Full text of Working Papers (Federal Reserve Bank of Richmond) : Payment System Disruptions and the Federal Reserve Following September 11, 2001, Working Paper 03-16

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Working Paper Series

Payment System Disruptions and the
Federal Reserve Following September
11, 2001

WP 03-16

This paper can be downloaded without charge from:
http://www.richmondfed.org/publications/

Jeffrey M. Lacker
Federal Reserve Bank of Richmond

Payment System Disruptions and the Federal Reserve
Following September 11, 2001†
Jeffrey M. Lacker*
Federal Reserve Bank of Richmond, Richmond, Virginia, 23219, USA
December 23, 2003
Federal Reserve Bank of Richmond Working Paper 03-16

Abstract
The monetary and payment system consequences of the September 11, 2001,
terrorist attacks are reviewed and compared to selected U.S. banking crises. Interbank
payment disruptions appear to be the central feature of all the crises reviewed. For some
the initial trigger is a credit shock, while for others the initial shock is technological and
operational, as in September 11, but for both types the payments system effects are
similar. For various reasons, interbank payment disruptions appear likely to recur.
Federal Reserve credit extension following September 11 succeeded in massively
increasing the supply of banks’ balances to satisfy the disruption-induced increase in
demand and thereby ameliorate the effects of the shock. Relatively benign banking
conditions helped make Fed credit policy manageable. An interbank payment disruption
that coincided with less favorable banking conditions could be more difficult to manage,
given current daylight credit policies. Keywords: central bank, Federal Reserve,
monetary policy, discount window, payment system, September 11, banking crises,
daylight credit.

†

Prepared for the Carnegie-Rochester Conference on Public Policy, November 21-22, 2003. Exceptional
research assistance was provided by Hoossam Malek, Christian Pascasio, and Jeff Kelley. I have benefited
from helpful conversations with Marvin Goodfriend, who suggested this topic, David Duttenhofer, Spence
Hilton, Sandy Krieger, Helen Mucciolo, John Partlan, Larry Sweet, and Jack Walton, and helpful
comments from Stacy Coleman, Connie Horsley, and Brian Madigan. Thanks are also due to Marla
Borowski, Marc Chumney, Selva Demiralp, Sherry Edwards, Amy Heffernan, Dan Herlihy, Ken Lamar,
Jamie McAndrews, Elaine Mandaleris, Trish Nunley, Lisa Oliva, Anne Porter, Julie Remache, Becky
Snider, John Walter, Bill Whitesell, Howard Whitehead, Heather Wiggins, Karen Williams, and Cyanne
Yates. Also, thanks to Andreas Hornstein for translation services. The views expressed are the author’s and
do not necessarily reflect those of the Federal Reserve Bank of Richmond or the Federal Reserve System.
*
Tel: +1-804-697-8279. Email address: jeffrey.lacker@rich.frb.org.

1. Introduction
One of the most visible effects of the terrorist attacks of September 11, 2001,
aside from the appalling loss of life and sizable loss of property, was the disruption to the
workings of the financial system. The destruction of the World Trade Center towers in
New York inflicted severe damage on banking and financial institutions in Lower
Manhattan; markets closed, participants relocated to backup sites, communications links
failed or were unreliable, settlement instructions were lost, payments were delayed, and
the Federal Reserve at one point injected more than $100 billion in additional liquidity,
an unprecedented sum. At the core of it all was the disruption of interbank payments.
This paper reviews the effects of the September 11 attacks on banking and
financial institutions, with a focus on the monetary and payment system consequences
and the Federal Reserve’s response. Government securities settlement was especially hurt
by the attacks. Cantor Fitzgerald, a key interdealer broker, was devastated, losing 658
employees. Many market participants were forced to relocate to backup sites, where
internal systems and communications were not as reliable. Several banks had difficulty
processing payment instructions, and the resulting accumulation of large balances drove
net balances in the remainder of the banking system negative, necessitating the Fed’s
huge injections.
The nature of the September 11 shock to the payment system is explored by
comparing the events to selected banking crises in U.S. history. Interbank payment
disruptions were a common element in all of the crises reviewed. In some cases,
insolvency concerns provided the primary spark for interbank payment disruptions. The
banking crises of the National Bank Era (1863-1914), the settlement problems during the
stock market crash of 1987, and the settlement strains after the failure of Bankhaus
Herstatt in 1974, all fit in this category. In other cases the trigger was a technological
shock, analogous to the damage resulting from the September 11 attacks. (I will use the
term “technological shock” to refer to a significant loss of operational capability due to
either the loss or malfunction of physical capital or the loss of staff.) Examples include
the 1985 software “glitch” at Bank of New York that led to a $22.6 billion advance from
the Federal Reserve Bank of New York, and to some extent the 1987 crash. But common
to all the crises reviewed, the central propagation mechanism was the difficulty of
transferring balances between banks.
A brief appraisal suggests that the probability of future interbank payment
disturbances is not negligible. Despite substantial investments in reliability and security
and an impressive record of performance and innovation, the heavy dependence of
interbank payment arrangements on automated payment processing and
telecommunications links makes occasional technological malfunctions reasonably likely.
Moreover, history suggests that credit quality scares are a recurring feature of the
financial system, despite substantial investments in supervisory activities, and from time
to time may cause banks to retreat from payments-related extensions of credit, thereby
impairing interbank payment arrangements. Of particular relevance in light of September
11, the possibility of sabotage aimed at damaging the operational capability of the
banking and financial sector cannot be ruled out. It is worth noting in this regard that,

2

despite the horrific nature of the destruction they achieved, the perpetrators of the World
Trade Center attacks seemed less interested in functional damage than in symbolic effect.
Gleaning lessons for central bank policy from the experience of September 11 and
its aftermath therefore appears worthwhile. Federal Reserve credit extension through
discount window advances and overnight overdrafts was quite successful in supplying the
additional bank balances necessitated by the disruption to interbank payment flows,
consistent with the lender of last resort principles articulated by Henry Thornton and
Walter Bagehot, as well as one of the purposes of the Federal Reserve Act: “to furnish an
elastic currency.” Fed credit extension at the end of each day was virtually preordained
by the Fed’s daylight overdraft policies, which assured that disruption-related increases in
the demand for balances manifest themselves as intraday overdrafts which then became
some form of overnight lending. These daylight credit policies, however, could make
managing credit extension difficult in a crisis if the Federal Reserve wished to be
selective about the account holders through which it was willing to channel reserve
injections. From this viewpoint, it was fortuitous that the banking system was in
relatively healthy condition on September 11. A confluence of technology shock,
whatever the cause, and banking sector weakness is not inconceivable, however.
The Federal Open Market Committee’s interest rate targeting procedures ensure
that account balances are supplied each day to satisfy demand at the target overnight rate.
Normally open market operations aim at supplying the banking system’s forecasted
reserve needs each day through purchases of U.S. government securities and short-term
repurchase agreements with government securities dealers. The discount window serves
as a backstop provider of funds. Following September 11, open market operations were
aimed at satisfying the financing needs of the severely disrupted government securities
dealer community, leaving to the discount window the task of elastically providing
balances to satisfy demand at the target rate. The huge additions of funds following
September 11 were therefore a by-product of operating procedures designed to target the
overnight funds rate. There is a sense, then, in which Bagehotian policy is built in to
Federal Reserve operating procedures.

2. The Financial and Monetary Effects of the September 11, 2001 Attacks 1
On the morning of September 11, 2001, two hijacked commercial jet airplanes
were flown into the World Trade Center. The two towers collapsed within hours,
destroying or damaging a number of nearby buildings and spreading dust and debris
across lower Manhattan. The devastating loss of life was concentrated in the financial
industry. Fatalities in that industry represented over 74 percent of the total civilian
casualties in the World Trade Center attacks, and one firm, Cantor Fitzgerald, a key
interdealer broker in the government securities market, lost 658 employees. (General
Accounting Office 2003, p. 31) Property damage was extensive; an insurance industry
1

For an extended report on the effect of the September 11 attacks on financial markets, see General
Accounting Office (2003).

3

group estimated that total insured claims would be about $40 billion. (Zolkos 2003) The
attacks caused major power outages and hazardous conditions that hampered activity in
the area for weeks.
Much of the telecommunications infrastructure of lower Manhattan was
unavailable for several days as a result of the attacks. (General Accounting Office 2003,
pp. 37-8) The collapse of the 7 World Trade Center building sent steel I-beams into the
adjacent Verizon communications center at 140 West Street. Switching equipment there
controlled over 40 percent of lower Manhattan’s phone lines and 20 percent of the lines
serving the New York Stock Exchange. (Krane 2001) Service was lost on voice, data,
PBX and internet lines affecting about 34,000 business and residential customers. 2 Other
telecommunications service providers had service disrupted as well, but virtually all of
the post-9/11 telecom outages in lower Manhattan resulted from the problems at 140
West Street. (Lower Manhattan Telecommunications Users' Working Group 2002)
Financial markets in New York generally ceased operations. The timing of the
attacks – around 9 a.m. eastern time – meant that many markets had not yet begun
trading. Many key market participants had substantial operations in or around the World
Trade Center that were destroyed or damaged in the attacks, and had to relocate to
backup facilities. The New York Stock Exchange and the Nasdaq Stock Market never
opened for trading the day of the attacks. The facilities of the New York Board of Trade
in Four World Trade Center were destroyed. Regional stock exchanges, the Chicago
Board of Trade, and the Chicago Mercantile Exchange all closed as well. European
markets remained officially open but “most traders found it difficult to do much
business.” (Schroeder 2001) Equity markets reopened on Monday morning, September
17.
The government securities market was hit particularly hard by the World Trade
Center attacks, in part because it opens earlier. Trading in U.S. government securities
starts at 8 a.m. in New York, and repo trading starts as early as 7 a.m. 3 “(T)he bulk of
government securities cash and repo trading takes place before 9:00 a.m., … so
September 11 was close to a full trading day.” (Green 2003, p. 3) According to Jeff
Ingber, the general counsel for the Government Securities Clearing Corporation, on the
morning of September 11, some $500 billion in repo transactions and about $80 billion in
government securities trades had already been executed when the planes hit. (Shephard
2002) Reconciling these trades would occupy back-office personnel for weeks. 4

2

According to Ivan Seidenberg, president and CEO of Verizon: “The network damage was equivalent to a
city the size of Reno or Cincinnati going out of service all at once. Two hundred thousand voice access
lines went out, 100,000 PBX/Centrex lines went out, 3.6 million data circuits went out, 10 cellular towers
were lost or damaged, and approximately 14,000 businesses and 20,000 residential customers were
affected.” (Rosenberg 2001)
3
A repurchase agreement (“repo” or “RP”), is a sale of securities coupled with an agreement to repurchase
the securities at a higher price on a later date. See Fleming and Garbade (2003), Lumpkin (1998) and
Shephard (2002).
4
See Shephard (2002) and the sequence of GSCC “Important Notices” following September 11, at
http://www.ficc.com/gov/gov.imp.notices.jsp?NS-query=.

4

The government securities market also was hit particularly hard because many
critical market participants were incapacitated. Dealers in U.S. government securities
trade with each other through “interdealer brokers” (IDBs). Cantor Fitzgerald, who
suffered tragic losses, was the largest IDB prior to the attack. Two other major IDBs were
located in the WTC towers – ICAP PLC, formerly known as Garban-Intercapital, and
Euro Brokers – but the only fatality was one Garban employee. Many other IDBs
suspended operations in the aftermath of the attacks. The interdealer market operates by
phone and screen-based trading systems. With phone contact through brokers disrupted,
traders turned to online platforms, including BrokerTec, a consortium of fourteen primary
dealers, and Cantor Fitzgerald’s own eSpeed Inc., which was able to continue operating
out of their London offices. (Parry 2001; Mackenzie 2001) The Bond Market Association
recommended that the market “be closed until further notice,” and then later
recommended reopening Thursday morning. (Bond Market Association 2001)
The failure of many communications links between government securities dealers
and the market’s clearing and settling institutions was also a source of major disruption.
The two main clearing banks for the government securities market, Bank of New York
(BoNY) and J.P. Morgan Chase (JPMC), operated just a few blocks from the World
Trade Center. JPMC was in the middle of migrating certain business operations to
Tampa, Florida, and were able to resume operations from there. (McLaughlin-Moore
2002) BoNY had more difficulty. 5 Their headquarters, One Wall Street, was untouched a
half a mile from the WTC, but had to be evacuated. BoNY’s main operations center at
101 Barclay Street, one block north of the World Trade Center, housed the bank’s funds
transfer and broker/dealer systems, including the bond clearing and settlement systems.
(MacRae 2001) Both facilities were evacuated on September 11, and operations were
established at contingency sites outside the city in New Jersey and New York. The
remainder of that week BoNY suffered intermittent connectivity problems that were not
resolved late Friday. By Monday, September 17, functionality had largely been restored,
though there was a tremendous backlog of transactions to reconstruct and reconcile.
(Beckett and Ip 2001; Cowan 2001)
A third major entity in clearing government securities trades was the Government
Securities Clearing Corporation, as it was then known. 6 Instructions from counterparty
government securities dealers are compared and confirmed by GSCC, which then
establishes a net position for each dealer in each security issue, along with a net cash
position, and interposes itself as counterparty to guarantee settlement. Positions are
settled using the Fedwire Securities Service or the clearing banks. 7 GSCC remained
operational after September 11, but many members were unable to deliver trade
instructions for the 11th, and thus GSCC had information from only one side of the trade.
5

For accounts of BoNY’s experience, see Beckett and Sapsford (2001), Guerra (2001), Gibbons (2001),
and MacRae (2001).
6
GSCC merged with the MBS Clearing Corporation on January 1, 2003, to form the Fixed Income
Clearing Corporation, a subsidiary of the Depository Trust & Clearing Corporation. See www.ficc.com.
For a description of repo clearing and settlement, see Fleming and Garbade (2002; 2003).
7
Government security trades are settled T+1, that is, the business day following the trade. For repos, GSCC
settles only the close (maturity) leg of the trade and the start leg for forward repos (repos beginning T+1 or
later); GSCC does not settle the start legs of RPs that start the same day.

5

GSCC’s connection to BoNY was lost for part of the week of the 11th and as a result they
did not know what securities and cash they had received, and were at times unable to
transmit settlement instructions to BoNY. (Costa 2001) Because the offices of so many
key market participants were destroyed, and because connectivity was problematic for
several days following September 11th, there was a dramatic increase in the volume of
settlement “fails” – failures to deliver U.S. government securities rose from $1.7 billion
per day the week of September 5 to $190 billion the week ending Wednesday, September
19. (Fleming and Garbade 2002)
BoNY’s role in clearing and settling government securities transactions placed it
at a critical node in interbank payment flows. The two clearing banks hold funds and
securities on behalf of government securities dealers. When counterparties both use the
same clearing bank, settlement involves offsetting transfers of securities and funds on the
books of that bank. When counterparties are customers of different clearing banks,
settlement involves an exchange of funds for securities between the two. The two
clearing banks process a substantial portion of the payments that flow across the Fedwire
system.
The communications and operations difficulties plaguing BoNY meant that not all
funds payment instructions were getting sent as intended. On the Federal Reserve’s
Fedwire Funds Transfer System, payments are initiated by the sender of funds. 8 A bank’s
inability to send funds transfer payment instructions results in funds accumulating in that
bank’s account. Balances in the rest of the banking system would be correspondingly
lower. At one point during the week after September 11, BoNY publicly reported to be
overdue on $100 billion in payments. (Beckett and Sapsford 2001) A handful of New
York banks found themselves in a similar situation – unable to make payments or loan
funds. (Markets Group of the Federal Reserve Bank of New York 2002, pp. 6, 24)
Balances accumulated in these banks’ accounts and resulted in a corresponding reserve
drain and large negative aggregate position for the remainder of the banking system.
The increase in account balances was more widespread than a few money center
banks, however. Figure 1 shows the account balance distribution from the beginning of
August through September 21. For each day, selected percentile balances are plotted. For
example, the top line shows, for each day, the account balance of the bank at the 99.9th
percentile; 99.9 percent of banks have smaller balances. Each line could represent a
different bank each day. The number of account holders was about 8,500 each day; so
there are eight or nine banks in each tenth of a percentile group. 9 Figure 1 shows that the
increase in account balances extended down to around the 90th percentile of the balance
distribution. In other words, about 800 banks experienced a noticeable increase in their
account balance. Interestingly, the increase was proportionally larger at the upper end of
8

On the Fedwire Securities Service, the party sending the security initiates the transaction, which results in
the immediate and simultaneous transfer of the security against offsetting payment – “delivery versus
payment.” Thus, for a security transfer the party receiving funds (and sending securities) initiates the
transaction.
9
Eight banks were assigned to the above-99.9-th percentile group, nine banks were assigned to the group
between the 99.9th and 99.8th percentiles, eight banks the next group, and so on. The figure for a given
percentile is the minimum balance within the group.

6

the distribution, as one would expect if those observations represented banks with larger
typical payment flows. For example, on a typical day in August, fewer than 8 banks held
balances greater than half a billion, whereas on September 11 and 17, more than 16 banks
held balances that large. Of course, there is no guarantee that banks retained their relative
position within the bank balance distribution. But Figure 1 indicates that the disruption to
payment flows affected far more than just a couple of New York banks.
It seems implausible that all of the institutions that showed higher account
balances following September 11 were directly damaged by the attacks. Some of the
accumulation of funds banks’ accounts appeared to have resulted from the breakdowns in
the fed funds market. Several federal funds brokers were disabled in the attacks and did
not resume operations until the following Monday. Banks with excess balances found it
difficult to locate borrowers. The general disruption in payment flows would also have
meant uncertainty for many banks about whether scheduled incoming payments would be
received as planned. This may have induced banks to delay or withhold payments.
(McAndrews and Potter 2002)
Other markets were affected by the World Trade Center attacks as well. Although
Fedwire and the Clearing House Interbank Payments System (CHIPS), operated by the
New York Clearing House (as it was known then – the name has since been changed to
The Clearing House), continued to function, payment processing was delayed at many
banks and closing times were pushed back. (McAndrews and Potter 2002; Goldenberg
and Stock 2001; New York Clearing House 2001) The majority of the commercial paper
that was scheduled for presentment on September 11 and 12 was not paid, but rolled over
and settled on Thursday. (Bond Market Association 2001) Issuance resumed fairly
quickly, however, and proved a relatively viable source of liquidity in the following days.
(Goldenberg 2001) The Moscow International Currency Exchange (MICEX), which used
BoNY as its dollar settlement bank, suspended trading on Thursday afternoon due to
BoNY’s problem but then resumed trading after switching to JPMC. (Russian rouble
firms on low liquidity 2001; Danielyan 2001)
Retail payment card networks – credit, debit, ATM cards, and the automated
clearinghouse networks – remained operational, except for scattered problems at bank
ATMs in New York City, and at BoNY’s ATM network, which crashed entirely on the
11th and wasn’t restored until the evening of September 19th. (In the Shadow of Tragedy,
U.S. Payment System Stands Strong 2001; The Bank of New York's ATM Network
Restored 2001; Mandaro 2001) BoNY announced it would refund ATM fees for
customers that used other banks’ ATMs. The grounding of airline flights seriously
hampered inter-regional check clearing for a time, as banks and the Federal Reserve
scrambled to arrange for substitute truck transport. (Edwards 2001; Mollenkamp,
Pinkston, and Schlesinger 2001) On Thursday, September 13, the Federal Aviation
Administration began reopening U.S. airspace and gave check air couriers approval to
resume its chartered flights. 10

10

The Federal Reserve Banks contract with private air couriers to transport checks overnight between
Federal Reserve Bank offices. A private air courier, Airnet, is a major transporter of checks for banks.

7

In New York and Washington, bank branch closings were widespread on
September 11th, but many banks outside those cities closed branches briefly as well.
Some state banking agencies and the Office of the Comptroller of the Currency issued
statements allowing banks to close at their discretion. The Banking Commissioner for the
State of Connecticut ordered all banks and credit unions to close. 11 Bank of America and
Wachovia closed their headquarters, which are housed in several tall towers that
dominate the Charlotte, North Carolina skyline. (Mollenkamp, Pinkston, and Schlesinger
2001) Wachovia and Chicago-based Bank One closed branches nationwide early in the
afternoon on Tuesday, but were open as usual the next day. The Chicago Tribune
reported that “a handful of bank branches in or around major landmarks such as the Sears
Tower” were closed. (Allison 2001)
Some banks discouraged cash withdrawals by customers. The Municipal Credit
Union, whose back offices were near the WTC, limited customers to $500 withdrawals
when their 11 branches reopened on Thursday. (Padgett 2001) For a day after the attack
Citibank recommended that customers limit cash withdrawals to $5,000. (Chaffin and
Silverman 2001) Wells Fargo, a San Francisco based bank, placed limits on per-person
cash withdrawals that reportedly varied across locations from $1,000 to $5,000, and
Washington Mutual, based in Seattle, imposed a limit of $2,500. (Ip, Sims, and Beckett
2001) Some armored carriers suspended operations in New York City, and transportation
difficulties impeded some deliveries of currency elsewhere in the country. Deliveries of
newly printed notes from the Bureau of Engraving and Printing to some Federal Reserve
Banks were delayed by the airline grounding, as were currency shipments to Alaska and
Hawaii. (Blackwell 2001)
Reports of increased cash withdrawals by bank depositors were common,
especially on the East Coast. (Mandaro 2001; In the Shadow of Tragedy, U.S. Payment
System Stands Strong 2001; Breitkopf 2001) Concord EFS, an ATM payment processor,
reported a surge in ATM card usage on their network Tuesday afternoon and Wednesday
morning. (Bills, Breitkopf, and Green 2001; Breitkopf 2001) At 8 p.m. volume was 31
percent higher than the previous week. The biggest surge was at point-of-sale terminals,
especially at gasoline retailers, perhaps reflecting the substitution of automotive for
airline travel. Concord’s network traffic was down in the hours immediately following
the attacks, however. Credit card networks reported lower volume for several days after
the attacks, consistent with the sharp drop in retail sales: see Figure 2. (Bills, Breitkopf,
and Green 2001)
Currency in circulation increased by $4.4 billion from Monday to Wednesday. In
comparison, total currency in circulation was $614 billion on September 5, of which
more than half was estimated to be overseas. See Figure 3, which shows the cumulative
change in currency in circulation following September 10. Nearly $3 billion of the $4.4
billion reflected an increase in banks’ vault cash holdings, consistent with heightened
cash shipments from the Federal Reserve banks. (Edwards 2001; Ip, Sims, and Beckett
11

Only in the order to reopen, issued at 7:15 p.m. on the 11th, did the Commissioner reassure consumers
that they “should remain confident that their savings in banks and credit unions are not at risk and are
insured by the Federal Deposit Insurance Corporation (FDIC) and the National Credit Union
Administration (NCUA), respectively.” (State of Connecticut Department of Banking 2001a, b)

8

2001) Currency held outside banks only rose by $1.6 billion. Some of the increase in
vault cash holdings probably reflected banking system preparations for a further expected
currency drain later in the week. In the event, there was a $2.6 billion increase on Friday.
The public’s currency holdings stabilized in the following weeks, falling during the week
and rising on the weekends, as is typical. Vault cash holdings trended downward.
The movements in currency demand following September 11 were pronounced,
but not unusually large. Bank vault cash typically rises during the week and then falls on
Friday as consumers and merchants withdraw cash to use over the weekend. The public’s
currency holdings display the opposite pattern, peaking over the weekend and falling
during the week. Figure 4 shows that the swings in currency demand around September
11 might have a been a bit elevated, but were well within typical ranges. Vault cash
holdings, however, were somewhat higher than trend over the weekend of September 1516. Banks appeared to have built up a large buffer of cash, but experienced a demand
surge that was smaller than anticipated.

3. The Federal Reserve’s Response
Virtually every instrument available to the Federal Reserve – open market
operations, discount window lending, payment services, supervision and regulation, and
communication – was pressed into service following the September 11 attacks. At first,
Fed communications were the most visible response. At 9:44 a.m. on the 11th, a
broadcast message was sent to banks over the Fedwire system stating that the system was
“fully operational at this time and will remain open until an orderly closing can be
achieved.” At 11:25 a.m. another broadcast stated that “The discount window is available
to meet liquidity needs.” Around noon the Board of Governors issued the following press
release.
The Federal Reserve System is open and operating. The discount window is
available to meet liquidity needs.
Later in the day the Boston Fed released a statement on behalf of the Fed’s financial
services functions with more specifics. Fed Governor Edward Gramlich, traveling in
Tucson, Arizona that day, was quoted as saying, “The Fed is the lender of last resort. If
credit is needed to make transactions go, the Fed will provide it.” (Gilbert and Thomas
2001) Reached by phone in Basle, Switzerland, New York Fed President William
McDonough said “I’m sure that central bankers everywhere will do everything possible
to maintain calm and seek to ensure the world economy functions smoothly in the face of
this horrendous deed.” (Ip and McKinnon 2001) 12
12

Vice Chairman Roger Ferguson was the only member of the Board of Governors in Washington on
September 11: Laurence Meyer was in China, Edward Gramlich was in Arizona, Edward Kelley was on
vacation, and two seats were vacant. (Ip and VandeHei 2001) Chairman Greenspan was flying from
Switzerland, where he had been attending an international meeting of central bankers, when his plane
turned back as a result of the closure of U.S. airspace. He returned to Washington Wednesday morning on a
military plane.

9

Conditions in Lower Manhattan affected operations at the Federal Reserve Bank
of New York, just three blocks to the east of the WTC. During the day on Wednesday,
staff began relocating to an operations center outside the city, and early Wednesday
evening remaining staff were forced to evacuate the Manhattan building due to concerns
about the structural integrity of One Liberty Plaza next door. The Board of Governors in
Washington, D.C., was evacuated along with many other government offices in the city,
although “about 100 staffers remained at work.” (VandeHei et al. 2001) Concern about
employee safety caused the Boston Fed to evacuate and led many other Federal Reserve
Banks to staff critical functions only.
The disruption to communications links impaired many institutions’ ability to
initiate payment instructions. The number of payments processed over the Fedwire funds
transfer and securities systems fell on September 11 and remained low for the rest of the
week; see Figure 5. (Coleman 2002; McAndrews and Potter 2002) Payments on those
systems occurred significantly later in the day, and intraday (“daylight”) overdrafts were
significantly larger. The Fed waived daylight overdraft fees from September 11 through
September 21. 13 To facilitate completion of payments processing, the Fed extended the
Fedwire closing times on the days following September 11. 14
Check collection, as noted above, was severely hampered by the lack of air
transport, and it was clear at the outset that the presentment of checks to paying banks
would be delayed significantly. This meant delays for the Fed in collecting funds from
paying banks. Fed policy is to credit banks that deposit checks according to a schedule
that replicates, on average, the schedule on which the Fed collects good funds, so that Fed
“check float” – the excess of credits to banks for deposited checks over the funds
collected from other banks on those checks – is typically near zero. Positive Fed check
float represents an implicit loan to the banking system as a whole; depositing banks are
credited before offsetting debits are made to paying banks. Collections are often delayed
by storms or other factors impeding transportation. The Fed often continues to grant
credit on normal schedules during such incidents, but sometimes delays credit until
presentments are actually made. On September 11, anticipating a general need for
liquidity in the banking system, a decision was made and announced late in the afternoon
to continue to extend credit on normal availability schedules. Fed check float, which
averaged $766 million per day in the first eight months of 2001, resulted in a net injection
of funds of $47 billion on Thursday and $44 billion on Friday: see Table 1, which
displays the factors affecting account balances for the days around September 11. In
comparison, the value of checks collected by the Reserve Banks averaged of $40 billion
per day in 2001, suggesting that the two-and-a-half day courier grounding delayed
roughly half the Fed’s checks each day. In addition, the Reserve Banks picked up check
volume from banks that found the Fed’s availability schedule attractive relative to the
availability they could obtain on their own. The reopening of U.S. airspace on Thursday
13

The Fed charges a small fee, equivalent to 36 basis points on an annualized basis, for overdrafts in excess
of a deductible for each bank. See Coleman (2002) for a discussion of the Fed’s daylight credit policies.
14
The funds transfer service, which normally closes at 6:30 p.m. ET, was extended to 9:00 p.m., 11:30
p.m., 11:00 p.m., and 8:30 p.m., Tuesday through Friday, respectively. The securities transfer system,
which normally closes at 3:15 ET, was extended to 7:15 p.m., 10:45 p.m., 8:30 p.m., and 6:30 p.m.

10

night prevented further increases in float on Friday, and the weekend allowed check
processors to catch up and work off the backlog.
Federal Reserve Bank staff contacted banks and armored carriers to reassure them
that ample currency was available for withdrawal, hours of operation were being
extended, and special deliveries could be arranged. In addition, banks were notified that
the Fed would waive the fees that applied to certain banks accessing Fed cash services
and would suspend “cross-shipping” rules aimed at preventing banks from depositing and
then quickly withdrawing the same currency. The availability of currency supplies was
never seriously in question; Reserve Bank inventories were $157 billion on September 5,
2001, versus outstanding Federal Reserve notes in circulation of $583 billion (much of
which was abroad). The increase in demand for cash turned out to be less than $5 billion:
see below.
As mentioned above, and as detailed by Fleming and Garbade (2002), there was a
dramatic increase in settlement fails following September 11. During normal times
settlement fails are often associated with a “scarcity” of a specific issue. To help alleviate
such scarcities and help limit fails, the Federal Reserve loans out securities from its
System Open Market Account, subject to self-imposed limits on the fraction of SOMA
holdings of any given issue that can be lent. The Fed relaxed the terms of the securities
lending program on September 11 by suspending per dealer limits, and then further
loosened terms on September 13. 15 Because settlement fails were still a problem two
weeks after the attacks, the Fed raised the security lending program limit from 45 percent
to 75 percent of each issue, beginning September 27 and continuing into October. Acute
settlement problems with the on-the-run ten-year note led the U.S. Treasury to reopen the
issue on October 4 and hold an unusual “snap” auction of new ten-year securities.
In the days following September 11, banking regulators realized that disruptions
were causing significant increases in the size of bank portfolios. Failures to settle various
transactions left offsetting payment and security delivery obligations sitting on the
balance sheets of market participants, along with the underlying cash or securities that
were awaiting delivery. These inevitably showed up on the banking system’s balance
sheet and reduced bank capital ratios. In addition, many firms drew on bank lines of
credit in response to operational difficulties rolling over commercial paper. On Friday
September 14, federal banking regulators issued a Joint Interagency Statement noting that
many banks may experience temporary balance sheet growth, and urged banks to contact
their regulators should they anticipate a resulting decline in their regulatory capital ratio.
The Federal Reserve later issued a Supervisory Letter allowing banks some flexibility in
calculating capital ratios for the third quarter of 2001. (Spillenkothen 2001) Bank
regulators also encouraged banks to lend to customers (“take prudent steps to make credit
available to sound borrowers”) affected by the events of September 11. 16 (Board of
Governors of the Federal Reserve System 2001c)
15

See http://www.newyorkfed.org/newsevents/news/markets/2001/omo010911.html for details.
The Comptroller of the Currency announced that “national banks that support recovery efforts in
communities affected by the September 11 terrorist attacks will receive credit under the Community
Reinvestment Act.” (Office of the Comptroller of the Currency 2001)
16

11

3.1. Monetary Injections I: Open Market Operations
Federal Reserve open market operations and credit extension injected
unprecedented quantities of funds into the banking system in the days following
September 11. Injections were necessitated by the accumulation of balances at
operationally constrained banks and the willingness of many banks to hold larger than
normal account balances under conditions of uncertainty about clearing and settlement
arrangements. Movements in autonomous factors also affected the magnitude of the
required injections.
Monetary policy operations during the week following September 11 had to adapt
to the disruption of financial markets and normal clearing and settlement activities. 17
Normally, operations aim to supply a level of bank balances each day that satisfies the
banking system’s demand at an overnight federal funds rate equal to the target rate set by
the Federal Open Market Committee. (Markets Group of the Federal Reserve Bank of
New York 2002) Banks’ demand for balances is determined by two factors: the amount
needed to meet reserve requirements and clearing balance commitments (“balance
requirement”), and the amount held in excess of those requirements to buffer against
unanticipated late-day payment flows. 18 The Trading Desk at the New York Fed normally
adds or drains balances each morning based on estimates of demand and autonomous
technical factors such as shifts in U.S. Treasury balances or changes in check float that
would otherwise affect supply. If demand turns out higher than expected or autonomous
factors unexpectedly lead to a deficiency, the result is usually either upward pressure on
the federal funds rate later in the day or bank requests at the discount window for
overnight loans, or both.
Because balance requirements must be met over a two-week period, banks have
some flexibility with regard to when they hold the necessary balances. This tends to limit
the pressure on the funds rate from unexpected movements in autonomous factors or
demand – that is, demand should be fairly elastic at the target. In fact, under certain
conditions the federal funds rate should be a martingale within a maintenance period, and
should remain close to the target as long as market participants expect the Desk to supply
balances on subsequent days to accommodate shifts in demand and autonomous factors.
(Hamilton and Jordà 2002) One qualification to this principle is that while banks can take
advantage of high rates by holding less than their balance requirement, overnight account
balances below zero are penalized. As a result, if aggregate balances are low enough,
banks will bid the funds rate up to avoid overnight overdraft penalties. In the opposite
direction, the funds rate can fall below the target prior to settlement day if enough banks
17

Due to the physical condition of Lower Manhattan, Trading Desk operations were relocated mid-day
Wednesday to a backup site outside the city.
18
Reserve requirements apply to a two week reserve maintenance period that ends on a Wednesday
(“settlement day”). Clearing balances earn credits that can be applied toward Federal Reserve Bank
financial services. Banks’ vault cash holdings can be applied toward reserve requirements; the remaining
reserve requirement is met with reserve account balances. See Meulendyke (1998) for a detailed description
of open market operations.

12

have already held more reserves than they anticipate they will need for the period and are
willing to offer their “surplus” reserves on the lending market at any positive rate. On
settlement days demand is less elastic since banks face penalties for failing to meet
balance requirements and they forego interest income if they hold excess balances.
Although the Desk’s intervention on settlement morning is aimed at minimizing the
expected deviation of the funds rate from the target over the remainder of the day,
unexpected changes in autonomous factors and unexpected shifts in bank demand can
push the funds rate away from the target.
The Trading Desk adds balances by lending in the repo market – that is, by
entering into repurchase agreements with dealers in U.S. government securities whereby
the Desk effectively advances funds against pledged collateral. 19 The rates on the Desk’s
repo transactions are set via auction. When the Desk wants to inject balances, it
announces the maturity and terms, requests bids specifying the interest rate dealers are
willing to pay and for how much money, and then selects the bids with the highest rates
down to the point at which it has accepted bids for the amount of funds it wishes to inject.
Settlement takes place on the books of BoNY or JPMC, since all of the primary dealers
clear through one of those banks. An injection of funds thus increases the banking
system’s balances in the first instance by increasing the account balances of BoNY or
JPMC. Those new balances are then reallocated throughout the banking system during
the day as banks send payments to other banks and borrow and lend in the interbank fed
funds market. In 2001 the Desk typically had a set of longer-term 28-day repos
outstanding. 20 Overnight and/or very short-term operations are used to accommodate dayto-day fluctuations in needs. Term repos totaling $22.755 billion were outstanding on
September 11: see Table 1. Short-term operations averaged close to $8 billion in 2001,
excluding the September 6-19 maintenance period. (Markets Group of the Federal
Reserve Bank of New York 2002)
Prior to the initial attack, the Desk had decided on the basis of available data to
arrange no open market operation for the day. In the wake of the attacks, the discount
window was considered the most effective means of providing any additional liquidity
that the banking system might need that day, consistent with the Board’s noon statement
(see above). By Wednesday morning the extent of the disruption to government securities
settlement was coming into view, as was the likelihood that the banking system’s needs
would be substantially elevated. Beginning Wednesday and continuing through the
following Monday, the Desk conducted open market operations with the aim of satisfying
dealer financing needs (Markets Group of the Federal Reserve Bank of New York 2002,
p. 22), a shift from the usual focus on bank needs.
During the week following September 11, the Desk accepted all propositions at or
above the target federal funds rate, which was then 3.5 percent: see Table 2. Operations
totaled $38.25 billion on Wednesday, $70.20 billion on Thursday, and $81.25 billion on
Friday, all for overnight funds. The $8.75 billion in term repos that matured Thursday the
19

Reserve drains are accomplished using reverse repurchase agreements in which the Fed effectively
borrows from dealers against collateral pledged from the System’s portfolio.
20
Effective September 18, 2003, the Trading Desk began using 14-day term RPs instead of 28-day RPs.
(Federal Reserve Bank of New York 2003)

13

13th were not rolled over, and thus Thursday’s overnight operation brought total RPs
outstanding to a little over $84 billion.
On Monday morning, September 17, the FOMC met by conference call at 7:30
a.m. eastern time and voted to lower the target for the federal funds rate by 50 basis
points to 3 percent. In its statement the Committee said:
The Federal Reserve will continue to supply unusually large volumes of liquidity
to the financial markets, as needed, until more normal market functioning is
restored. As a consequence, the FOMC recognizes that the actual federal funds
rate may be below its target on occasion in these unusual circumstances.
The Desk added $57.25 billion in funds later that morning via overnight repos, accepting
every proposition at or above the stop-out rate of 3 percent, the new funds target rate.
Term repos for about $2 billion rolled off Monday and were not replaced, bringing total
outstanding repos to $69.25 billion. The next two days, the Desk accepted bids below the
target but added a smaller quantity of balances each day. On Wednesday, the Desk also
executed term repos worth $22.75 billion with maturities of 14, 21 and 28 days, for
commencement and settlement the following day, the beginning of the next maintenance
period, in order to reduce the level of intervention that would be necessary. (Markets
Group of the Federal Reserve Bank of New York 2002, p. 24) Open market operations
were closer to normal scale over the following days.

3.2. Monetary Injections II: Federal Reserve Credit
The second half of the Federal Reserve’s two-sentence press release following the
attacks was devoted to lending: “The discount window is available to meet liquidity
needs.” While this might have been seen as an unmistakable implication of first
sentence’s message that the Fed was “open and operating,” it was widely interpreted as
evidence that the Fed was willing to lend to ease payment strains in the aftermath of the
attacks. The statement echoed the Fed’s reaffirmation of readiness on the morning of the
stock market crash of October 20, 1987. 21 In fact, Governor Gramlich indicated that his
statement (quoted above) was “in the spirit” of a statement by Chairman Greenspan in
1987. (Gilbert and Thomas 2001) Several other Fed officials reinforced that message
over the course of the next few days.
Discount window borrowing is generally arranged at the end of the day,
sometimes after the close of business, although borrowing can be arranged earlier in the
day and banks often are in contact with their Federal Reserve Bank lending officers
before a formal request is made. 22 For many years prior to January 2003, the interest rate
applicable to discount window loans was generally below the federal funds rate target
and thus below the general level of short-term interest rates, giving banks an incentive to
21

“The Federal Reserve, consistent with its responsibilities as the nation’s central bank, affirmed today its
readiness to serve as a source of liquidity to support the financial and economic system.”
22
See www.frbdiscountwindow.org for details on the Federal Reserve’s discount window policy.

14

borrow at the window to exploit the spread. (Board of Governors of the Federal Reserve
System 2002a) Regulations governing the use of the discount window limited borrowing
by requiring that an institution first exhaust other available sources of funds and explain
its need for adjustment credit. In addition, banks were prohibited from using discount
window credit to finance lending in the federal funds market. (Under current policy the
“primary rate” is usually 100 basis points over the target funds rate, and the
administrative conditions intended to limit borrowing have been eased.) In practice,
interpretation of the regulations by Reserve Banks effectively limited most borrowing to
situations in which banks experienced unanticipated late-day drains and were unable to
cover their shortfall in the fed funds market. In this context, a noon-time statement
emphasizing the availability of the discount window was likely seen as a distinct regime
shift.
The discount window serves as an outlet for unmet demand for Reserve Bank
account balances. The Trading Desk estimates autonomous factors before they undertake
open market operations in the morning, but unanticipated variations in these factors can
occur. For example, banks may make unusually large currency withdrawals from Federal
Reserve Banks, or there could be an unanticipated flow of tax receipts into the Treasury’s
account, draining private sector balances. In such cases, the balances added by the Desk
might be insufficient to satisfy the banking system’s demand for balances at the target
funds rate. Prior to settlement day, banks may be willing to hold less than their balance
requirement, reckoning that they will be able to hold additional balances later in the
maintenance period. On settlement day, however, their opportunity to substitute
intertemporally is much more limited. Banks would tend to bid up the funds rate in an
attempt to obtain the balances they desire, but if aggregate balances were insufficient,
then some banks would be left short at the end of the day. The Fed’s discount window
thus serves to limit the tendency of the funds rate to rise above the target.
The Federal Reserve has an additional channel by which banking system balances
can increase. Most banks are entitled to incur intraday overdrafts, also known as
“daylight” credit, subject to a system of caps and fees. 23 The daylight overdraft cap is set
as a multiple of capital for a depository institution that qualifies through a self-assessment
of its creditworthiness, credit policies, and operational controls and procedures. Some
depository institutions qualify for smaller caps, and five percent of depository institutions
have a zero daylight overdraft cap. 24
When the Fedwire funds transfer system closes (normally at 6:30 p.m.) banks
effectively are no longer able to obtain funds from other banks. A negative account
23

The Federal Reserve’s Payment System Risk Policy governs the extension of daylight credit: see
Coleman (2002) for details. A daylight overdraft fee of 36 basis points at an annual rate applies throughout
the day to overdrafts over a deductible set at 10 percent of an institution’s capital. In other words, if the fee
was compounded for 24 hours a day over the course of an entire year it would come to 36 basis points.
Currently, Fedwire is only open for 18 hours a day however, so the effective maximum fee works out to 27
basis points annually.
24
Multiples vary from 1.125 to 2.25 times (risk-based) capital. Some depository institutions qualify for a de
minimus cap of 20 percent of capital; others have a zero cap. These net debit caps apply to any single day’s
daylight overdraft. A separate cap applies to the two-week average daylight overdraft. See Coleman (2002).

15

balance at the end of the day becomes an overnight overdraft if no further action is taken,
and a penalty is incurred equal to the effective federal funds rate plus 400 basis points.
Alternatively, the bank could request a discount window loan at the discount rate, which
they presumably would prefer because the cost is lower. Another difference between
overnight overdrafts and discount window loans is that the latter are explicitly
collateralized while the former are not necessarily collateralized. 25 In addition, some
institutions with Federal Reserve accounts are ineligible for discount window credit. 26
For the period from September 11 through September 21, the Federal Reserve waived
daylight overdraft fees for all account holders and eliminated the penalty on overnight
overdrafts for depository institutions; depository institutions were charged the effective
federal funds rate on overnight overdrafts while non-depository institutions were charged
the “extended credit” rate, which was 4 percent at the time, plus 55 basis points.
Foreign financial institutions operating in the U.S. also faced payment flow
disruptions, and some experienced balance deficiencies in the days after September 11.
“In some cases, however, these institutions encountered difficulties positioning the
collateral at their U.S. branches to secure Federal Reserve discount window credit.”
(Board of Governors of the Federal Reserve System 2002b, p. 142) The Federal Reserve
arranged new or expanded swap lines with three central banks in order to help meet those
needs. These swap lines entitled the foreign central banks to receive dollars from the
Federal Reserve in exchange for their respective currencies. The lines were for $50
billion with the European Central Bank, $30 billion with the Bank of England, and an
increase of $8 billion (from $2 billion to $10 billion) with the Bank of Canada. When a
foreign central bank drew on one of these lines, the New York Fed credited them with
dollar balances, which they then lent on to foreign financial institutions. This mechanism
had the effect of adding dollar balances to the banking system while interposing foreign
central banks between the Federal Reserve Banks and foreign financial institutions. Table
1 lists the total amount drawn on these swap lines around September 11.
Following September 11, the demand for overnight Federal Reserve credit –
discount window advances and overnight overdrafts – reflected the banking system’s
residual need for funds. The Desk’s repo operations added substantial amounts of
25

Discount window borrowing requires that a bank execute the lending agreement contained in Operating
Circular 10, which secures any borrowing, including daylight and overnight overdrafts, with the collateral
pledged to the Federal Reserve. See www.frbdiscountwindow.org. Discount window lending must be
secured “to the satisfaction of the Reserve Bank,” which normally requires that the Reserve Bank obtain a
perfected security interest in the pledged collateral. If a lending agreement is in place and sufficient
collateral has been pledged, then overnight overdrafts are implicitly collateralized, but a lending agreement
is not required for access to daylight and overnight overdrafts. For a bank without a lending agreement in
place and which has not posted collateral, overnight overdrafts would be uncollateralized.
26
The Board of Governors has defined “depository institution” for purposes of Regulation A (governing
extensions of credit by Federal Reserve Banks) as ruling out financial institutions that are not required to
meet reserve requirements. These include banker’s banks, corporate credit unions, and the government
sponsored enterprises. Some of these institutions can waive their exemption from reserve requirements and
become eligible for regular access to the discount window. In certain cases, Federal Reserve Banks may
extend credit to nondepositories: “In unusual and exigent circumstances and after consultation with the
Board of Governors, a Federal Reserve Bank may extend credit to an individual, partnership, or corporation
that is not a depository institution if, in the judgment of the Federal Reserve Bank, credit is not available
from other sources and failure to obtain such credit would adversely affect the economy.” 12 C.F.R. 201

16

balances on Wednesday, Thursday, and Friday. Check float added additional billions on
those days as well. But as the end of the day approached, the banking system’s net
demand for balances had to be brought into line with supply. Some banks were able to
adjust their holdings to a desired level. Others were constrained by disruptions in
communications and processing, and were unable to send payments and reduce their
holdings to a planned or desired level. The sum of banks’ positive balances, either
intended or constrained, exceeded the Fed’s earlier injections by large amounts.
The banking system’s net end-of-day deficit meant that many banks were short on
balances. Banks with deficiencies had two options; they could either bid for borrowed
funds in the market or turn to the Federal Reserve for overnight credit. Federal Reserve
statements were likely interpreted as implying a fairly elastic supply of funds at the
discount rate. Given that assumption, there was no need to bid up the funds rate. And the
Fed ultimately followed through with the provision of credit as expected. Discount
window borrowing rose from an average of about $200 million in the year prior to the
week of September 11 to $37 billion that night and overnight overdrafts rose from an
August 2001 average of $9 million to around $2 billion: see Table 1. Overnight credit of
over $38 billion was required by the banking system because there were no open market
injections that day and check float had added only $4 billion. Moreover, currency
withdrawals had drained $2 billion from bank balances. On Wednesday, the Desk
injected over $38 billion, check float added another $23 billion, but there was still an
additional $50 billion in unsatisfied demand for balances at the end of the day. Balance
injections picked up on Thursday, with check float adding $47 billion and the Desk
adding a net of $61 billion (after letting $9 billion in term repos roll off), only $8 billion
in overnight Fed credit was required; overdrafts fell below $500 million and discount
window advances fell to $8 billion. Check float was still substantial Friday night and the
Desk added even more balances that day, finally satisfying demand; overnight credit
extensions was negligible. Float receded Monday, but the Desk’s generous balance
provision kept Fed overnight credit to a minimum. Borrowing rose on Wednesday, the
last day of the maintenance period, but then subsided.

3.3. Interest rates
Short term interest rates generally declined in the days following September 11.
The FOMC cut the target overnight funds rate by 50 basis points, from 3.5 percent to 3
percent, at a special meeting convened before markets opened on Monday morning the
17th, and then cut another 50 b.p. at its next scheduled meeting on October 2nd. The
week before, numerous news stories after the attacks had carried conjectures about
imminent Fed rate cuts, although market participants were unsure of the timing.
Overnight rates were steady the week of September 11 – see Figure 6 – but the fall in the
implied rate on the September federal funds futures contract to below 3.4 percent on
September 13 – see Table 3 – indicates that some market participants expected the funds
rate to decline before the end of the month. They may have expected a move before the
October 2 meeting. Alternatively, they may have anticipated that with the large amount
of reserves injected by the Fed, banks with excess balances at the end of the maintenance

17

period would drive the funds rate down. The fact that the October contract closed at 2.87
on the 13th suggests that rate cuts were expected. By the close on Friday, the market was
pricing in a 2.8 percent average funds rate for the month of October, suggesting
expectations of one 50 b.p. cut and a substantial probability of another. After the Monday
morning move, markets quickly priced in a higher probability of a further rate cut in
October. Eurodollar futures prices indicated a downward revision in expected mid-2002
short rates of about a half a percent in the week after September 11. The downward shift
in the yield curve accompanying the FOMC’s rate cuts following September 11 is
consistent with market perceptions that the cuts would not be soon reversed.
Overnight interest rates sagged at the start of the week after the attacks due to the
overhang of balances that had been added the week before. The first three days of that
week were the last three days of the maintenance period. Because many banks had been
forced to hold large balances the previous week, the banking system ended the
maintenance period with a large excess reserve position – $38 billion, compared to a
maximum excess position of $1.7 billion over the period between Y2K and September
11. As a result, the Desk aimed to leave relatively low levels of balances each day.
Autonomous factors were draining reserves, however, (see Table 1, second panel, column
labeled “Other”) and so the size of the needed operations remained large. The Desk “had
to accept the vast majority of propositions – even those offered at rates well below the
new 3 percent target level – in order to arrange RPs of sufficient size.” (Markets Group of
the Federal Reserve Bank of New York 2002, p. 24) On Wednesday, the Desk accepted
all propositions submitted, the lowest of which was ¾ percent: see Table 2. The effective
federal funds rate sank to 1¼ percent on Tuesday and below that on Wednesday: see
Table 3. Rates returned to normal a couple of days into the next maintenance period.

4. The Nature of the Shock
Drawing lessons for central bank policy from the events of 9/11 calls for an
understanding the nature of the shock to the interbank payment system and the nature of
the economy’s response. Because the events of September 11 were characterized by a
tangible sense of crisis, and because the Federal Reserve’s lender of last resort activities
figured so prominently, I will look for useful analogies to other U.S. banking crises. The
comparisons are instructive, both for the similarities and for the differences that emerge.
While different initial shocks set events in motion, disrupted interbank payments are
central to all the crises examined here. This suggests two lessons. First, interbank
payments disruptions have more in common with other banking crises than previously
appreciated. Second, the role of interbank payment arrangements in the instigation and
propagation of banking crises deserves more attention.
Historical banking crises are numerous, and the literature on banking crises is
large. My review is selective, and starts with the banking panics of the National Bank Era
(1863-1914). These helped motivate the founding of the Federal Reserve, and disruptions
to arrangements for interbank clearing and settling payments figured prominently.
National Bank Era panics generally occurred in New York City, then, as now, the central

18

node in payments settlement, and spread to the interior (1893 was an exception). They
were distinguished less by bank failures (apart from private banks that were better
classified as brokerages) than by suspensions of payment by New York banks to banks in
the “interior”– that is, New York banks refused to ship currency to redeem interior banks’
deposits. (Sprague 1910; Wicker 2000) These deposits served as the reserves used to
settle interregional payments prior to the founding of the Federal Reserve, and the New
York banks’ suspensions impaired their use as a settlement medium. (Watkins 1929) The
ensuing loss of confidence in New York banks led the country banks to withhold
remittances and withdraw reserves. Sprague speaks of the “dislocation of domestic
exchanges” – that is, local interbank markets for correspondent balances at New York
banks (pp. 293-4). Suspensions in New York thus added costs to local interbank
payments elsewhere.
Wicker (2000, p. 146) observes that in the major panics of 1893 and 1907,
“insufficient reserves were not the problem; it was the unequal distribution of those
reserves among the large [New York] Clearing House banks. Reserve equalization would
have forestalled the two panics.” Similarly, Sprague (1910, p. 276) says of the suspension
of 1907: “It was directly due to uneven distribution of the reserves held by the New York
banks and to the use of clearing-house loan certificates as the sole medium of settlement
of balances at the clearing house.” In 1873, in contrast, the pooling of reserves by the
New York Clearing House made possible the “ultra-liberal policy of continuing to pay
out cash to the interior,” which was not observed in subsequent panics. (Wicker 2000, p.
33) Timberlake (1984), however, argues that the pooling of reserves was unnecessary
because clearinghouse certificates could have accomplished the same goal of “equalizing
reserves.” He agrees that the interbank distribution of reserves was the issue. The shockinduced maldistribution of reserve balances and the technological impediments to
redistributing them were precisely the problem for the banking system following
September 11.
There were significant differences between September 11 and nineteenth century
banking panics, however. First, as noted above, the public’s movement from bank
deposits into currency was relatively minor: see Figure 4. A slight pick up in currency
withdrawals occurred, but this was as much banks’ preparation for possible consumer
demand as it was an actual increase in consumer currency holdings. And some portion of
the increase in the public’s currency holdings could well have reflected the transaction
demand associated with the substitution of long-distance ground travel for airline travel
in the two days after the attacks, rather than currency hoarding in response to concerns
about potential bank closures. Bank runs were a common feature of National Bank Era,
especially in 1893. (Wicker 2000) The absence of bank runs following September 11
could be attributable to the presence of federal deposit insurance, or it could have
reflected widespread confidence in banks’ solvency.
Second, and related, insolvencies were not a crucial feature of the aftermath of
September 11. In contrast, the initial trigger of a nineteenth century bank crisis was
invariably an instance of default, insolvency or some other event that either causes or
raises questions about the solvency of a particular bank or set of banks. (Calomiris and
Gorton 1991) The banking industry was in relatively healthy condition prior to the
19

attacks. Despite the economic downturn of early 2001 and a gradual uptrend since the
mid-1990s in noncurrent commercial and industrial loans, particularly at larger banks,
capital ratios were relatively good, the number of problem banks was steady at a level
that was low by historical standards, and the industry’s ability to absorb losses appeared
strong. 27 To be sure, the loss of property and life was widely expected to result in large
claims for many insurance companies, but estimates available immediately after the event
indicated that claims were likely to be well within the industry’s capacity to absorb. 28 The
airline industry was hit hard by the closing of the national airspace. And after the
grounding was lifted, the stiffening of airport security controls and the increase in the
perceived risk associated with air travel raised the implicit cost of airline service and cut
sharply into demand; traffic was more than 20 percent below year-earlier levels in
October 2001, compared to 4 percent above year earlier levels in August. Midway
Airlines Corp., a North Carolina-based carrier listing $318 in assets, announced on
September 12 that it would permanently cease operations; it had filed for bankruptcy on
August 13, 2001, but this was the only significant failure announced that week and it
represented a pre-existing problem made worse by the attacks. The tourism and
hospitality industries saw demand fall off as well, but the banking sector’s exposure to
these industries was relatively small. The two major government securities clearing
banks, BoNY and JPMC, held enough capital to qualify as “well capitalized,” according
to the most recent call reports. In short, there were no obvious threats on the horizon that
seemed capable of damaging the condition of the banking sector.
A corollary to the minimal concern regarding bank insolvencies was a lack of
“contagion” effects following September 11. There was no evidence that any institution’s
financial health was adversely affected by payment difficulties at other banks. Thus, there
appeared to be none of the “knock-on” or “domino” effects that are a central
preoccupation of the literature on payment system risk.
Perhaps the most striking difference between nineteenth century bank panics and
the September 11 episode is the behavior of interest rates. Bank panics were typically
associated with spikes in short-term money market interest rates. To prevent such spikes,
Walter Bagehot recommended that the central bank lend freely but at a “penalty rate,”
that is, a rate above prevailing market rates. In contrast, the Fed lent at below market
rates to depository institutions. Since then, alterations in the discount window program
have raised the interest rates on advances to above overnight market rates. (Board of
Governors of the Federal Reserve System 2002a) Perhaps Fed lending would have looked
more Bagehotian had such a program been in place then.
27

See the Federal Deposit Insurance Corporation’s Quarterly Banking Profiles for the third and fourth
quarters of 2001. (Federal Deposit Insurance Corporation 2001) The Shared National Credit review for
2001, based on second quarter analysis and released on October 5 of that year, reported a continued
deterioration in the quality of syndicated bank loans, but showed quality measures still significantly better
than during the years of the recession of 1990-91. (Board of Governors of the Federal Reserve System,
Federal Deposit Insurance Corporation, and Office of the Comptroller of the Currency 2001)
28
An insurance industry group estimated that total insured claims for property damage would be about $40
billion. (Zolkos 2003) As one analyst put it: “this is not an overly worrisome event, since the industry has
the capital to absorb a hit of this magnitude.” (Fitch Insurance Conference Call Summary 2001) Comments
like this were typical in the weeks after September 11.

20

The proximate initial trigger of the disruptions following September 11 was the
malfunctioning of automated systems for recording, processing and communicating
payment obligations, together with the loss of many specialized professionals. These
were by all accounts operational difficulties, not unanticipated financial losses. In this
regard, the payment system events following September 11 resemble other late-twentieth
century instances in which technological malfunctions disrupted interbank payments,
perhaps the most noteworthy of which was BoNY’s November 1985 overdraft, which
resulted in a $22.6 billion discount window advance to the Bank of New York. (The
Federal Reserve Bank of New York Discount Window Advance of $22.6 Billion Extended
to the Bank of New York 1985) Faulty software prevented BoNY from sending securities
to counterparties at times during the day on Thursday, November 21. This also prevented
them from collecting the offsetting payments on those securities. (The exchange of a
security and an offsetting payment are simultaneous under Fedwire’s delivery-versuspayment regime.) They continued to receive securities and funds transfers but the
inability to send securities meant that they ran up an overdraft during the day. Although
the Fedwire securities system was kept open late to provide time to attempt to solve the
problem, these attempts were unsuccessful and the securities system was closed at 1:30
a.m. on Friday morning and the funds system was closed at 2:15 a.m. BoNY borrowed
$700 million from the private sector after midnight, but ended the day overdrawn by
$23.6 billion. 29
The 1985 episode was due to software problems at a single institution. In contrast,
available evidence indicates that payment processing problems affected a number of
depository institutions besides BoNY following September 11. This may be because the
technological effects of the September 11 attacks were far more widespread. Many banks
and market participants were forced to operate from back up sites, and many
communications links were problematic. In addition, the nature of the holdup in
payments flows following September 11 was that many institutions had incomplete
information on payments obligation incurred just before the attacks but not yet cleared
and settled. Many government securities trades had been executed, but the confirmation
and trade matching process had not occurred. The two clearing banks and the GSCC had
many “half-legs” – trades for which they had received only one party’s confirmation.
Reconciliation was time-consuming and fraught with difficulties.
One other noteworthy difference between November 1985 and September 2001 is
that in 1985 the direct effect of the disruption was a large negative balance at BoNY. In
2001, the direct effect was an accumulation of large positive balances at several
institutions. Large positive balances imply large negative balances at other institutions,
29

BoNY’s initial estimate of its position at 2:15 was the basis for the discount window loan of $22.6
billion, but that estimate was incorrect and BoNY incurred an additional $1 billion in the form of an
overnight overdraft. See the prepared statement of E. Gerald Corrigan, p. 6, in (The Federal Reserve Bank
of New York Discount Window Advance of $22.6 Billion Extended to the Bank of New York 1985). At the
time of the most recent call report on September 30, 1985, BoNY had total assets of $15.9 billion. Under
the terms of the special loan agreement covering the advance, negotiated earlier on Thursday, the advance
was secured by all of the assets of the bank as well as all of the customer securities they were empowered
to pledge. The Federal Reserve Bank of New York estimated the book value of the assets securing the
advance at $36 billion.

21

absent Fed credit extension, so from that viewpoint the two incidents are similar. The
difference is that Fed credit was extended to the institution at which the disruption
occurred in 1985, while in 2001 credit was particularly necessary at institutions other
than those whose outgoing payments were impeded.
One striking similarity between the two episodes is that the effects of the
disruption to interbank payments were focused within the banking system. Some bank
customers experienced difficulties in transferring funds, but in neither case did there
appear to be runs on banks or other evidence of loss of depositor confidence. 30 Again, the
presence of deposit insurance and the fact that the disruptions were largely seen as
independent of bank insolvency concerns probably helped reassure depositors. Related,
neither incident resulted in any net outlay from the federal financial safety net, ex post,
apart from the implicit subsidy of below-market discount window advances and the
potential underpricing of credit risk. 31
Two other postwar episodes of interbank payment disruption have more in
common with the nineteenth century panics. During the sharp fall in stock market prices
the week of October 19, 1987, the unusually large trading volume caused confirmation
problems for several equities-related options markets. (The Presidential Task Force on
Market Mechanisms 1988; Bernanke 1990) Communications difficulties, including a two
and a half hour Fedwire outage on October 20, interrupted the flow of payments between
New York and options-market accounts in Chicago. (Garsson 1988) On top of the
technological problems, the fall in asset prices had eroded the net worth of many traders
and specialist firms that acted as market makers, leading creditors, including banks, to
decline to advance funds as usual to cover settlement and margin-call payments.
Incomplete information contributed to the difficulties; the volatility of market prices and
the delay in price reporting made it difficult for banks to know if their customers were
solvent. Although payment system disruptions did not have the starring role in the 1987
market break – the inability of market makers to absorb heavy selling on October 19 and
20 took center stage – they did contribute to the impairment of the efficiency of market
mechanisms.
The Federal Reserve’s response to the 1987 market break in many ways paralleled
the response following September 11. At 8:15 a.m. on Tuesday morning, October 20, the
Board of Governors issued a brief statement:

30

Bank customers were directly affected by the BoNY’s ATM system outage on September 11 and after
and the destruction of some bank branches, but these were not the result of the disruptions to interbank
settlement.
31
Corrigan’s testimony stated that the 1985 BoNY discount window advance was made at 54 basis points
below that day’s prevailing fed funds rate. With the penalty rate on BoNY’s overnight overdraft, and the
fact that BoNY had to finance its holdings of securities it owed to the Federal Reserve Bank of New York
but the latter was entitled to the accrued interest on those securities, Corrigan calculated that “the net
financial result of all transactions between Bank of New York and the Federal Reserve Bank of New York”
was in favor of the latter. (Corrigan 1990, p. 8) The Federal Reserve’s discount window program has been
revised. (Board of Governors of the Federal Reserve System 2002a) Among other changes, the interest
rates on advances are now above the federal funds target rate.

22

The Federal Reserve, consistent with its responsibilities as the nation’s central
bank, affirmed today its readiness to serve as a source of liquidity to support the
financial and economic system. 32
The Fed’s September 11 statement echoes this affirmation of reserve supply intentions. In
1987, the Fed reportedly “encouraged” banks to lend to their customers, including stock
specialist firms, paralleling bank regulatory agencies’ statements following September
11. (Bernanke 1990) In 1987, the Fed relaxed firewall restrictions on transactions
between banks and affiliates, and allowed Continental Illinois to inject funds into First
Options, its options-clearing subsidiary. 33
The failure of Bankhaus Herstatt on June 26, 1974, led to the disruption of
interbank payments to settle foreign exchange trades. (Remolona et al. 1990) Herstatt, a
medium-sized German commercial bank active in the foreign exchange market, was
closed by banking authorities at 4:00 p.m. local time in Köln – mid-morning in New
York. Chase Manhattan, Herstatt’s U.S. correspondent bank, refused to honor about $620
million in payment orders and checks drawn on the account. Many of these were
settlement for the dollar leg of foreign exchange trades on which Herstatt had already
received the deutschemark leg. At that time, settlement for foreign exchange trades was
accomplished by having each party initiate the side of the trade on which they were the
payer; for example, a trader selling dollars sent instructions to their U.S. correspondent to
pay the counterparty, while the trader selling deutschemarks sent payment instructions to
their German correspondent. 34 Settlement takes a number of days, and often one payment
becomes irrevocable before the other payment can be confirmed. This leaves
counterparties vulnerable to the risk that one party irrevocably delivers one leg but the
counterparty fails to deliver the other leg, a scenario that is now known as “Herstatt
risk.”35
Following the Herstatt failure, New York banks would not make payments on
foreign exchange trades until they received confirmation that the other payment had been
received. Large balances accumulated and other payments were delayed as a result. The
following week, the New York Clearing House banks changed the terms on their CHIPS
net settlement system. 36 Prior to that, CHIPS payments had been irrevocable – the sender
could not recall them once they were entered into the system. Under the special
temporary procedure adopted in the wake of Herstatt’s failure, payments were conditional
until 10:00 a.m. the next morning. This made banks more comfortable sending payments,
but reduced the extent to which banks could rely on funds that came through CHIPS.
There were no incidents of runs caused by the Herstatt failure. (Kaserer 2000)

32

Quoted in The Presidential Task Force on Market Mechanisms (1988).
See Walter (1996) for an introduction to bank firewall regulations.
34
Foreign exchange settlement still takes place in this fashion, except for participants in the Continuous
Linked Settlement system. See Kahn and Roberds (2001) and Lacker (2001).
35
Non-German creditors of Bankhaus Herstatt recovered between 58.78 and 86 percent of their claims
under a settlement agreed to on February 24, 1975. Payments were made June 9, 1975. (Becker 1976)
36
See www.theclearinghouse.org/.
33

23

In both the Herstatt and the 1987 cases, risk considerations led to a pull back in
the private extension of payment system credit. A wide variety of payment processing
activities are conducted in a way that involves the extension of credit, at least briefly,
reflecting a choice regarding the trade-offs between credit risk on the one hand and the
resource cost and timeliness advantages of credit-intensive arrangements on the other.
When events lead to revisions in assessments of the associated credit risks, banks
sometimes have an incentive to rein in credit exposures and shift to less credit-intensive
arrangements. These are generally more time consuming, and the result is generally
payment delays and increased demand for reserves. A banking sector retreat from credit
exposure in payment arrangements can be seen as an instance of the widely observed
propensity for bank lending standards to vary with aggregate economic conditions.
Weinberg (1995) argues that such variations are to be expected in well functioning credit
markets, and need not represent market failure. In both the Herstatt and the 1987 stock
market cases, it seems plausible that an increase in the lending standards applied to
payments-related credit extensions was warranted on objective risk-assessment grounds.
To summarize this review, late nineteenth century U.S. banking crises, the
Bankhaus Herstatt failure, the 1985 BoNY software glitch, 1987 stock market crash, and
the September 11 events were all characterized by interbank payments disruptions. In the
BoNY 1985 and the September 11 episodes technological difficulties rather than
heightened credit risk served as the initial trigger. In the nineteenth century and the
Herstatt episodes, insolvencies and potential credit losses were the initial trigger. During
the stock market crash credit concerns were the trigger and a technological malfunction
occurred in the midst of the troubles as well, although there is dispute about the extent to
which it had significant independent effects. In all of the episodes reviewed, however, the
propagation of the shocks was similar. The redistribution of reserves among banks was
impeded and payments were delayed. Moreover, bank runs either did not occur or were
secondary; the main event in all was in the interbank payment system.
Two types of interbank payment disruptions have been identified, therefore –
credit-shock induced and technology-shock induced – based on the nature of the initial
triggering events. They might seem different enough to warrant treatment as separate
phenomena. But to the extent that the choice of an interbank payment arrangement is
viewed as balancing a trade-off between the costs of credit risk exposure during clearing
and settlement, and the costs of minimizing that exposure by increasing payment speed,
both types of shocks can be seen unanticipated disturbances to the frontier of feasible
arrangements.

5. Future Shocks
If September 11 and the three other episodes cited above are taken as instances of
a single class of events – interbank payment disruptions – then the occurrence of at least
four over the last thirty years (about as often as recessions over that time period) suggests
that these deserve policy-makers attention. Indeed, central banks around the world have

24

devoted significant attention to the risks associated with the operation of interbank
payment systems. 37
My selective review suggests that technology-shock-induced interbank payment
disruptions are a phenomenon of the last few decades. This seems reasonable, given the
significant changes that have taken place in the record-keeping and communications
technologies employed in the banking industry over the last century, particularly for
interbank payments. One hundred years ago, the New York Clearing House functioned as
the central node for U.S. interbank payments, and its operations were largely as they had
been since the 1850s. 38 Bank clerks met daily and exchanged paper, with net obligations
settled by payment of reserves in the form of specie or paper notes. Modern interbank
payments are made by electronic transfers, with bank computers connecting directly with
the computers of the Federal Reserve Banks or clearinghouse systems such as CHIPS or
CLS. For a bank of any size, the internal processing of payment instructions, even before
they result in the issue of interbank payment instructions, is also highly automated.
Reliance on electronic processing and communications makes critical payments
activities vulnerable to periodic interruptions in the normal functioning of these
technologies. It seems reasonable to conjecture that, in contrast, the paper-based systems
of the nineteenth century were less vulnerable to malfunctions. The reliability and
security of electronic payments processing arrangements are presumably the implicit
result of a wide array of benefit-cost decisions. At a fundamental level, the likelihood of
future technological malfunctions rests on prospects for these benefit-cost trade-offs.
Experience strongly suggests that such calculations are not likely to lead to the complete
elimination of the risk of malfunctions.
Perhaps the most obvious potential source of payment system disruption is the
malfunctions to which any automated system is subject. The software outage that
afflicted BoNY in 1985 is an archetypal case, but power outages, storms and other
unanticipated infrastructure break downs would have to be included in this category. For
example, the Northeast blackout of August 14, 2003 affected the financial system,
although disruptions were minimal. (Financial and Banking Information Infrastructure
Committee 2003; Kite 2003) The stock exchange and government securities trading had
already closed for the day when the outage occurred at around 4:15 p.m. All but one
reopened the next day for a full day of trading. Most banks switched to backup power
generators and there were few reports of processing delays. The federal funds market was
affected by the lack of power at several funds brokers, but trading continued with higher
volatility that afternoon. Several foreign banking organizations reportedly suffered failed
backup power facilities and/or lost telecommunications links. (Financial and Banking
Information Infrastructure Committee 2003) Excess reserve holdings during the twoweek maintenance period that included the blackout averaged $5.9 billion per day, more

37

Witness the series of publications by the Committee on Payment and Settlement Systems:
www.bis.org/cpss/cpsspubl.htm.
38
Compare Cannon (1901) and Gibbons (1859).

25

than twice as high as the next largest two-week excess reserve figure from September 11,
2001, through October 2003, suggesting that some payment flows were impeded. 39
Power outages and storm-related disruptions recur so often that preparing for
them is a routine business activity. It appears that disruptions in this category would have
to be unusually severe to result in a significant dislocation in the payment systems. But
again, benefit-cost calculations are not likely to lead to complete elimination of the
possibility of dislocations due to technical malfunctions.
Deliberate terrorist attacks on physical infrastructure are obviously capable of
interrupting normal payment functions, and September 11 was not the first such attack.
At about noon on September 16, 1920, a horse-drawn wagon carrying hundreds of
pounds of explosives was detonated at the corner of Wall and Broad Streets in Lower
Manhattan, killing thirty people instantly and causing the stock exchange to close. The
exchange reopened “defiantly” the next day, however, and banking and financial activity
appeared to return to normal quickly. (Brooks 1969) Both the 1920 and the 2001 attacks
seem to have been aimed at killing as many people as possible connected with the
financial and commercial activity symbolized by the physical location of their targets,
rather than inflicting as much long-term damage as possible to the functioning of the
financial system, consistent with a strategy of maximizing the publicity value of the
attack. Thus the payments system problems following September 11 can be seen as
collateral damage rather than the direct objective of the attacks. Nonetheless, the iconic
status of financial markets seems enduring and could draw terrorists to financially-related
targets in the years ahead.
The vulnerability of the payments system to a physical attack that directly targets
operational capabilities is unclear. One observer of the damage sustained by the Verizon
building on September 11 noted that “(i)f they really wanted to do damage, they would
have taken out the telco building. They went after the thing that had press potential.”
(Coffield 2001). An article in the trade publication, American Banker, in December 2001
discusses potential terrorist threats to the financial system, and quotes market participants
that claim that “with only a little extra effort” terrorists could have crippled the financial
system. (Bach 2001) The attack on the World Trade Center seemed designed for maximal
symbolic value rather than maximal functional impairment, despite videotape in which
Osama bin Laden cites quantitative measures of the losses due to the attacks. (Bin Laden
hails economic losses from Sept. 11 attacks 2002)
The vulnerability of interbank payments infrastructures to attack is not limited to
physical assaults. The scenario of a computer virus debilitating critical payments
processing systems is not beyond the realm of possibility. Most viruses target personal
computers and travel over the internet, while interbank payments usually flow from
mainframe to mainframe over dedicated lines. As a result, one might think that interbank
payments are far better insulated from virus attacks than one’s home PC. Nonetheless,
recent incidents have demonstrated the vulnerability of critical computer networks,
39

See the Board of Governors of the Federal Reserve System H.3 release at
www.federalreserve.gov/releases/h3/hist/.

26

including those of the banking sector. For example, Bank of America’s ATM network
was knocked out by the SQL Slammer internet worm in January, 2003; several other
large banks were affected as well, though to a smaller degree. (Breitkopf 2003; Lee 2003)
In June of the same year, the BugBear virus specifically targeted financial institutions.
(Weiss 2003) During an attack by the Blaster worm in August, 2003, CSX temporarily
stopped railroad service and Nordia, Scandinavia’s largest bank, closed 80 branches
across Finland. (Guth 2003) Shortly thereafter, two Baltic banks shut down their systems
after attacks by the Sobig.F virus. Recent reports suggest that the planners of the
September 11 attacks have recently sought the ability to launch assaults targeting digital
devices that allow remote control of services such as fire dispatch and pipeline
equipment, and “had ‘far more interest’ in cyber-terrorism than previously believed.”
(Gellman 2002)
Computer virus attacks are purely destructive in the sense that they do not result
in a direct wealth transfer to the attacker. Electronic payments systems would appear to
be inviting targets for attacks that aimed instead at transferring wealth. Citibank’s funds
transfer system was penetrated in August 1995 by a Russian hacker who was able to
move at least $10 million in funds to accounts at other banks. (Caldwell 1995) Such
attacks obviously trigger upgrades in security to prevent against similar future attacks,
but in the resulting “arms race,” successful future attacks cannot be ruled out. A hacker
attack could conceivably force a bank to shut down important payments processing
systems in order to prevent further losses, despite the attendant disruption to legitimate
payments, similar to the credit-risk provoked pull-back from payment system credit
extension in the Herstatt and 1987 market crash episodes.
To summarize this brief review then, it appears plausible to assign at least some
positive probability to future interbank payment disruptions roughly similar in effect to
that of September 11.

6. Implications for Central Bank Policy
6.1. Central bank credit policy
The Federal Reserve System’s announcement that “the discount window is
available to meet liquidity needs,” along with other reinforcing communications by
Federal Reserve officials, appeared to have a calming effect during the first two days
after the attacks, judging by the frequency with which the statement was repeated in news
coverage of the event. In short order, market participants were able to see evidence of
Fed action. Late Thursday afternoon, the Board of Governors released the H.4.1, which
reports data on the Fed’s balance sheet for the night before. It showed the discount
window borrowing of $46 billion for Wednesday. 40 Coming on the heels of Wednesday’s
40

The September 13, 2001, H.4.1 release also showed average daily figures for the week ending September
12. These showed discount window borrowing averaged $11.7 billion over that week, from which market
participants could infer that discount window borrowing was likely to have been about $36 billion on
Tuesday the 11th, if borrowing was about average earlier in the week.

27

super-sized open market operation of $38.25 billion, the figures showed the Fed
following through on the commitment implicit in Tuesday’s announcement. Aggregate
funds injection figures were widely cited in the days following the attacks as measures of
the Fed’s provision of funds.
The Fed’s clear communications regarding its intention to supply balances as
needed conforms well to classic lender of last resort principles. Bagehot (1991, p. 85)
argued that a central bank ought to provide assurance of support in advance, in order to
alleviate uncertainty in the minds of the public regarding the central bank’s course of
action in a crisis, thereby contributing to stability and helping to avert panic. (Humphrey
and Keleher 1984) It also conforms well to more recent analyses that emphasize
limitations on a central bank’s ability to commit to a desired lending policy. (Goodfriend
and Lacker 1999) Articulating a willingness to lend eliminated an important potential
source of uncertainty in the minds of market participants. Confidence in the Fed’s
willingness to carry through on that commitment seemed warranted on September 11,
because the condition of the banking system, through which lending would be channeled,
was known to be relatively healthy going into the day.
Federal Reserve credit extension was very effective at increasing the supply of
bank balances. Banking system balances went from $13 billion on September 10 to over
$120 billion on the 13th: see Table 1. The end-of-day need for balances could not have
been forecast with any degree of precision at the time of day that open market operations
were conducted, even with the delay of operations on some days. Moreover, open market
operations were focused on meeting the financing needs of the government securities
dealer community, which had been severely affected by communications breakdowns and
the difficulties at BoNY, rather than on supplying the balance needs of the banking
system, although that was a welcome by-product. If the Fed had not accommodated the
increased demand for balances, either through open market operations or overnight
overdrafts, banks likely would have bid up the funds rate in the open market. A sharp rise
in overnight interest rates could have significantly changed the tenor of that week in
financial markets. A scramble for balances could have led banks to liquidate assets,
driving down financial asset prices, driving up longer-term interest rates, and perhaps
could have led to a 1987-style impairment of market mechanisms, further exacerbating
the scarcity of funds. Federal Reserve credit extension after September 11 thus served its
Bagehotian function of providing an elastic supply of central bank liabilities, preventing a
spike in interest rates, and thus limiting the secondary repercussions of the initial
disturbance that gave rise to the increase in demand. 41
Federal Reserve credit extension following September 11 was unsterilized, in the
sense that it resulted in a net increase in the monetary base. Goodfriend and King (1988)
argue that unsterilized central bank lending is best thought of as “monetary policy” and is

41

In Bagehot’s doctrine, the central bank is typically viewed as preventing a shock-induced fall in the
money stock. (Humphrey and Keleher 1984) In September 11, the task was to respond to a shock-induced
increase in the demand for reserves. One purpose of the Federal Reserve Act, according to its preamble, is
“to furnish an elastic currency.”

28

unnecessary given the ability of open market operations to achieve the same result. 42 The
Fed’s current operational policies make open market operations an imperfect substitute
for end-of-day credit extension, however. The Trading Desk policy generally intervenes
only a limited number of times a day, usually once or twice in the morning when repo
markets are liquid. Moreover, the close of the Fedwire securities transfer system at 3:30
p.m. limits same-day repo trading after that time. Thus the Desk generally does not, under
current policies, respond to events after the typical morning intervention time, although
on September 12 the Desk announced that it might intervene again that day should it be
necessary. (Ip, Sims, and Beckett 2001) Although the Desk intervened somewhat later in
the day following September 11, the Fedwire funds transfer system was closing late in
the evening that week. Arranging repos near midnight would have been difficult, if not
impossible. (Groshen et al. 2002)
The difficulty of injecting the requisite balances after September 11 without endof-day Fed credit extension does not refute Goodfriend and King’s thesis, however.
Eschewing credit extension would have required different operating procedures on the
part of the Trading Desk. Current Desk practice presumes the availability of end-of-day
Fed credit. In fact, the Desk’s strategy following September 11 was predicated on the
availability of discount window lending to meet what was expected to be a large and
difficult to forecast demand for balances. Thus to argue that foregoing credit extension
would not have supplied sufficient balances given the Desk’s actions doesn’t take into
account how differently the Desk might have acted had credit extension been unavailable.
Moreover, one could argue that the appropriate question concerns how private sector
banks would behave under an alternative regime for supplying central bank liabilities. In
other words, are there alternative Desk procedures which could substitute for unsterilized
credit extension? Although it is beyond the scope of this paper to answer this question,
this would appear to be an open question. Thus Fed credit extension following September
11 would not appear to have a direct bearing on the Goodfriend-King thesis.
One feature of the Fed’s end-of-day credit extension following September 11 that
went unnoticed in popular accounts is that it was virtually preordained by the Fed’s
intraday credit policies. As mentioned above, depository institutions with accounts at a
Federal Reserve Bank can apply for daylight overdraft privileges. Seventy five percent of
account holders incurred daylight overdrafts – 5,300 out of 8,500 institutions – at some
point during the third quarter of 2001, according to Coleman (2002). When payment
processing disruptions caused several banks to accumulate large account balances after
September 11, other account balances necessarily were driven down by a commensurate
amount. Open market operations and check float added balances, but a substantial
“financing gap” remained. In the absence of discount window lending, and if banks had
been unable to reallocate reserves among themselves, banks would have been overdrawn
by the amount of this gap. To illustrate, consider Wednesday September 12. Account
balances ended the day $95 billion higher than they had been on Monday. See Table 1.
42

Sterilized lending is “banking policy”, that is, redirecting the allocation of credit toward a particular
institution, holding the monetary base as given. Goodfriend and King would argue that Bagehot’s
principles pertain to unsterilized lending, i.e. for the conduct of monetary policy, in an institutional setting
without other means (i.e. OMOs) for injecting reserves.

29

Suppose we take this as a measure of the balances whose transfer was blocked by
payment processing disruptions, plus any possible increase in precautionary holdings – in
other words, as a measure of the disruption-induced increase in demand. Check float
added $23 billion toward meeting the increased need, foreign central bank draws on FX
swap lines added $5 billion, and overnight repos added another $38 billion. Currency
withdrawals by banks had drained $4 billion by Wednesday, and “other” factors (chiefly
the net effect of the foreign RP pool) drained an additional $16 billion. At the end of the
day, reserve demand was up by $95 billion, but supply was up by a net of only $46
billion. That left a $50 billion gap to filled by end-of-day Reserve Bank lending, either as
discount window credit or as overnight overdrafts. Ultimately, $46 billion of the gap was
met at the discount window and $4 billion was met through overnight overdrafts.
Depository institution behavior could well have been different had they believed
that the Reserve Banks would not make discount window advances to cover their
overdrafts, and this might have mitigated the extent to which end-of-day Fed lending was
preordained. Normally, prospects of an overnight overdraft would give a bank an
incentive to borrow on the fed funds market at rates up to the overnight overdraft penalty
rate (four percentage points above the overnight funds rate), if the discount window was
unavailable to them. In that case, some banks that were holding additional balances after
September 11 might have been willing to lend those balances at a higher fed funds rate.
The additional supply of reserves lending called forth by the funds rate increase would
have reduced the amount of balances the Fed needed to provide through end-of-day
credit. The Federal Reserve rescinded overnight overdraft penalties, however, for the
days immediately following September 11; overnight overdrafts by depository
institutions were assessed at the fed funds rate, while non-depositories were assessed the
extended credit rate plus 55 basis points. Banks would have had little incentive to bid up
the funds rate to elicit additional balances had the discount window been unavailable.
For the Reserve Banks to not extend end-of-day credit following September 11,
they would have had to shut down the discount window and prevent daylight overdrafts.
The Federal Reserve’s Payments System Risk Policy controls banks’ use of daylight
credit. 43 The Fed’s Account Balance Monitoring System tracks banks’ balances in real
time during the day, and can be used to control a bank’s intraday overdrafts by rejecting
any payment with settlement-day finality (e.g. funds transfers) that would cause the bank
to exceed its pre-agreed net debit cap. “Real-time monitoring,” as it is called, is used to
reduce the risks posed by “institutions in deteriorating financial condition or institutions
with a history of excessive overdraft activity.” (Board of Governors of the Federal
Reserve System 2001a) Approximately five percent of banks were monitored in reject
mode as of early 2001. (Board of Governors of the Federal Reserve System 2001b) For
the rest, ABMS tracks their daylight position, but does not prevent breaches of their net
debit cap; for these banks intraday payments could result in daylight overdrafts in excess
of their caps. Preventing an overnight overdraft to a bank requires shutting off intraday

43

See www.federalreserve.gov/paymentsystems/psr/default.htm.

30

credit by putting the bank on the monitor in reject mode at a zero net debit cap. 44 Without
taking such action across the board, end-of-day credit extension was virtually automatic.
Despite the equivalence between the reserves injection capability of overnight
overdrafts and discount window lending, there were good reasons to prefer to advance
credit through the discount window. As mentioned above, discount window lending is
generally collateralized, providing greater security to the Reserve Bank, while overnight
overdrafts may or may not be collateralized. (See note 25) Goodhart argues that requiring
collateral also has the effect of helping to screen out insolvent borrowers. (Goodhart
2002) In addition, the Federal Reserve clearly wanted to encourage banks to consider
availing themselves of the discount window should the need arise. Following September
11, many banks had difficulty operating in the federal funds market because of problems
with communications links and processing systems. Anticipating that the Federal Reserve
would likely grant a request for a discount window loan, may have encouraged some
banks to forego strenuous efforts to obtain funds. Fed lending thus helped reduce the
deadweight costs of trading in a difficult situation.
A corollary to the observation that end-of-day credit extension would have
injected sufficient balances automatically is that the quantity of balances injected after
September 11 was independent of the Reserve Banks’ decision to provide credit for
check deposits on the normal schedule, a decision that resulted in over $40 billion in
check float late in the week. If policy had kept check float within historical bounds by
delaying availability in accordance with actual check presentment, the difference, all else
equal, would have emerged as an additional $40 billion demand for end-of-day credit.
Federal Reserve Bank lending could have presented far more difficult policy
issues had the shock coincided with significant weakness in the U.S. banking sector,
either due to pre-existing financial distress or with shock-induced losses that threatened
banks’ solvency. Central bank lending policy, to the extent that it provides qualifying
institutions with a backstop line-of-credit service on terms more advantageous than
private lines, inevitably seeks to balance the benefits of emergency liquidity provision
against the costs of “distortions in the price signals that are used to allocate resources,
induced excessive risk-taking, and, to limit the resultant moral hazard, greater
government supervision and regulation.” (Greenspan 2001) At times this will require
denying credit extension when to lend would postpone a warranted winding up and foster
future moral hazard. This can be difficult, because the short term calculus is generally
tilted toward leniency; the cost of enhanced moral hazard lies out into the future, while
the pressures to assist are immediate. (Goodfriend and Lacker 1999) At other times the
optimum is at the opposite corner solution, and lending is clearly in the public interest.
Arguably, September 11 was just such a case.
Had the banking sector been in some difficulty at the time of an interbank
payment disruption, it might have been difficult to sort out the deserving institutions –
44

A policy of putting every bank on the monitor in reject mode, although not necessarily with a zero cap, is
referred to as “universal real-time monitoring.” This policy was considered in the early 1990s and then
again in 2000. See Board of Governors of the Federal Reserve System (2001b).

31

solvent but illiquid – from the insolvent. Goodhart (2002) argues that it is not generally
possible to distinguish between illiquidity and insolvency, and that “nowadays illiquidity
implies at least a suspicion of insolvency.” He grants that there are some exceptions,
however, and cites the 1985 BoNY incident as an example because the bank was clearly
solvent, at least to supervisors. Prior to the incident, the bank’s condition was well
understood, and there was no apparent reason to believe that the incident itself had
directly caused losses large enough to threaten BoNY’s capital cushion. Just a few years
later, during the banking and savings and loan crisis, however, the chance that a
randomly chosen depository institution was in danger of serious distress was surely
nonnegligible. Had such a technological malfunction occurred at a bank to which the
Reserve Banks would not otherwise have wished to lend, policy makers would have
faced a choice between ameliorating the effects of the disruption and exacerbating moral
hazard or minimizing moral hazard but allowing the disruption to be resolved without
central bank assistance. The funds transfer outage in the midst of the 1987 stock market
crash demonstrates that such a confluence of events is not out of the question.
Lending policy could be difficult to manage during a payment system disruption
if the central bank wants to selectively control which institutions are granted credit.
Under the Federal Reserve’s current policy, this would require rejecting payments that
would otherwise send a bank into overdraft. Federal Reserve Bank officials continually
monitor the condition of individual account holders, particularly institutions in
deteriorating financial condition, and adjust daylight credit policy accordingly – for
example, they can place an institution on the real-time monitor, so that payments that
would cause an overdraft above a specified threshold are rejected. One problematic
scenario could occur if a significant number of institutions are in weakened financial
condition, but not yet weak enough to be on the monitor. An interbank payment
disruption that is associated with further deterioration in the financial condition of these
institutions, to the point that real-time monitoring would be warranted, could pose
problems. Newly restricting the payments of a number of deteriorated institutions in the
midst of the event could contribute to the disruption. The resulting rejected payments
would be added to the already elevated flow of items that required special managerial
attention. The Fed’s payments system regime relies fairly heavily on extensions of
intraday credit – larger by far than banks’ typical end of day balance. 45 And payments
processing relies heavily on standardized practice; payment flows are automated to
optimize given an established policy regime. The availability of intraday credit is a key
determinant of a bank’s payments processing choices; how much credit to grant
customers and counterparties, who to use as correspondent, who to clear through, and so
on. The sudden withdrawal of a bank’s intraday credit could require an abrupt
reorganization of the bank’s payments processing.
Because withdrawing intraday credit can impose costly adjustments on banks,
central banks will often find it a hard step to take. This will be particularly true in a crisis,
45

Aggregate average daylight overdrafts – the sum of all depository institutions’ average daylight
overdrafts – in August 2001 were $32.8 billion. (Coleman 2002) Peak daylight overdraft – the largest of the
aggregate banking system overdraft during a given day – averaged $92.9 billion across the month of
August 2001. Account balances averaged $15 billion.

32

when a central bank will be naturally hesitant to add any strain to an already strained
banking system. From the point of view of ex ante efficiency, a central bank might want
to have banks believe that should their condition deteriorate significantly enough,
daylight credit would be withdrawn, consistent with a general policy of withdrawing
safety net support from failing institutions. Providing daylight credit on liberal terms
during normal times, however, encourages banks to make investments and adopt
arrangements which commit them to large adjustment costs should daylight credit be
withdrawn. In the event of a disruption, the central bank then finds it difficult to impose
adjustment costs on the bank. Anticipating leniency, banks make investment decisions ex
ante that assume continued access to central bank credit ex post, an assumption that is,
because of their investments, confirmed.
This is a particular instance of the dilemma that Goodfriend and Lacker
(Goodfriend and Lacker 1999) discuss. They draw the parallel between central bank
lending and private line of credit lending, and note that private credit line contracts are
structured to provide the parties with incentives ex post to take actions that, from an ex
ante point of view, balance expected costs and benefits. How does central bank lending
differ from presumptively efficient private line of credit arrangements? In private
lending, the lender’s profit motive provides an incentive to withdraw the line when the
borrower’s credit worthiness has deteriorated. To prevent overly opportunistic (and
inefficient) exercise of this option, the contract prevents the lender from doing so unless
specific covenants have been violated or there has been a “materially adverse change” in
the financial condition of the borrower. And since withdrawal is not required in those
cases, the borrower and lender are free to renegotiate terms when it is mutually
beneficial. A central bank, in contrast, is not solely profit motivated. Heightened aversion
to being accused of mismanaging a crisis, Goodfriend and Lacker argue, can tilt central
bank incentives away from the long-run moral hazard concerns and toward minimization
of crisis-related dislocations.
The Federal Reserve’s daylight credit policy well illustrates this dilemma. The
current policy, with relatively limited use of real-time monitoring, cedes effective control
of lending decisions to account holders, who can obtain central bank credit in a crisis
simply by running up an intraday overdraft. Granted, Reserve Bank credit officers can
and do place troubled institutions on the monitor as their condition deteriorates. 46 During
a payment system disruption, however, events unfold quickly and there will be a natural
reluctance, for the reasons described above, for a Reserve Bank to place a bank newly on
the monitor to limit credit extension.
Note that moral hazard has a particularly concrete manifestation in this setting.
Operational preparedness is central to coping with payment system disruptions, and for
private sector banks it’s a classic risk-return issue. To the extent that central bank crisis
response eases the costs of the disruption, banks’ incentives to reduce the risk of such
46

Certain provisions of the Federal Deposit Insurance Corporation Improvement Act of 1991 (FDICIA) are
designed to discourage Federal Reserve lending to critically undercapitalized institutions and in some

circumstances it imposes losses on the Fed in the event a borrower fails. This encourages the use of
the monitor for such institutions.

33

disruptions will be to some degree muted. This tension between central bank crisis
response and private sector preparedness is at the heart of decades of payment system risk
policy around the world. Strenuous central bank pressure on the private sector to step up
efforts to control payment system risk suggests that central bankers believe that private
reliance on central bank crisis intervention tends to blunt incentives to mitigate risk.
Exacerbating the dilemma for the Federal Reserve is the fact that the prohibition
of interest on reserves dramatically enhances the value of daylight credit. 47 The current
regime of high costs for overnight credit and the relatively low cost (36 basis points at an
annual rate) of daylight credit makes it less costly for banks to minimize the foregone
interest cost of overnight reserve balances than if daylight credit were unavailable. This
reduces the deadweight societal loss banks incur to avoid the tax on overnight balances.
(Lacker 1997; Zhou 2000; Martin 2002) As a result, limiting daylight credit would be
costly in normal times; perhaps less so if interest were paid on reserves, however.
The notable lack of a significant flight to currency in the aftermath of September
11 limited the stress on the banking system, but this could turn out differently in a future
disruption episode. Bank runs have occurred as recently as the 1980s, and as in late
nineteenth and early twentieth century runs, they tended to be local affairs associated
with the insolvency of state-level deposit insurance funds or just the lack of deposit
insurance. Most interbank payment disruptions are not associated with run-like behavior
by consumers, which suggests that official assurances that technological malfunctions are
at fault have been viewed as credible. Should a disruption occur at a time of banking
sector weakness, consumers may have a more difficulty disentangling insolvency from
disruption-induced illiquidity. This would amplify the difficulties facing central bank
policy makers.
Nonpublic central bank information about a payment system participant, gleaned
either from bank regulatory agencies or from the central bank’s direct supervisory role,
can play an important role in the resolution of payment system disruptions. Bank
supervisors often have better information about a bank’s condition than market
participants. One element of the folklore surrounding the 1985 BoNY incident is the story
that their counterparties were skeptical of their claims that a software problem was to
blame, and obtained direct assurances from senior Federal Reserve officials that it was.
One can easily imagine why a supervisory relationship might provide sharper incentives
for truthful revelation than would an arm’s length relationship with market
counterparties. Information channeled through supervisory staff might therefore be more
credible than direct communications, and in a crisis this might facilitate efficiencyenhancing collaboration among market participants that would otherwise be discouraged
by mistrust.

47

Required clearing balances earn credits that can be applied to the purchase of Reserve Bank payments
services. The amount is set before the maintenance period begins, so marginal intra-period reserves do not
earn interest.

34

6.2. Interest rate policy
The role of interest rate policy in the aftermath of payment system disruptions has
received little attention. The direct role of interest rate reductions in addressing the
immediate payment system problems is probably fairly limited. The additional amount of
balances that were required to achieve a 50 basis point reduction in the overnight funds
rate was probably at least two orders of magnitude smaller than the amounts added the
previous week. 48 Fulfilling the need for additional balances to offset the effects of the
payments processing disruptions did not, by itself, require reducing interest rates.
The interest rate cuts following September 11 are probably best viewed as
addressing the medium- and longer-term macroeconomic consequences. The FOMC had
already cut the federal funds rate target in seven steps over the course of 2001 from 6.5
percent down to 3.5 percent in August. In the days immediately following the attacks, it
became apparent that they had had “considerable adverse repercussions on an already
weak economy.” (Board of Governors of the Federal Reserve System 2002b) Uncertainty
about the economic outlook widened significantly. In the subsequent weeks incoming
reports showed that consumer confidence dropped sharply, manufacturing activity
contracted further, and unemployment rapidly increased. (Goodfriend 2002)
Consequently, the FOMC cut its target for the federal funds rate by ½ percent steps on
September 17, October 2, and November 6, and then by ¼ percent on December 11,
bringing the rate down to 1¾ percent.
The first rate cut, adopted on a special conference call at 7:30 a.m. September 17,
was noteworthy. Unscheduled intermeeting policy moves are not common, but economic
developments sometimes warrant re-evaluating policy rather than waiting for the next
meeting. The September 17 meeting clearly fit that pattern. The timing of the meeting
was relatively unusual, however – intermeeting decisions are generally released during
regular business hours. Equity markets were scheduled to reopen Monday morning,
however, after having been closed since the attacks. Widespread commentary debated the
extent to which prices were likely to fall, given intervening declines on overseas markets.
The early hour of the call had the advantage of providing market participants with
information on the FOMC’s policy move sooner than if the Committee had waited until
normal business hours to convene. This may have helped dampen the fall in asset prices
Monday morning.
A tension arises when interest rate policy responds with alacrity to sudden
payment disturbances like September 11 or the market turmoil in September 1997
following the Russian debt default. The real economic effects of the shock, as opposed to
the immediate perturbation in the demand for reserves, are highly uncertain at first. And
yet financial asset prices respond rapidly. In the event that significant real effects
subsequently become evident, an immediate sequence of rate cuts ends up paying off. In
48

Estimated responses of excess reserves to changes in the target rate are typically $100 million or less.
Thus the amount of additional reserves needed to reduce the target on September 17 was unlikely to have
been larger than $100 million. In comparison, overnight credit, open market operations, check float, and
FX swap draws added well over $100 billion each day on September 12-14, three orders of magnitude
larger.

35

the event that adverse real effects turn out to be of smaller magnitude than expected,
some unwinding of the initial cuts may be warranted. If so, the desire to reverse field may
conflict with a desire to continue to foster market expectations that the central bank
“smooths” interest rate changes. (Goodfriend 1991) Central banks typically change their
target rate “through a series of small adjustments in the same direction, drawn out over a
period of months, rather than through an immediate once-and-for-all response to the new
development.” (Woodford 2003) As Goodfriend emphasizes, this increases the central
bank’s influence over longer term interest rates, but, as Woodford emphasizes, it requires
history-dependence in interest rate settings. That is, the ability of a change in the
overnight target rate to carry with it much of the short end of the yield curve requires that
market participants believe that it is not likely to be reversed. The tension following a
sudden shock, is that if adverse effects prove smaller than expected, the central bank will
be forced to choose between (1) accommodating expectations of smoothing but risking an
overly accommodative policy, and (2) responding to emerging economic developments
but eroding the central bank’s reputation for smoothing. (Cook and Korn 1991;
Goodfriend 1993)
Enhanced communication would appear to offer a means of mitigating this
tension. If the central bank could convince the public of the unusual nature of the shock,
then it could act more flexibly in the aftermath without altering expectations regarding
their behavior in normal times. For example, the statement accompanying an initial rate
cut in response to a liquidity disturbance could note that there is an unusual degree of
uncertainty regarding the warranted policy response, and that if conditions warrant a
reversal might be forthcoming relatively quickly. In fact, the minutes of the FOMC’s
September 17 meeting state:
These actions were taken against the backdrop of heightened concerns and
uncertainty created by the recent terrorist attacks and their potentially adverse
effects on asset prices and the performance of the economy.
This would seem to have encouraged market participants to believe that because a
relatively singular event had occurred, little by way of precedent was being set.

6.3. The Path to Recovery
One key issue policy makers faced in the days immediately following September
11 was how fast to attempt to restore normal market functioning. This issue arose most
critically in regard to the New York equity markets. Reopening would require costly
work to restore telecommunications functioning, and the safety of the work environment
in Lower Manhattan. Just as the equity markets “defiantly” resumed trading following the
1920 bombing, there was a powerful psychological desire to see the markets return to
normal functioning after September 11. For example, on the day of the attacks, Gary
Gensler, a former Treasury department official, said: “Keeping the markets closed shows
that terrorists brought you to bay, and it also creates more uncertainty.” (Schroeder 2001)
At a news conference on Wednesday, Richard A. Grasso, chairman of the New York

36

Stock Exchange, vowed that U.S. stock trading would resume no later that Monday. “‘It
is clearly the goal to bring this market up as quickly as is humanly possible,’ Grasso
said.” (Blustein and Day 2001) But there was a risk of bringing the markets back too
soon. If too few participants were functioning again, poor liquidity could hamper trading
and exacerbate the expected price declines. Moreover, physical conditions in Lower
Manhattan were unpleasant and potentially harmful. The symbolic value of a return to
normalcy was very attractive, however.
The Federal Reserve faced a resumption timing issue with regard to open market
operations. Supplying reserve needs was a high priority, but the restoration of normalcy
could have a reassuring effect. It took some time to engineer. The overhang of the
previous week’s reserve injections weighed down the funds rate at the beginning of the
week of the 17th: see Table 1. As the close of the maintenance period on Wednesday
approached, payment processing returned to normal and the movement of balances
around the banking system was less constrained. The enormous balance balances some
banks had held the week before meant that they had already met balance requirements
and now had an incentive to loan excess funds aggressively. The funds rate sagged below
the target until the maintenance period was over.
One striking aspect of the financial market’s path to recovery is the range of
cooperative behavior that was observed. Press reports on financial institutions’ efforts to
restore their operations were replete with descriptions of firms providing office space and
logistical support to help competitors get back on their feet. Cooperation was suggestive
of collusion at some points however. Accounts of foreign exchange market activity in the
two days immediately following the attacks reported on a “gentlemen’s agreement”
among large banks not to sell the dollar and to discourage such trades by speculative
accounts. Similar agreements were said to have been in effect in the federal funds and RP
markets on Wednesday. On Friday, September 14, the Washington Post reported:
In advance of the planned reopening of U.S. stock markets on Monday, major
securities firms and corporations have reached an extraordinary agreement to prop
up prices by buying shares if a flood of sell orders threatens to send markets into a
free fall, industry and government sources said yesterday. Federal securities
regulators have made it clear they will permit these and other market practices
that might raise legal questions in ordinary circumstances, the sources said. (Day
and Berry 2001)
Such seemingly altruistic behavior could have stemmed from the patriotic spirit of the
days after the attack, or it could have represented self-interested efforts to prevent asset
price dips that they believed would be fear-driven and temporary. In the foreign exchange
case, what was effectively achieved was a phased reopening, in which the market
reopened for everyone but speculative short sellers at first. The stock market agreement,
if true, would represent a collective market-making effort, a coordinated, SECsanctioned, version of the widespread stock buy backs following the 1987 stock market
crash, somewhat analogous to the coordination of collective efforts through
clearinghouse actions in response to nineteenth century U.S. banking crises.

37

7. Conclusion
The September 11, 2001 terrorist attacks, whose human consequences were so
horrendous, had monetary and payment system consequences that are worth examining.
This review has brought together disparate source material pertaining to these
consequences. Just two years after the events there is a wide array of information
available, but this treatment is hardly exhaustive. It relied mainly on published sources,
especially press reports, for nonquantitative information, and on nonproprietary official
statistics for quantitative data. More surely will emerge as time passes, memoirs are
published, and further confidential material is made public.
A key idea that emerges from this study is that disruptions to interbank payment
arrangements, whether due to technological impediments or credit quality concerns, have
been central to several banking crises and are likely to recur. Putting interbank payments
at center stage offers an informative view of past and present banking crises and
payments system disruptions, which are often thought of as distinct phenomena. It takes
us beyond somewhat amorphous discussions of “liquidity,” and casts light on the
functioning of specific economic arrangements. A focus on interbank payments also
highlights the common issues for lender of last resort policy that come into play in both
types of events, particularly the difficulties associated with limited commitment. The
fine-grained structure of central bank credit policy is critical to how an economy
responds to a shock to the monetary and financial systems of this magnitude and
destructiveness.

38

Table 1. Factors Affecting Account Balances of Depository Institutions, September 10-21, 2001
End of day balances, billion $
Date
Sep 10
Sep 11
Sep 12
Sep 13
Sep 14
Sep 17
Sep 18
Sep 19
Sep 20
Sep 21

Repos
Term
Overnight
23
0
23
0
23
38
14
70
14
81
12
12
12
33
33

Check float
1
4
23
47
44

Swap draws
0
0
5
20
9

Currency
-611
-613
-616
-615
-615

Other
601
595
585
577
578

12
9
4
3
1

0
0
0
0
0

-615
-616
-615
-614
-612

579
578
584
583
588

Check float
0
3
23
47
43

Swap draws
0
0
5
15
-11

Currency
0
-2
-4
-4
-3

Other
0
-6
-16
-24
-24

11
8
4
2
0

0
0
0
0
0

-4
-5
-4
-2
-1

-22
-23
-17
-18
-14

57
36
28
7
1

Overnight credit
Discount
Overdrafts
0
0
37
2
46
4
8
0
0
0
0
0
3
1
2

Balances
13
47
109
121
111

0
0
0
0
0

45
19
15
13
12

Overnight credit
Discount
Overdrafts
0
0
37
2
46
4
8
0
0
0

Balances
0
33
95
108
98

Cumulative change from September 10, billion $
Date
Sep 10
Sep 11
Sep 12
Sep 13
Sep 14
Sep 17
Sep 18
Sep 19
Sep 20
Sep 21

Repos
Term
Overnight
0
0
0
0
0
38
-9
70
-9
81
-11
-11
-11
10
10

57
36
28
7
1

0
0
3
1
2

0
0
0
0
0

32
6
2
0
-1

Source: L.6.1, Board of Governors of the Federal Reserve System. Balances are the deposits of depository institutions with the Federal
Reserve Banks, and include required clearing balances. Check float equals "Items in process of collection" minus "Deferred availability cash
items" from the Consolidated Statement of Condition of the Federal Reserve Banks. Swap draws are the amounts foriegn central bank utilized
under the foriegn exchange swaps announced during the week of September 11. Currency is currency in circulation; it drains reserves and
thus has a negative effect on reserve balances. The "Other" factor aggregates the net effect of other Federal Reserve assets and liabilities,
and consists predominantly of the System's holdings of U.S. government securities.

39

Table 2. Summary Results of Repurchase Financing, September 12-19, 2001
Date
Sep 12
Sep 13
Sep 14
Sep 17
Sep 18
Sep 19

Total Props

Accepted Props

(Billion $)

(Billion $)

46.25
70.20
81.25
59.55
37.75
27.60

38.25
70.20
81.25
57.25
36.25
27.60

High Bid

Low Bid

Stop-Out

Weighted Avg

(Financing Rates for Overnight Repurchase Agreements)

3.60
4.00
3.75
3.15
2.30
1.20

3.25
3.50
3.50
2.90
1.75
0.75

3.50
3.50
3.50
3.00
2.00
0.75

3.54
3.60
3.54
3.07
2.16
1.00

Table 3. Selected Interest Rates, September 6 to 21, 2001

Date
Sep 6
Sep 7

Effective Fed
Funds Rate
3.52
3.44

Fed Funds Futures Contract Rates*
Sept. 2001 Oct. 2001 Nov. 2001
3.515
3.330
3.250
3.455
3.215
3.240

Sep
Sep
Sep
Sep
Sep

10
11
12
13
14

3.5
3.5
3.56
3.31
3.13

3.485
3.470

3.195
3.130

3.110
3.050

3.370
3.340

2.870
2.800

Sep
Sep
Sep
Sep
Sep

17
18
19
20
21

2.13
1.25
1.19
2.22
3.11

3.200
3.020
2.800
2.805
2.890

2.680
2.615
2.460
2.530
2.510

EuroDollar Futures Contract Rates^
Dec. 2001 Mar. 2002 Sept. 2002
3.440
3.535
4.160
3.265
3.360
4.020

2 Yr US
10 Yr US
Govt Yield Govt Yield
3.637
4.873
3.505
4.790

3.360
3.150
3.395
2.945
2.840

4.035
3.890
4.065
3.665
3.545

3.504
3.517

4.835
4.809

2.800
2.725

3.280
3.100
3.270
2.875
2.795

2.984
2.866

4.623
4.553

2.585
2.510
2.360
2.390
2.410

2.680
2.640
2.440
2.525
2.455

2.765
2.735
2.555
2.650
2.605

3.580
3.615
3.455
3.535
3.510

2.939
2.947
2.814
2.879
2.867

4.623
4.707
4.691
4.742
4.691

* Source: Chicago Board of Trade; Rates are end of day
^ Source: Chicago Mercantile Exchange; Rates are end of day

40

Figure 1
Account Balance Distribution, August 1 - September 20, 2001
Millions $

2,000
9/11

1,500

1,000

500

0
8/1

8/8
Percentiles:

8/15

8/22

99.9th

8/29

99.8th

99.7th

9/5
99.6th

9/12

9/19

99.5th

Thousands $
900
800
9/11
700
600
500
400
300
200
100
0
8/1

8/8
Percentiles:

8/15

8/22
90th

80th

8/29
60th

9/5
40th

Source: Daylight Overdraft Reporting and Pricing System and author's calculations.

41

9/12

9/19

Figure 2
Retail Chain Store Sales Index
1977 = 100

405

400

395

390

385

380

375
4-Aug

11-Aug 18-Aug 25-Aug

1-Sep

8-Sep

Source: Bank of Tokyo-Mistubishi, Ltd and UBS Warburg.

42

15-Sep

22-Sep

29-Sep

6-Oct

13-Oct

20-Oct

27-Oct

Figure 3
Cumulative Changes in Components of Currency in Circulation from
September 10, 2001
Billion $
5

Currency in circulation
Currency outside banks

4
3
2
1
0
-1
-2
-3

Vault cash
-4
-5
9/10 9/11 9/12 9/13 9/14 9/15 9/16 9/17 9/18 9/19 9/20 9/21 9/22 9/23 9/24 9/25 9/26 9/27 9/28 9/29 9/30
Source: FAME database and EDDS, Board of Governors. Vault cash: held at large domestic commercial banks. Currency outside
banks is Currency in circulation minus vault cash.

Figure 4
Currency in Circulation, 2001
Billion $

Billion $

630

120

620

110
Currency in Circulation, left axis

610

100

600

90
9/11

590

80

580

70

Currency Outside Banks, left axis

570

60

560

50
Vault Cash, right axis

550

40

540

30

530
8/5

8/12

8/19

8/26

9/2

9/9

9/16

Source: FAME Database and EDDS, Board of Governors.
Vault cash: held at large domestic commercial banks.

43

9/23

9/30

10/7

10/14

10/21

20
10/28

Figure 5
Daily Fedwire funds and book-entry securities transfers and daylight overdrafts for September
10-21, 2001, compared with August 2001
Transactions
500,000

Billion $
2,500

Funds

450,000
400,000

2,000

350,000
300,000

1,500

250,000
200,000

1,000

150,000
100,000

500

50,000
0

0
AUGUST

9/10

9/11

9/12

9/13

9/14

9/17

9/18

9/19

9/20

9/21

Securities
Transactions
180,000

Billion $
2,500

160,000
140,000

2,000

Volume, left axis
Value, right axis

120,000

1,500

100,000
80,000

1,000

60,000
40,000

500

20,000
0

0
AUGUST

9/10

9/11

9/12

9/13

9/14

9/17

9/18

9/19

9/20

9/21

Daylight Overdrafts
Billion $
Peak
Average

160

160

140

140

120

120

100

100

80

80

60

60

40

40

20

20

0

0
AUGUST

9/10

9/11

9/12

9/13

9/14

9/17

9/18

9/19

9/20

9/21

Notes: The PEAK daylight overdraft is determined by adding the account balances of all depository institutions in a negative position for each minute during the day
and then selecting the largest negative end-of-minute balance. The aggregate AVERAGE daylight overdraft is the sum of all DI's average daylight overdrafts. A DI's
average daylight overdraft is the sum of negative end of minute balances divided by total minutes in the standard Fedwire operating day. August is a monthly average
of the daily data. Source: Coleman (2002).

44

Figure 6
Federal Funds Rates around September 11, 2001: High, Low, and Effective Rates
Percent

7

6

5
High Rate
Effective Rate
4

3
Target Level
2

1

Low Rate
0
8/23 8/24 8/27 8/28 8/29 8/30 8/31 9/4 9/5 9/6 9/7 9/10 9/11 9/12 9/13 9/14 9/17 9/18 9/19 9/20 9/21 9/24 9/25 9/26 9/27 9/28 10/1 10/2 10/3
Source: Markets Group, Federal Reserve Bank of New York

45

References
Allison, Melissa. 2001. Fed Stresses it's open; banks report no big woes. Chicago
Tribune, September 12.
Bach, Deborah. 2001. Will the Next Strike Aim at the Financial System? American
Banker, December 26, 9.
Bagehot, Walter. 1991. Lombard Street: A Description of the Money Market. New York:
Orion. Original edition, New York: Scribner, Armstrong, 1873.
The Bank of New York's ATM Network Restored. 2001. PR Newswire, September 20.
Becker, Joseph D. 1976. International Insolvency: The Case of Herstatt. American Bar
Association Journal 62:1290-95
Beckett, Paul, and Greg Ip. 2001. Reopening Wall Street: Bank of New York Clears
Systems That Had Failed. Wall Street Journal, September 17, C2.
Beckett, Paul, and Jathon Sapsford. 2001. Rebuilding Wall Street: How Wall Street's
Nervous System Caused Pain. Wall Street Journal, September 21, C1.
Bernanke, Ben S. 1990. Clearing and Settlement During the Crash. Review of Financial
Studies 3 (1):133-51
Bills, Steve, David Breitkopf, and Jeff Green. 2001. Concord: Americans' ATM Usage
Spiked After Attacks. American Banker, September 18, 8.
Bin Laden hails economic losses from Sept. 11 attacks. 2002. The Miami Herald, April
18, F1.
Blackwell, Rob. 2001. Task Force Established. American Banker, September 18, 1.
Blustein, Paul, and Kathleen Day. 2001. Industrialized Nations Act to Reassure World.
Washington Post, September 13, E01.
Board of Governors of the Federal Reserve System. 2001a. Guide to the Federal
Reserve's Payments System Risk Policy. May. Washington, D.C.
———. 2001b. Policy Statement on Payments System Risk: Potential Longer-Term
Policy Direction. May 30. Washington, D.C.
———. 2001c. Press Release. September 14. Washington, D.C.: Board of Governors of
the Federal Reserve System.

46

———. 2002a. Extensions of Credit by Federal Reserve Banks. Final rule, technical
amendment. October 31. Washington, D.C.
———. 2002b. Monetary Policy Report to Congress. Federal Reserve Bulletin 88
(3):141-172
Board of Governors of the Federal Reserve System, Federal Deposit Insurance
Corporation, and Office of the Comptroller of the Currency. 2001. Joint Press
Release: Shared National Credit Review. October 5. Washington, D.C.
Bond Market Association. 2001. Minutes of Emergency Meetings of The Bond Market
Association: September 11-21, 2001. New York, NY: Bond Market Association.
Breitkopf, David. 2001. Big Banks, Networks Saw No Cash Stampede. American
Banker, September 21, 1.
———. 2003. Who Patched? B of A Hardest Hit By Virus. American Banker, 1.
Brooks, John. 1969. Once in Golconda. New York: Harper & Row.
Caldwell, Bruce. 1995. Hacking spree targets Citibank. InformationWeek, 20.
Calomiris, Charles W., and Gary Gorton. 1991. The Origins of Banking Panics: Models,
Facts, and Bank Regulation. In Financial Markets and Financial Crises, edited by
R. G. Hubbard. Chicago: University of Chicago Press.
Cannon, James G. 1901. Clearing-Houses: Their History, Methods and Administration.
London: Smith, Elder, & Co.
Chaffin, Joshua, and Gary Silverman. 2001. Assault on America Corporate Impact.
Financial Times, September 21.
Coffield, Dana. 2001. Your network is at risk. Interactive Week from ZDWire, October
19.
Coleman, Stacy Panigay. 2002. The Evolution of the Federal Reserve's Intraday Credit
Policies. Federal Reserve Bulletin 88 (2):67-84
Cook, Timothy, and Steven Korn. 1991. The reaction of interest rates to the employment
report: The role of policy anticipations. Federal Reserve Bank of Richmond
Economic Review 77 (5):3-12
Corrigan, E. Gerald. 1990. Statement. U.S. Senate, 101st Congress, Second Session,
Committee on Banking, Housing, and Urban Affairs. May 3. Washington, D.C.:
U.S. Government Printing Office.

47

Costa, Thomas F. 2001. Important Notices. September 12 and following. New York, NY:
Government Securities Clearing Corporation.
Cowan, Lynn. 2001. Bank of New York Still Had Some Systems Problems Monday.
Dow Jones News Service, September 17.
Danielyan, Artyon. 2001. Russian rouble slips, trade upset by U.S. woes. Reuters News,
September 13.
Day, Kathleen, and John M. Berry. 2001. Stocks to trade Monday With Special Rule; Big
Firms Can Buy Shares Back to Halt Plunge. Washington Post, September 14,
E01.
Edwards, Will. 2001. Fed banks Deliver Checks By Truck, See Demand for Cash Rise.
Bloomberg News, September 13.
Federal Deposit Insurance Corporation. 2001. FDIC Quarterly Banking Profile.
Washington, D.C.
Federal Reserve Bank of New York. 2003. A Technical Change to Certain Temporary
Open Market Operations. Press releasae. September 4. New York: Federal
Reserve Bank of New York.
U.S. Congress, House of Representatives. 1985. Subcommittee on Domestic Monetary
Policy of the Committee on Banking, Finance and Urban Affairs. The Federal
Reserve Bank of New York Discount Window Advance of $22.6 Billion Extended
to the Bank of New York. Ninety-Ninth Congress, First Session. December 12,
1985.
Financial and Banking Information Infrastructure Committee. 2003. Impact of the Recent
Power Blackout and Hurricane Isabel on the Financial Sector. October 20: U.S.
Department fof Treasury, President's Working Group on Financial Markets.
Fitch Insurance Conference Call Summary. 2001. Business Wire, September 21.
Fleming, Michael J., and Kenneth D. Garbade. 2002. When the Back Office Moved to the
Front Burner: Settlement Fails in the Treasury Market after 9/11. Federal Reserve
Bank of New York Economic Policy Review 8 (2):35-57
———. 2003. The Repurchase Agreement Refined: GCF Repo®. Federal Reserve Bank
of New York Current Issues in Economics and Finance 9 (6)
Garsson, Robert M. 1988. Payment-System Glitch Complicated Stock Market Crash.
American Banker, January 14, 1.

48

Gellman, Barton. 2002. Cyber-Attacks by Al Qaeda Feared. Washington Post, June 27,
A01.
General Accounting Office. 2003. Potential Terrorist Attacks: Additional Actions Needed
to Better Prepare Critical Financial Market Participants. February. Washington,
DC: U.S. General Accounting Office.
Gibbons, J.S. 1859. The Banks of New York, their Dealers, the Clearing-house, and the
Panic of 1857. New York: D. Appleton and Co.
Gibbons, Todd. 2001. Emerging Stronger from 9/11/01: An Interview with Todd
Gibbons. Journal of Lending & Credit Risk Management 84 (4):8
Gilbert, Mark, and Bethan Thomas. 2001. Attack on America: Economic Outlook. South
Florida Sun-Sentinel: Bloomberg News, September 13, 3D.
Goldenberg, Liz. 2001. Companies Sell Commercial Paper After the Attack: Rates of
Return. Bloomberg News, September 26.
Goldenberg, Liz, and Helen Stock. 2001. U.S. Interbank Payment Systems Running;
Some Delays (Update 2). Bloomberg News, September 12.
Goodfriend, Marvin. 1991. Interest Rates and the Conduct of Monetary Policy. CarnegieRochester Conference Series on Public Policy 34:7-30
———. 1993. Interest rate policy and the inflation scare problem: 1979-1992. Federal
Reserve Bank of Richmond Economic Quarterly 79 (1):1-24
———. 2002. The Phases of U.S. Monetary Policy: 1987 to 2001. Federal Reserve Bank
of Richmond Economic Quarterly 88 (4):1-17
Goodfriend, Marvin, and Robert G. King. 1988. Financial Deregulation, Monetary
Policy, and Central Banking. Federal Reserve Bank of Richmond Economic
Review 74 (May/June):3-22
Goodfriend, Marvin, and Jeffrey M. Lacker. 1999. Limited Commitment and Central
Bank Lending. Federal Reserve Bank of Richmond Economic Quarterly 85
(Fall):1-27
Goodhart, Charles. 2002. Myths About the Lender of Last Resort. In Financial Crises,
Contagion, and the Lender of Last Resort, edited by C. Goodhart and G. Illing.
Oxford: Oxford University Press. Original edition, International Finance 2 (3),
1999.
Green, Micah S. 2003. Testimony of Micah S. Green, President, The Bond Market
Association. In Hearing Before the Subcommittee on Capital Markets of the

49

Committee on Financial Services, United States House of Representatives,
Recovery and Renewal: Protecting the Capital Markets Against Terrorism Post
9/11. February 12.
Greenspan, Alan. 2001. The Financial Safety Net. In 37th Annual Conference on Banking
Structure and Competition of the Federal Reserve Bank of Chicago. May 10.
Chicago Il.
Groshen, Erica L., Linda S. Goldberg, James A. Kahn, Hamid Mehran, Donald P.
Morgan, and Carol L. Osler. 2002. Introduction. Federal Reserve Bank of New
York Economic Policy Review 8 (2):1-4
Guerra, Anthony. 2001. The Buck Stopped Here: BONY's Disaster Recovery Comes
Under Attack. Wall Street + Technology, November 1, 48.
Guth, Robert A. Fields, Gary. 2003. FBI Averts Fresh Computer Attack. In Wall Street
Journal - Eastern Edition: Dow Jones & Company Inc.
Hamilton, James D., and Òscar Jordà. 2002. A Model of the Federal Funds Rate Target.
Journal of Political Economy 110 (5):1135-67
Humphrey, Thomas M., and Robert E. Keleher. 1984. The Lender of Last Resort: A
Historical Perspective. CATO Journal 4 (1). Reprinted in Charles Goodhart and
Gerhard Illing, Eds. Financial Crises, Contagion, and the Lender of Last Resort:
A Reader. Oxford, Oxford University Press: 2002.
In the Shadow of Tragedy, U.S. Payment System Stands Strong. 2001. EFT Report,
September 19.
Ip, Greg, and John D. McKinnon. 2001. Terrorists Destroy World Trade Center, Hit
Pentagon in Raid With Hijacked Jets -- Attacks Raise Fears of a Recession. Wall
Street Journal, September 12, A1.
Ip, Greg, G. Thomas Sims, and Paul Beckett. 2001. Fed Pumps Cash to Banking System
to Help Meet Spiking Demand. Wall Street Journal, September 13, C1.
Ip, Greg, and Jim VandeHei. 2001. Economic Front: How Policy Makers Regrouped to
Defend the Financial System. Wall Street Journal, September 18, A1.
Kahn, Charles M., and William Roberds. 2001. The CLS Bank: A Solution to the Risks
of International Payments Settlement? Carnegie-Rochester Conference Series on
Public Policy 54:191-226
Kaserer, Christoph. 2000. Der Fall der Herstatt-Bank 25 Jahre danach. Vierteljahrschrift
fur Sozial- und Wirtschaftsgeschichte 86 (2):166-192

50

Kite, Shane. 2003. In Blackout, BCP Focus Pays Off. Securities Industry News, 1.
Krane, Jim. 2001. Laboring to Reconnect Wall Street to the world. Associated Press
Newswires, September 13, 2001.
Lacker, Jeffrey M. 1997. Clearing, Settlement and Monetary Policy. Journal of Monetary
Economics 40 (2):347-81
———. 2001. The CLS Bank: A Solution to the Risks of International Payments
Settlement? A Comment. Carnegie-Rochester Conference Series on Public Policy
54 (0):227-33
Lee, W.A. 2003. Post-Worm, Louder Call for United Solution. American Banker, 2003,
9.
Lower Manhattan Telecommunications Users' Working Group. 2002. Building a 21st
Century Telecom Infrastructure. August.
Lumpkin, Stephen A. 1998. Repurchase and Reverse Repurchase Agreements. In
Instruments of the Money Market, edited by T. Q. Cook and R. K. Laroche.
Richmond, Virginia: Federal Reserve Bank of Richmond.
Mackenzie, Michael. 2001. Rebuilding Treasury Market Infrastructure: A Work in
Progress. Capital Markets Report, September 16.
MacRae, Desmond. 2001. Looking Into the Abyss. Global Custodian, Winter, 70-6.
Mandaro, Laura. 2001. Fed, FDIC, OCC: System is Sound. American Banker, September
13, 1.
Markets Group of the Federal Reserve Bank of New York. 2002. Domestic Open Market
Operations During 2001. February. New York: Federal Reserve Bank of New
York.
Martin, Antoine. 2002. Optimal Pricing of Intra-Day Liquidity, Federal Reserve Bank of
Kansas City, Kansas City.
McAndrews, James J., and Simon Potter. 2002. Liquidity Effects of the Events of
September 11, 2001. FRBNY Economic Policy Review 8 (1):59-79
McLaughlin-Moore, Linda. 2002. One Year Later. Interview by Ivan Schneider. Bank
Systems & Technology, September, 40-46.
Meulendyke, Ann-Marie. 1998. U.S. Monetary Policy and Financial Markets. New York,
NY: Federal Reserve Bank of New York.

51

Mollenkamp, Carrick, Will Pinkston, and Jacob Schlesinger. 2001. U.S. Bank Branches
Close in Aftermath of Terror Attacks -- Check Processing Disrupted As All Air
Traffic Stops. Asian Wall Street Journal, September 13.
New York Clearing House. 2001. New York Clearing House Issues Statement on Stability
of Private Sector Electronic Payments Networks. Press release. September 13.
New York, NY.
Office of the Comptroller of the Currency. 2001. News Release. October 5. Washington,
D.C.: U.S. Department of Treasury.
Padgett, Tania. 2001. Attack Creates Problems for Banks' ATM Services. American
Banker, September 20, A79.
Parry, John. 2001. Online Bond Systems Given Chance to Shine After Attacks. Capital
Markets Report, September 24.
Remolona, Eli M., Richard Cantor, M. Ellen Gaske, Lauren A. Hargraves, Lisa B.
Schwartz, and Victoria E. Stein. 1990. How Safety Nets Work. Central banking
(2):61-72
Rosenberg, Martin. 2001. Verizon Reconnects. Utility Business, October 1.
Russian rouble firms on low liquidity. 2001. Reuters News, September 18.
Schroeder, Michael. 2001. Attack Shuts Down U.S. Markets and Causes Global Declines.
Wall Street Journal, September 12, B1.
Shephard, Bill. 2002. Unsnarling the Mess. Investment Dealers Digest, May 13.
Spillenkothen, Richard. 2001. Supervisory Letter SR 01-24. October 14. Washington,
D.C.: Board of Governors of the Federal Reserve System.
Sprague, Oliver M. 1910. History of Crises Under the National Banking System, U.S.
National Monetary Commission, Senate Document No. 538, 61 Cong. 2 Sess.
Washington, D.C.: Government Printing Office.
State of Connecticut Department of Banking. 2001a. Banks to Re-Open. September 11.
———. 2001b. Commissioner Requests that Banks Close. September 11.
The Presidential Task Force on Market Mechanisms. 1988. Report of the Presidential
Task Force on Market Mechanisms. January 8. New York, N.Y.

52

Timberlake Jr., Richard H. 1984. The Central Banking Role of Clearinghouse
Associations. In Journal of Money, Credit & Banking: Ohio State University
Press.
VandeHei, Jim, Jeanne Cummings, David Rogers, Shailagh Murray, and John Harwood.
2001. A Day of Terror: Government Officials Relocate, Keep Working. Wall
Street Journal, September 12, A20.
Walter, John R. 1996. Firewalls. Federal Reserve Bank of Richmond Economic Quarterly
82 (4):15-39
Watkins, Leonard L. 1929. Banker's Balances. Chicago: A. W. Shaw.
Weinberg, John A. 1995. Cycles in Lending Standards? Federal Reserve Bank of
Richmond Economic Quarterly 81 (3):1-18
Weiss, Todd R. 2003. New Bugbear Virus Targets Financial Institutions. In
Computerworld: Computerworld.
Wicker, Elmus. 2000. Banking panics of the gilded age. Cambridge: Cambridge
University Press.
Woodford, Michael. 2003. Optimal Interest-Rate Smoothing. Review of Economic
Studies 70 (4):861-886
Zhou, Ruilin. 2000. Understanding intraday credit in large-value payment systems.
Federal Reserve Bank of Chicago Economic Perspectives:29-44
Zolkos, Rrodd. 2003. Multibillion loss estimates remain firm. Business Insurance 37
(37):34

53