for Bank A would be zero.92 With a large
enough deposit base (a large number of coin tosses) the observed result
over the clearing day should approach that expected zero.
In its basic form, the application of the LLN to clearing is insightful.
Surely, the LLN is significant to banks’ ability to engage in maturity
transformation. At the same time, the way authors use the LLN has a basic
limitation insofar as it goes the extra step to assume the LLN works to
prevent clearing drains all together. This position is rarely stated
explicitly, but it is heavily implied. Expositions often describe the
enormous power of the LLN in enabling maturity transformation, then
move on to state that in a run, banks can no longer count on the LLN—
because deposits transfers are no longer random. The general notion
conveyed is that outside of crisis, the LLN compresses the net clearing
position to a figure so small it is nearly trivial. 93 As discussed below, the
90. See Richard Routledge, Law of Large Numbers, BRITANNICA, https://

www.britannica.com/science/law-of-large-numbers [https://perma.cc/3ALZ-M4HU]
(last visited Jan. 29, 2022). The LLN applies to events that are identically and
independently distributed (“i.i.d.”). It is an interesting question whether or not clearing
flows can actually be characterized as i.i.d., but this inquiry is beyond the scope of this
Article.
91. The example is highly stylized in that the real-life banking system has multiple
banks, of various sizes, and the size of deposits and potential transfers within each bank
varies considerably as well.
92. =0.5*$1 + 0.5*(-$1)
93. See RICKS, supra note 7, at 58; KEYNES, supra note 30.
pre-GFC practice, when banks held reserves of under 1 percent of their

deposits, has surely contributed to that impression.
One cause for suspicion is the lack of detailed analysis specifying
how small exactly the reserve can be. Here, note that the LLN does not
speak of the observed value (the net clearing position) being identical to
the expected value (zero). It speaks merely of convergence given a “large
number” of observations. But what counts for a large enough number of
deposit transfers that would make the clearing position sufficiently close
to zero? And how close exactly does it have to be? And at what level of
confidence can we expect it to be so close? Expositions of bank money
creation rarely engage with these questions, let alone provide empirical
tests of real-life clearing positions.94 The lack of empirical analysis is
partly attributable to data limitations. Public Fedwire data is provided on
a highly aggregated basis, often consolidating the clearing activity of all
participating banks, and averaging it across long periods of time.95 The
clearing position of an individual bank on an individual day is proprietary
information and is not generally observable to researchers.96 Without this
kind of empirical calibration, it is difficult to rule out that on some days,
some banks would be subject to clearing drains that would put them under
serious pressure. What is more, even if such data were readily available,
the staggering institutional complexity of the modern system would have
made it somewhat difficult to interpret.
An important benefit of the historical case study is that the NYCHA
ledger offers data on the individual clearing positions of member banks on
individual days. Further, the relative institutional simplicity of the system
around that time makes the data simpler to interpret. As discussed below,97
the empirical findings from the 1905 ledger demonstrate large clearing
drains did in fact occur. The findings, in my view, do not necessarily
indicate a problem with the application of LLN to clearing per se, but with
94. For discussion of earlier theoretical (rather than empirical) work on these issues,
see generally J.H.G. Olivera, The Square-Root Law of Precautionary Reserves., J. POL.
ECON. 1095 (1971).
95. See Fedwire Funds Service – Annual Statistics, FED. RSRV., https://www.
frbservices.org/resources/financial-services/wires/volume-value-stats/annual-stats.html
[https://perma.cc/UPV6-A8FS] (last visited Jan. 29, 2022).
96. There are exceptions with respect to data availability, especially for researchers
working in, or collaborating with, central banks. See discussion of studies in note 142,
infra. Despite data availability, these studies did not utilize institution–level data for the
purposes envisioned here.
the specific way in which it is currently used. As discussed below, a

different way of applying the LLN to clearing actually supports the
conclusion that large drains are inevitable given a sufficiently large number
of banks and clearing days.
C. INTERBANK MARKETS CAN SOLVE DRAINS
The third assumption used to abstract from clearing is the confidence

in well-functioning interbank markets. Where the second assumption
asserts that the law of large numbers prevents drains, the third assumption
acknowledges the occurrence of drains but asserts these drains can be
effectively handled through contracting between banks.98 In practice, the
two assumptions are often intermingled, namely, clearing drains are
rarely a concern, but if they are, interbank markets stand ready to handle
them.99
In the United States, since the GFC, the Federal reserves’ program
of Large-Scale Asset Purchases (colloquially known as quantitative
easing, or “QE”) has brought reserves up to $4 trillion, around 20 percent
of commercial bank deposits. This large amount of reserves is a sharp
departure from the pre-GFC era, where reserves were often below 1
percent of deposits.100 In an environment with such little reserves, the
interbank market—the so-called federal funds market—was essential to
the system’s ability to clear. As previously explained, over the day, banks
needing to make payments in excess of their available reserves would use
daylight overdrafts, but come evening, any remaining balances had to be
paid back to the Fed.101 To acquire the necessary reserves, deficit banks
turned to the federal funds market, where they could borrow from banks
with excess reserves.102 In the decade preceding the GFC, about $200 to
98. See, e.g., Zoltan Jakab & Michael Kumhof, Banks Are Not Intermediaries of
Loanable Funds–And Why This Matters 10 (Bank of Eng., Working Paper No. 529,
2015).
99. For a summary of the literature, see Admati & Hellwig, supra note 89, at 5.
100. See Reserves, supra note 54, and Deposits, supra note 1 (e.g., for Jan. 2007, 45
billion in reserves over $6.1 trillion in deposits).
101. The discussion in these paragraphs builds on Perry Mehrling's microstructure
approach to monetary policy implementation. See Perry Mehrling, Monetary Policy
Implementation: A Microstructure Approach, in DAVID LAIDLER’S CONTRIBUTIONS TO
ECONOMICS 212, 224-25 (Robert Leeson ed., 2010).
102. As Mehrling discusses, the Eurodollar market and the repo market provide
additional mechanisms to “resolve a payment imbalance at the end of the day.” Id.
$340 billion worth of transactions were negotiated in the federal funds

market every day.103
This interbank market brings up the alluring image of a banking
system that is autonomous from the central bank, and the public money it
issues (reserves). After all, if one bank needs to make a payment, another
bank must be receiving it. It follows that by accounting identity, for every
bank short of reserves there is a bank with excess reserves somewhere in
the banking system. For an interest rate, the deficit and surplus banks can
arrive at a mutually beneficial trade (often with the help of a broker),
thereby minimizing the need for reserves provided by the central bank.104
For example, Bank A in need of $10 million in reserves could borrow that
amount from Bank B with a surplus of the same amount (see example in
Balance Sheet 4).105
If we follow this line of thinking to its logical conclusion, the demand
for reserves at the end of the day would be a perfect zero. And with a zero
demand for reserves, it is market trading among banks, not public money
supplied by the central bank that is doing the daily work of clearing. 106
The central bank can of course force banks to hold reserves by imposing
regulatory reserve requirements, or by use of other tools with similar
effects.107 But experts highlight that reserve requirements are just an
103. See Morten Bech & Enghin Atalay, The Topology of the Federal Funds Market,
389 PHYSICA A: STAT. MECHS. & ITS APPLICATIONS 5223, 5227 (2010). A large
proportion of trading volume–about half—reflected banks acting as dealers, borrowing
funds earlier in the day, and lending them (at a mark-up) later in the day. See Gara Afonso
& Ricardo Lagos, An Empirical Study of Trade Dynamics in the Federal Funds Market
19 (Fed. Rsrv. Bank of N.Y., Working Paper No. 550, 2010).
104. See Timo Henckel et al., Central Banking Without Central Bank Money 15-16
(Int’l Monetary Fund, Working Paper No. 99/92, 1999).
106. A 1999 IMF Working Paper discusses this exact hypothetical:
Should the money market always clear at the end of the day as a closed
system (i.e., without net liquidity injections or contractions by the
central bank) a private broker–rather than a central bank—could, in
principle, facilitate the recycling of liquidity and post for such
transactions an interest rate band that might differ from that of the
central bank.
Henckel et al., supra note 104, at 24.

107. These other tools include open market purchases of securities and clearing of
incoming payments to the Treasury account.
intervention into a well-functioning market for reserves, something that a

private system of banks left to their own devices could likely do
without.108 This view of the efficiency of interbank markets predictably
leads to the conclusion that reserves are not truly necessary for settlement.
But can banks, left to their own devices, truly trade away their
clearing drains in an efficient market? Here again, the modern system
makes it difficult to meaningfully address that question. The reason is that
the modern interbank market is in fundamental ways the product of
central bank design and receives ongoing central bank support. The
remarkable efficiency of interbank markets could be attributed to that
ongoing support rather than the power of markets per se. Specifically,
modern central banks target the overnight rate at which banks trade
reserves with each other (in the United States, that is the “fed funds
rate”).109 This targeting of overnight rates is the traditional tool through
which the central bank implements its monetary policy. 110 It is well
understood that maintaining the target rate requires the central bank to
accommodate any unmet demand in the market for reserves.111 When such
demand causes the overnight rate to exceed the central bank’s target rate,
the central bank would increase the supply of reserves so as to bring the
overnight rate back down to its official target. Notice the implications of
this policy for banks’ ability to trade in an interbank market. Banks are
trading in an environment where an adequate supply of reserves is
essentially guaranteed, and the rate of borrowing reserves is predictable.
108. See Henckel et al., supra note 104, at 27. Mervyn King, then Deputy Governor
(later Governor) of the Bank of Eng. (2003-2013), has cited this point in agreement. See
Mervyn King, Deputy Governor, Speech: Challenges for Monetary Policy: New and Old
(Aug. 27, 1999) at 26, https://www.bankofengland.co.uk/-/media/boe/files/speech/1999
/challenges-for-monetary-policy-new-and-old.pdf [https://perma.cc/B58J-KS3B].
109. Since the rise in reserves during the GFC, the supply of reserves is overabundant,
and the targeting of the overnight rate takes place primarily through the payment of
interest on reserves. See Jane E. Ihrig et al., Monetary Policy 101: A Primer on the Fed’s
Changing Approach to Policy Implementation (Bd. of Governors of the Fed. Rsrv. Sys.,
Working Paper No. 2015-047, 2015); Federal Reserve Act, 12 U.S.C. § 19(b)(12).
110. See Ihrig et al., supra note 109.
111. The insight that central banks must accommodate banks’ demand for reserves
originated in the highly innovative work of Post-Keynesian economists in the 1980s,
especially BASIL J. MOORE, HORIZONTALISTS AND VERTICALISTS: THE
MACROECONOMICS OF CREDIT MONEY (1988). It has since become integral to the central
banking literature. See, e.g., Ulrich Bindseil & Philipp J. König, Horizontalists and
Verticalists: An Appraisal 25 Years Later, 1 REV. KEYNESIAN ECONS. 383 (2013). And it
has more recently been included in legal scholarship. See Hockett & Omarova, supra
note 6, at 1162.
That is hardly the kind of setting that would usually pass for a real-life
private market.
An important benefit of the historical case study here is the way it
allows us to examine whether banks would trade in an interbank market
in the absence of ongoing support by a central bank.112 As discussed
below,113 the case study answers this question in the negative. There is no
evidence NYCHA traded reserves in an interbank market. On the other
hand, these banks heavily relied on non-market mechanisms: high levels
of reserves of public money (coin and legal tender notes) and a carefully
coordinated system of “secondary reserves” they created (by internally
recycling call loans to stockbrokers). The historical practice is therefore
fundamentally at odds with the conventional view that casts bank money
creation as the product of individual decisions.
We can now summarize our discussion up to this point. While clearing
is essential to bank money, the conventional view’s account has no clearing
in it. The gap between the practical significance and theoretical disregard
is filled by the three assumptions. These assumptions—that netting is freely
available, that the LLN prevents drains, that interbank markets can handle
drains—are meant to explain why bank money creation can be understood
“as if” clearing did not exist. The result is a vision of bank money creation
that in ordinary times revolves entirely around market forces. The Article’s
argument in this Part was that data limitations, and even more basically,
modern central banks’ deep involvement in the payments system, make
these assumptions extremely difficult to assess. In the next Part, we move
to examine these assumptions through the historical case study of the pre-
Fed NYCHA.
III. ADDRESSING THE QUESTION: DAILY GOVERNANCE IN THE NEW

YORK CLEARING HOUSE ASSOCIATION
The New York Clearing House Association was established in 1853

by private agreement between the city’s main banking institutions. 114 As
stated in its Constitution, the goal of the NYCHA was “effecting at one
112. This question has not been at the center of discussions in the Post-Keynesian
literature, nor has it been discussed in the legal literature on money building on it. See
also discussion infra notes 200 and 201.
114. 1 MARGARET MYERS, THE NEW YORK MONEY MARKET: ORIGINS AND
DEVELOPMENT 95 (Benjamin Haggott Beckhart ed., 1931).
place of the daily exchange between the various Associated Banks, and
the payment at the same place of balances resulting from such
exchanges.”115 That Constitution is an important legal source, providing
the detailed rules governing the clearinghouse over some twenty pages.
The discussion in this Part focuses on the operation of the NYCHA at the
turn-of-the-century, with quantitative data drawn from 1905.116 That
period was chosen for its proximity to the creation of the Federal Reserve
System in 1913, while still preceding the 1907 Panic and the changes it
brought into New York banking practice.117
While financial historians have fruitfully studied the pre-Fed
NYCHA, their work has focused on the NYCHA banks’ collective action
during periods of crisis.118 This approach is analogous to the conventional
view that limits the role of governance to crisis time, while assuming that
in ordinary times bank money creation occurs through simple private
transactions. In distinction, the history presented in this Part seeks to
identify the challenges that clearing presented to the NYCHA banks on a
daily level, and the way these challenges were addressed in practice. The
discussion is organized around the three “as if” assumptions used to justify
the abstraction from clearing: the free availability of netting, the LLN’s
prevention of drains, and the capacity of interbank markets. As discussed
below, the history calls all three assumptions into question.
During 1905, the ranks of the NYCHA included 52 banks, holding
national and state charters, with total deposits of about $1,100 million.119
115. CONST. OF THE N.Y. CLEARING HOUSE ASSOC., § 2, https://babel.hathitrust.org/

cgi/pt?id=wu.89099412611&view=1up&seq=3&skin=2021 [hereinafter NYCHA
CONSTITUTION].
116. The case study is intended primarily as a snapshot of NYCHA practice around
1905. It is not generally meant to describe its historical origins.
117. Some of these changes, like the rise of outside lending in the call loan market,
have already been on the rise since around 1900.
118. See, e.g., GARY GORTON, SLAPPED BY THE INVISIBLE HAND: THE PANIC OF 2007
(2010); Gary Gorton, Clearinghouses and the Origin of Central Banking in the United
States, 45 J. ECON. HIST. 277 (1985); Ellis W. Tallman & Jon R. Moen, Liquidity
Creation Without a Central Bank: Clearing House Loan Certificates in the Banking
Panic of 1907, 8 J. FIN. STABILITY 277 (2012); Ellis W. Tallman & Jon R. Moen, Why
Didn’t the United States Establish a Central Bank Until After the Panic of 1907 (Fed.
Rsrv. Bank Atlanta, Working Paper No. 99-16, 1999). In distinction from the prevalent
focuson crisis dynamics, my emphasis on daily routines builds on works like MYERS,
supra note 114, and Mehrling, supra note 101.
119. New York Clearing House Association, Statement of the Associate Banks of the
City of New York from Reports to the New York Clearing House --- for Week Ending,
Harvard Social History/Business Preservation Microfilm Project, Project 3; 90037 (Jan.
The size of these banks could be divided into three tiers: the “Big Six”
banks (deposits of $50 to $190 million) held about one half of total
members’ deposits,120 with the remainder divided roughly equally between
15 medium-sized banks (deposits of $15 to $35 million) and 31 smaller
banks (deposits below $10 million). The NYCHA was part of a broader
system of clearinghouses that existed in dozens if not hundreds of cities
across the United States. Like other clearinghouses, membership in the
NYCHA was limited to local city banks, but the significance of the
NYCHA went far beyond this local scope. Over half of NYCHA banks’
deposits—nearly $600 million—were not held by individuals, but by out-
of-town banks for whom the NYCHA banks acted as correspondents.121
The out-of-town banks used these so called “bankers’ balances” to settle
their inter-regional payments.122 The NYCHA therefore stood at the apex
of the U.S. payments systems. The only modern counterpart in its sense
of significance would be Fedwire.
A. NETTING WAS NOT FREELY AVAILABLE
When the Constitution of the NYCHA speaks of “effecting at one

place of the daily exchange between the various Associated Banks, and
the payment at the same place of balances resulting from such exchanges”
it is speaking of a system of multilateral netting.123 Described in Part I,
such a system would allow banks to net their obligations throughout the
day, and across all clearinghouse members. Prior to the creation of the
Fed, this system was the legal infrastructure that actually allowed banks
to enjoy the benefits of the law of large numbers. This Section argues that
netting in this way was not freely available as we have come to assume,
but involved the banks taking on substantial risk exposure towards one
7, 1905) statement [hereinafter NYCHA Weekly Statements]. The size of deposits waned
and waxed with seasonal cycle with the 1905 peak being around $1,200 million and
through being around $1,000 million.
120. The Big Six were National City (#8), National Bank of Commerce (#23),
Hanover National Bank (#33), National Park Bank (#54), First National Bank (#65), and
Chase National Bank (#74). The numbers in parentheses represent the banks’ identifying
numbers on the clearinghouse ledger and weekly statements discussed below.
121. See MYERS, supra note 114, at 241.
122. See LEONARD LYON WATKINS, BANKERS’ BALANCES BEFORE AND SINCE THE
FEDERAL RESERVE SYSTEM 102 (1926).
123. NYCHA CONSTITUTION, supra note 115 § 2.
another. We can understand this risk exposure by following the daily

routine of netting, as well as the procedures set out in the event of default.
In 1905 netting was a corporeal experience. Every morning around
09:45 a.m., clerks of the various NYCHA banks would enter the gallery of
the elegant clearinghouse building at the corner of Nassau and Cedar Street.
At 10:00 a.m. sharp, the clearinghouse manager sounded the gong, and a
curious ritual began: “four columns of young men moving simultaneously
like a military company in step.”124 At the center of the clearinghouse
gallery were 52 desks, each assigned to one of the member banks. Standing
behind the desks were “settling clerks” that each bank had sent to receive
packages from the other banks. Rotating around these same desks were
“delivery clerks” that each bank had sent to deliver packages to the various
settling clerks. The contents of these packages were all of the checks drawn
against the settlement clerk’s bank that were received in the delivery clerk’s
bank within the past 24 hours. All in all, each morning some 2,652 packages
between 1,326 pairs of banks were handed this way from the delivery to
the settlement clerks.125
To each package was appended an “exchange slip” summing the
total dollar amount of checks delivered.126 These exchange slips were
important because they represented credits to the delivery clerk’s bank,
and amounts owed by the settling clerk’s bank. A bank’s clerks would
rapidly sum all the exchange slips given and received to and from the
other 51 banks. In jargon, one speaks of the difference between credits
and debits within each pair of banks (one of the 1,326 pairs) as a “bilateral
net.” As discussed in Part 1, in the absence of a clearinghouse, each one
of these bilateral net positions would have had to be settled
124. J.G. CANNON, CLEARING-HOUSE METHODS AND PRACTICES, S. Doc. No.

491(1910) reprinted in CLEARING HOUSES AND CREDIT INSTRUMENTS 195 (1911). The
author, J.G. Cannon, was a contemporary New York banker who had started his career
as a clearinghouse clerk. The description that follows draws heavily on his account in
Chapter XIV of the book.
125. Where n is the number of NYCHA banks (52), the number of pairs (1,326) is
denoted by n(n-1)/2. While not a member of the NYCHA, the U.S. Treasury also
participated in the daily exchanges of the NYCHA.
126. To the best of my knowledge, daily exchange slips are not available at the
NYCHA archive at Columbia University. Rare Book & Manuscript Library Acquires
Archive of the Clearing House Association, COLUM. U. LIBRS. https://library.columbia
.edu/about/news/libraries/2014/rare-book---manuscript-library-acquires-archive-of-the-
clearing-.html [https://perma.cc/Q3XR-BVTA] (last visited Jan. 29, 2022).
individually.127 The key advantage of the clearinghouse system was that

on regular days that bilateral net was meaningless. The agreement to
engage in multilateral netting allowed NYCHA banks to setoff all credits
from all banks against amounts owed to all banks. Only the difference
between the two, the so called “daily balance,” would then have to be
settled in actual reserves. Figure 1 provides photographs of the NYCHA
building, the morning exchanges, and settlement in the “Cash Room.”
Figure 1: NYCHA Building on Cedar Street (left); The 10:00 a.m.
Exchanges in the Clearinghouse Gallery (top right); Settlement by 01:30
p.m. in the Cash Room128
127. To reduce the cost and complexity involved in such settlement, New York banks
prior to 1853 often resorted to a system of drafts banks drew on each other. See CANNON,
supra note 125, at 150 for discussion of that system and its disadvantages.
128. WILLIAM JAY GILPIN & HENRY E. WALLACE, CLEARING HOUSE OF NEW YORK
CITY: NEW YORK CLEARING HOUSE ASSOCIATION, 1854-1905 6, 13, 17 (Moses King ed.,
1904).
A look at 1905 data helps appreciate how substantial the

compression of the clearings to daily balances was. For 1905, the total
NYCHA bank deposits of $1,100 million produced a daily average of
$302 million (some 27 percent of deposits) in checks between customers,
representing the contents of the boxes handed by the delivery clerks. On
the highest days, this so-called “clearings” figure rose to nearly $700
million—63 percent of total deposits.129 And this is still an average across
the 52-member banks. Individual banks may experience gross amounts that
constitute an even larger percentage of their deposits. With multilateral
netting, this enormous clearing figure was reduced to an average total
daily balance of $13 million, just 1 percent of total NYCHA deposits, and
4.3 percent of average daily clearings.130 The highest recorded total daily
balance was $42 million, still only 3.5 percent of total deposits, and under
7 percent of the very highest clearing figure.131
At first sight, this data is consistent with the conventional view’s
assertion that netting reduces the need for settlement in actual reserves.
What is left out of the conventional view, however, is the mutual risk
exposure that clearing banks were subject to as part of the netting
process.132 The Constitution of the NYCHA133 required banks with negative
daily balances to settle their indebtedness by 1:30 p.m., and to settle it in
reserves (specie and legal tender notes) delivered to the clearinghouse
manager in a specially designated room.134 That Constitution also provided
procedures for the expulsion or suspension of members by a simple
majority vote.135 It is likely that such procedures were initiated against
members failing to settle by 1:30 p.m., thus making it practically impossible
129. See CANNON, supra note 124, at 217, 222 ($686 million figure for Jan. 3, 1906).
For total deposits of $1,100 million see supra note 119.
130. See CANNON, supra note 124, at 221.
131. Id. at 222.
132. To clarify, my argument is not that multilateral netting increases risk exposures
compared to a bilateral netting system as its baseline. Rather, my argument is that
multilateral netting, considered on its own, involves considerable risk that is not
sufficiently discussed and analyzed in the literature on bank money creation.
133. NYCHA CONSTITUTION, supra note 115 at § 12.
134. To avoid the inconvenience in movement and counting of specie and legal
tender, balances were often settled in clearinghouse certificates that were backed 100
percent by the above. See NYCHA CONSTITUTION, supra note 115, at § 17. These regular
100 percent backed certificates are not to be confused with clearinghouse loan certificates
discussed in Section III.C.1 below.
135. NYCHA CONSTITUTION, supra note 115 at §§ 20-21.
for them to continue business operations.136 Even more importantly for our
purposes are the immediate consequences of default to the other (non-
defaulting) members. Section 13 of the Constitution provided the basic rule
on default, in pertinent part:
Should any one of the Associated Banks fail to appear at the Clearing
House at the proper hour prepared to pay the balance against it, the
amount of that balance shall be immediately furnished to the Clearing
House by the several banks exchanging at that establishment with the
defaulting bank, in proportion to their respective balances against that
Bank, resulting from the exchange of the day . . . .137
The meaning of this highly technical rule is that a default would have
exposed banks owed a bilateral net balance from the defaulting bank to
loss of that balance. To take a stylized example, assume that on a given
day, Bank A defaulted on its $10 million daily balance. On most days, the
2,652 exchange slips produced by the clerks would be totaled and could
then be thrown away (after all, multilateral netting means that only grand
totals matter). But with a default occurring, these slips now determine
how the loss will be allocated. Each bank will calculate its bilateral net
towards the defaulting bank, reflecting the difference between the
exchange slip given to that bank, and the one received from it. Assume
that it was found that three of the banks—Bank B, Bank C and Bank D—
had an excess of checks delivered to the defaulting Bank A over checks
received from it, e.g., $1 million positive net balance to Bank B, $2
million to Bank C, and $7 million to Bank D. Each of these banks would
have to then “immediately furnish” these amounts to the clearinghouse.
It is important to appreciate the distinct ways in which this default
management procedure affects the capital and the liquidity position of the
surviving banks now subject to Section 13. Capital reflects the difference
between a bank’s assets (mainly loans) and liabilities (mainly deposits),
and is a safety cushion protecting creditors in the event of a loss to assets.
136. While it seems that banks occasionally defaulted in the clearinghouse, detailed
discussions of defaults are difficult to come by. See, e.g., CANNON, supra note 125, at
209. Future research on this matter will likely require work in the NYCHA Archive. See
supra note 126.
137. In a strange coincidence of legal history, the original section in the Federal
Reserve Act providing lender-of-last-resort authority was also numbered Section 13 of
the Federal Reserve Act. See discussion supra note 53. The original discount window
authority in § 13(2) has become antiquated, and since the New Deal the two main
authorities are found in §§ 10B and 13(3).
When Section 13 is triggered, banks like B, C, and D have to pay the

balance immediately (a new liability), but they can try to recover that
amount against the defaulting bank outside the clearinghouse, e.g., by
filing an insolvency claim (a new asset). Whether or not the surviving
banks will suffer a capital loss ultimately depends on the success of those
collection efforts, which might play out over a significant period of time.
Far more immediate are the liquidity implications of the default. This
means bringing these amounts in actual reserves to the clearinghouse
manager, just as if the amount had to be settled as a regular negative daily
balance. If that bank happens to be flush with reserves on that day, that
might not be a problem. But if that bank was already in the position of
having to settle a negative daily balance, it might simply be unable to
settle the new and aggravated balance. If so, that bank too would have to
default. And once it defaults, Section 13 would be activated yet again,
leading to a new group of banks that have to “immediately furnish” their
net bilateral balance. In modern jargon, this is the risk of a “cascading
default.”138
The default management procedure laid out in Section 13 shows us
that while multilateral netting shrinks the obligations between banks,
bilateral risk exposures are not eliminated entirely, and resurface in the
event of default.139 Defaults not only impose losses on individual
surviving banks, but also risk a cascading default that threatens
clearinghouse members as a whole. The threat of a cascading default in
multilateral netting systems like CHIPS emerged as a growing source of
anxiety in central banking circles since the rise of financial globalization
in the 1970s and 1980s.140 While that story is beyond the scope of the
current Article, the anxiety felt by central bankers in the late twentieth
century informs our understanding of the concern such cascades have
138. See generally Fariba Karimi & Matthias Raddant, Cascades in Real Interbank
Markets, 47 COMPUTATIONAL ECON. 49 (2016).
139. In modern practice, it is common for clearinghouses—especially derivatives
clearinghouses, or “CCPs”—to have guarantee funds that mutualize losses across
clearinghouse members in the event of default. See Richard Squire, Clearinghouses as
Liquidity Partitioning, 99 CORNELL L. REV. 857 (2014). For reasons that go beyond this
Article, I do not believe loss mutualization solves the risk exposure involved in
multilateral netting.
140. See, e.g., BANK FOR INT’L SETTLEMENTS, REPORT OF THE COMMITTEE ON
INTERBANK NETTING SCHEMES OF THE CENTRAL BANKS OF THE GROUP OF TEN
COUNTRIES 29 (1990), https://www.bis.org/cpmi/publ/d04.pdf [https://perma.cc/4U9Z-
YARQ]; BANK FOR INT’L SETTLEMENTS, REPORT ON NETTING SCHEMES (1989),
https://www.bis.org/cpmi/publ/d02.pdf [https://perma.cc/3387-UDKE].
posed at the turn of that century.141 Modern work in economics

demonstrates the enormous risk that cascading defaults presents.142 These
concerns were so significant they led central banks around the world to
reform their payments infrastructure over the ensuing decades.143 In many
ways, the system we inhabit today is a creature of those reforms.
The turn away from netting systems underscores that cascading defaults
are an existential risk to clearinghouse members, and to the broader financial
system. The understanding that netting is not freely available—that it comes
with considerable fragility—is a good starting point for a theory of bank
money creation. It focuses our attention on the drains that can lead to the
default, and on the actual means banks have to meet those drains. At an even
deeper level, the threat of a cascading default starts to build the intuition that
clearing banks cannot approach each other as profit maximizing individuals
in arms-length transactions. They have to consider the way that their actions
141. It is worth noting that while default procedures in the NYCHA and 1970s CHIPS
were similar, they were not identical. CHIPS’ so called “unwind” rule provided for
complete removal of the defaulter’s positions from the day’s clearings, followed by a
recalculation of all positions. See Robert T. Clair, The Clearing House Interbank
Payments System: A Description of its Operation and Risk Management, FED. RSRV.
BANK DALLAS’ HIST. LIBR. 130 (1989). By reducing the default amount to the multilateral
net, the historical NYCHA rule seems to slow the reach of a cascade compared to the
CHIPS unwind rule. While interesting to pursue in future work, a more detailed
comparison between the two rules (and potential reasons for the difference) is beyond the
scope of this Article.
142. See David B. Humphrey, Payments Finality and Risk of Settlement Failure, in
TECHNOLOGY AND REGULATION OF FINANCIAL MARKETS: SECURITIES, FUTURES, AND
BANKING 97 (Anthony Saunders & Lawrence J. White eds., 1986). Later studies have
arrived at different conclusions. See, e.g., P. Angelini, G. Maresca & D. Russo, Systemic
Risk in the Netting System, 20 J. BANKING & FIN. 853 (1996); Morten L. Bech & Kimmo
Soramäki, Systemic Risk in a Netting System Revisited (draft, 2005) https://
www.suomenpankki.fi/globalassets/en/financial-stability/payment-and-settelement-
system-simulator/events/sempaper04-bech-sormaki.pdf [https://perma.cc/5JJR-JRQL].
All of these studies (including the Humphrey study) focus on a bank’s capital position
rather than their liquidity position. This focus, in my view, leads to a considerable
underestimation of the risk of cascades.
143. One strand of these reforms included a shift away from traditional netting systems
like the NYCHA to RTGS operated by the central bank. See supra Section I.A. The situation
in the United States was different, given that Fedwire has been in operation for decades
prior to the global shift in perspective. A second strand of these reforms made any remaining
netting systems (like CHIPS) safer. This was achieved through fundamental changes to their
legal architecture and an increase of their reliance on reserves. See supra note 83 and
accompanying text.
affect the possibility of another bank’s default. The next Sections of the case
study further continue to explore these themes.
B. CLEARING DRAINS DID OCCUR
This Section presents evidence that clearing drains did in fact occur in
the NYCHA. In so doing, it calls into question the conventional view’s
assumption that the law of large numbers prevents drains in ways that make
clearing seem easier than it is. Drains exposed banks to the risk of default,
and as discussed above, the default of any one member bank could pose a
risk to others through the liability imposed by Section 13. The
underestimation of drain risk results in an underestimation of the challenges
involved in clearing, and the governance designed to address them. This
Section studies two distinct types of drains to which banks were subject:
daily and weekly. A daily drain is as an abnormally large negative daily
balance at the clearinghouse, while a weekly drain reflects the
accumulation of smaller but persistent negative daily balances throughout
the clearing week. Daily drains can be studied empirically through the
NYCHA daily ledger. Weekly drains are far more challenging to study
empirically, but as discussed below, we can gain much insight into their
occurrence through a simple theoretical model.
Before immersing in the study of drains, it is important to gain an
intuitive understanding as to why the conventional view underestimates
their significance. The problem is not with the LLN per se, but with the
specific way in which the LLN is applied.144 As we saw above, multilateral
netting was able to reduce the $302 million average in clearings to an
average total daily balance of $13 million, just 1 percent of total NYCHA
deposits. At first sight, that might sound close enough to the notion that the
LLN eliminates drains. But on closer look, such figures are highly limited
by virtue of being averages: an average across days, and an average across
banks. Averages mean little in practice, because each individual bank must
stand ready to meet its own individual balance, and it must stand ready to
meet it on every single clearing day. Looking at the LLN at the level of the
single bank on a single day is fundamentally different from looking at the
LLN across banks and across time. At the level of the single day, it is
indeed correct that the LLN predicts small negative balances. Our stylized
bank in a two-bank system (each with $500 in deposits) is expected to
144. My assumption here is that payment flows by depositors are i.i.d., and generally
do follow the LLN. An additional line of inquiry could examine that assumption. See
discussion supra note 90.
lose under 4 percent of deposits 9 out of 10 days, and its probability of

losing 10 percent or more of deposits on a given day is 1 in 1,000.145 But
that risk dramatically grows once we consider that multiple banks need to
clear on multiple days. The NYCHA had 52 banks that each cleared about
300 days a year. With 15,600 observations, the expected number of 1 in
1,000 drains (i.e., drains that amount to 10 percent of a bank's deposits) is
15.6. That is more than once monthly, i.e., not very rare at all.146 With the
intuition in place, we turn to look at the drains more closely.
1. Daily Drains
Assessment of real-life drains requires disaggregated data for

specific banks on specific days. Today, such data is extremely difficult to
find due to its proprietary nature. Even if found, it would be difficult to
interpret given the institutional complexity of the modern financial
system.147 In contrast, the ledger of the New York Clearing House provides
exactly such disaggregated data and to the best of my knowledge, has not
been previously studied with this purpose in mind. The ledger is
challenging to work with. Its handwriting and sheer physical size make it
difficult to digitize, and the nearly 16,000 entries per year make it difficult
to study without digitization. Nevertheless, with the help of some
heuristics, we can look for evidence of some drains, knowing the imperfect
nature of the search could easily miss out on others. To be sure, much future
145. See supra Section I.I.B; see also Maciej Kowalski, Coin Flip Probability
Calculator, OMNI CALCULATOR (Aug. 17, 2021), https://www.omnicalculator.com/
statistics/coin-flip-probability [https://perma.cc/BF6M-QV6P] [hereinafter OMNI
CALCULATOR]. “Number of flips” set to 1,000. First calculation: “I want to have at least”
set to 480 heads; chances of success are 90.3 percent. Second calculation: “I want to have
at most” set to 450 heads; chances of success are 0.0865 percent.
146. 15.6 = 0.001*300 clearing days *52 banks. Similarly, the probability of any one
of the 52 banks experiencing a 1-in-1,000 drain on a given day is about 5 percent (0.05
= 1- (1-0.001)52); The probability of any one of the 52 banks experiencing a 1-in-1,000
drain on a given week is about 26 percent (0.26 = 1 - (1 - 0.05)6); The probability of any
one of the 52 banks experiencing a 1-in-1,000 drain over the course of a year (52 weeks)
is greater than 99.9 percent (0.999999 = 1 - (1-0.26)52 weeks) essentially certain.
147. Modern studies, like those mentioned in note 140, supra, have used datasets that
are conceptually similar to the ledger (often through central bank affiliation). The goal of
these studies was different, namely, assessing the risk of cascades as opposed to the
occurrence of drains per se. Using these datasets to study drains would require extensive
controls for the wide array of modern liquidity management tools that obscure the
underlying flow of depositor payments.
work remains to be done with the ledger, including through more refined
research methods and controls (see Figure 2 for an image of a page from
the ledger).
Figure 2: Page from the NYCHA Ledger148
My study into daily drains in the ledger involved several steps. The
first step was to define the size of the negative daily balance that would
qualify as a drain. The size chosen for these purposes was a daily balance
of about 10 percent or more149 of a bank’s deposits (deposit data is drawn
148. Columbia University, Rare Book & Manuscript Library Acquires Archive of the
Clearing House Association, COLUM. U. LIBRS.
149. The heuristic used in the initial phase of the search was to normalize drains by
deposits and circulating notes on the first week of the quarter. Once an entry has been
identified, Tables 2-4 in the Appendix notes the level of deposits and circulating notes on
the week prior to the drain (W-1). This two-stage process led to a few entries where the
proportion is slightly lower than 10 percent, but these entries do not change the overall
results.
from NYCHA weekly statements).150 While the percentage is somewhat

arbitrary, this threshold was chosen for two reasons. First, as noted above,
losing 10 percent or more of deposits on a single day should be a small,
one-in-a-thousand probability.151 It’s precisely the kind of small probability
event that the conventional view abstracts from. Second, NYCHA banks
held reserves of slightly over 25 percent of their deposits, which they were
required to maintain under the National Banking Act.152 A daily drain
greater or equal to 10 percent of deposits means losing at least 40 percent—
nearly one half—of the required reserve amount. This is a serious liquidity
event, about one additional drain from default. The occurrence of such daily
drains would go a long way in explaining the need for NYCHA banks’
relatively high reserve ratios. Importantly, it will do so in ways that are
based on the obligations banks routinely incur to each other, rather than
demand for cash from depositors, or the binding legal nature of the reserve
requirements.153 Daily drains of this magnitude also highlight banks’ need
to promptly replenish their reserve to pre-drain levels, an important
question to be addressed in the next Section.
The second step was to search 1905 ledger entries for these drains.
As noted above, the search had to be incomplete due to the use of
heuristics, and the non-digitized format. A notable limitation of the study
in this respect is the focus on medium-sized and larger banks, to the
exclusion of the smaller banks that represent about 25 percent in total
deposit volume, and about 30 out of 52 of the total banks in number. This
is a conservative decision that understates the true number of drains in the
ledger. Searched in this fashion, the ledger produced a total of 45
observations for daily drains. These observations are summarized in
Table 1 in the Appendix.
150. See supra note 120 and accompanying text. Note that deposits reported in those
statements reflect weekly average rather than a snapshot of a specific day.
151. See supra note 145 and accompanying text.
152. See National Banking Act of 1864, 12 U.S.C. § 38; Sess. 1 ch. 106 §§ 6, 7; 13
Stat. 99 38th Congress, 1st Session, ch. 106, § 31, 13 Stat. 99 (codified as amended in
scattered sections of 12 U.S.C.). Reserve requirements applied to both deposits and
national bank notes issued under N.B.A. § 21 (these notes are not to be confused with
legal tender notes discussed in note 168, infra). For completeness, the 10 percent
threshold discussed above includes these notes in the denominator. For simplicity, this
Section abstracts from discussion of reserve requirements under New York state law,
because the share of deposits by state banks in NYCHA was small (around 10 percent).
153. While runs decrease the total reserve available to banks, clearinghouse balances
reallocate a given total reserve across different banks.
Importantly, however, not all 45 observations—and not even the

majority of them—represent true daily drains. This is where the third and
final step comes in. The main analytical challenge of working with the
ledger is the realization that daily balances do not represent only the
underlying flows between deposits, but can also reflect proactive attempts
by clearing banks to affect their daily liquidity. These proactive attempts,
and the system of governance necessary to sustain them, will be discussed
in detail below.154 For now, what is important to understand is that drains
resulting from this kind of proactive action are not a threatening shock to
the bank (like a true drain), but intentional attempts to reduce
overabundant reserves. To make the results conservative, and prevent
overestimation, these entries had to be controlled for. The control stage
of the study included classifying the 45 entries into three groups:
Table 2: Results of Search in the Ledger
# of
Significance of finding
observations
High level of confidence that the
Group 1 7
drain is a true drain155
Indication of a true drain, but lesser
Group 2 6
level of confidence156
154. See infra, Section III.C.2.

155. “High level of confidence” reflects instances where (a) there was no positive balance
on the day preceding the drain (T-1) that exceeded about 50 percent of the drain amount; (b)
at least 70 percent of the drain was recouped in the day following the drain (T+1; in one of the
seven cases, T+2 was used as well); (c) T and T+1 both fall within the same week, and the
same month, thereby virtually eliminating the possibility that the drain was motivated by
reserve cycle management concerns. See infra, Section III.C.1.
156. “Indication of a true drain, but lesser level of confidence” reflects instances
where: Group 1, Condition (a) was met; (b) a considerable portion (75 percent or more)
of the drain was recouped within a period of up to 6 days or shorter from the drain (T+6);
(c) there is no strong indication that the negative drain on T was incurred to offset positive
balances in the preceding days (T-3; the highest sum of T-1, T-2, and T-3 for Group 2 is
about 54 percent of the negative balance on T); ; (d.) because recoupment periods are
longer, and sometimes straddle two different clearing weeks, or go across months, it is
possible (though not necessary) that reserve cycle maintenance has played a role. Also
note that end-of-month periods may also involve heightened clearing volume due to
heightened payments activity (e.g., payment of wages).
The drain is unlikely to be a true

Group 3 32
drain157
Total 45
Classified this way, Group 1 (high level of confidence) includes

seven findings, from five distinct banks: four medium-sized banks, and
one large bank. Taken together, these banks account for $310 million in
deposits, some 27 percent of total NYCHA deposits.158 In their relative
size, the drains vary from 36 percent to 55 percent of the relevant bank’s
required reserves on the week prior to the drain (average of 45 percent).159
These Group 1 findings go a long way in corroborating the intuitive
argument above, that even if the probability of an individual drain is
small, a very large number of clearing events would produce a meaningful
number of these drains (in that highly stylized model, 15.6 events).160 In
addition to these seven Group 1 observations, Group 2 includes six
additional observations with indications of a true drain, while at the same
time acknowledging a lesser level of confidence given their more
complex nature. These observations include four additional banks, not
included in Group 1, and represent an additional $122 million in deposits
over the Group 1 banks.161 The size of these drains varies from 38 percent
of required reserves to a staggering 80 percent (average of 58 percent).
The 13 entries from Groups 1 and 2 are summarized in Tables 3-4 in the
Appendix. These findings suggest a substantial share of NYCHA banks,
by number, and deposit size, experienced a drain over the period of only
one year. They stand in stark contrast to the image conveyed by the annual
averages across the year and across banks where the daily balance is only
4 percent of required reserves (1 percent of total deposits). By looking at
157. Group 3 is a residual category for observations that did not meet the Group 1 or
Group 2 conditions. For a large subset of this group, there is an indication that the negative
balance was not a true drain, with the T-1 balance being relatively high, and the T+1 balance
being relatively low. Readers may detect these entries by looking through Table 1 in the
Appendix.
158. See Appendix, Table 3.
159. For methodological aspects, see discussion supra notes 155, 156, and 157.
160. To be clear, the model is so stylized that the empirical findings are not meant as
any kind of corroboration. This Section merely indicates the relative similarity of the
results.
161. See Appendix, Table 3. One of the banks, Bank #1 (Bank of New York), appears
twice in Group 1, and once in Group 2. One of the reasons for this might be that, given
its place on the first line of the ledger, the entries more easily jump out.
individual ledger entries we see the average underestimates the stress to

which a specific bank may be subject by a factor of 10 or more.
2. Weekly Drains
Where daily drains represent an abnormally large negative balance on

a given day, weekly drains are the accumulation of several smaller negative
balances over the week. Unlike daily drains, weekly drains are not directly
observable. To find a weekly drain, one cannot simply look at a given
bank’s daily balances over the course of a week, and sum-up the result. One
reason is practical—the difficulties of digitizing the ledger make such study
extremely labor intensive. More fundamentally, even if the data were
conveniently available, the results would likely not be reflective of the
underlying flows of deposits between bank customers. The reason is that—
as discussed below162—NYCHA banks developed a system to proactively
influence their daily balances across the week through call loans to brokers.
A bank facing a negative daily balance could improve its next day balance
by calling back loans, and narrative accounts suggests that such practice
was routine. To this extent, a simple summation of balances across the week
will bring results far closer to zero than to the “real” weekly drain, leading
to a serious underestimation of the risk they posed.163
While weekly drains are difficult to observe directly, we can use a
highly stylized model to understand their basic logic. If we set aside
proactive actions by banks, a weekly drain could be modeled using a simple
binomial distribution. Each day of the week, a bank goes to the
clearinghouse where, with equal probability, it may “win” by having a +$x
positive daily balance, or “lose” by having -$x negative daily balance. We
may define a draining week as one of six consecutive losses, leading to a
total loss of $6x in reserves. While somewhat arbitrary, this six-day period
stands for the proposition that in the very short-run a bank has very little
ability to improve its balances by reducing loans to clients or selling assets
at acceptable prices.164 In this basic model, the probability of a bank facing
162. See infra Section III.C.3.

163. Call loan data is not available at the individual bank level, so the original clearing
position cannot be reverse engineered.
164. This assumption is common in modeling demand for reserves. See, e.g., William
Poole, Commercial Bank Reserve Management in a Stochastic Model: Implications for
Monetary Policy, 23 J. FIN. 769, 770 (1968):
In the very short-run, the banks reserve manager takes as given the
fluctuations in deposits and reserves caused by the clearing of checks
a draining week on any given week is 1-in-64 (0.56) or 0.015 (see

Appendix, Figure 3). The next question is how severe would a $6x drain
be to a bank, that is, how does it scale to the size of its deposits and reserve.
A point of departure for estimating x would be to look at average negative
balances at the clearinghouse as a whole. As noted above, the 1905 average
was $13 million. To keep things simple, assume (counterfactually) that all
clearinghouse members had the identical average deposit size of $23
million (=$1,200/52). Each day, 26 members receive a negative balance (x)
of $0.5 million (=$13 million/26) or a positive balance of the same. With x
as $0.5 million, the weekly drain is $3 million, which is 13 percent of the
average bank’s deposits and about half of the required reserves. In other
words, the weekly drain has about the same impact noted as the daily drain
discussed above.
This model, and the parameter used are admittedly very stylized. On
the one hand, the probability for a weekly drain is likely to be lower than
1-in-64. This is because the daily balance is a continuous rather than
discrete variable.165 On the other hand, the potential size of a weekly drain
is likely to be somewhat greater. This is because the variance in daily
balances to which individual banks are exposed is greater than the
clearinghouse-wide annual average used as the parameter for x.166 While
the development of a more nuanced model is beyond our current scope, it
makes the basic logic of weekly drains intuitive to grasp. As discussed
below, an important historical indication of the existence and magnitude of
these weekly drains was the call loans held by NYCHA banks, a total of
and the activities of the bank’s loan department. The reserve

manager’s job is to adjust to these fluctuations. The model presented
here concentrates on these very short-run adjustments. However, it is
obvious that the bank must make further adjustments if it experiences
persistent reserve drains or accretion.
165. Payment flows between depositors are not a discrete, but a continuous variable,
and likely one following a normal distribution. On most days, the daily balance would
fall relatively close to the mean (zero); occasionally, the balance would be substantial,
like the x’s that make up a weekly drain; very rarely, it would be very large, like the daily
drain. As a result, a more nuanced model would show that the probability of a weekly
drain is lower than 1-in-64 in the binomial distribution. Another consideration is that a
continuous model would need to account for weekly drains resulting from a combination
of daily balances of various sizes (some even positive). All things equal, such
combination would have the effect of increasing the probability of a weekly drain, though
it is difficult to estimate by how much.
166. This is a result of the law of large numbers. See supra Section I.C.2.
about $400 million. With its $3 million weekly drain across 52 banks, our
stylized model can account for $156 million, a substantial portion of that
amount.
C. INTERBANK MARKETS WERE ABSENT
So far, our discussion has shown that default at the clearinghouse

posed a risk to other members, and that clearing drains were large,
highlighting the possibility of default. These findings challenge the
conventional view, but they would be of little significance if drains could
be handled through simple trading in interbank markets. The history
declines to put that assumption into effect. On the one hand, there is no
evidence of an interbank market between NYCHA member banks. On the
other hand, there is strong evidence these banks relied on two alternative
tools to handle clearing drains: reserves and call loans. These tools
followed a strict division of labor. Large reserves of state money were
used to avoid immediate default by allowing settlement of daily drains.
The system of call loans to brokers was used to recover reserves following
a daily drain, as well as to avoid weekly drains. We take these issues in
turn.
1. The Use of Reserves
As we have seen, the Constitution of the NYCHA required that daily

balances be settled in actual reserves. Reserves consisted of two forms of
publicly issued money: specie and legal tender notes. Specie referred to
gold coin minted by the U.S. Mint, a bureau within the Treasury
Department.167 Legal tender notes, also known as “greenbacks,” referred
to paper money issued directly by the Treasury. 168 Despite their apparent
differences, both types of public money enjoyed formal legal tender status
under the law. That is, they were both legally defined to redeem debt at
its nominal value, the same legal attribute Federal Reserve notes and
reserves (account balances with the Fed) enjoy today.169 As a private
167. See Coinage Act of 1873, ch. 131, 17 Stat. 424.

168. See Legal Tender Act of 1862, 12 Stat. 345 (1862); discussion supra note 137.
Legal tender notes are not to be confused with a second type of paper money, national
bank notes issued under the National Bank Act of 1864. National bank notes did not enjoy
legal tender status under the law, did not qualify for reserve requirements, and were not
used in settlement of clearinghouse balances.
169. See, e.g., 17 Stat. 424 §§ 14-15; 12 Stat. 345 § 1; Knox v. Lee, 79 U.S. 457
(1871). Since 1879 legal tender notes have been convertible into coin on demand at the
association, the NYCHA Constitution could have theoretically opted for

a broader definition of money to settle daily balances in. It did not, opting
instead for the very narrowest definition of money under the law.
The ongoing use of reserves in settlement marks a fundamental
reliance of clearing on publicly supplied money. This role has not been
sufficiently appreciated in the financial history literature for two reasons.
The first reason concerns the literature’s intense focus on the exceptional
use of an instrument known as “clearing house loan certificates.”170 The
issuance of clearinghouse loan certificates was a measure NYCHA banks
took during financial panics. In these periods, the NYCHA banks
experienced large withdrawals of reserves from their depositors, first and
foremost, the out-of-town that they served as correspondents who were
experiencing liquidity pressures themselves due to seasonal strains.171 To
economize on their reserves, the NYCHA banks issued the loan
certificates that could be formally used in settlement of the daily balance
in lieu of reserves. Banks short of reserves could procure these certificates
by providing a special clearinghouse committee with collateral that would
back the loan certificates. The loan certificates were jointly guaranteed by
all NYCHA banks, so the banks that held them were not exposed to credit
risk from any individual bank. Financial historians have correctly
interpreted the issuance of loan certificates as an early form of lending-
of-last-resort, preceding the creation of the Federal Reserve. But while
financial historians have studied loan certificates in great detail, they have
spent little time studying the use of actual reserves during ordinary times.
The result is an unbalanced account that understates the difficulties of
clearing in ordinary times, and the way that reserves—not loan
certificates—allowed banks to meet those difficulties day-in and day-out.
The second reason for underestimation of the role of reserves
concerns the significance of regulatory reserve requirements. There is a
prevalent conception that reserves are held primarily to meet these
regulatory requirements. In this view, reserves are a relatively idle stock.
Their role is to allow banks to redeem deposits in the rare occasions
demand for cash rises, so a run can be avoided.172 For the NYCHA, these
Treasury, but their legal tender status clearly did not depend on redemption. See Specie
Payment Resumption Act of 1875, ch. 15, 18 Stat. 296 (1875).
170. See, e.g., GORTON, supra note 118; Awrey, supra note 9.
171. See Orian Peer, supra note 9, at 380.
172. See, e.g., BARR ET AL., supra note 75, at 238; Joshua N. Feinman, Reserve
Requirements: History, Current Practice, and Potential Reform, 1993 FED. RSRV. BULL.
depositors were, again, the out-of-town banks who subjected the NYCHA
to seasonal strains around fall. The role of the seasonal cycle is certainly
important to understanding crisis dynamics under the National Banking
Era, but here again, it results in an unbalanced picture on the role of
reserves. Reserve requirements were not only about run mitigation, but
about meeting the far more common clearing drains to which NYCHA
banks were subject. Regulatory reserve requirements actually promoted
this role through the particular way in which they were calculated. 173
NYCHA banks’ regulatory reserve requirements were not binding at
every moment, but reported as an average across time. The standard
regulatory practice was for averaging of reserve requirements over 30
days.174 The NYCHA Constitution followed an even tighter schedule by
publishing a statement of members’ average weekly reserves every
Saturday.175 This time signature explains how NYCHA banks could use
reserves to meet clearing drains.
Assume a bank began its clearing week with reserves of 25 percent
of deposits. On Tuesday that bank suffers a large drain of 15 percent of
deposits, i.e., 60 percent of its reserves. The reserve was large enough to
absorb that drain, and averaging across the week meant the bank could
build back its reserve position in time for the Saturday statement. A 15
percent decline in reserves on a single day would reduce the weekly
average by only 2.5 percent. A bank could recoup this drop by inducing
569, 569 (1993). As Feinman notes, with the creation of the central bank, the significance
attributed to reserve requirements has shifted from run prevention (that could now be
cured by central bank lending-of-last-resort) to monetary policy. The role reserve
requirements played in monetary policy has been the subject of some controversy. For
discussion, see Scott T. Fullwiler, The Social Fabric Matrix Approach to Central Bank
Operations – An Application to the Federal Reserve and the Recent Financial Crisis
(2009), https://ssrn.com/abstract=1874795; Ihrig et al., supra note 109. In contrast to
earlier views on a “reserve multiplier,” the accepted view today is that the only
significance of the quantity of reserves is in helping the Fed implement its target policy
rate. That role has changed in the aftermath of the GFC—see discussion supra note 83–
and the transition to abundant reserves. Since March 26, 2020, reserve requirements have
been set at zero, making them ineffective. See Federal Reserve Actions to Support the
Flow of Credit to Households and Businesses, BD. OF GOVERNORS OF THE FED. RSRV.
SYS., (Mar. 15, 2020).
173. My analysis here builds on Fullwiler, supra note 174, on the importance of
reserve maintenance periods in modern liquidity management.
174. See, e.g., GEORGE MATHEWES COFFIN, HAND-BOOK FOR BANK OFFICERS 22
(McGill & Wallace eds., 1896) (“In each report of condition, a [national] bank is required
to state its average reserves on deposits for the preceding 30 days.”).
175. NYCHA CONSTITUTION, supra note 115, at § 16.
positive daily balances on the day or days subsequent to the drain.176 The
mechanism to induce the adjustment of balances in this way was the use
of call loans, which we move to discuss in the next Section. As far as
reserves are concerned, the key is to appreciate their role as an immediate
shock absorber against daily drains. This shock absorber quality depends
on their legal tender status. A bank pressed to settle a large balance within
a couple of hours does not need to trade or contract in order to avoid
default. It simply transfers the one asset under the law that would
unconditionally redeem its debt.
2. The Use of Call Loans
Reserves enabled NYCHA banks to settle a daily drain, but to

survive, a bank must ensure its reserve position over time. This includes
replenishing the reserve following a daily drain, and avoiding the
accumulation of medium-sized negative balances into a weekly drain.
Here again, there is no evidence to suggest NYCHA banks resorted to
interbank markets, but there is ample evidence of the system of
“secondary reserves” they developed. That system was designed around
call loans that the NYCHA banks made to brokers on the New York Stock
Exchange (NYSE) at the corner of Nassau Street, just a few moments’
walk from the clearinghouse building. Banks never used call loans to
borrow, not from each other, and not from the brokers. The banks were
always lenders to the brokers, and they devised an elegant way to shift the
loans among themselves.
In its basic terms, the historical call loan resembles the modern-day
repurchase agreement or “repo.”177 The call loan was an overnight loan,
secured by collateral (stocks and bonds) actively trading on the stock
exchange.178 The NYSE brokers used these loans as a cheap and abundant
source of funding to purchase securities for their own account, and
176. In the empirical study (III.B.) the difference between these two possibilities
(recoupment the subsequent day, and recoupment over several days) would largely map
on to the difference between Group 1 and 2.
177. For a discussion of the repo market, see Perry Mehrling, supra note 101.
178. Like modern repo, call loans had a “haircut” being a small excess in the value of
collateral over the amount of the loan. Also like modern repo, call loans had a market-to-
market provision that required borrowers to ensure a sufficient amount of collateral in
the face of price movements. For discussion, see Nadav Orian Peer, A Constitutional
Approach to Shadow Banking: The Early Shadow System, Ch. 2 (2016) (unpublished
S.J.D. dissertation, Harvard Law School) (on file with author).
especially, for their brokerage customers.179 At about $400 million, these

loans constituted over one third of NYCHA banks’ total loans, a
staggering proportion.180
Contemporary accounts by Wall Street figures highlight the role of
the call loan market as the grounds for the daily recycling of reserves
among the banks. In 1908, Thomas Woodlock, member of the NYSE and
editor of the Wall Street Journal remarked:
At about 11 o’clock the banks in New York City know more or less
what their balances are as a result of that day’s clearings, and by that
time they have called such loans as they need to call in order to meet
their requirements if they have such requirements.181
A congressional testimony by J.G. Cannon (Fourth National Bank) a

few years later clearly states the daily logic of this recycling. “Deposits
decrease, and you call in your money. It is the leeway, you might say, of
the banks.”182 The same principle held for the surplus banks:
Mr. Untermeyer[:] On the other hand, when you have a surplus

outside183 your credit balance in the clearing house, you lend it in the
Street from day to day, do you not?
Mr. Cannon[:] We lend it in the Street from day to day; yes, sir.184
179. In this respect, the NYSE brokers were like modern day broker-dealers who
borrow in the repo market to lend to the hedge funds they serve as prime brokers. See id.
180. See MYERS, supra note 114, at 270 (call loans were one third of NYCHA banks’
total loans), 272 (graph showing NYCHA’s banks’ call loans at about $400 million for
1905); NYCHA Weekly Statements, supra note 120(total NYCHA deposits at a range of
$1,000-$1,200 million during 1905).
181. Thomas F. Woodlock, The Stock Exchange and the Money Market, in THE
CURRENCY PROBLEM AND THE PRESENT FINANCIAL SITUATION: A SERIES OF ADDRESSES
DELIVERED AT COLUMBIA UNIVERSITY 1907–1908 19, at 31 (Edwin R.A. Seligman ed.,
1908).
182. See WASH. GOV’T PRINTING OFF., MONEY TRUST INVESTIGATION:
INVESTIGATION OF FINANCIAL AND MONETARY CONDITIONS IN THE UNITED STATES
UNDER HOUSE RESOLUTIONS NOS. 429 AND 504 BEFORE A SUBCOMMITTEE OF THE
COMMITTEE ON BANKING AND CURRENCY, HOUSE OF REPRESENTATIVES, 62D CONG.
1865–1974 348-49 (1913), https://fraser.stlouisfed.org/title/money-trust-investigation-
80/part-5-23662 [https://perma.cc/Q3HM-NX4L].
183. The word “outside” seems to be an error in the transcription. From the context,
it is clear that Untermeyer is referring to a bank’s clearinghouse balance.
184. See Investigation of Financial and Monetary Conditions in the United States:
Hearing on H.R. 429 and H.R. 504 Before a Subcomm. of the H. Comm. on Banking and
By mapping such narrative accounts to the mechanics of the morning

exchanges we can draw a detailed picture of the role that call loans played.
Call loans could never help an NYCHA bank settle today’s balance. That
role was strictly fulfilled by reserves. What call loans could do was to
proactively improve tomorrow’s daily balance in light of today’s loss.
Here is how it worked (see Appendix, Balance Sheet 5; Figure 4).
Every morning, after the daily balance at the clearinghouse became
known, banks could use their call loans to lean against the wind and
essentially offset the day’s loss or gain in the next morning’s exchanges.
Banks facing a negative daily balance—”deficit banks”—could improve
the next day’s balance by “calling in” loans from brokers. This meant
demanding the brokers repay their loan by check drawn on a New York
bank before the end of the day. Consider the case where a deficit bank
demanded payment of $x in call loans from the brokers. The next day at
the clearinghouse, that bank would include a check for $x in the packages
its delivering clerk handed over, thereby increasing its total credits, and
its daily balance by that amount. Meanwhile, banks learning of a positive
daily balance—"surplus banks”—would do the opposite. Consider the
case where a surplus bank incurred a $x million positive balance. That
bank would instruct its representatives on the NYSE to increase its
lending to the brokers by $x.185 This means the surplus banks would cause
the brokers’ bank accounts to be credited with newly created balances for
that amount. The brokers would use these newly lent balances to write
checks to repay the loans called back by the deficit banks. The next
morning at the clearinghouse, the checks would be handed by the deficit
bank’s delivery clerk to the surplus bank’s settling clerk. This would
increase the surplus bank’s due to’s and reduce its daily balance by the
same amount. This carefully coordinated system achieved something
remarkable. It allowed the NYCHA banks to avoid weekly drains, and
recover from daily drains, without having to face each other in any kind
of interbank market.186
Used this way, call loans provided a powerful tool for liquidity
management. A bank facing a daily drain could replenish its reserve the
very next day by calling in loans for the same amount. Just as importantly,
Currency, 62d Cong. 348-49 (1913) [hereinafter Monetary Conditions Hearings]

(statement of James G. Cannon, President, Fourth National Bank).
185. See text accompanying supra note 184.
186. This made the kind of interactions described in Balance Sheet 4 in the Appendix
unnecessary.
through daily leaning against the wind, the NYCHA banks could have
eliminated much of the weekly variance in their reserve position. Loss of
deposits that would have otherwise caused a decrease in reserves could
be absorbed as a decrease in call loans instead. To see how, we can return
to our stylized model187 of the daily balance as a discrete variable +$x, or
-$x with equal probability. Imagine it is Monday 10:40 a.m., and Bank A
just learned of a negative daily balance of -$x at the clearinghouse. Bank
A knows that tomorrow’s daily balance would be determined by deposit
transfers that are, by definition, random. But by calling in a call loan for
$x, Bank A could proactively increase tomorrow’s daily balance by that
amount, no matter what the forces of chance bring. If Tuesday’s random
flows from deposit transfers produced another -$x, Bank A’s Tuesday
clearinghouse balance would not show -$x, but a flat zero (-$x from
random payment flows, +$x from yesterday’s call loans). Conversely, if
random flows brought +$x, Tuesday’s balance would show +$2x (being
+$x from random flows, and +$x from call loans). Following this way of
thinking it could be shown that the maximum change in a bank’s reserve
position over the week could be limited to the narrow bend of -$x/+$x,
rather than the far wider -$6x/+$6x bend it would have otherwise been
subject to (Figure 5).188 Used this way, call loans address the problem of
the weekly drain much the same way that reserves address the problem of
the daily drain.
The role that call loans played in liquidity management required
considerable discipline from surplus banks in daily (or close to daily)
recycling of their reserves back into the market. That is, it required a kind
of implicit day-to-day cooperation between the various banks who
happened to have surpluses and deficits on any given day. 189 This is the
case for two distinct reasons. The first reason has to do with the
relationship between funding liquidity and market liquidity that has been
at the center of post-GFC scholarship in finance.190 In a nutshell, funding
liquidity refers to the availability of debt finance to actors wishing to
purchase securities, while market liquidity refers to the ability to trade in
securities markets without causing rapid disruption in a security’s price.
187. See supra Section III.B.1.

188. See Appendix, Figure 5.
189. I am not necessarily suggesting that reserves were mechanistically recycled
every single day. On days when balances were relatively small, banks may have behaved
more flexibly. Nevertheless, the low amount of excess reserves and the weekly reporting
of average weekly reserves likely meant recycling operated on a near daily basis.
190. See Markus K. Brunnermeier & Lasse Heje Pedersen, Market Liquidity and
Funding Liquidity, 2008 REV. FIN. STUD. 22; MEHRLING, supra note 18.
For much of its history, the NYCHA banks were the main lenders in the
call loan market, i.e., the main providers of funding liquidity to the NYSE
brokers.191 If surplus banks had not recycled their reserves, the total
quantity of call loans would have gone down. The size of a single day’s
average clearing balance of $13 million was small in comparison to the
roughly $400 million in themarket.192 But that daily balance often ran in
the tens of millions, and even average daily balances, if accumulated over
a single week, would amount to $78 million, a substantial portion of total
call loan funding.
As highlighted by the modern literature, such a large drop of funding
liquidity could have had serious implications.193 Like disruptions in the
modern repo market during 2009, call loan funding enabled trading that
was central to day-to-day price formation and liquidity at the stock
exchange. The evaporation of funding liquidity could have led traders to
liquidate positions, putting considerable pressure on prices, and leading
to an evaporation of market liquidity. 194 In turn, dislocations in the stock
exchange would have had serious consequences for the NYCHA banks.
Recall that a third of these banks’ assets were secured by stock exchange
collateral. The call loan system was premised on the assumption that if a
broker happened to default on their loan, the lending NYCHA bank could
quickly sell the collateral on the NYSE and increase its daily balance the
very next day. If a drop in call loan funding led to dislocations in the stock
market, there would be no way for a bank to liquidate the collateral, so
the banks’ call loans would become frozen. Financial historians Jon Moen
and Ellis Tallman emphasized that the fear of such scenarios provided a
strong incentive for NYCHA to cooperate and “insure” the call loan
market during crisis time.195 While that is correct, it is equally important
191. For the rise in “outside lending” starting the 1890s, see generally discussion in
Jon Moen & Ellis Tallman, Outside Lending in the New York City Call Loan Market:
Evidence from the Panic of 1907, 26 FIN. HIST. REV. 43 (2019).
192. This is taking call loans by NYCHA banks. As noted by Moen and Tallman, the
overall size of the call loan market is difficult to estimate. Moen & Tallman, supra note
186.
193. See Brunnermeier & Pedersen, supra note 189.
194. For historical call loans as well as modern repo, a key piece of the transmission
between funding and market liquidity are market-to-market provisions mentioned in
supra note 178.
195. Moen & Tallman, supra note 191. Moen & Tallman argue that 1907 marked a
break with earlier tradition because by that time the presence of “outside lenders” (non-
to recognize that discipline in the recycling of reserves prevented these

same dynamics from occurring day-in and day-out.
The second reason why prompt recycling by surplus banks was
necessary is subtler and takes us back to the mechanics of clearing.
Assume that an NYSE broker had their loan called by a deficit bank and
was somehow able to secure an alternative loan from some outside lender,
not a NYCHA surplus bank. In this case, the outside loan allows the
broker’s funding to remain stable, so the risk is not that the securities
market would suffer a dislocation of the kind outlined above. Rather, the
problem is that the deposits the outside lender is lending to the broker
would not necessarily be drawn on an account with a surplus bank
(whether the outside lender’s bank happens to be a surplus or deficit bank
on a given day is entirely random). Now, if the funds happen to be
transferred from a bank that is already a deficit bank, that bank would
increase its due to’s at the clearinghouse the next morning, thereby
reducing its daily balance. This means that while one deficit bank (the one
that called the loan) was able to improve its position, another deficit
bank’s position (the outside lender’s bank) has deteriorated as a result.
The deficit as a whole is not being eliminated, but merely moved around.
For the call loan system to help bring reserves to equality, banks could
not have left things to chance, meaning surplus banks would have had to
cooperate.
Financial historians, like many contemporaries of the call loan
market, have criticized the panic prone and seemingly irrational reliance
of the NYCHA banks on call loans for their liquidity. 196 That view
highlights the Hobson’s choice that NYCHA banks faced when met with
redemptions of bankers’ balances by out-of-town banks. One option
individual NYCHA banks faced was calling in their loans from brokers,
but of course, the improvement in one bank’s reserve position came at the
expense of amother bank, so the overall reserve position remained largely
the same, while a shrinking quantity of call loans threw the stock
exchange into disarray.197 The second option (often following the failure
of the first) was to suspend payments. This had the effect of throwing the
banking apparatus into disarray and causing fractures in the interregional
NYCHA banks) in the call loan market was too large for NYCHA banks to handle
effectively. Id.
196. See id. at 55; Orian Peer, supra note 9, at 397-406.
197. For a more nuanced account on the mechanism, see CHARLES GOODHART, THE
NEW YORK MONEY MARKET AND THE FINANCE OF TRADE, 1900–1913 (1969).
payments system.198 The desire to avoid these recurring banking crises

was an important part of the impetus for the creation of the Federal
Reserve. Under its original design, the system of commercial paper
discounting by the Federal Reserve was in large part meant to replace the
reliance on call loans.199 Told from this perspective, the story of the call
loan market is one about the failure of call loans in preventing occasional
runs, highlighting the necessity for a public lender-of-last-resort. While
this story is true, it is also looking at the glass half-empty. Day-in and
day-out, the call loan market, and more specifically, NYCHA banks
cooperation in recycling reserves through the call loan market, provided
NYCHA with extraordinary liquidity. This system addressed a very real
challenge these banks were facing in their daily affairs. As with the case
of NYCHA banks’ dependence on reserves issued by the state, the call
loan market helps us appreciate bank money creation as a daily
governance project, not merely a decentralized private market subject to
occasional failure.
We can now summarize our discussion of NYCHA practice, and
assess on its significance. The first finding concerned the workings of
netting. In the conventional view, netting of obligations between banks
appears as a simple affair, a kind of “free lunch” that allows them to
engage in fractional–reserve banking. In contrast, the case study
demonstrates that netting involved a structural risk-exposure between the
banks. A default in the clearinghouse would trigger the liability of other
banks, and such defaults could potentially cascade through the
clearinghouse. The second finding concerned the occurrence of drains. In
the conventional view, the law of large numbers makes clearing drains
disappear, obviating the need to consider how banks meet those drains.
Meanwhile, the case study provides evidence that drains occurred with
considerable frequency: 7 to 15 times in 1905 alone, and likely more,
given the conservative nature of the study. This finding is not in tension
with the LLN, but actually underscores it. We can count on small–
probability events to occur given a large-enough number of trials. Daily
clearing among many banks generates that large number of trials. Finally,
the conventional view considers that drains, if they do occur, could be
managed through simple trading in interbank markets. For an interest rate,
a surplus bank would lend balances to a deficit bank. This would make
198. See Ellis Tallman & Jon Moen, The Transmission of Financial Crisis in 1907:
An Empirical Investigation, 12 CLIOMETRICA 277, 278 (2018).
199. See Orian Peer, supra note 9, at 406-12.
reserves unnecessary for interbank settlement, though they might still be

required to avoid bank runs, meeting regulatory requirements, or other
extraneous purposes. In practice, the case study demonstrated interbank
markets between NYCHA banks are nowhere to be found, while their
reliance on large amounts of reserves and carefully recycled call loans
was ubiquitous.
This last finding requires a brief explanation. Why were interbank
markets so conspicuously absent from NYCHA practice? While a
detailed response to this question lies beyond the scope of this Article, the
following provide the contours of such response.200 Interbank markets
were absent from NYCHA practice because in the absence of a central
bank, interbank markets would have been ridden with strategic behavior.
A deficit bank could easily find itself hostage to a competitor bank, whose
surplus position can give it an effective monopoly over the funds.201
Meanwhile, the default of a deficit bank exposes the clearinghouse to the
risk of a cascading default, and could itself generate a hostage dynamic
where surplus banks are pressed to lend or else risk the consequences of
a cascade. This double-bind—the monopoly power of surplus banks, and
the moral hazard of deficit banks—makes a very poor environment for an
interbank market to develop in. Meanwhile, the reliance on reserves on
call loans overcomes these problems by creating clear guidelines as to the
maximum size of deficits that banks can incur without requiring the
consent of their competitors. Outlining the extraordinary challenges that
interbank markets would have to face in the absence of a central bank
highlights that bank money creation cannot ultimately be understood as
originating from pairwise decisions between banks. Bank money creation
is inherently and fundamentally about the availability of a medium in
which obligations can be settled, and the provision of that medium is
unavoidably the stuff of governance.
CONCLUSION
Every day, trillions of dollars of payments between depositors flow

through the banking system. These payments become obligations
between banks and require banks to clear them within hours. My
argument here has been that clearing is not the simple and easy process it
is so often taken for. Rather, clearing presents serious risks and depends
200. For an extended version of this argument, see Orian Peer, supra note 178, at 102-
106.
201. See Henckel et al., supra note 104, at 19.
on governance for its daily completion. In the historical practice of the

NYCHA, that governance was provided through the state (as the issuer of
reserves used in settlement) and through cooperation among the NYCHA
banks (in carefully recycling these reserves). In modern times, we see
analogous functions provided by central banks, whether in operating the
RTGS systems, providing daylight liquidity, backstopping interbank
markets, or (since the GFC) issuing a large supply of reserves.
The historical case study teaches us these are not mere interventions
in an otherwise viable system of decentralized private action. Instead, the
central bank’s daily role in clearing addresses fundamental challenges
that have historically been addressed through other forms of governance.
To be sure, much remains to be studied and theorized about these
challenges. The history of the NYCHA provides the seeds from which a
more general theory can be developed: the threat of cascading defaults,
the high probability of drains, and the inadequacy of interbank markets,
likely due to strategic behavior among banks. While developing that
theory remains beyond the scope of this Article, the apparent consistency
between historical and modern practice is striking. The specific
institutions evolve over time; the challenges and the need for governance
remain the same.
How should this proposed understanding of clearing inform our
understanding of banking regulation? The methodological individualism
of the conventional approach—the notion that bank money creation is the
sum of individual decisions and preferences—limits banking regulation
to the role of market failure collection. The goal, in that view, is to
minimize intervention into otherwise efficient private arrangements. In
contrast, accounting for the governance that is integral to clearing casts
bank money creation in a more public light. The daily clearing
arrangements on which bank money creation depends arise not from
individual preferences, but from collective decisions. This calls into
question the presumption of market efficiency that is so central to the
conventional view, and its narrow vision of bank regulation.
What, then, would banking regulation become if instead of
methodological individualism, we started with the premise that bank
money creation is to an important degree a creature of governance?
Starting from such premise, it would be natural to think of how banking
is implicated, or can play a constructive role in, the pressing challenges
society is facing in financial stability, civil rights, climate policy, and
financial technology. And it would be natural to borrow concepts from
legal areas like utility regulation, public–private partnerships, or (in
political theory terms) social contract theory. In other words, such a view
of banking regulation would resemble the one gradually being developed
by the new legal literature on bank money creation. Reasonable people
will of course divide—at times, sharply—about a given proposal under
consideration. That is natural and desirable. An understanding of the
public aspects of bank money creation does not require the adoption of
any particular proposal. It carves out the room for substantive discussion
that has long been preempted by an over-simplistic theory of bank
money’s private origins.
APPENDIX
1. BALANCE SHEETS
*(Millions of dollars are abbreviated to “MM”.)
Balance Sheet 1: The Typical Deposit Creation Process

Balance Sheet 2: Payments Between Depositors Become

ObligationsBetween their Banks
Balance Sheet 3: Payments in a Single Bank

Balance Sheet 4: The Intuition that Interbank Markets Could Eliminate

the Need for Reserves
Balance Sheet 5: Daily Reserve Recycling in the Call Loan Market

II. FIGURES
Figure 3: A “Draining” Clearing Week for an Individual Bank (Stylized

Example)
See Article III.C.2. for discussion of assumptions.

Green arrows represent a daily balance of +$0.52MM; red arrows
represent a daily balance of -$0.52MM; “the cumulative clearing
position” (x axis) marks the aggregation of daily balances across days of
the week. Probabilities for each cumulative position are encircled.
Figure 4: Clearing Timeline for Use of Call Loans
Figure 5: Using Call Loans to Prevent Weekly Drains

III. TABLES
Table 1: Initial observations of large negative balances (1905 ledger).
6.
5. Deposits 7. Specie 8. 9. Actual 10.
11. T-1 13. T+1
1. Image 2. Date 3. 4. Bank Balance + + Legals Required reserve Balance 12. T
balance balance
number (1905) Weekday # (negative, Circulatio (W-1, reserves Ratio (W- /Required (millions)
(millions) (millions)
millions) n (W-1, millions) (W-1) 1) Reserves
millions)
364 4-Jan Wednesday 2 $ 3.8 $ 34.7 $ 10.2 $ 8.68 29% 44% $1.3 ($3.8) $1.0
364 4-Jan Wednesday 21 $ 5.1 $ 28.8 $ 6.8 $ 7.20 24% 71% $2.4 ($5.1) $1.3
365 5-Jan Thursday 4 $ 2.9 $ 23.4 $ 6.3 $ 5.85 27% 50% ($1.0) ($2.9) $2.1
367 11-Jan Wednesday 3 $ 2.3 $ 16.7 $ 4.3 $ 4.18 26% 55% $0.2 ($2.3) $2.0
367 11-Jan Wednesday 53 $ 2.1 $ 21.9 $ 5.5 $ 5.48 25% 38% $0.5 ($2.1) ($0.2)
371 19-Jan Thursday 6 $ 2.3 $ 26.8 $ 7.5 $ 6.70 28% 34% ($1.1) ($2.3) $1.0
375 26-Jan Thursday 3 $ 2.9 $ 19.3 $ 5.1 $ 4.83 26% 60% $2.8 ($2.9) $1.1
376 28-Jan Saturday 3 $ 2.4 $ 19.3 $ 5.1 $ 4.83 26% 50% $1.1 ($2.4) ($1.9)
376 30-Jan Monday 80 $ 2.2 $ 16.5 $ 4.0 $ 4.13 24% 53% $0.2 ($2.2) $0.1
379 6-Feb Monday 6 $ 2.3 $ 30.8 $ 8.7 $ 7.70 28% 30% $2.9 ($2.3) $0.4
384 18-Feb Saturday 3 $ 3.8 $ 18.9 $ 4.7 $ 4.73 25% 80% ($0.1) ($3.8) $0.7
388 28-Feb Tuesday 23 $ 18.0 $ 158.5 $ 37.1 $ 39.63 23% 45% $2.2 ($18.0) ($3.5)
388 1-Mar Wednesday 21 $ 2.6 $ 26.0 $ 5.4 $ 6.50 21% 40% $3.5 ($2.6) ($0.9)
401 28-Mar Tuesday 21 $ 2.5 $ 25.0 $ 5.4 $ 6.25 22% 40% ($0.2) ($2.5) $2.0
404 4-Apr Tuesday 6 $ 2.6 $ 24.7 $ 5.9 $ 6.18 24% 42% $5.1 ($2.6) ($0.1)
404 4-Apr Tuesday 21 $ 2.1 $ 25.6 $ 5.6 $ 6.40 22% 33% $2.5 ($2.1) $0.8
406 6-Apr Thursday 1 $ 1.7 $ 19.0 $ 4.6 $ 4.75 24.2% 36% $0.6 ($1.7) $2.0
408 11-Apr Tuesday 21 $ 2.4 $ 25.1 $ 5.2 $ 6.28 21% 38% $2.4 ($2.4) ($0.2)
412 21-Apr Friday 4 $ 2.4 $ 22.1 $ 5.8 $ 5.53 26% 43% $2.7 ($2.4) $0.4
414 26-Apr Wednesday 1 $ 2.1 $ 18.3 $ 4.2 $ 4.58 23% 46% $0.7 ($2.1) $1.6
417 2-May Tuesday 4 $ 3.6 $ 24.8 $ 7.2 $ 6.20 29% 58% $2.0 ($3.6) ($0.7)
418 5-May Friday 80 $ 3.5 $ 15.2 $ 3.6 $ 3.80 24% 92% $0.2 ($3.5) ($0.0)
422 15-May Monday 21 $ 2.9 $ 30.4 $ 7.7 $ 7.60 25% 38% $3.0 ($2.9) ($0.1)
424 18-May Thursday 21 $ 2.1 $ 30.4 $ 7.7 $ 7.60 25% 28% $6.3 ($2.1) ($0.1)
425 20-May Saturday 32 $ 2.2 $ 14.9 $ 3.8 $ 3.73 26% 59% ($1.4) ($2.2) ($0.0)
426 23-May Tuesday 21 $ 2.9 $ 31.4 $ 8.6 $ 7.85 27% 37% ($0.5) ($2.9) ($0.3)
427 26-May Friday 21 $ 2.4 $ 31.4 $ 8.6 $ 7.85 27% 31% $5.0 ($2.4) ($0.3)
429 1-Jun Thursday 3 $ 2.1 $ 16.9 $ 4.4 $ 4.23 26% 50% $2.8 ($2.1) $3.3
430 2-Jun Friday 8 $ 20.2 $ 223.7 $ 57.7 $ 55.93 26% 36% ($9.1) ($20.2) $19.3
430 3-Jun Saturday 3 $ 2.7 $ 16.9 $ 4.4 $ 4.23 26% 64% $3.3 ($2.7) $0.6
430 3-Jun Saturday 61 $ 2.2 $ 22.8 $ 5.9 $ 5.70 26% 39% $0.2 ($2.2) ($0.1)
442 1-Jul Saturday 1 $ 2.2 $ 18.0 $ 4.7 $ 4.50 26% 49% $1.0 ($2.2) $2.4
443 3-Jul Monday 4 $ 6.8 $ 25.6 $ 8.1 $ 6.40 32% 106% ($1.7) ($6.8) ($0.7)
443 5-Jul Wednesday 21 $ 3.5 $ 27.6 $ 6.3 $ 6.90 23% 51% $1.1 ($3.5) $0.5
454 31-Jul Monday 80 $ 2.2 $ 14.4 $ 3.6 $ 3.60 25% 61% ($0.1) ($2.2) ($0.3)
455 2-Aug Wednesday 4 $ 3.1 $ 23.3 $ 5.8 $ 5.83 25% 53% $0.3 ($3.1) $0.6
468 29-Aug Tuesday 23 $ 15.5 $ 159.9 $ 37.0 $ 39.98 23% 39% $3.5 ($15.5) $3.0
469 31-Aug Thursday 21 $ 4.2 $ 25.0 $ 5.5 $ 6.25 22% 67% $1.9 ($4.2) $1.7
470 2-Sep Saturday 4 $ 2.5 $ 23.2 $ 5.6 $ 5.80 24% 43% $2.1 ($2.5) ($0.9)
471 6-Sep Wednesday 21 $ 2.5 $ 27.8 $ 6.5 $ 6.95 23% 36% $0.7 ($2.5) $1.0
484 3-Oct Tuesday 23 $ 37.7 $ 140.3 $ 32.0 $ 35.08 23% 107% $6.0 ($37.7) ($0.8)
485 5-Oct Thursday 6 $ 3.5 $ 25.3 $ 6.2 $ 6.33 25% 55% $1.0 ($3.5) $2.4
498 2-Nov Thursday 4 $ 3.9 $ 19.5 $ 4.9 $ 4.88 25% 80% $1.6 ($3.9) $0.8
510 1-Dec Friday 21 $ 3.9 $ 25.6 $ 5.1 $ 6.40 20% 61% $5.2 ($3.9) ($1.9)
514 12-Dec Tuesday 3 $ 2.3 $ 14.1 $ 3.8 $ 3.53 27% 65% $0.4 ($2.3) ($0.2)
Table 3: Group 1 Summary (“high level of confidence that the drain is a

true drain”)
1. 2. Date 3. 4. 5. 6. 7. 8. Specie 9. 10. 11. 12. T-1 13. T 14.

Image (1905) Weekda Bank Balance Deposits Balance/ + Legals Required Actual Balance balance (million T+1
number y # (negative, + (Deposit (W-1, reserves reserve /Required (millions) s) (million
millions) Circulatio s+Circul millions) (W-1) Ratio Reserves s)
n (W-1, ation) (W-1)
367 11-Jan Wed 3 $ 2.3 millions)
$ 16.7 13.8% $ 4.3 $ 4.18 25.7% 55% $0.2 ($2.3) $ 2.0
401 28-Mar Tues 21 $ 2.5 $ 25.0 10.0% $ 5.4 $ 6.25 21.6% 40% ($0.2) ($2.5) $ 2.0
406 6-Apr Thurs 1 $ 1.7 $ 19.0 8.9% $ 4.6 $ 4.75 24.2% 36% $0.6 ($1.7) $ 2.0
414 26-Apr Wed 1 $ 2.1 $ 18.3 11.5% $ 4.2 $ 4.58 23.0% 46% $0.7 ($2.1) $ 1.6
426 23-May Tues 21 $ 2.9 $ 25.6 11.3% $ 5.6 $ 6.40 21.9% 45% $0.5 ($2.9) $ 5.0
430 2-Jun Fri 8 $ 20.2 $ 223.7 9.0% $ 57.7 $ 55.93 26% 36% ($9.1) ($20.2) $ 19.3
485 5-Oct Thurs 6 $ 3.5 $ 25.3 13.8% $ 6.2 $ 6.33 25% 55% $1.0 ($3.5) $ 2.4
*As noted in the Article (note 155), the T+1 entry for image #426 is a T+2
balance.
Table 4: Group 2 Summary (“Indication of a true drain, but lesser level

of confidence”)
1. Image 2. Date 3. 4. 5. 6. 7. 8. Specie 9. 10. 11. 12. T-3 13. T-2 13. T-1 14. T 15. T+1 16.
number (1905) Weekday Bank Balance Deposit Balanc + Legals Required Actual Balance balance balance balance (million balance Recoup
ment
# (negativ s + e/(Dep (W-1, reserves reserve /Require (million s) (millions
e, Circulat osits+C millions) (W-1) Ratio d s) )
millions) ion (W- irculati (W-1) Reserve
1, on) s
millions
)
By T+5
364 4-Jan Wed 2 $ 3.8 $ 34.7 11.0% $ 10.2 $8.7 29% 44% $0.6 ($0.4) $1.3 ($3.8) $1.0
75%
within
367 11-Jan Wed 53 $ 2.1 $ 21.9 9.6% $ 5.5 $5.5 25% 38% $0.4 $0.2 $0.5 ($2.1) 0.2 T+4
Full by
384 18-Feb Sat 3 $ 3.8 $ 18.9 20.1% $ 4.7 $4.7 25% 80% $0.9 $1.0 ($0.1) ($3.8) 0.7 T+2
Full by
425 20-May Sat 32 $ 2.2 $ 14.9 14.8% $ 3.8 $ 3.7 25.5% 59% ($0.7) $2.3 ($1.4) ($2.2) ($0.0) T+2
Full by
442 1-Jul Sat 1 $ 2.2 $ 18.0 12.2% $ 4.7 $ 4.5 26% 49% ($0.6) $0.0 $1.0 ($2.2) $2.4
T+1
Full by
498 2-Nov Thurs 4 $ 3.9 $ 19.5 20.0% $ 4.9 $ 4.9 25% 80% ($0.3) $0.2 $1.6 ($3.9) 0.8 T+6

Money Creation and Bank Clearing

Uploaded by

Copyright:

Available Formats

Money Creation and Bank Clearing

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Money Creation and Bank Clearing

Uploaded by

Copyright:

Available Formats

Fordham Journal of Corporate & Financial Law

Volume 28 Issue 1 Article 2

Money Creation and Bank Clearing

Follow this and additional works at: https://ir.lawnet.fordham.edu/jcfl

Nadav Orian Peer*

possible: the New York Clearing House Association prior to the

pressing challenges: financial instability, racial and wealth inequality,

10. For proposals, see notes 11-14 infra.

difficulties to banks, and is therefore not integral to bank money

20. See infra Part II.

banks. Instead, these banks were able to successfully clear by relying on

27. See infra Section III.B.

to bank money creation. It then surveys the conventional view’s account

I. THE KEY QUESTION: HOW DO BANKS CLEAR THEIR DAILY

A. CLEARING IS ESSENTIAL TO BANK MONEY CREATION

Every day, the commercial banking system stands ready to clear

transfers, and no bank money creation. To develop our understanding of

is through a deposit transfer.37 Here, the depositor is not interested in

described in Balance Sheet 2.43 Consider a Buyer of widgets (the payor)

43. See Appendix, Balance Sheet 2.

business day, banks use the Fed’s payment system—Fedwire—to transfer

“mutuality” between parties.58 Maturity means that both obligations are

(meaning that every $1 in deposits supports, on average, about $80 in

B. THE CONVENTIONAL VIEW ABSTRACTS FROM CLEARING

The conventional view’s account of bank money creation has no

for example, to speak of the origins of banking as a warehousing

reserve for occasional withdrawals. When these conditions are met,

cash against collateral, thereby helping meet the immediate obligation to

difficulties quantifying the total proportion of bank resources dedicated

II. SIDESTEPPING THE QUESTION: ASSUMPTIONS BEHIND

The conventional view’s abstraction from clearing is an “as if”

A. NETTING IS FREELY AVAILABLE TO BANKS

B. THE LAW OF LARGE NUMBERS PREVENTS DRAINS

While the first assumption is that netting is freely available to banks,

90. See Richard Routledge, Law of Large Numbers, BRITANNICA, https://

pre-GFC practice, when banks held reserves of under 1 percent of their

the specific way in which it is currently used. As discussed below, a

C. INTERBANK MARKETS CAN SOLVE DRAINS

The third assumption used to abstract from clearing is the confidence

$340 billion worth of transactions were negotiated in the federal funds

Henckel et al., supra note 104, at 24.

intervention into a well-functioning market for reserves, something that a

III. ADDRESSING THE QUESTION: DAILY GOVERNANCE IN THE NEW

The New York Clearing House Association was established in 1853

115. CONST. OF THE N.Y. CLEARING HOUSE ASSOC., § 2, https://babel.hathitrust.org/

A. NETTING WAS NOT FREELY AVAILABLE

When the Constitution of the NYCHA speaks of “effecting at one

another. We can understand this risk exposure by following the daily

124. J.G. CANNON, CLEARING-HOUSE METHODS AND PRACTICES, S. Doc. No.

individually.127 The key advantage of the clearinghouse system was that

A look at 1905 data helps appreciate how substantial the

When Section 13 is triggered, banks like B, C, and D have to pay the

posed at the turn of that century.141 Modern work in economics

B. CLEARING DRAINS DID OCCUR

lose under 4 percent of deposits 9 out of 10 days, and its probability of

Assessment of real-life drains requires disaggregated data for

Figure 2: Page from the NYCHA Ledger148

from NYCHA weekly statements).150 While the percentage is somewhat

Importantly, however, not all 45 observations—and not even the

154. See infra, Section III.C.2.