Full text of Review (Federal Reserve Bank of St. Louis) : January/February 1992

View original document

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

3 I m p l ic a t io n s o f N e ttin g A r r a n g e
m e n t s f o r B a n k R is k in F o r e i g n
E x ch a n g e T ra n sa ctio n s
1 7 I n s t i t u t i o n a l D e v e l o p m e n t s in t h e
G lo b a liz a t io n o f S e c u r i t i e s a n d
F u tu re s M a rk e ts
3 1 D a t a E n v e l o p m e n t A n a ly s is a n d
C o m m e rcia l B a n k P e r fo r m a n c e : A
P r i m e r W i t h A p p lic a t io n s to
M isso u ri B a n k s

THE
FEDERAL
A RESERVE
RANK of
A r ST.IjOIIIS

F e d e r a l R e s e r v e B a n k o f St. L o u is

R e v ie w
January/February 1992

In T h is Issu e . . .

In the first article of this Review, “Implications of Netting A rrange
m ents for Bank Risk in Foreign Exchange T ransactions,” R. Alton Gilbert
describes th e risks assum ed by banks in settling foreign exchange tran s
actions w ith other banks. The risks involve default by the other parties
to the transactions. As the author notes, the volume of transactions in
the foreign exchange m arket is very high, and banks commonly engage
in transactions w ith counterparties headquartered in other countries.
Thus, a default by a m ajor participant in the foreign exchange m arket
could affect the operation of paym ents systems in several countries.
Central banks have interest in the design of any arrangem ents am ong
banks th at m ight reduce their risk in settling foreign exchange transac
tions. One way that banks may be able to reduce transaction costs and
risks is through netting arrangem ents. The central banks of 10 deve
loped countries recently issued a report on netting arrangem ents, w hich
included a list of guidelines for their design. Gilbert examines the impli
cations of netting arrangem ents for risk assum ed by banks in settling
foreign exchange transactions and indicates w hy some of the guidelines
are im portant if netting arrangem ents are to reduce risk.
***
Financial transactions, like the buying and selling of securities, com
modities, foreign exchange and bonds, have increasingly involved in
dividuals and firm s from different countries. In the second article of the
Review, "Institutional Developm ents in the Globalization of Securities and
Futures M arkets,” Jodi G. Scarlata describes recent institutional develop
m ents in this globalization and discusses the advantages and disadvan
ta g e s of th e s e c h a n g e s . S u b s ta n tia l b e n e fits , s h e n o te s , are o c c u r r in g
because of these developm ents. At the same time, dom estic rules and
regulations are not sufficient safeguards for m any international trades.
In particular, she stresses how some risks at various stages of the clear
ing and settlem ent process are m ore im portant in an international set
ting than in a strictly dom estic setting.
W eaknesses in the clearing and settlem ent system have prom pted
w orld financial leaders to w ork tow ard global coordination. Scarlata
concludes th at significant steps rem ain in integrating the w orld’s grow
ing securities and futures m arkets.
***
In the third article in this issue, “Data Envelopm ent Analysis and Com
m ercial Bank Perform ance: A Prim er w ith Applications to Missouri
Banks,” Piyu Yue discusses a relatively new m ethodology for evaluating
JANUARY/FEBRUARY 1992

the technical or productive efficiency of business enterprises. The
m ethodology is called Data Envelopm ent Analysis or DEA. After discuss
ing the distinction betw een technical and economic efficiency, the
author explains w hat DEA is and how it can be used to partition a
group of firm s into those that are DEA-efficient and those th at are not.
She illustrates the usefulness of the technique w ith data for 60 Missouri
com m ercial banks for the period 1984-90.
***

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

R. Alton Gilbert
R. Alton Gilbert is an assistant vice president at the Federal
Reserve Bank of St. Louis. Richard I. Jako provided research
assistance.

Implications of Netting
Arrangements for Bank Risk
in Foreign Exchange Trans
actions

H HE MAJOR FINANCIAL institutions of m any
nations are active participants in the m arket for
foreign exchange. The exchanges of currencies
that take place through this m arket facilitate in
ternational trade and the international flow of
capital for investm ents.
The volume of transactions in the foreign ex
change m arket—already very large—has grow n
rapidly in recent years. As of April 1989, the
date of the last international survey, foreign ex
change transactions had an average value of
$640 billion per business day.
W ith dollar am ounts in this lofty range, p ar
ticipants in the foreign exchange m arket could
incur substantial losses if the other parties to
their transactions w ere to default on the pay
m ents required to settle their side of the tran s
'Netting agreements between pairs of banks may apply to
payments in settlement of transactions other than foreign
exchange. This paper, however, limits analysis to the net
ting of foreign exchange transactions. All participants in
the foreign exchange market are called banks to simplify
exposition. In some markets, the important participants in
clude firms that are not banks. See Federal Reserve Bank

actions. To reduce the costs of transactions and
limit the size of these possible losses, some
banks engage in bilateral netting of their for
eign exchange transactions .1 In bilateral netting,
tw o banks exchange daily only the net units of
currencies in the transactions betw een them .
Some groups of banks have also studied the
possibility of m ultilateral arrangem ents for net
ting foreign exchange transactions, though none
are in operation at this tim e .2 M em bers of a
m ultilateral netting arrangem ent w ould settle
transactions w ith each other by m aking pay
m ents to a clearing house for th eir net position
in each currency w ith the other m em bers.
As part of their responsibility to avoid disrup
tions in the operation of paym ent systems, cen
tral banks have a strong interest in such netting
of New York (1989) and Bank of England (1989). See glos
sary on page 14 for definition of netting and other terms
used in this paper.
2See Deeg (1990), Duncan (1991), Luthringhausen (1990)
and Polo (1990).

JANUARY/FEBRUARY 1992

Table 1
Minimum Standards for the Design and Operation of CrossBorder and Multi-Currency Netting and Settlement Schemes
I.

Netting schemes should have a well-founded legal basis under all relevant jurisdictions.

II.

Netting scheme participants should have a clear understanding of the impact of the particular
scheme on each of the financial risks affected by the netting process.

III.

Multilateral netting systems should have clearly defined procedures for the management of credit
risks and liquidity risks which specify the respective responsibilities of the netting provider and
the participants. These procedures should also ensure that all parties have both the incentives
and the capabilities to manage and contain each of the risks they bear and that limits are placed
on the maximum level of credit exposure that can be produced by each participant.

IV.

Multilateral netting systems should, at a minimum, be capable of ensuring the timely completion
of daily settlements in the event of an inability to settle by the participant with the largest single
net-debit position.

Multilateral netting systems should have objective and publicly disclosed criteria for admission
which permit fair and open access.

VI.

All netting schemes should ensure the operational reliability of technical systems and the availa
bility of back-up facilities capable of completing daily processing requirements.

SOURCE: Bank for International Settlements (1990c).

arrangem ents .3 Since foreign exchange transac
tions often involve parties headquartered in
different countries, a default by one participant
is likely to affect those in other countries. Banks
adversely affected by such defaults typically
w ould tu rn to their central banks for assistance
in coping w ith liquidity problem s.
In recent years, prom oters of interbank net
ting arrangem ents have requested the views of
central banks individually on projects that ap
peared to have implications for a num ber of
countries. The central banks of 10 m ajor indus
trialized countries recently issued a joint state
m ent, through the Bank for International Set
tlem ents, about the netting of foreign exchange
transactions. This is commonly called the "Lamfalussy Beport,” nam ed after the com m ittee
chairm an w ho drafted the report. The com m it
tee expressed concern about the risks involved
in settling foreign exchange transactions and
discussed the potential benefits and draw backs
of netting arrangem ents. The central bankers
listed m inim um standards for the design of net
3See Summers (1991) for a discussion of the role of central
banks in the operation of payment systems.
4Bank for International Settlements (1990c).
5Cody (1990) also provides an introduction to the risk in
settling foreign exchange transactions and the implications

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

ting arrangem ents for bankers w ho may develop
them (see table l ).4
This paper illustrates the risk in settling for
eign exchange transactions and the risk implica
tions of netting, using a hypothetical example of
transactions am ong three banks. This exercise
illustrates how netting m ay reduce risk, if net
ting arrangem ents conform to the guidelines in
the Lamfalussy R eport .5

THE MARKET FOR FOREIGN EX
CHANGE
A foreign exchange transaction is an agree
m ent by tw o parties (generally large banks) to
exchange currencies on a given date, called the
value date of the transaction. The m ost com m on
type of transaction betw een participants in the
foreign exchange m arket, a spot transaction, is
an agreem ent betw een tw o parties to exchange
units of currencies tw o business days from the
date the transaction is negotiated. A transaction
w ith a value date m ore than tw o days after the
of netting. See Juncker, Summers and Young (1991) for a
general discussion of the issues raised by netting ar
rangements.

date of negotiation is called a forw ard transac
tion. Some forw ard transactions have value
dates m ore than a year into the future, but
m ost call for settlem ent w ithin a m onth. Several
other types of transactions, including futures
contracts, options and swaps, have been deve
loped to m ore effectively limit the effects of
changes in exchange rates on the w ealth of
banks and their custom ers .6
Large com m ercial banks are the m ajor p ar
ticipants in the foreign exchange m arket. The
latest international survey of foreign exchange
activity, in April 1989, indicates that the three
m ost active centers are London, New York and
Tokyo (table 2). The value of foreign exchange
transactions has been grow ing faster than inter
national trade in goods and services (table 3).
Such grow th reflects m ore than the grow th of
international trade; it also reflects international
capital flows and transactions by banks and
their custom ers to m anage exchange rate risk.
Transactions in the foreign exchange m arket
link the m ajor financial institutions of the world.
In the London m arket, for instance, 80 percent
of the value of foreign exchange transactions in
April 1989 was by firm s w ith headquarters out
side of England .7 In the survey of foreign ex
change m arket activity in New York, 40 percent
of the value of transactions was reported by
offices of foreign banks .8 Thus, one of the im
portant ways in w hich a m ajor financial institu
tion can affect institutions in other countries is
by defaulting on foreign exchange transactions.

THE CONFIRMATION AND SET
TLEMENT OF FOREIGN EX
CHANGE TRANSACTIONS
The process of confirm ation and settlem ent
begins after traders at tw o banks agree on the
term s of a transaction. Each bank sends the
other a m essage specifying the term s of the
transaction, using a variety of m ethods, includ
ing telephone calls. If the details of the m es
sages m atch, the transaction is considered
confirm ed.
The next step depends on the value date of
the transaction. If it is a forw ard transaction,
w ith a value date several weeks or m onths into
the future, the inform ation is stored for future
6For a more detailed discussion of the foreign exchange
market, see Chrystal (1984).

Table 2
Foreign Exchange Market Activity in
April 1989 (billions of U.S. dollars)1
Countries and items
United Kingdom
United States
Japan
Switzerland3 [85°/o]
Singapore
Hong Kong
Australia
France3 [95%]
Canada
Netherlands
Denmark3 [90°/o]
Sweden
Belgium3 [90%]
Italy3 [75%]
Other countries4

Value of transactions per day
$

1872
1292
115
57
55
49
30
262
15
132
13
13
10
10
22
744

Total
Adjustment for
cross-border
double-counting

-204

Total reported net
turnover

540

Estimated gaps in
reporting

100

Estimated global
turnover

640

’ Value of transactions in currencies other than U.S. dol
lar converted to dollars at prevailing exchange rates.
The figures for individual countries indicate turnover
net of double-counting arising from local interbank busi
ness. The totals at the foot of the table are estimates of
turnover net of double-counting arising from both local
and cross-border interbank business.
2Based on estimates of domestic and cross-border inter
bank business arranged through brokers.
3No adjustment for less than full coverage; estimated
market coverage is given in square brackets.
4Bahrain, Finland, Greece, Ireland, Norway, Portugal
and Spain.
SOURCE: Bank for International Settlements (1990a).

settlem ent. On th e value date, banks transm it
inform ation to initiate paym ent. The steps to in
itiate paym ent depend on the paym ent system
used in the country issuing the currency and
the relationship of the paying bank to th at pay8Federal Reserve Bank of New York (1989).

7Bank of England (1989).

JANUARY/FEBRUARY 1992

Table 3
Growth of Foreign Exchange Market Transactions, Foreign
Trade and International Banking Activity

Countries

Value of foreign exchange
transactions: percentage
change between March 1986
to April 1989 net turnover

Exports and imports of
goods and services:
percentage change
from 1/1986 to 1/1989

United Kingdom

108%

62%

United States

120

Japan

140

116

Canada
Total

SOURCE: Bank for International Settlements (1990a).

m ent system. For a bank paying in a currency
other than th at of its hom e country, paym ent
generally is m ade by a correspondent headquar
tered in the foreign country. A correspondent is
a bank th at holds deposits and provides services
for other banks. The paying bank commonly
sends a m essage over SWIFT, instructing its cor
respondent to m ake paym ent to the counter
party in the foreign exchange transaction .9
Suppose, for instance, th at a bank headquar
tered in the United States m ust pay Germ an
m arks to a counterparty to settle a foreign ex
change transaction. The U.S. bank instructs its
Germ an correspondent to m ake paym ent to the
counterparty (or the counterparty’s Germ an cor
respondent). The G erm an correspondent debits
the account of the U.S. bank denom inated in
m arks and transfers the m arks to the counter
party. Suppose a U.S. bank is obligated to pay
dollars. It w ould send a m essage over CHIPS to
m ake paym ent to the counterparty, either di
rectly if it is a m em ber of CHIPS, or through a
correspondent in New York w ho is a m em ber
of CHIPS.10
9SWIFT (Society for Worldwide Interbank Financial
Telecommunication) is an electronic system, located in
Brussels, Belgium, for sending messages among the
world’s major banks.
10See Bank for International Settlement (1990b) for a
description of payments systems in various countries.
CHIPS (Clearing House for Interbank Payments System) is
an electronic payments system operated by the New York
Clearing House Association. CHIPS participants (131 as of
the end of 1990) exchange payment messages during
each business day and settle for the net amounts at day-

FEDERAL RESERVE BANK OF ST. LOUIS

THE RISKS INVOLVED IN
SETTLING FOREIGN EXCHANGE
TRANSACTIONS: AN ILLUS
TRATION
Banks assum e the risk that their counterpar
ties will default on paym ents on their side of
foreign exchange transactions. Effects on coun
terparties of default on settlem ent obligations
depend on the financial condition of the bank
that defaults. A solvent bank m ay default for a
variety of reasons. O perating problem s (for ex
ample, com puter failure) may prevent them
from executing their paym ent instructions. A
solvent counterparty may not have funds in the
proper currency on the value date, or simply
may forget to send paym ent orders to settle
some of their transactions.
Defaults by solvent banks on settlem ent obli
gations may have systemic effects, preventing
other banks from settling their obligations.
These banks may tu rn to their central banks
for short-term loans denom inated in the currenend with transfers of reserve balances at the Federal
Reserve. See Federal Reserve Bank of New York (1991).
A large share of CHIPS messages involve payment for the
dollar side of foreign exchange transactions. Given that
most foreign exchange transactions involve the U.S. dollar,
CHIPS has a major role in the settlement of foreign ex
change transactions. See Federal Reserve Bank of New
York (1987).

Table 4
Payments in Settlement of Foreign Exchange Transactions under Gross Settlement and Bilateral Netting_________________________________________
Bilateral netting

Gross settlement

Counterparties

Transaction
number

Direction of
payment

Bank A to Bank C
£ 100
Bank C to Bank A
$170
(Profit of $5.00 for Bank A)

Bank B to Bank C
£ 150
Bank C to Bank B
$262.50
(Profit of $15.00 for Bank B)
Bank B to Bank C
$85.00
Bank C to Bank B
£ 50
(Profit of - $2.50 for Bank B)

cies necessary to settle their obligations. Thus,
central banks have a collective interest in mini
mizing the chances of such liquidity problems.
Most liquidity problem s are often only tem
porary. Bankruptcy and liquidation of a p ar
ticipant in the foreign exchange m arket, how
ever, pose a m ore serious threat to individual
counterparties and create the potential for sys
temic disruptions in the paym ent system (default
by one bank causing default by others). U nder a
general definition of bankruptcy, the value of
liabilities exceeds the value of assets. Some large
bankrupt banks have been reorganized w ith as
sistance of their hom e governm ents. The reo r
ganized banks continue to operate as going con
cerns, making paym ents in settlem ent of their
obligations. Such reorganizations impose no loss
es on their counterparties.
In other cases, however, bankrupt banks
cease to operate as going concerns. The courts
appoint receivers to liquidate the bankrupt

Bank A to Bank B
£ 50
Bank B to Bank A
$90
(Profit of $7.50 for Bank A)

Bank A to Bank C
$262.50
Bank C to Bank A
£ 150
(Profit of -$ 1 5 .0 0 for Bank A)

Units of
currencies

Bank A to Bank B
$85
Bank B to Bank A
£ 50
(Profit of -$ 2 .5 0 for Bank A)

Bank B and
Bank C

Bank A to Bank B
£ 100
Bank B to Bank A
$175
(Profit of $10.00 for Bank A)

Bank A and
Bank C

Bank A and
Bank B

Direction of
payment

Units of
currencies

Bank A to Bank C
$92.50
Bank C to Bank A
£ 50
(Profit of -$ 1 0 .0 0 for Bank A)

Bank B to Bank C
£ 100
Bank C to Bank B
$177.50
(Profit of $12.50 for Bank B)

banks’ assets and m ake paym ents to their credi
tors. The receivers may impose losses on other
banks that w ere counterparties to foreign ex
change transactions. Such losses depend on the
legal principles followed by bankruptcy courts
and the nature of netting agreem ents betw een
counterparties.
The effects of the liquidation of a participant
in the foreign exchange m arket on its counter
parties are illustrated below. Legal assum ptions
are specified along the way as the example
raises questions about the principles followed
by bankruptcy courts. In each case in w hich a
bank is assum ed to go bankrupt, it is also as
sum ed to be liquidated by a court-appointed
receiver.

The Exam ple

Suppose three banks (A, B and C) engage in
foreign exchange transactions in tw o currencies:
the U.S. dollar and the British pound. Each
bank has foreign exchange transactions w ith the
other two. Table 4 lists the transactions between
JANUARY/FEBRUARY 1992

the counterparties to be settled on the same
value date. Each pair of banks has tw o transac
tions to settle. In one transaction, a bank pays
dollars in exchange for pounds; in the other, a
bank pays pounds in exchange for dollars.
The exchange rate on the value date is $1.65
per British pound. Transactions to be settled on
the value date w ere negotiated a few days earli
er w hen the exchange rate was higher: some
transactions w ere negotiated w ith an exchange
rate of $1.70; others, w ith an exchange rate of
$1.75. Transactions are of varying size, creating
im balances in the flows of currencies betw een
counterparties.
The exam ple is designed to be as simple as
possible and yet illustrate the risk involved in
netting arrangem ents. T here m ust be at least
tw o transactions betw een a pair of banks if
bilateral netting is to reduce the volume of pay
m ents and settlem ent risk. T hree is the mini
m um num ber of banks for m ultilateral netting.

The E ffects o f Bilateral Netting on
the N um ber and Value o f Trans
actions
Figure 1 illustrates how bilateral netting af
fects the flows of currencies betw een Banks A
and B in settling the transactions listed in table 4.
U nder gross settlem ent, banks m ake paym ents
to each other to settle each transaction betw een
them . To settle transaction num ber 1, Bank A
pays £ 100 to Bank B, receiving $175 in turn.
Since the exchange rate is $1.65 on the value
date, this exchange of currencies yields a profit
of $10 to Bank A. (Bank A receives $175, w hereas
the £ 100 paid by Bank A has a value of $165
on the value date). Bank A pays $85 to Bank B
in settlem ent of transaction num ber 2 , receiving
£ 50. This exchange yields a loss of $2.50 for
Bank A on the value date.
Banks A and B can econom ize on transactions
by netting their paym ents flows. As illustrated
in the bottom half of figure 1, Bank A could
pay £ 50 to Bank B and receive $90 from Bank
B. Bilateral netting reduces the num ber of pay
m ents from four to tw o and the value of pay
m ents, converted to dollars at the exchange rate
of $1.65, from $507.50 to $172.50.
"B a n k for International Settlements (1989), pp. 13-14.

FEDERAL RESERVE BANK OF ST. LOUIS

The R isk in Settling Foreign Ex
change Transactions w ithout a
Netting A rrangem ent
To illustrate how netting arrangem ents affect
risk, one m ust first understand the risk that
banks assum e w ithout a netting agreem ent.
— This section specifies
several assum ptions about the legal principles
that the bankruptcy court follows w hen banks
settle their transactions w ithout netting arran ge
m ents. W hile these principles are not applied in
all cases, they are com m on and they simplify
the analysis.

L e g a l A s s u m p tio n s

One assum ption concerns the application of le
gal rights o f set-off perm itted by the court. Un
der the legal rights of set-off, the counterparty
of a failed bank may settle its obligations w ith
the receiver by paying the net am ount of the
transactions betw een them . If on net the failed
bank owes a solvent counterparty, the counter
party is a general creditor of the failed bank for
the net am ount. Applying the rights of set-off to
the foreign exchange transactions betw een a
pair of banks yields the same loss to the solvent
counterparty as it w ould un d er bilateral netting.
Applying the legal rights of set-off, how ever, is
uncertain and varies am ong the courts of differ
ent countries .11 In this paper, rights of set-off
are assum ed not to apply in bankruptcy. Each
transaction is treated separately, not linked to
other transactions betw een the same parties.
The court w ith jurisdiction in a bankruptcy
case is assum ed to appoint a receiver. In m ak
ing paym ents to settle foreign exchange transac
tions or defaulting on transactions, the receiver
acts to maximize the retu rn to all creditors of
the failed bank, w ithout regard for the counter
parties to foreign exchange transactions as a
particular group of creditors.
A final issue concerns the status of claims
against a bankrupt bank th at result from its
default on foreign exchange transactions. Sol
vent counterparties are assum ed to have the
status of general creditors. In our example, loss
es are calculated under the assum ption that
general creditors receive nothing. All proceeds
from the liquidation of assets go to creditors
w ith m ore senior claims.

Figure 1
Flow of Currencies Between Banks Under Gross Settlement and Bilateral Netting
Gross Settlement

£100

£ 50
$85

Transaction #2

Profit to Bank A:
-$2.50

Number of Payments: 4
Dollar Value: $507.50

Bilateral Netting

£ 50

Profit to Bank A:
$7.50

$90
Number of Payments: 2
Dollar Value: $172.50

JANUARY/FEBRUARY 1992

These legal assum ptions yield the maximum
losses to counterparties. Thus, the losses calcu
lated in particular cases can be viewed as the
m aximum, not necessarily the most likely, losses.
Ex
p o s u r e — Suppose Banks A and B agree to set
tle their transactions as illustrated in the top
half of figure 1 , the gross settlem ent m ethod.
Suppose also that Bank A goes bankrupt before
the four paym ents are executed on the value
date. The possible loss to Bank B depends on
the timing of the bankruptcy of Bank A.
In one situation, called pre-settlement failure,
Bank A goes bankrupt before the value date,
and Bank B knows about this event by the
opening of business on the value date. In the
other situation, called settlement failure, a bank
m akes paym ent on the value date for its side of
foreign exchange transactions bu t does not re
ceive paym ent from a counterparty.
One feature of the foreign exchange m arket
that makes banks vulnerable to settlem ent fail
ure is the difference in the tim e zones of cen
tral banks. The failure of the H erstatt Bank in
1974 illustrates the relationship betw een time
zones and settlem ent failure. On June 26, 1974,
Germ an banking authorities closed H erstatt as
of the close of business in Germany. H erstatt
had received paym ent in m arks during Germ an
banking hours for its foreign exchange transac
tions w ith th at value date. It w as closed, how
ever, before the tim e for m aking paym ents in
dollars in New York. C ounterparties of H erstatt
w ere left w ithout the dollars they expected, af
ter paying m arks to H erstatt earlier in the day.lz
O ur example of settlem ent failure in this paper
reflects the implications of differences in time
zones. One bank is assum ed to go bankrupt af
ter the time for paym ents in pounds but before
the tim e for m aking paym ents in dollars.
P re-S ettlem en t F ailure — Suppose Bank A
goes ban krup t before the value date. W ithout a
netting agreem ent betw een Banks A and B, the
legal obligations of each bank are those speci
fied in the individual transactions betw een them .
W ith an exchange rate of $1.65 on the value
date, transaction num ber 1 is profitable to Bank
A. The receiver of Bank A will pay £ 100 to
Bank B to settle transaction num ber 1. Bank B is
T im in g o f B a n k r u p t c y a n d S ize o f

12Moore (1974).

FEDERAL RESERVE BANK OF ST. LOUIS

obligated to pay $175 to Bank A to settle this
transaction. Since transaction num ber 2 is not
profitable to Bank A on the value date, the re
ceiver will default on transaction num ber 2 .
Bank B anticipated a profit of $2.50 on the value
date from transaction num ber 2. Thus, the bank
ruptcy of Bank A im poses a loss of $2.50 on
Bank B. Table 5 shows the loss to each bank
due to the bankruptcy of its counterparty be
fore the value date, under both gross settlem ent
and netting arrangem ents.
S ettlem en t F ailure — Suppose Bank A goes
bankrupt on the value date after paym ent in
pounds bu t before paym ent in dollars. Bank A
defaults on its paym ent of $85 to Bank B on the
value date. U nder gross settlem ent of transac
tions, how ever, Bank B is obligated to pay the
$175 to Bank A. Bank B becom es a general cre
ditor of Bank A for $85. The m aximum loss to
Bank B, as table 6 indicates, is $85.
Settlem ent failure can create liquidity problem s
for the counterparties of a failed bank. Suppose
Bank B pays the $175 to Bank A before discov
ering that Bank A is bankrupt. The cash bal
ances of Bank B denom inated in dollars will be
$85 below the level it had projected for the
value date. Bank B m ight request a discount
w indow loan from the Federal Reserve to cover
the $85 shortfall in its reserve account.

H ow B ilateral Netting A ffects the
Losses in Settlem ent o f Foreign Ex
change Transactions

If Banks A and B engage in bilateral netting, the
effects of the bankruptcy of Bank A on Bank B
depend on w hether paying the net am ount dis
charges the obligations betw een counterparties.
L e g a l A s s u m p tio n s — U nder one type of
agreem ent called position netting, tw o banks
agree to net their paym ents to reduce transac
tions costs, but the agreem ent has no effect on
their legal obligations. U nder the legal assum p
tions in this paper, the position netting agree
m ent w ould not prevent the receiver from
m aking paym ents in settlem ent of some transac
tions but defaulting on others w ith the same
counterparty. The bankruptcy court w ould treat
the paym ent obligation of Banks A and B as
though they had no netting agreem ent. The
bankruptcy of one party has the same implica-

Table 5
Bank Losses from Pre-Settlement
Failure
Losses to
Failure of

Bank A

Bank B
Gross settlement
Bilateral netting
Multilateral netting

5.00
0.00
0.00

$15.00
10.00
2.50

$10.00
7.50
0.00

Bank C
Gross settlement
Bilateral netting
Multilateral netting

BankC

$ 2.50
0.00
0.00

Bank A
Gross settlement
Bilateral netting
Multilateral netting

Bank B

2.50
0.00
0.00

15.00
12.50
2.50

Table 6
Bank Losses from Settlement Failure
Losses to
Failure of

Bank A

Bank B
Gross settlement
Bilateral netting
Multilateral netting

170.00
0.00
0.00

$262.50
92.50
2.50

$175.00
90.00
0.00

Bank C
Gross settlement
Bilateral netting
Multilateral netting

Bank C

$ 85.00
0.00
0.00

Bank A
Gross settlement
Bilateral netting
Multilateral netting

Bank B

85.00
0.00
0.00

262.50
177.50
85.00

tions for the counterparty as if they settled
transactions using the gross settlem ent m ethod.
13Bank for International Settlements (1989), p. 13.
14One firm that offers legal advice and a communications
network for bilateral netting by novation is FXNET. The
netting contract drafted by FXNET includes netting by no
vation and closeout. See Bartko (1990). For further refer
ence to FXNET, see Scarlata (1992), this Review. Plans

Netting agreem ents th at reduce this exposure
to loss m andate th at banks discharge th eir obli
gations by paying the net am ount of the tran s
actions betw een them . The legal language for
such agreem ents is netting by novation. This
paper assum es th at bankruptcy courts recognize
a contract for netting by novation as the only
contract betw een counterparties for settlem ent
of foreign exchange transactions.
A provision of bilateral netting contracts that
reduces risk is called closeout, w hich becom es
effective w hen a receiver or liquidator is ap
pointed after a bank declares bankruptcy .13 A
netting agreem ent includes a form ula that con
verts all outstanding transactions betw een a
pair of counterparties, for all value dates, into
one am ount payable immediately. The closeout
provision prohibits the receiver of a bankrupt
bank from m aking paym ents in settlem ent for
transactions w ith some value dates b u t default
ing on transactions w ith other value dates .14
B ankruptcy courts are assum ed to recognize
closeout provisions as valid parts of netting a r
rangem ents.
P re - s e ttle m e n t F a ilu re — As the bottom
half of figure 1 illustrates, th e one contract be
tw een Banks A and B under netting by novation
calls for Bank A to pay £ 50 and receive $90. At
the exchange rate of $1.65 on the value date,
this contract is profitable for Bank A. Thus, the
receiver of Bank A w ould pay the £ 50 to Bank
B to settle the contract. The bankruptcy of Bank
A prior to the value date w ould impose no loss
on Bank B, since Bank B had anticipated honor
ing its contract w ith Bank A before discovering
th at Bank A was bankrupt. In each case of presettlem ent failure illustrated in table 5, the loss
es are sm aller un d er bilateral netting by nova
tion than under gross settlem ent.
S e ttle m e n t F a ilu re — The bankruptcy of
Bank A after paym ents in pounds but before
paym ents in dollars imposes no loss on Bank B
since, under the netting agreem ent, Bank A had
no obligation to pay dollars to Bank B. As table
6 indicates, in settlem ent failure, the loss to a
bank from the bankruptcy of its counterparty is
for multilateral netting include similar closeout provisions
in contracts between individual members and the clearing
houses that would act as paying agents for the netting ar
rangements. See Duncan (1991). These closeout provi
sions limit the losses of solvent banks resulting from
default by counterparties.

JANUARY/FEBRUARY 1992

sm aller under bilateral netting by novation than
u n d er gross settlem ent for each com bination of
failed bank and counterparty.

The Im portance o f the Legal Basis
f o r Netting A greem ents

The assum ptions in this paper concerning the
principles that bankruptcy courts follow yield
the m aximum reductions in losses from netting.
These reductions in losses could be sm aller un
der alternative assum ptions.
The Lamfalussy Report indicates that bilateral
netting could increase risk in settling foreign ex
change transactions if netting arrangem ents do
not have a sound legal basis. If netting "obscures
the level of exposures, then netting arrange
m ents have the potential to contribute to an in
crease in systemic risk .”15 The argum ent that
bilateral netting may pose greater risks is based
on assum ptions about how banks that are active
in the foreign exchange m arket set credit limits
w ith counterparties. Banks w ith bilateral netting
agreem ents may set credit limits w ith each other
based on their net positions rath er than the
gross value of the underlying transactions be
tw een them . If a bankruptcy court requires
paym ents by a solvent counterparty based on
the value of the underlying transactions rather
than the netting agreem ent, the exposure of the
solvent counterparty w ould be larger th an ex
pected. This point indicates w hy the Lamfalussy
Report em phasizes the legal basis for netting ar
rangem ents (table 1 ).

M ultilateral Netting

Banks m ay be able to fu rth er reduce their
transaction costs and their exposure to loss by
engaging in m ultilateral netting. No m ultilateral
netting arrangem ents are in operation at this
time. This section examines the implications of a
m ultilateral netting arrangem ent modeled after
a draft of the plans of the ECHO NETTING sys
tem in London .16
L e g a l A s s u m p tio n s — In th e contract for
m ultilateral netting, m em bers of a netting ar
rangem ent establish a clearing house, which
receives and pays out currencies in settlem ent
of foreign exchange transactions. The clearing
house is the counterparty for each transaction
betw een m em bers of the m ultilateral netting ar
rangem ent. Each m em ber settles its legal obliga
15Bank for International Settlements (1990c), p. 3.

FEDERAL RESERVE BANK OF ST. LOUIS

tions w ith the others by m aking paym ents to
the clearing house. The clearing house assum es
responsibility for paying all net am ounts due to
m em bers, even if a m em ber defaults on its pay
m ents to the clearing house.
The contract in a m ultilateral netting arrange
m ent is assum ed to include a closeout provision.
If a m em ber of the clearing house goes bank
rupt, its receiver has only one decision to m ake
about the foreign exchange transactions that the
failed bank negotiated w ith other m em bers:
m ake the paym ents to settle the one contract
w ith the clearing house or default.
P a y m e n t s F lo w s a n d L o s s S h a rin g —

Figure 2 presents the paym ents betw een m em
bers of the netting arrangem ent and the clear
ing house, derived from paym ents that w ould
be m ade un d er bilateral netting in table 4. The
calculation of the num bers in figure 2 is illus
trated for Bank A. U nder bilateral netting, Bank
A pays the other banks £ 50 (Bank B) and $92.50
(Bank C) and receives $90 (Bank B) and £ 50
(Bank C). U nder m ultilateral netting, therefore,
Bank A owes the clearing house $2.50 and the
clearing house owes Bank A nothing on the
value date. Figure 2 also indicates the paym ents
betw een the clearing house and Banks B and C.
Any clearing house losses resulting from the
default of a m em ber are allocated to the other
m em bers in proportion to the losses they would
have in curred under bilateral netting. This for
mula gives each m em ber of the arrangem ent an
incentive to avoid transactions w ith m em bers it
considers to be in danger of going bankrupt.
P r e -S e ttle m e n t F a i lu r e — If Bank A goes
bankrupt before the value date, its receiver will
default on the paym ent of $2.50 to the clearing
house. The loss of $2.50 is allocated to Bank C,
since only Bank C w ould have a loss under
bilateral netting.
If Bank B goes ban krup t before the value
date, its receiver will m ake the paym ent of £ 50
to settle the contract w ith the clearing house,
since it yields a profit of $5 to Bank B. As table
5 indicates, the bankruptcy of Bank B before
the value date imposes no loss on the other
banks. The bankruptcy of Bank C imposes a loss
of $2.50 on Bank B. In each case in table 5, the
16Duncan (1991).

Figure 2
Payments Between Members of a Multilateral Netting Arrangement and
the Clearing House

BANK A

BANK B

Profit to Bank A: -$2.50
Profit to Bank B: $5.00

CLEARING
HOUSE

Profit to Bank C: -$2.50

N um ber of Payments: 5
Dollar Value: $340

JANUARY/FEBRUARY 1992

A Glossary of Terms
B i l a t e r a l n e ttin g

C le a r in g h o u s e

C o u n te rp a rty

F o r e ig n e x c h a n g e
tra n s a c tio n
G ro ss s e ttle m e n t

L egal rig h ts of
s e t-o ff
M u ltila te r a l n e t t i n g

N e ttin g

N e ttin g b y
n o v a tio n

P o s i t i o n n e ttin g

P re - s e t t l e m e n t
f a ilu r e
S e ttle m e n t

S e ttle m e n t f a ilu r e

S y s te m ic r i s k

V a lu e d a te

An arrangem ent betw een tw o parties in w hich they exchange only
the net units of the currencies specified in the transactions betw een
them , ra th e r than the currencies for each transaction individually.
An institution established by a group of banks to facilitate the settle
m ent of obligations am ong them selves. Each bank settles its obliga
tions w ith the others by m aking paym ent to the clearing house for
the net am ount ow ed the other m em bers.
The other party in a transaction. In a foreign exchange transaction,
one party agrees to m ake paym ent in one currency and its counter
party agrees to pay in another currency.
An agreem ent by tw o parties (generally large banks) to exchange cu r
rencies on a given date.
A m ethod of m aking paym ents betw een a pair of parties in w hich
each party m akes a separate paym ent in settlem ent of each transac
tion betw een them .
U nder bankruptcy law, a right to net obligations w ith a bankrupt
counterparty.
An arrangem ent betw een th ree or m ore parties in w hich each m em
b er m akes paym ents to a clearing house for the net paym ents due to
the other m em bers and receives from the clearing house th e net
am ounts due from the other m em bers.
An arrangem ent by w hich parties w ith m ore th an one transaction to
settle on a given date exchange only the net am ounts of the transac
tions betw een them .
The replacem ent of tw o existing contracts betw een tw o parties for
delivery of a specified currency on the sam e date by a single net
contract for that date, so that the original contracts are satisfied and
discarded.
The netting of paym ent obligations betw een tw o or m ore parties
w hich neither satisfies no r discharges the original obligations that
w ere netted.
Bankruptcy of a bank prior to the value date of transactions w ith a
counterparty.
Completion of a paym ent betw een tw o parties discharging an obli
gation.
Default by a bank on paym ent in one currency after the bank and
its counterparty had m ade paym ents in the other currency.
The risk th at the inability of one institution w ithin a paym ent system
to m eet its obligations w hen due will cause other participants to be
unable to m eet th eir obligations w hen due.
The date on w hich banks exchange currencies in settlem ent of a for
eign exchange transaction.

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

loss un der m ultilateral netting is either zero or
sm aller than under bilateral netting.
S e ttle m e n t F a i l u r e — Suppose Bank A goes
bankrupt after the paym ent of pounds but be
fore the paym ent of dollars. The loss to be
borne by m em bers of the clearing house is
$2.50, the paym ent obligation of Bank A. This
loss is im posed on Bank C, w hich w ould have a
loss of $92.50 under bilateral netting (table 6).
Bank B has no obligation to pay dollars to the
clearing house. Thus, the bankruptcy of Bank B
after the paym ent of pounds but before the
paym ent of dollars imposes no losses on other
m em bers of the clearing house. The bankruptcy
of Bank C after paym ent in pounds imposes a
loss of $85 on Bank B. The loss in each case u n
der m ultilateral netting in table 6 is either zero
or less than the loss under bilateral netting .17

L iquidity R equirem ents o f the
Clearing H ouse
One of the concerns central bankers have
about m ultilateral netting is w hether the clear
ing house w ould have access to sufficient li
quidity to m ake paym ents to other m em bers if
one of them defaults. The Lamfalussy Report in
dicates th at a clearing house should "be capable
of ensuring the timely com pletion of daily settle
m ents in the event of an inability to settle by
the participant w ith the largest single net-debit
position.” In figure 2, Banks A and C each have
net debit positions of $2.50. The clearing house
w ould need access to at least $2.50 to m eet the
minim um liquidity requirem ent of the Lamfal
ussy Report. This requirem ent is a cost of oper
ating the clearing house, either as the oppor
tunity cost of liquid assets held by the clearing
house or the cost of credit lines. Bilateral net
ting, in contrast, involves no such costs.

CONCLUSIONS
Banks assum e risk in settling foreign exchange
transactions. This paper exam ines the implica
tions of netting by using a hypothetical exam
17The generality of the result that netting reduces losses to
solvent counterparties can be investigated by simulating
the losses resulting from default in an example with more
banks and more transactions and with some terms of the
transactions chosen at random. Our simulation includes 10
banks. Each has 10 transactions to settle with each of the
other nine banks. The size of the transactions and ex
change rates are chosen at random. The multilateral net-

ple. The exam ple shows how netting schem es
can reduce the size of losses to counterparties
w hen a bank goes bankrupt and is liquidated.
A com m ittee of central bankers from the m a
jor developed countries recently exam ined the
implications of netting arrangem ents for risk.
The com m ittee's report indicates that netting ar
rangem ents may either increase or decrease
risk, depending on w hether they m eet certain
m inim um standards listed in the report.
Bilateral netting could reduce the loss w hen a
counterparty defaults, if the bankruptcy court
w ould recognize the paym ent of the net am ount
betw een the counterparties as a settlem ent of
the transactions betw een them . It could increase
risk in settlem ent of foreign exchange transac
tions, however, if counterparties set credit
limits based on their net exposures but the
court requires paym ent in settlem ent for each
underlying transaction betw een counterparties.
M ultilateral netting can reduce the losses re
sulting from default even m ore than bilateral
netting, if the clearing house created to settle
transactions has access to the liquid assets ne
cessary to com plete the settlem ent. Lack of
sufficient liquidity for the clearing house could
create a m ajor disruption in the operation of
the paym ent system.

REFERENCES
Bank for International Settlements. Report on Netting Schemes
(BIS, February 1989).
_______ . Survey of Foreign Exchange Market Activity (BIS,
F eb ru ary 1 9 9 0 a ).

_______ . Large-Value Funds Transfer Systems in the Group
of Ten Countries (BIS, May 1990b).
_______ . Report of the Committee on Interbank Netting
Schemes of the Central Banks of the Group of Ten Coun
tries (BIS, November 1990c).
Bank of England. “ The Market in Foreign Exchange in Lon
don,” Bank of England Quarterly Bulletin (November 1989),
pp. 531-35.

ting arrangement uses the loss-sharing formula described
above. Each of the 10 banks is assumed to go bankrupt,
imposing losses on nine counterparties. In each of the 90
cases of losses to counterparties, losses are smaller under
bilateral netting than under the gross settlement method,
and losses under multilateral netting are either zero or
smaller than under bilateral netting.

JANUARY/FEBRUARY 1992

Bartko, Peter. "Foreign Exchange and Netting by Novation,”
Payment Systems Worldwide (Spring 1990), pp. 48-51.
Chrystal, K. Alec. “A Guide to Foreign Exchange Markets,”
this Review (March 1984), pp. 5-18.
Cody, Brian J. “ Reducing the Costs and Risks of Trading
Foreign Exchange,” Federal Reserve Bank of Philadelphia
Business Review (November/December 1990), pp. 13-23.
Deeg, Ernst. “A Proposal for a Multi-Currency Netting Sys
tem,” Payment Systems Worldwide (Spring 1990),
pp. 40-45.
Duncan, G. M. "ECHO NETTING: Multilateral FX Netting in
Europe,” mimeo, Barclays Bank (March 1991).
Federal Reserve Bank of New York. “A Study of Large-Dollar
Flows Through CHIPS and Fedwire,” (December 1987).
_______ . “ Summary of Results of U.S. Foreign Exchange
Market Survey Conducted in April 1989” (September 13,
1989).

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

_______ . The Clearing House Interbank Payments System
(January 1991).
Juncker, George R., Bruce J. Summers and Florence M.
Young. “A Primer on the Settlement of Payments in the
United States,” Federal Reserve Bulletin (November 1991),
pp. 847-58.
Luthringshausen, Wayne P. “ Characteristics of a Compre
hensive Clearing System,” Payment Systems Worldwide
(Spring 1990), pp. 38-39.
Moore, A. E. “ Foreign Exchange,” Bankers' Magazine
(August 1974), pp. 49-51.
Polo, Renato. “A Perspective on the Euronetting Project,”
Payment Systems Worldwide (Spring 1990), pp. 46-47.
Scarlata, Jodi G. “ Institutional Development in the Globaliza
tion of Securities and Futures Markets,” this Review (Janu
ary/February 1992), pp. 17-30.
Summers, Bruce J. “ Clearing and Payment Systems: The
Role of the Central Bank,” Federal Reserve Bulletin (Febru
ary 1991), pp. 81-91.

Jodi G. Scarlata
Jodi G. Scarlata was a visiting scholar at the Federal Reserve
Bank of St. Louis when this paper was written. David H. Kelly
and Lynn Dietrich provided research assistance.

Institutional Developments in
the Globalization of Securities
and Futures Markets

r [NANCIAL TRANSACTIONS such as the buying and selling of securities, commodities, foreign
exchange and bonds, have increasingly involved
individuals and firm s from different countries.
For example, a Japanese resident m ight purchase
U.S. dollars w ith Japanese yen (a foreign ex
change transaction) to buy shares of IBM on the
New York Stock Exchange (a securities transac
tion). To accom m odate such transactions, futures
and securities exchanges have expanded the
services they offer their users, adding num erous
financial instrum ents, engaging in cooperative
efforts across exchanges and introducing computer-based technologies.
The globalization of w orld m arkets provides
significant benefits, including greater opportuni
ties for investors to diversify risk, and access to
broader m arkets for dem anders of funds. Inter
national trading in financial instrum ents, how
ever, does pose risks, some of w hich can be
m itigated by coordination betw een global finan
cial m arkets.
This paper describes recent institutional de
velopm ents in the globalization of financial m ar
kets and discusses the advantages and disadvan
tages of these innovations. The paper opens
w ith a brief overview of the various transn a
1Abken (1991), p. 3.

tional developm ents th at are occurring in world
securities and comm odities m arkets. It then ad
dresses both the benefits of expanding financial
m arkets and the costs that accom pany the move.
Risk factors and standardization of procedures
are highlighted as issues of concern as financial
centers globalize. The paper closes w ith a dis
cussion of the Group of T hirty proposal for the
coordination of clearing and settlem ent in w orld
securities m arkets.

INTERNATIONAL DEVELOP
MENTS IN FINANCIAL AND
COMMODITY MARKETS
The tren d tow ard internationalization of finan
cial m arkets can be illustrated by highlighting
the rapid increases in transactions in a few
m arkets. For example, cross-country activity,
w hen m easured as the volume of foreign tran s
actions in securities of U.S. firm s (aggregate
purchases and sales), grew from $75.3 billion in
1980 to $361.4 billion in 1990.1 Similarly, U.S.
transactions in securities of foreign firm s (aggre
gate purchases and sales) grew from $17.85 bil
lion to $253.4 billion betw een 1980 and 1990.2 In
futures and options m arkets, 20 new exchanges
2lbid.

JANUARY/FEBRUARY 1992

w ere established w orldw ide betw een 1985 and
1989, bringing the total to 72.3 Likewise, nearly
40 million futures and options contracts w ere
traded w orldw ide in 1988, an increase of ap
proxim ately 186 percent since 1983.4 Eurodollar
interest rate futures saw an especially large
change, increasing alm ost 70 percent annually
betw een 1983 and 1988.5

ing the position book betw een time zones is ac
tually to transfer the handling instructions be
tw een traders. An exam ple is a New York
currency trad er w ho instructs the trad er at his
Singapore office to track the price of a currency
during evening hours in New York. W hen the
m arket reaches a particular price, the Singapore
trad er will buy or sell, depending on instructions
from New York.

The globalization of financial and com m odity
m arkets involves num erous activities and institu
tional developm ents th at facilitate access to for
eign m arkets, w h eth er by a trad er or a security.
One of these activities is the cross-listing of
securities in several countries. Cross-listing sim
ply m eans, for example, that a firm in the United
States lists its stocks on a London exchange. In
1990, the International Stock Exchange (ISE) of
London had one of the highest percentages (23
percent) of foreign com pany stock listings .6
A nother tren d is cross-country hedging and
portfolio diversification. A U.S. trader, for exam
ple, can diversify a portfolio com posed of U.S.
stocks by buying stocks of a U.K. firm in Lon
don through a London broker. Globalization can
also m ean holding m em bership in another coun
try ’s exchanges. For example, after “The Big
Bang” of 1986 in London, m any U.S. securities
firm s and banks applied to buy seats on London
exchanges .7
A th ird tren d in the internationalization of
financial m arkets is called “passing the book,”
w hereby control of trading is passed betw een
traders at exchanges around th e globe. This
enables 24-hour trading of a financial instrum ent.
An exam ple of this w ould be a U.S. investm ent
firm trading from New York during U.S. and
Japanese hours and from its London desk during
U.K. hours. The m ore com m on practice of pass

One tren d that does not involve actual trading
is the underw riting of corporate securities
through offices outside the hom e country. An
u n derw riter is a firm th at buys an issue of
securities from a com pany, th en resells it to in
vestors. For the com pany issuing the securities,
underw riting provides a guarantee th at a certain
am ount of m oney will be derived from the sale
of the securities that can be used for capital ex
penditure. A large stock issue may have un d er
w riters from several countries, for example, to
com pensate for a country w hose capital m arket
does not have the depth to handle large securi
ties offerings .8 The distribution of underw riters
across several countries provides the issuing
firm w ith a w ider access to funding .9 Investors,
on the other hand, obtain a broader selection of
securities.

Illustrations o f G lobalization

3Baer, Evanoff and Pavel (1991), p. 11.
‘'Ibid.
E uro do lla r deposits are dollar-denominated deposits out
side the United States. Eurodollar interest rate futures con
tracts are futures contracts on the interest rates on these
deposits. The figures are from Baer, Evanoff and Pavel
(1991), p. 11.
6For the New York Stock Exchange, the figure was 3.7 per
cent. U.S. Congress, Office of Technology Assessment
(1990), p. 29.
TThe Big Bang was a deregulation effort for British financial
markets which began on October 27, 1986. Examples of
changes to the London equity markets are the end of fixed
commission rates; barriers between order-taking brokers
and risk-taking market makers were broken down; and, the
self-regulating Securities and Investment Board (SIB) was

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

D evelopm ents in A utom ated Trad
ing S ystem s
M ore recently, the developm ent of autom ated
trading has received substantial attention. Auto
m ated or electronic trading system s allow agents
to m ake trades via com puter, w ithout the "open
outcry” or pit auction system .10 Interestingly,
the developm ent of m uch of the cu rren t auto
m ation is an extension of technological innova
tions originally developed for dom estic m arkets.
It is clear, how ever, that this autom ation has afestablished. The SIB is a non-governmental version of the
Securities and Exchange Commission (SEC) in the United
States. Khoury (1990), p. 129.
8Depth means that there are enough buyers and sellers in
a market that a large transaction will not affect the price.
9An example is the privatization of French companies in
1986, where the value of these newly privatized companies
was approximately $30 billion, but the total value of list
ings on the Paris Bourse was only about $80 billion. U.S.
Congress, Office of Technology Assessment (1990), p. 34.
10Open outcry occurs on an organized exchange when ord
ers between buyers and sellers are traded between third
parties in anonymity. The buyer/seller enters into a con
tract with the exchange or its representative
clearinghouse.

fected the globalization of financial m arkets con
siderably.
The National Association of Securities Dealers
Autom ated Quotations (Nasdaq) was one of the
earliest developm ents in financial m arket com
puterization, beginning in 1971. Nasdaq p ro
vides com puter listings of price inform ation for
several thousand companies. By 1982, the Na
tional Association of Securities Dealers (NASD)
had produced the National M arket System, w hich
provided investors w ith inform ation as sales
w ere com pleted, and by 1991, had developed
the Private Offerings, Resales, and Trading
through Autom ated Linkages (PORTAL) system .11
From a com puter term inal, PORTAL enables
users to trade in unregistered domestic and for
eign debt and equity securities .12
Nasdaq has established com puter telephone
linkage as well as autom ated trade execution
and international clearing and settlem ent with
the International Stock Exchange of London and
the Stock Exchange of Singapore. Nasdaq has
since becom e a significant m arket for the listing
of foreign securities, trading approxim ately $6
billion in foreign securities as of 1991, up from
the $2.6 billion in 1985.13 Thus, Nasdaq provides
the cross-listing of securities, together w ith the
rapid trade execution of an autom ated system.
The grow th of international trading has also
affected futures and options exchanges in the
United States. The fact th at traders could access
instrum ents and overseas m arkets after norm al
U.S. trading hours had ended, provided a m otiva
tion for m any of the extended-hour and 24-hour
trading initiatives (see shaded insert for examples
of extended trading hours and table 1 for auto
m ated trading systems).
A significant portion of U.S. financial in stru
m ents, futures and options is traded at exchanges
throughout the world. That is, foreigners do not
have to use the Chicago M ercantile Exchange to
trade Eurodollar contracts, a CME staple. For
"P O R T A L uses a book entry settlement system with no
physical delivery, eliminating the problem of unmatched
trades. PORTAL is currently the only fully automated clear
ing and settlement system in the United States. Clearing
and settlement issues, book entry and matching trades will
be discussed in more detail later in the text.
12Unregistered securities are not registered with the Securi
ties and Exchange Commission (SEC). They are issued in
a limited volume or by small companies.

example, in 1989, a th ird of th e trade in con
tracts offered by the CME originated outside of
N orth Am erica .14 In 1989, 10 percent of the
CME’s daily volume w as transacted overnight in
an overseas exchange while the CME was
closed .15 F urtherm ore, in 1985, the CME and
Chicago Board of T rade together accounted for
83 percent of all futures volume. By 1990, the
figure had fallen to 55 percent .16 The attem pt to
regain m arket share instigated such CME expan
sions as extended trading hours and autom ated
trading systems.
Perhaps the m ost am bitious project in auto
m ated trading is Globex, an attem pt to create a
24-hour trading m arket originally proposed in
1987 by the CME .17 Globex is an electronic trade
execution system w hereby traders enter buy
and sell orders th at are m atched autom atically
according to price and tim e priority .18 Originat
ing as a strictly off-hours trading system, the
purpose of Globex is to enable continued active
trading beyond the CME's regular trading
hours. The CME intends to use Globex to access
m arkets after its ow n close of business and re
gain some of the m arket share lost to foreign
exchanges.

POTENTIAL GAINS AND RISKS OF
INTERNATIONALIZATION
The significant changes accom panying the in
ternationalization of financial trading systems al
low for th e realization of substantial gains; at
the same time, globalization also exacerbates the
risks already present in financial trading. The
m ost significant of these gains and risks are
described below.

B enefits o f G lobalization

One of the m ost im portant areas of progress
is the speed w ith w hich inform ation is processed
and dissem inated to m arket participants. In
creased flow of m arket data provides greater
15lbid.
16Chesler-Marsh (1991), p. 33.
17The Chicago Board of Trade has since become a par
ticipant with the CME in the Globex project.
18For detailed reading on automated trading systems, see
Domowitz (1990).

13See Nasdaq (1991), p. 16, for the 1991 figure, and NASD
(1991), p. 15, for the 1985 figure.
14Hansell (1989), p. 187.

JANUARY/FEBRUARY 1992

“After-Hours" Trading
The following is a listing of some of the de
velopm ents in trading after regular operating
hours.

N ew York S tock Exchange

Has tw o after-hours trading sessions, begin
ning June 13, 1991. One 45-m inute, orderm atching session and a one-hour-and-15-minute session for crossing baskets of stocks.

A m erican Stock Exchange

Plans for an after-hours session betw een
4:15 p.m. and 5 p.m. w here buyers and
sellers can trade at regular session’s closing
price. C urrently, talks w ith R euters Holdings
PLC, the Chicago Board Options Exchange
and the Cincinnati Stock Exchange for a
global after-hours trading system for stocks
and their options and derivatives such as
stock-index w arrants.

N asdaq

Has filed w ith the SEC to start trading on
Nasdaq International at 3:30 a.m. EDT.

P acific S tock Exchange
Has filed plans w ith the SEC to extend tra d
ing in listed stocks in a regular auction m ar
ket lasting until 4:50 p.m. EDT. Currently,
PSE closes at 4:30 p.m. EDT. Will m atch buy
and sell orders in a 5 p.m. EDT session at the
NYSE closing price.

Philadelphia S tock Exchange

T rades currency options and futures for
20.5 hours a day. Has filed plans w ith the
SEC for an after-hours order m atching sys
tem similar to the Big Board. Also plans to fill
custom er orders based on after-hours activity
on the Big Board.

B oston S tock Exchange

Will not have its ow n session, but will fill
custom er orders based on after-hours activity
on the Big Board.

Chicago B oard O ptions Exchange

M idw est Stock Exchange

Has no cu rren t plans to trade stock options
before or after norm al trading hours. Involved
w ith Amex in electronic system.

Will not have its ow n after-hours session,
but will fill custom er orders based on afterhours activity on the New York Stock Ex
change.

An electronic trading system ow ned by
Reuters Holdings, operates up to 16 hours a
day.

Instinet

’ “ Big Board After-Hours Trading” (1991).

accessibility to foreign m arkets. In turn, the
larger the num ber of participants using a m arket,
the greater the liquidity of the m arket and, thus,
its desirability for investors .19
The level of m arket activity and the transm is
sion of data on prices, m arket supply and de
m and conditions are a few examples of inform a
tion relevant to traders. Yet, this inform ation
technology has less to do w ith im proving the ef
ficiency of forecasting techniques than it does
to shaving seconds off the receipt of up-to-the19Liquidity is the depth of the market (for example, securities
or futures) and its ability to absorb sudden shifts in supply
and demand without excessive price fluctuation.

FEDERAL RESERVE BANK OF ST. LOUIS

m inute m arket events. Inform ation like the an
nouncem ent of a com m odity quota or a cor
porate m erger provides the im petus for the rapid
decision-making th at characterizes financial
trading.
Access to foreign stocks provides investors
w ith opportunities for diversification; investors
need inform ation about the foreign firm (for ex
ample, its financial stability or successful m anage
ment), how ever, in order to m ake an investm ent
decision. The rapid rise of inform ation technolo-

gy increases the fam iliarity of foreign corpora
tions and their operations. This spread of infor
m ation reduces one of the traditional obstacles
to foreign investm ent and opens up both savings
and investm ent opportunities for firm s and in
dividuals. The payoff is a m ore efficient alloca
tion of capital and, thus, a stim ulus to production
and real output.
Likewise, the advent of alm ost im m ediate
transfer of inform ation around the globe reduces
the inform ational discrepancies betw een m arket
participants. A rbitrageurs, of course, attem pt to
profit from price discrepancies. The m ore people
have access to the same inform ation, how ever,
the m ore likely price discrepancies will be spot
ted and acted upon. The divergence of prices
from th eir no-arbitrage relationship (which p ro
vides profit potentials), will be quickly arbitraged
away as both the quantity and speed of infor
m ation transfer is enhanced .20
A nother benefit of internationalization is ac
cess to m arkets otherw ise inaccessible. As m en
tioned earlier, the Chicago M ercantile Exchange
will be accessible after regular trading hours
through Globex. Not only will U.S. traders now
be able to operate after U.S. trading hours, but
foreign traders can also use Globex to operate
during th eir ow n regular trading hours. New
m arkets enable investors to introduce diversity
into their portfolios in both the type of in stru
m ent and the country from w hich it is issued.
Just as com puterized systems, such as Globex,
facilitate diversification and accessibility, so too
can other m ethods, such as cross-exchange list
ings and cross-m em berships.

R isks in G lobalization

Trading in financial assets, w h eth er done
domestically or across national boundaries, in
volves risk. Some of these risks are m ore im por
tan t in an international setting than a strictly
dom estic setting. They occur prim arily at vari
ous stages of the clearing and settlem ent
process. Unlike risks comm only associated w ith
price uncertainty, the risks in clearing and set
tlem ent procedures involve uncertainty about
the timely paym ent of funds and the transfer of
assets in financial trades.
20An example of a no-arbitrage condition is that the differ
ence between the cash and futures price of a storable
commodity, at any point in time, should reflect carrying
costs of storing the commodity until maturity. If the price
differential exceeds carrying costs, then there exists an in-

An exam ple of a typical securities transaction
can provide a clear illustration. Once a securities
trade is executed, the m em ber firm s involved
subm it the trade inform ation for confirm ation
to the clearing agent. The trade is th en com
pared and m atched by com puter for accuracy
and the inform ation on the trade is sent to the
relevant m em bers on either the day of the
trade or the day after. If both parties concur
w ith the conditions of the trade, the trade is
ready for settlem ent. At present, settlem ent in
securities occurs five days after the trade in the
United States.
Using this example, the next section will briefly
discuss the concepts and institutions in clearing
and settlem ent procedures before introducing
the specific risks of globalization.

Clearing and Settlem ent
P rocedu res
“Clearing” a trade involves th e confirm ation of
the type and quantity of the financial in stru
m ent being traded, th e transaction date and
price, and the identification of th e buyer and
seller. “Settlem ent” m eans the fulfillm ent of the
obligations of the transaction. In equities and
bonds, for example, settlem ent m eans paym ent
to the seller and delivery of the security certifi
cate or transfer of ow nership to the buyer.
The clearing and settlem ent process depends
on the institutions th at facilitate transactions.
Commodities and securities exchanges provide
the facilities for trad ers to conduct their busi
ness, establish and enforce trading rules, collect
and distribute m arket and economic inform ation
about prices, and provide an institutional fram e
w ork for arbitrating and settling disputes.
A nother institution—the clearinghouse—
com pares trades betw een parties and can re
move risk from the settlem ent process .21 A
clearinghouse places itself betw een the buyer
and seller, ensuring th at the buyer receives the
instrum ent purchased and the seller receives
paym ent. T hat is, by becom ing the counterparty
to all trades, the clearinghouse guarantees every
trade. Each participant has a net obligation w ith
the clearinghouse to buy or sell th e security
centive to enter the market, that is, to buy today and sell
at the higher futures price.
21Trade comparison involves confirming and matching the
terms of the trade to ensure accuracy.

JANUARY/FEBRUARY 1992

Table 1
Automated Trading Systems
System’

Sponsor

Purpose

Instruments

Access

New York Mercantile
Exchange (NYMEX)

Computerized screen trading
system to automatically match
trade on a first-in-first-out
basis. Allows traders to
select a standing bid or offer
(but are blind to the
counterparty chosen).

Energy futures and futuresoptions for crude oil, heating
oil, gasoline, propane and
natural gas.

Automated Pit Trading
(APT)

London International
Financial Futures
Exchange (LIFFE)

Intended to extend trading
hours to cover European
trading day. Its aim is to
copy the life of the trading
floor on to a computer screen.

FT-SE 100 index futures and
most of LIFFE's main
contracts.

EJV (Electronic Joint
Venture)

Collaboration between
First Boston, Goldman
Sachs, Morgan Stanley,
Salomon Brothers,
Shearson Lehman and
Citibank

Trading system that allows
dealers to buy and sell
securities electronically
using voice-activated computer
technology. Also provides
price and analytic services.

Currently restricted to
Treasury bills and notes with
maturity of less than three
years. Once established, it
expects to extend coverage
to all maturities.

Euroquote

European Community (EC)
national stock exchanges

A European-wide share trading
system. Will combine price infor
mation from 12 EC exchanges
into an electronic feed for sub
scribers. Eventually, may be
come a full trading system and
integrate Euroquote with a set
tlement system to decrease the
cost and difficulties of settling
cross-country transactions. (This
proposal has since been aban
doned by the chairmen of Eu
rope’s national stock
exchanges.)

EC stocks

Globex

Chicago Mercantile
Exchange (CME), Reuters
and Chicago Board of
Trade (CBOT)

An automated trading system
with anonymous buy and sell
orders that are matched by
price and time. (See text
for details.)

Traded instruments will be
introduced in three separate
waves: 1) financial futures
and options, e.g., Eurodollar
futures, futures and options
on Eurodollar currencies—
Deutschemark, yen, pound
sterling, Canadian and Aus
tralian dollars, Swiss franc,
LIBOR, and U.S. Treasury
bond and note futures and
options; 2) equity-related
products; and 3) agriculturerelated futures.

FEDERAL RESERVE BANK OF ST. LOUIS

Table 1 (continued)
Automated Trading Systems
System1

Sponsor

Purpose

Instruments

Quotron System Inc.

Current testing of the
prototype involves The
Bank of America, Barclays
Bank, Chase Manhattan
Bank, Chemical Bank,
Citibank, Credit Suisse,
Lloyds Bank, Midland Bank,
Morgan Guaranty, National
Westminster, Swiss Bank
Corp. and Union Bank of
Switzerland.

Joint project to develop a
computer system that
automatically matches and
executes foreign exchange
trades.

Trades in foreign exchange.

Swiss Options and
Financial Futures
Exchange (SOFFEX)

Owned by Switzerland’s
three leading stock
exchanges and five
largest banks.

Fully automated trading and
clearing system, where
quotes and orders are
recorded, sorted and
matched automatically.
The computer screen is com
posed of five segments, each
with different types of informa
tion and automatically updated
throughout the day.

Futures and options on the 13
underlying stocks and the
Swiss Market Index (SMI), a
basket of Switzerland's 24
leading stocks.

10 f these systems, only LIFFE’s APT system and SOFFEX are currently in operation.
NOTE: For a more extensive survey of automated systems, see Peter A. Abken, “ Globalization of Stock, Futures, and
Options Markets,” Federal Reserve Bank of Atlanta Economic Review (July/August 1991).

based on h er net position w ith other participants
in the clearinghouse.
A third institution—the depository—is an or
ganization (not necessarily p art of an exchange)
that holds stocks and bonds for safekeeping on
behalf of their ow ners. It has a com puterized
accounting system to record and transfer ow ner
ship of securities betw een participants by in
tegrating a book-entry system w ith a m oney
transfer system .22
The procedures for clearing and settlem ent
vary across countries. At present, there are
th ree com m on m ethods of clearing and settle
m ent. Each involves various com binations of the
th ree central institutions involved in futures and
securities m arkets.
22A book-entry system means a credit or debit to a cus
tomer’s account will transfer securities between buyer and
seller. A money transfer system transfers the funds be
tween the parties to the trade, such as a wire transfer.

The first model is exemplified by the United
Kingdom’s equities m arket. In this model, there
is neither a central depository nor a separate
clearinghouse. Instead, the stock exchange itself
is responsible for trade m atching and confirm a
tion as well as providing a location for the deliv
ery and receipt of securities and paym ents be
tw een trad ers .23
The second model, exem plified by G erm any’s
D eutscher Kassenverein depository system, has
no independent clearinghouse, b u t does have a
centralized depository and a stock exchange that
provides th e m atching and confirm ation of
transactions. Once m atched and confirm ed, the
trade inform ation is sent to the depository for
settlem ent .24
23U.S. Congress, Office of Technology Assessment (1990),
p. 58.
24lbid.

JANUARY/FEBRUARY 1992

The th ird model, as seen in the U.S. equities
m arket, contains all th ree institutions: a stock
m arket, a central depository and an indepen
dent clearinghouse. For example, the National
Securities Clearing Corporation (NSCC), w hich
processes 95 percent of all equities trades in the
United States, is jointly ow ned by the New York
Stock Exchange, the Am erican Stock Exchange,
and NASD .25 The m ajority of the securities for
NSCC m em bers, in turn, are held by the Deposi
tory T rust Company. The stock m arket and clear
inghouse together m atch and confirm transac
tions. The clearinghouse also places itself be
tw een counterparties to trades, then passes
trade inform ation to the depository .26

R isks in Clearing and Settlem ent

Credit (or counterparty) risk occurs w hen one
side of the transaction does not settle in full,
either w hen due or on a later date. The exis
tence of counterparty risk, w hich is of minimal
significance in m any U.S. m arkets because of a
clearinghouse, can be critical in an international
transaction. The clearinghouse, generally well
capitalized, guarantees th at all trades will be honnored. In m any international transactions, how
ever, no clearinghouse exists. Thus, a trad er
lacks inform ation about the counterparty’s relia
bility. Varying regulations on foreign trading
m ay m ake it even m ore difficult to ascertain the
safeguards available to a trad er in that m arket.
Closely related to credit risk is liquidity risk,
w hich is the risk that trades will not be settled
at the appointed time, bu t at some undeterm ined
tim e th ereafter .27 At settlem ent, counterparties
are exposed to both credit and liquidity risks.
Liquidity risk occurs because settlem ent may
not occur on the specified date; credit risk occurs
because the other party may not deliver at all.
Thus, at settlem ent, th e parties m ay not know
w h eth er the problem will be one of liquidity or
credit. The settlem ent of international trades
25U.S. Congress, Office of Technology Assessment (1990),
p. 81.
26For related readings on clearance and settlement systems,
see the monographs prepared by the Payment System
Studies Staff of the Federal Reserve Bank of New York in
the references to this paper.
27A temporary inability to convert assets into cash is often
associated with liquidity risk while bankruptcy of a counter
party is associated with credit risk. For a more detailed
description, see Federal Reserve Bank of New York
(February 1989).
28For further reading on market risks, see Baer and Evanoff
(1990).

FEDERAL RESERVE BANK OF ST. LOUIS

can exacerbate the problem of sim ultaneously
exchanging securities for paym ent because of
tim e zone differences .28
A nother risk, replacem ent cost risk, occurs
w hen the price of th e security changes betw een
trade and settlem ent. W hen one party has de
faulted and the price of the instrum ent changes,
then one of the parties involved would be ad
versely affected by th e price change and suffer
a loss in replacing the transaction. In foreign
m arkets, the potential for adverse changes in
the exchange rate can exacerbate this risk.
O perational risk occurs because of the possible
failure of com puter systems, telecom m unications
or institutionalized procedures during trading.
Given the heavy reliance on technology in ac
cessing financial m arkets abroad, this issue is ex
trem ely im portant in determ ining the success or
failure of new trading systems. The precautions
taken by th e CME for its Globex system —an im
portant p art of its initial proposal to the Com
modity Futures Trading Commission, a govern
m ent regulator of futures exchanges—are a good
example of this .29
Yet another risk, especially w orrisom e to regu
lators, is systemic risk. Systemic risk occurs
w hen credit risks stem m ing from operational or
financial problem s result in agents exiting the
m arket, which, in tu rn , threatens the industry.
The inability of one financial institution to m ake
its paym ents can cause other participants to be
unable to m eet their financial obligations in a
timely m anner. In the banking sector, this is
typified as a ru n from deposits to currency. In
futures and options, it occurs w hen agents no
longer trade through standard channels like an
exchange. For example, if m em bers of an ex
change begin trading elsew here, the financial
stability of the exchange is threatened as m em
bers w ithdraw their financial collateral .30
29Examples of CME precautions include measures to pre
vent unauthorized individuals from accessing the system,
such as four different identification codes; termination of a
computer operator’s session if nonstandard instructions
are entered; and, in the failure of the central computer,
recovery would involve automatic switchover to a back-up
mainframe, taking approximately 60-90 seconds. See
CFTC (1989), pp. 125-32.
30For further reading on systemic risks, see OECD (1991).

GLOBAL COORDINATION
The O ctober 1987 stock m arket crash, w ith
w orldw ide repercussions, revealed w eaknesses
in the clearing and settlem ent system. Many
feared th at the default of a m ajor m arket player
could threaten the financial systems of m any
countries. This prom pted w orld financial lead
ers to w ork tow ard global coordination. The
clearing and settlem ent of trades w as consi
dered one of the m ost crucial aspects of this
coordination.
In 1989, the Group of Thirty issued a report,
Clearing and Settlement System s in the W orld’s
Securities M arkets.31 Based on its examination,
five critical deficiencies in the clearing and set
tlem ent systems across countries w ere iden
tified:
[1] Absence of com patible trade confirm ation
and m atching systems for both dom estic and
international trades;
[2] Varying settlem ent periods across the differ
ent m arkets;
[3] Absence of delivery versus paym ent in some
m arkets;
[4] Absence of standardized trade guarantees;
[5] Absence of book entry processing for settle
m ent of securities transactions in several
m arkets.

Trade C onfirm ation and M atching

T rade confirm ation and m atching, also know n
as trade com parison, is the process of confirm
ing and m atching the term s of a trade to ensure
accuracy (for example, the issue, price, quantity
and counterparties) and is usually done by a
clearinghouse (although som etim es by an ex
change or by the parties them selves, in the for
w ard foreign exchange m arket). If not confirm ed
and m atched, a chain reaction of failed trades is
possible as subsequent trades are m ade on the
assum ption that earlier trades will be success
fully completed.
Rapid trade com parison shortens the am ount
of time betw een w hen the trade is m ade and
w hen it is successfully m atched. This reduces
credit risk by reducing the am ount of tim e an
31The Group of Thirty is a private sector organization that
takes its membership from financial sectors such as ex
changes, banks and investment houses.

agent has to opt for defaulting on a trade. In
the international context, delays of hours in a
dom estic m arket may result in a delay of days
for international trades. Requiring all investors
to obtain m em bership in a trade com parison
system and achieving a com patible system across
international m arkets can reduce the delays and
credit risks involved in diverse systems.

Settlem ent P eriods

The second deficiency is unequal settlem ent
periods, w hich can increase settlem ent risk and
potential default. Settlem ent risk occurs w hen
there are gaps in the timing of paym ents and
receipts on settlem ent date .32 The harm of dif
ferent settlem ent periods is that, as m entioned
earlier, traders or investors w ho are active p ar
ticipants in the m arket m ake later trades contin
gent on th e assum ption of the successful settle
m ent of earlier trades. Hence, the harm is tw o
fold—the default of an earlier counterparty and
the dependence on this trade that could
jeopardize subsequent trades. As w ith m any
trade issues th at require timeliness, delays in
settlem ent can be exacerbated if spread across
different trading hours and tim e zones.
While this is costly in a dom estic m arket, the
investm ent of a U.S. agent dealing in interna
tional m arkets can be even m ore costly because
it is also subject to the econom ic conditions of
foreign countries and exchange rates. Adverse
changes in the exchange rate can tu rn a m inor
loss into a significant one in the presence of
currency risk. Thus, for agents moving betw een
international m arkets, an uncertain settlem ent
period com bined w ith an uncertain exchange
rate can increase financial losses.
The grow ing volume of trades has led to a
num ber of techniques w here, to reduce the
num ber of settlem ent transactions, trades are
not processed one at a time. “Netting” is a sys
tem w hereby transactions are aggregated, so
th at debit and credit positions offset each other,
leaving a participant w ith one final position in
the m arket of owing or being owed. Netting
greatly increases the liquidity of the m arket and
the trad er’s flexibility because, rath er than post
ing collateral for every trade, the trad er is
responsible only for th e net settlem ent debit .33
33Board of Governors of the Federal Reserve System and
the Federal Reserve Bank of New York (1990), p. 40.

32Settlement risk encompasses both liquidity risk and credit
risk.

JANUARY/FEBRUARY 1992

T here are th ree m ain choices for a netting
system. The first is bilateral netting, w hereby
all trades in the same security and betw een the
same parties to the trade are netted to one final
delivery versus paym ent (DVP).34 For example, if
Ralph sells 100 shares of British M ohair to Sam,
th en buys back 75 shares of British M ohair from
Sam, the net position is that Ralph m ust deliver
25 shares to Sam. This is the narrow est of the
th ree netting options.
The second is m ultilateral netting (or daily
netting), which, unlike bilateral netting, allows
for different counterparties in the netting
schem e. In this instance, all trades in the same
security are netted to a final debit or credit po
sition for each participant.
The last option is continuous net settlem ent,
w hereby all trades in a particular security are
pooled by issue to a final debit or credit posi
tion for the day and any unsettled trades are
carried over and offset against the next day’s
trades. In practice, the clearing corporation sub
stitutes as the counterparty to the trade in con
tinuous net settlem ent.
The type of netting system im plem ented de
pends on the volume of the m arket. Establishing
a m ultilateral or continuous net settlem ent sys
tem is a costly procedure, requiring a risksharing arrangem ent am ong m em bers, a clear
ing corporation (as w ith continuous net settle
ment) and pow erful com puter systems to handle
the volume of trades. The costs of such a sys
tem may exceed the costs of operating w ith
only a bilateral system. This is especially tru e in
low-volume m arkets w here bilateral netting can
be a feasible and less costly alternative. A pro
posal for a m ultilateral netting system in the
high-volume foreign exchange m arket w as exa
m ined in 1988 by m em bers of FXNET, a bilater
al netting system .35 R epresentatives of leading
international banks, responding to FXNET’s ques
tionnaire, felt that a m ajor benefit would be to
reduce processing costs .36
This is especially relevant in m arkets expand
ing their foreign m em bership. If netting is desir
able because it reduces the num ber of trades to
s^DVP is a payment system whereby the debits and credits
of a trade are applied to the parties’ accounts simul
taneously.
35For further reading on the netting of foreign exchange
transactions, see Gilbert (1992).
36Minutes of FXNET Multilateral Netting Steering Committee,
(1989).

FEDERAL RESERVE BANK OF ST. LOUIS

process, it becom es even m ore so as m arkets
service no longer just domestic, bu t a grow ing
num ber of foreign clients. In the FXNET ques
tionnaire, respondents stated that, "Cross-border
aspects of m ultilateral netting should be consi
dered early in the process, as they will be m ore
im portant th an w ith bilateral netting .”37 W ith
the addition of cross-border trad ers increasing
the transactions volume a m arket handles, a
netting system would simplify the re
peated paym ents th at w ould be introduced.
W hichever netting system is chosen, the desir
able settlem ent tim e fram e is a rolling settle
m ent system. In such a system, trades settle on
all business days of the week, scheduled the
same num ber of days after the trad e .38 Thus,
the presence of a standardized settlem ent period
and a netting system is a crucial aspect to mov
ing betw een international m arkets w ith security
of settlem ent.

D elivery Vs. P aym ent

The th ird finding by the Group of Thirty is
the absence of delivery versus paym ent (DVP) in
some m arkets. DVP is a tw o-sided paym ent sys
tem th at sim ultaneously debits or credits the
cash account of one m em ber and m akes the
corresponding entry on the securities side of
the transaction. This reduces th e settlem ent risk
that occurs w hen th ere is a discrepancy be
tw een the timing of paym ents and receipts on
settlem ent date.
The Group of Thirty, arguing the need for
prom pt two-sided paym ents, has recom m ended
interim procedures: risk can be reduced by de
livering securities only against a certified check
or by employing a m echanism w hereby delivery
and paym ent are done sim ultaneously although
through different systems. In either case, net
settlem ent of cash and securities is com pleted
by the end of the day.
Even w ithout a form alized DVP, m ethods can
be developed to minimize settlem ent risk by
having both parties to a trade settle their ac
counts sim ultaneously. W ith m arkets in different
tim e zones and, thus, different operating hours,
allowing each side of a trade to settle at a differ37lbid.
38The Group of Thirty recommends the implementation of a
rolling settling system by 1992 so that final settlement oc
curs three days after the trade.

ent time could result in a next-day paym ent, not
one w ithin the hours of the first settlem ent.

S tan dardized Trade Guarantees
The fourth deficiency is the absence of stan
dardized trade guarantees. A trade guarantee
ensures th at all com pared or netted trades will
be settled, based on the conditions on w hich
they w ere com pared, even in the event of coun
terparty default. To assure trade guarantees,
each m em ber of the com parison and netting
systems assum es the default risk of the system.
A standard m ethod of providing a guarantee
is to establish a general clearing fund based on
m em ber contributions. W hen a default occurs,
the losses are first extracted from the defaulting
party’s clearing fund contribution. If that contri
bution does not m eet the full am ount of the loss,
the rem ainder is charged against the clearing
corporation’s general clearing fund.
The international environm ent adds an extra
facet to these guarantees. Since m em bership is
becom ing increasingly international, a m ajor
financial loss can strain the capacity of the cor
poration to handle the failure immediately. Ob
taining perm ission for access to additional
funding, for example, could cause unnecessary
delays. Thus, the m aintenance of additional
sources of funds, like m em ber deposits or ac
cess to bank lines, becom es crucial in an inter
national setting. To ensure the integrity of the
corporation and, thus, the m arket, trade guaran
tees provide a m easure of security and stability
in the face of potential failures.

B ook E ntry P rocessing

The last issue to be addressed in global coor
dination is the absence of book entry processing
for settling securities transactions in several
m arkets. Before addressing book entry, however,
other institutions surrounding this process
should be introduced.
The first of these is a central securities
depository (CSD).39 The prim ary activity of a
CSD is to immobilize and dem aterialize securi
ties so th at they can be processed in the m ore
efficient book entry m ethod. Immobilization of
39The strict definition of a CSD requires that a country
should have only one depository. In practice, however,
more than one may exist. This type of system can be ef
fective as long as there is linkage between the entities to
coordinate trade information. The United States has sever
al depositories.

securities m eans that the physical docum ents
(for example, share certificates) are stored at
the depository, eliminating th eir actual m ove
m ent w hen ow nership changes. Dem aterializa
tion m eans th at no physical securities w ith title
of ow nership are issued. Securities exist solely
as com puter records.
T ransfers of certificates are done by book en
try, w here a simple credit and a balancing debit
to custom ers’ com puterized accounts on the
books of the CSD will transfer securities from
one account to another. Immobilization and de
m aterialization replace the m ore risky and timeconsum ing process of transferring the securities
in paper form w henever a transaction is made.
T ransfers of stocks trade-by-trade introduce a
needless com plication to the clearing and settle
m ent system, w hich becom es even m ore compli
cated if it involves delivering them to investors
w orldw ide.

B ecom m endations b y the Group o f
Thirty
The Group of T hirty has proposed nine
recom m endations found in table 2 to correct
the preceding five deficiencies. The status of the
Group of Thirty recom m endations are listed in
table 3. This table depicts the extent to w hich
21 countries have m ade progress on these
recom m endations. While th e United States has
accom plished m ore th an m ost of the countries
surveyed, the fact th at so m any countries have
not finalized these policies, and m ay not by
1992, has implications for the eventual tim etable
of global coordination .40
Currently, th ere is no well-defined regulatory
structure for the global m arketplace. While
regulatory authorities exist in specific countries—
for example, the Securities and Exchange Com
mission has the regulatory authority, oversight
and arbitration of securities disputes in the U.S.
stock m arket—the international arena has no
similar agency to govern global financial rela
tions. In its absence, voluntary coordination of
clearing and settlem ent system s can help reduce
the risks th at lead to defaults, failures and po
tential disputes betw een legal and regulatory
authorities. Thus, th ere are potential gains if
40ln addition to the Group of Thirty proposal, other groups,
such as the Working Group on Financial Markets have
studied clearing and settlement issues.

JANUARY/FEBRUARY 1992

Table 2
Group of Thirty Recommendations________________________
The following are nine recommendations put forth by the Group of Thirty to correct the deficiencies
it finds in the coordination of clearing and settlement systems.'The numbers in brackets are the rele
vant deficiencies presented in the section, Global Coordination, that these recommendations address.
• By 1990, all comparisons of trades between direct market participants (i.e., brokers, brok
er/dealers and other exchange members) should be accomplished by T+1, (the first day after
the trade). [1]
• Indirect market participants (such as institutional investors or any trading counterparties that
are not broker/dealers) should, by 1992, be members of a trade comparison system that
achieves positive affirmation of trade details. [1]
• Each country should have an effective and fully developed central securities depository, or
ganized and managed to encourage the broadest possible industry participation (directly
and indirectly) by 1992. [3, 5]
• Each country should study its market volume and participation to determine whether a trade
netting system would be beneficial in terms of reducing risk and promoting efficiency. If a
netting system would be appropriate, it should be implemented by 1992. [2]
• Delivery versus payment (DVP) should be employed as the method for settling all securities
transactions. A DVP system should be in place by 1992. [3, 5]
• Payments associated with the settlement of securities transactions and the servicing of
securities portfolios should be made consistent across all instruments and markets by adopt
ing the “ same day” funds convention. [2]
• A “ Rolling Settlement” system should be adopted by all markets. Final settlement should oc
cur on T+3 by 1992. As an interim target, final settlement should occur on T+5 by 1990 at
the latest, save only where it hinders the achievement of T+3 by 1992. [2]
• Securities lending and borrowing should be encouraged as a method of expediting the set
tlement of securities transactions. Existing regulatory and taxation barriers that inhibit the
practice of lending securities should be removed by 1990.2 [4]
• Each country should adopt the standard for securities messages developed by the Interna
tional Organization for Standardization [ISO Standard 7775]. In particular, countries should
adopt the ISIN numbering system for securities issues as defined in the ISO Standard 6166,
at least for cross-border transactions. These standards should be universally applied by
1992.
’ Group of Thirty (March 1989).
2For information on securities lending, see Paul C. Lipson, Bradley K. Sabel, and Frank Keane, “ Secu
rities Lending” (Federal Reserve Bank of New York, August 1989).

the clearing and settlem ent systems operating in
dom estic m arkets are coordinated am ong global
financial m arkets.

CONCLUDING REMARKS
This discussion has attem pted to present an
overview of issues th at are currently of interest
in the globalization of financial m arkets. The
linkage am ong international m arkets is of in
terest both to private investors and to national
governm ents, w ho desire stable dom estic and
international financial sectors.

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

International com petition am ong financial
m arkets is grow ing rapidly and has produced
benefits such as new financial instrum ents, new
m arkets and extended trading hours. These
changes, how ever, are not w ithout costs. Domes
tic rules and regulation are not sufficient
safeguards for a system th at operates in an in
creasingly international environm ent. Financial
and governm ental com m unities are addressing
the need to integrate international expansion to
facilitate the continued safe and profitable
grow th of financial instrum ents and the im por
tan t functions these m arkets serve. It is clear
that m uch w ork rem ains.

Table 3
Current Status of the Group of Thirty’s Recommendations for International
Settlement—Equities
1

Institutional

Recommendation No.

Central

Delivery

Rolling

Comparison Comparison Securities
Country

Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Hong Kong
Italy
Japan
Korea
Netherlands
Norway
Singapore
Spain
Sweden
Switzerland
Thailand
United Kingdom
United States

Securities

versus

Settlement

Same-Day

on T+11

System

Depository

Netting

Payment

on T+5*

Funds

ISO/ISM

Securities
Lending

Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes

No
No
No
Yes
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Yes
Yes

No
Yes
Yes
Yes
No
No
Yes
Yes
No
Yes
No
Yes
Yes
No
No
No
Yes
Yes
Yes
No
Yes

No
No
No
Yes
No
No
No
No
No
No
Yes
No
Yes
No
No
No
No
No
Yes
Yes
Yes

Yes
No
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
No
Yes

Open
Weekly
Fortnightly
T+5
T+3
T+5
Monthly
T+2
T+1
Monthly
T+3
T+2
T+5
T+6
T+5
Weekly
T+5
T+3
T+4
Fortnightly
T+5

No
Yes
Yes
Yes
Yes
No
Yes
Yes
No
Yes
No
No
No
Yes
No
No
No
Yes
Yes
No
No

No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No

Yes
No
Yes
Yes
Yes
No
Yes
No
Limited
Limited
Yes
No
Yes
No
Yes
Limited
Yes
Yes
No
Limited
Yes

SOURCE: Updated by the Office of Technology Assessment, July 1990, from A Comparative View: The Group of Thirty’s
Recommendations and the Current U.S. National Clearance and Settlement System (Morgan Stanley & Co., June 1989).
1T+1 means the first day after the trade.
2T+5 means five days after the trade.

REFERENCES
Abken, Peter A. “ The Globalization of Stock, Futures, and
Options Markets.” Federal Reserve Bank of Atlanta Eco
nomic Review (July/August, 1991), pp. 1-22.
Baer, Herbert L., and Douglas D. Evanoff. “ Payments Sys
tem Risk Issues in a Global Economy,” Federal Reserve
Bank of Chicago Working Paper 90-12 (August 1990).
Baer, Herbert L., Douglas D. Evanoff, and Christine A.Pavel.
“ Payments System Issues in a 24-Hour Global Economy,”
Research in Financial Services: Public and Private Policy
(vol. 3) (JAI Press, Inc., 1991).
“ Big Board After-Hours Trading May Lead to a Two-Tier
Market,” Wall Street Journal, June 13, 1991.
Board of Governors of the Federal Reserve System and the
Federal Reserve Bank of New York. Clearance and Settle
ment in U.S. Securities Markets, Prepared for the Commit
tee on Payment and Settlement Systems, Bank for
International Settlements. (Basle, Switzerland: December
5-6, 1990).

Chesler-Marsh, Caren. “ Globex Countdown,” Euromoney
(March 1991), pp. 33-35.
Chicago Mercantile Exchange. Globex Report: An Update for
the Global Trader (November 14, 1990).
Chicago Mercantile Exchange, Business Development Group.
The Challenges of the 24-Hour Financial Marketplace (May
1989).
Commodity Futures Trading Commission, Division of Trading
and Markets. Chicago Mercantile Exchange’s Proposed Glo
bex Trading System (February 2, 1989).
Diamond, Barbara B., and Mark P. Kollar. 24-Hour Trading:
The Global Network of Futures and Options Markets (John
Wiley & Sons, 1989).
Domowitz, Ian. “ The Mechanics of Automated Trade Execu
tion Systems,” Journal of Financial Intermediation (June
1990), pp. 167-94.
Federal Reserve Bank of New York. "An Overview of the
Operations of the Options Clearing Corporation” (April
1989).

JANUARY/FEBRUARY 1992

________“ Clearing and Settlement Through the Board of
Trade Clearing Corporation” (February 1990).

Hansell, Saul. “ The Computer that ate Chicago,” Institutional
Investor (February 1989), pp. 181-88.

________“ Clearing and Settling the Euro-Securities Market:
Euro-Clear and Cedel” (March 1989).

Khoury, Sarkis J. The Deregulation of the World Financial
Markets: Myths, Realities, and Impact (Greenwood Press,
Inc., 1990).

_______ .“ Exchanges and Clearinghouses for Financial Fu
tures and Options in the United Kingdom” (March 1989).

Nasdaq Stock Market, Business Development. Nasdaq: The
Stock Market for the Next 100 Years (Washington, D.C.,
1991).

_______ .“ The Clearing House Interbank Payments System"
(January 1991).
________Bank for International Settlements. Report on Net
ting Schemes (February 1989).
FXNET Multilateral Netting Steering Committee. Documenta
tion (February 1, 1989).
Gilbert, R. Alton. “ Implications of Netting Arrangements for
Bank Risk in Foreign Exchange Transactions,” this Review
(January/February 1992), pp. 3-16.

National Association of Securities Dealers (NASD). Nasdaq
Fact Book & Company Directory (Washington, D.C., 1991).
Organization for Economic Cooperation and Development
(OECD). Systemic Risks in Securities Markets (Paris: OECD
Publications Service, 1991).
Soffex. Soffex Swiss Options and Financial Futures Exchange
Ag, reprinted from a Supplement to Euromoney (June
1990).

Group of Experts on Payment Systems, Bank for Internation
al Settlements (BIS). Report on Netting Schemes (Basle,
Switzerland, February 1989).

U.S. Congress, Office of Technology Assessment. Trading
Around the Clock: Global Securities Markets and Informa
tion Technology-Background Paper, OTA-BP-CIT-66 (Govern
ment Printing Office, July 1990).

Group of Thirty. Clearance and Settlement Systems in the
World’s Securities Markets (New York and London, March
1989).

Working Group on Financial Markets. Clearance and Settle
ment Reform: The Stock, Options, and Futures Markets are
Still at Risk (Government Printing Office, April 1990).

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

Piyu Yue
Piyu Vue, a research associate at the 1C2 Institute, University of
Texas at Austin, was a visiting scholar at the Federal Reserve
Bank of St. Louis when this article was written. Lynn Dietrich
provided research assistance. The author would like to thank A.
Charnes, Rolf Fare and Shawna Grosskopf for their constructive
comments and useful suggestions. Their DEA computer code led
to a significant improvement of the paper.

Data Envelopment Analysis and
Commercial Bank Performance:
A Primer With Applications to
Missouri Banks
C -JO M M E R C IA L BANKS PLAY a vital role in the

econom y for tw o reasons: they provide a m ajor
source of financial interm ediation and their check
able deposit liabilities represent the bulk of the
nation’s m oney stock. Evaluating th eir overall
perform ance and m onitoring th eir financial condi
tion is im portant to depositors, ow ners, potential
investors, m anagers and, of course, regulators.
Currently, financial ratios are often used to
m easure the overall financial soundness of a bank
and the quality of its m anagem ent. Bank regu
lators, for example, use financial ratios to help
evaluate a bank’s perform ance as part of the
CAMEL system .1 Evaluating the economic p erfo r
m ance of banks, how ever, is a com plicated
process. Often a num ber of criteria such as
'For more details, see Booker (1983), Korobow (1983) and
Putnam (1983).
2The name DEA is attributed to Charnes, Cooper and Rhodes
(1978), for the development of DEA, see Charnes, et al.(1985)
and Charnes, et al. (1978); for some applications of DEA, see
Banker, et al. (1984), Charnes, et al. (1990) and Sherman and
Gold (1985).

profits, liquidity, asset quality, attitude tow ard
risk, and m anagem ent strategies m ust be consi
dered. The changing nature of the banking
industry has m ade such evaluations even m ore
difficult, increasing the need for m ore flexible
alternative form s of financial analysis.
This paper describes a particular methodology
called Data Envelopm ent Analysis (DEA), that has
been used previously to analyze the relative effi
ciencies of industrial firms, universities, hospitals,
m ilitary operations, baseball players and, m ore
recently, com m ercial banks .2 The use of DEA is
dem onstrated by evaluating the m anagem ent of
60 M issouri com m ercial banks for the period from
1984 to 1990.3
Ehlen (1983), Korobow (1983), Putnam (1983), Wall (1983)
and Watro (1989)), actual banking efficiency has received
limited attention. Recently, a few publications have used DEA
or a similar approach to study the technical and scale efficien
cies of commercial banks (e.g., Sherman and Gold (1985),
Charnes et al. (1990), Rangan et al. (1988), Aly et al. (1990),
and Elyasiani and Mehdian (1990)).

3Although there is vast literature analyzing competition and
performance in the U.S. banking industry (e.g., Gilbert (1984),

JANUARY/FEBRUARY 1992

DATA ENVELOPMENT ANALYSIS:
SOME BASICS
DEA represents a m athem atical program m ing
m ethodology that can be applied to assess the effi
ciency of a variety of institutions using a variety of
data. This section provides an intuitive explana
tion of the DEA approach. A form al m athem atical
presentation of DEA is described in appendix A; a
slightly different nonparam etric approach is
described in appendix B.

The DEA Standard f o r Efficiency
DEA is based on a concept of efficiency th at is
widely used in engineering and the natural
sciences. Engineering efficiency is defined as the
ratio of the am ount of w ork perform ed by a
m achine to the am ount of energy consum ed in the
process. Since m achines m ust be operated
according to the law of conservation of energy,
their efficiency ratios are always less than or equal
to unity.
This concept of engineering efficiency is not
im m ediately applicable to economic production
because the value of output is expected to exceed
the value of inputs due to the "value added” in
production. Nevertheless, under certain circum
stances, an economic efficiency standard—similar
to the engineering standard—can be defined and
used to com pare the relative efficiencies of
econom ic entities. For example, a firm can be said
to be efficient relative to another if it produces
either the sa m e level of output w ith few er inputs
or m ore output w ith th e same or few er inputs. A
single firm is considered "technically efficient” if it
cannot increase any output or reduce any input
w ithout reducing other outputs or increasing
other inputs .4 Consequently, this concept of tech
nical efficiency is similar to the engineering
concept. The som ew hat b roader concept of
"economic efficiency,” on the other hand, is
achieved w hen firm s find the com bination of
inputs th at enable them to produce the desired
level of output at m inim um cost .5
4See Koopmans (1951).
5This is also named “ allocative efficiency” because a profit
maximizing firm must allocate its resources such that the
technical rate of substitution is equal to the ratio of the prices
of the resources. Theoretical considerations of allocative effi
ciency can be found in the articles by Banker (1984) and
Banker and Maindiratta (1988).
6lt is common to estimate production functions using regres
sion analysis. When cross-section data are used, the esti-

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

DEA and Technical Efficiency
The discussion of the DEA approach will be
undertaken in the context of technical efficiency
in the m icroeconom ic theory of production. In
microeconomics, the production possibility set
consists of the feasible input and output com bina
tions th at arise from available production tech
nology. The production function (or production
transform ation as it is called in the case of m ultiple
outputs) is a m athem atical expression for a
process th at transform s inputs into output. In so
doing, it defines the frontier of the production
possibility set. For example, consider the wellknow n Cobb-Douglas production function:
(1)
Y = AKaLll_al,
w here Y is the m axim um output for given quanti
ties of tw o inputs: capital (K) and labor (L). Even if
all firm s produce the same good (Y) w ith the same
technology defined by equation 1 , they may still
use different com binations of labor and capital to
produce different levels of output. Nonetheless, all
firm s w hose input-output com binations lie on the
surface (frontier) of the production relationship
defined by equation 1 are said to be technologi
cally efficient. Similarly, firm s w ith input-output
com binations located inside the frontier are tech
nologically inefficient.
DEA provides a similar notion of efficiency. The
principal difference is th at the DEA production
frontier is not determ ined by some specific equa
tion like th at show n in equation 1 ; instead, it is
generated from the actual data for the evaluated
firm s (which in DEA term inology are typically
called decision-making units or DMUs ).6 Conse
quently, the DEA efficiency score for a specific
firm is not defined by an absolute standard like
equation 1. Rather, it is defined relative to the other
firm s under consideration. And, similar to engi
neering efficiency m easures, DEA establishes a
"benchm ark” efficiency score of unity th at no
individual firm ’s score can exceed. Consequently,
efficient firm s receive efficiency scores of unity,
while inefficient firm s receive DEA scores of less
than unity.
mated production function represents the average behavior of
firms in the sample. Hence, the estimated production function
depends upon the data for both efficient and inefficient firms.
By imposing suitable constraints, these statistical procedures
can be modified to orient the estimates toward frontiers. In
this manner, the frontier of the production set can be esti
mated econometrically.

In m icroeconom ic analysis, efficient production
is defined by technological relationships w ith the
assum ption th at firm s are operated efficiently.
W hether or not firm s have access to the same
technology, it is assum ed th at they operate on the
frontier of their relevant production possibilities
set; hence, they are technically efficient by defini
tion. As a result, m uch of m icroeconom ic theory
ignores issues concerning technological ineffi
ciencies.
DEA assum es that all firm s face the same
unspecified technology w hich defines their
production possibilities set. The objective of DEA
is to determ ine w hich firm s operate on their effi
ciency frontier and w hich firm s do not. T hat is,
DEA partitions the inputs and outputs of all firm s
into efficient and inefficient com binations. The
efficient input-output com binations yield an
implicit production frontier against w hich each
firm ’s input and output com bination is evaluated.
If the firm ’s input-output com bination lies on the
DEA frontier, the firm m ight be considered effi
cient; if the firm ’s input-output com bination lies
inside the DEA frontier, the firm is considered
inefficient.
An advantage of DEA is that it uses actual
sam ple data to derive the efficiency frontier
against w hich each firm in the sample can be
evaluated .7 As a result, no explicit functional form
for the production function has to be specified in
advance. Instead, the production frontier is gener
ated by a m athem atical program m ing algorithm
w hich also calculates the optimal DEA efficiency
score for each firm.
To illustrate the relationship betw een DEA and
economic production in its simplest form,
consider the example shown in figure i , in w hich
firm s use a single input to produce a single output.
In this example, there are six firm s w hose inputs
are denoted as Xj and w hose outputs are denoted
7DEA has two theoretical properties that are especially use
ful for its implementation. One is that the DEA model is
mathematically related to a multi-objective optimization
problem in which all inputs and outputs are defined as
multiple objectives such that all inputs are minimized and
all outputs are maximized simultaneously under the tech
nology constraints. Thus, DEA-efficient DMUs represent
Pareto optimal solutions to the multi-objective optimization
problem, while the Pareto optimal solution does not neces
sarily imply DEA efficiency.
Another important property is that DEA efficiency scores
are independent of the units in which inputs and outputs
are measured, as long as these units are the same for all
DMUs. These characteristics make the DEA methodology
highly flexible. The only constraint set originally in the
CCR model is that the values of inputs and outputs must
be strictly positive.

a s y (i = 1 , 2 ,..., 6); their input-output com binations
are labeled by Fs(s =1,2,..., 6). While the produc
tion frontier is generated by th e input-output
com binations for the firm s labeled F1( F3, F5 and F6,
the efficient portion of the production frontier is
shown by the connected line segments. F2 and F 4
are clearly DEA inefficient because they lie inside
the frontier; F6 is DEA inefficient because the
same output can be produced w ith less input.

The Im portance o f Facets in DEA
“Facets” are an im portant concept used to
evaluate a firm ’s efficiency in DEA. The efficiency
m easure in DEA is concerned w ith w hether a firm
can increase its output using the same inputs or
produce the same output w ith few er inputs.
Consequently, only p art of the entire efficiency
frontier is relevant w hen evaluating the efficiency
of a specific firm. The relevant portion of the effi
ciency frontier is called a facet. For example, in
figure 1, only the facet from F, to F3 is relevant for
evaluating the efficiency of the firm designated by
F2. Similarly, only the facet from F3 to F5 is used to
evaluate the firm denoted by F 4.8
The use of facets w ith DEA enables analysts to
identify inefficient firm s and, through com parison
w ith efficient firm s on relevant facets, to suggest
ways in w hich the inefficient firm s m ight im prove
their perform ance. As illustrated in figure 1, F2
can becom e efficient by rising to some point on
the Fi-F3 facet. In particular, it could move to A by
simply using less input, to B by producing m ore
output or to C by both reducing input and
increasing output. Of course, in this example, the
analysis is obvious and the recom m endation
trivial. In m ore com plicated, m ultiple inputm ultiple output cases, how ever, the appropriate
efficiency recom m endations w ould be m uch m ore
difficult to discover w ithout the DEA
m ethodology .9
This constraint, however, has been abandoned in the new
additive DEA formulation. As a consequence, the additive
DEA model is used to compute reservation prices for new
and disappearing commodities in the construction of price
indexes by Lovell and Zieschang (1990).
8ln a multiple dimensional space, the efficiency frontier
forms a polyhedron. In geometry, a portion of the surface
of a polyhedron is called a facet; this is why the same
term is used in DEA. These facets have important implica
tions in empirical studies, such as identification of compe
titors and strategic groups in an industry. See Day, Lewin,
Salazar and Li (1989).
9For alternative measures of efficiency, see appendix B.

JANUARY/FEBRUARY 1992

F igu re 1

Production Frontier and Efficiency Subset
Output Y

tionate decrease in th eir input and o utput places
them inside the production frontier. A propor
tionate increase in th eir input and output is im pos
sible because it w ould move them outside of the
frontier.
Constant retu rn s to scale occur if all p rop or
tionate increases or decreases in inputs and
outputs move the firm either along or above the
production frontier. In figure 2, for example, F3
exhibits constant retu rn s to scale because propor
tionate increases or decreases w ould place it
outside the production frontier.
Since the facets are generated by efficient firms,
the scale efficiency of these firm s is determ ined by
the properties of th eir particular facet. Scale effi
ciencies for inefficient firm s are determ ined by
their respective reference facets as well. Thus, F2
and F4 in figure 1 exhibit increasing and
decreasing retu rn s to scale, respectively.

DEA and Econom ic E fficiency

Scale E fficiency
In addition to m easuring technological effi
ciency, DEA also provides inform ation about scale
efficiencies in production. Because the m easure of
scale efficiency in DEA analysis varies from model
to model, care m ust be exercised. The scale effi
ciency m easured for the DEA m odel used in this
study, how ever, corresponds fairly closely to the
m icroeconom ic definition of econom ics of scale in
th e classical theory of production .10
To illustrate, consider the F!-F3 facet in figure 2.
Firms located on this facet exhibit increasing
retu rn s to scale because a proportionate rise in
their input and output places them inside the
production frontier. A proportionate decrease in
th eir input and output is impossible because it
w ould move them outside of the frontier. This is
illustrated by a ray from the origin th at passes
through the F,-F3 facet at F’2.
Firms located on the F3-Fs facet exhibit
decreasing retu rn s to scale because a prop or
10See Fare, Grosskopf and Lovell (1985). Different DEA
models employ different measures of scale efficiency. See
appendixes A and B for details.

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

While the discussion of DEA in the context of
technological efficiency of production is useful for
illustrative purposes, it is far too narrow and
limiting. DEA is frequently applied to questions
and data th at transcend the n arro w focus of tech
nical efficiency in production. For example, DEA is
frequently applied to financial data w hen
addressing questions of economic efficiency. In
this regard, its application is som ew hat m ore
problem atic. For example, w hen firm s face
different m arginal costs of production due to
regional or local wage differentials, one firm may
a p p e a r inefficient r e la tiv e to a n o t h e r . G iv en th e
potential differences in relative costs th at a firm
may face, how ever, it m ight be equally efficient.
Alternatively, differences th at appear to be due to
econom ic inefficiencies m ay in fact be due to cost
differences directly attributable to the non
hom ogeneity of products. Because of problem s
like these, DEA m ust be applied judiciously.

DEA W indow Analysis
To this point, the discussion of DEA has been
concerned w ith evaluating the relative efficiency
of different firm s at the same time. Those w ho use
DEA, how ever, frequently em ploy a type of sensi
tivity analysis called “w indow analysis.” The
perform ance of one firm or its reference firm s

F igure 2

An Illustration of Scale Efficiencies
Output Y

for a firm over time. Of course, com parisons of
DEA efficiency scores over extended periods may
be misleading (or w orse) because of significant
changes in technology and the underlying
economic structure.

APPLYING DEA TO BANKING:
AN EVALUATION OF 6 0 MISSOURI
COMMERCIAL BANKS
To dem onstrate DEA’s use, it is applied to
evaluate relative efficiency in banking. Financial
data for 60 of the largest M issouri com m ercial
banks for 1984 (determ ined by th eir total assets in
1990) are used. Initially, the relative efficiency of
these banks is exam ined using tw o alternative
DEA models: the CCR m odel and the additive DEA
model. A discussion of these alternative DEA
models appears in appendix A. In extending the
discussion and analysis, how ever, we focus solely
on the CCR model.

M easuring Inputs and Outputs
may be particularly “good” or "bad” at a given tim e
because of factors th at are external to the firm ’s
relative efficiency. In addition, the num ber of
firm s th at can be analyzed using the DEA m odel is
virtually unlimited. T herefore, data on firm s in
different periods can be incorporated into the
analysis by simply treating them as if they
represen t different firms. In this way, a given firm
at a given tim e can com pare its perform ance at
different tim es and w ith the perform ance of other
firm s at the same and at different times. Through
a sequence of such "w indows,” the sensitivity of a
firm ’s efficiency score can be derived for a partic
ular year according to changing conditions and a
changing set of reference firm s . 11 A firm th at is
DEA efficient in a given year, regardless of the
window, is likely to be truly efficient relative to
other firms. Conversely, a firm th at is only DEA
efficient in a particular w indow may be efficient
solely because of extraneous circum stances.
In addition, w indow analysis provides some
evidence of the short-run evolution of efficiency

Perhaps the m ost im portant step in using DEA to
exam ine the relative efficiency of any type of firm
is the selection of appropriate inputs and outputs.
This is partially tru e for banks because there is
considerable disagreem ent over the appropriate
inputs and outputs for banks. Previous applica
tions of DEA to banks generally have adopted one
of tw o approaches to justify their choice of inputs
and outputs .12
The first ‘‘interm ediary approach” views banks
as financial interm ediaries w hose prim ary busi
ness is to borrow funds from depositors and lend
those funds to others for profit. In these studies,
the banks’ outputs are loans (m easured in dollars)
and their inputs are the various costs of these
funds (including interest expense, labor, capital
and operating costs).
A second approach views banks as institutions
that use capital and labor to produce loans and
deposit account services. In these studies, the
banks’ outputs are th eir accounts and transac
tions, while their inputs are th eir labor, capital
and operating costs; th e banks’ interest expenses
are excluded in these studies.

’ ’ This is called “ panel data analysis” in econometrics.
12Some studies have adopted the simple rule that if it
produces revenue, it is an output; if it requires a net ex
penditure, it is an input. For example, see Hancock (1989).

JANUARY/FEBRUARY 1992

O ur analysis of 60 Missouri banks uses a variant
of the interm ediary approach. The banks’ outputs
are interest incom e (IC), non-interest incom e (NIC)
and total loans (TL). Interest incom e includes
interest and fee incom e on loans, incom e from
lease-financing receivables, interest and dividend
incom e on securities, and other income. Non
interest incom e includes service charges on
deposit accounts, incom e from fiduciary activities
and other non-interest income. Total loans consist
of loans and leases net of unearned income. These
outputs represen t the banks’ revenues and major
business activities.
The banks' inputs are interest expenses (IE),
non-interest expenses (NIE), transaction deposits
(TD), and non-transaction deposits (NTD). Interest
expenses include expenses for federal funds and
the purchase and sale of securities, and the inter
est on dem and notes and other borrow ed money.
Non-interest expenses include salaries, expenses
associated w ith prem ises and fixed assets, taxes
and other expenses. Bank deposits are disaggre
gated into transaction and non-transaction depos
its because they have different turnover and cost
structures. These inputs represen t m easures for
the banks’ labor, capital and operating costs. De
posits and funds purchased (m easured by their
interest expense) are the source of loanable funds
to be invested in assets .13

Evaluation o f M issouri Bank
M anagem ent P erf orm ance in 1984
The DEA scores and retu rn s to scale m easures
resulting from applying the CCR and additive DEA
m odels are presented in table l .14 Although the
overall results are similar across the tw o models,
th ere are m inor differences in the individual effi
ciency scores th at may provide inform ation about
the relative efficiency of these banks.
The tw o models differ fundam entally in their
definition of the efficiency frontier. In particular,
the CCR model assum es constant retu rn s to scale,
while the additive m odel allows for the possibility
of constant (C), increasing (I) or decreasing (D)
13This is controversial, however. Some researchers specify
deposits as outputs, arguing that treating deposits as inputs
makes banks that depend on purchased money look artifi
cially efficient (see Berg et al., 1990).
14The results from solving the DEA model also include informa
tion about DEA scale efficiencies, the efficient projection on
the efficiency frontier, slack variables sr+ andS;- and the dual
variables iu, and v:. The “ dual” variables represent “ shadow
prices” for each input and output. That is, they represent the

FEDERAL RESERVE BANK OF ST. LOUIS

returns. Because of this, banks th at are efficient in
the CCR m odel m ust also be efficient in the addi
tive model. As table 1 illustrates for our Missouri
banks, the converse, how ever, is not true.
The overall efficiency score is com posed of
"pure” technical and "scale” efficiencies. In the
CCR model, a firm w hich is technologically effi
cient also uses the m ost efficient scale of opera
tion. In the additive model, how ever, the score
represents only “p u re” technical efficiency. By
com paring the results of the CCR and additive
models, w e can see th at while five of our M issouri
banks w ere technologically efficient, they w ere
not operating at the m ost efficient scale of opera
tion. The reader is cautioned, how ever, th at this
analysis excludes a num ber of factors (such as
dem ographic characteristics of the m arkets in
w hich they operate) th at m ay be im portant in
determ ining the m ost economically efficient scale
of operation.
Since the efficiency scores are defined differ
ently in the CCR and the additive DEA models, it is
not possible to generate a m easure of scale ineffi
ciency using the results in table 1. Nevertheless,
the fact th at the efficiency scores from the tw o
models are quite similar suggests that the scale
inefficiency is not a m ajor source of overall ineffi
ciency for these banks. It appears th at the ineffi
cient banks simply used too m any inputs or
produced too few outputs rath er than chose the
incorrect scale for production .15

A Further A nalysis o f the CCB M odel
An illustration of the use of DEA analysis can be
obtained by considering the data for the bank
w ith the lowest efficiency score, bank 59. The
results for this bank are sum m arized in table 2 .
The reference banks m aking up the facet to w hich
bank 59 is com pared and "lam bda,” a m easure of
the relative im portance of each reference bank in
the facet, are given. The table shows th at three
reference banks com pose the facet for bank 59.
Banks 51 and 39 play the m ajor role and the other
bank is relatively unim portant.
marginal effects of the input and output variables on the
bank’s DEA efficiency score. See appendix A for details.
15Similar results of insignificant scale-inefficiency of U.S. banks
have been reported by Aly et al. (1990).

Table 1
Overall Performance of 60 Missouri Commercial Banks
Evaluated by the CCR and Additive DEA Models (1984)________
________ Efficiency Ratio________

________ Efficiency Ratio________

Bank
no.

CCR
model

Additive
model

Type of
scale1

Bank
no.

CCR
model

Additive
model

Type of
scale

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

.8545
.9228
.9033
.8588
1.0000
.8766
.8709
.8841
.8735
.8115
.9086
.7852
.8338
.9739
.8937
.8292
.8705
.9684
.8439
.9527
.9746
.8681
.9744
.9003
1.0000
.8714
1.0000
1.0000
.8753
.9003

.8825
1.0000
.9129
.9498
1.0000
.9042
.9144
.9323
.9857
.9116
.9856
.8388
.9927
.9024
.9829
.8492
.8211
.9783
1.0000
.9930
1.0000
.8888
.9642
.9646
1.0000
.8406
1.0000
1.0000
.9351
.9319

D
I
I
I
C
I
I
I
I
I
I
I
I
I
I
I
I
I
D
I
I
I
I
I
C
I
c
c
I
D

31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

.8568
.9305
.8509
.8392
.8596
1.0000
.8712
.8707
1.0000
1.0000
.8500
.8867
.8220
.8254
1.0000
.9124
1.0000
1.0000
.9507
1.0000
1.0000
1.0000
.8992
.9443
.9303
.8889
.8434
1.0000
.7600
.8614

.9310
.9537
.8642
.9554
.8986
1.0000
.9813
.9150
1.0000
1.0000
.9453
.9656
.8965
.9069
1.0000
.9889
1.0000
1.0000
.9890
1.0000
1.0000
1.0000
.9705
1.0000
.9931
1.0000
.9338
1.0000
.7824
.9541

D
D
D
I
I
C
I
I
C
C
I
I
I
I
c
I
c
c
I
c
c
c
I
I
I
D
I
c
I
I

Scale efficiency is measured by the CCR model.
C = constant returns to scale
I = increasing returns to scale
D = decreasing returns to scale
’ Determined by the CCR model.

The value m easure in the first colum n in the
low er half of the table gives the value of the
outputs and the inputs for bank 59 in 1984. The
second colum n gives the value m easure th at bank
59 w ould have to achieve in order to be DEA effi
cient. The difference betw een these num bers is
presented in the third colum n .16 Bank 59 should
increase its total loans by 143 percent and its non
interest incom e by 6 percent. Bank 59 should
reduce its four inputs by 26.6 percent of interest
expenses and by 24 percent of the other inputs.

Table 2 also presents a m easure for bank 59
denoted as the “dual.” This m easure is im portant
because the ratio of th e duals for outputs and
inputs shows the tradeoff of increm ents or decre
m ents in inputs and outputs to DEA efficiency.
This is w ith the assum ption th at the bank is free to
vary all of its inputs and outputs. The fact th at the
dual for NIE is large relative to the others suggests
th at the biggest efficiency gains for bank 59 will
come from decreasing non-interest expenses. A
similar analysis can be conducted for each ineffi-

16ln the case of outputs, this difference is a measure of
“ slack.” In the case of inputs, however, the slack variable
is more complicated.

JANUARY/FEBRUARY 1992

cient bank to determ ine its reference banks and
the w ay in w hich it can becom e DEA efficient.

A W indow A nalysis

Table 2
Detailed Results for Bank 59

The available data cover a seven-year span from
1984 through 1990. A three-year period w as
chosen to allow five windows. The w indow s and
the periods they cover are as follows:
window 1 1984 1985 1986
window 2
1985 1986 1987
window 3
1986 1987 1988
window 4
1987 1988 1989
window 5
1988 1989 1990
In each w indow, the num ber of banks is tripled
because each bank at a different year is treated as
an independent firm . Repeating the procedure
discussed above for each window, inform ation
about the evolutions of DEA efficiencies of every
bank during the seven-year period w as obtained.
Table 3 lists the DEA scores of th ree banks by year
in each w indow. The average of th e 15 DEA effi
ciency scores is presented in the colum n denoted
“m ean.” The colum n labeled GD indicates the
greatest difference in a bank’s DEA scores in the
same year b u t in different w indows. The colum n
labeled TGD denotes the greatest difference in a
bank's DEA scores for the entire period.
A bank can receive a different DEA efficiency
score for the same year in different w indows. This
variation in the DEA scores of each bank reflects
both the perform ance of that bank over tim e as

Efficiency Score = .7600
Facet
51
39
27
Lambda = .315 .188 .037
Value if
efficient

Difference

9,627.0
371.9
54,599.8

.0
21.9
32,157.8

.7895E-04
1000E-08
.3704E-10

7,887.0
2,182.0
19,915.0
77,005.0

1C
NIC
TL

Value
measures
9,627.0
350.0
22,442.0

Outputs

5,784.3
1,658.4
15,136.0
58,526.1

2,102.7
523.6
4,779.0
18,478.9

.4762E-09
.2277E-03
.2780E-05
.5815E-05

Dual

Inputs
IE
NIE
TD
NTD

well as th at of other banks. The distribution of
banks by th eir average efficiency over the five
window s is presented in table 4.
Bank 48 w as the only one th at w as efficient for
every year in every w indow over the 1984-90
period. Its average efficiency of 1.00 indicates th at
bank 48 w as a superb bank in the sample DEA
evaluation.
Bank 41, on the other hand, began in the first
w indow w ith scores of 0.84 in 1984, 0.85 in 1985
and 0.89 in 1986. In the second w indow, bank 41
had scores of 0.86 in 1985, 0.90 in 1986 and 0.94 in
1987. Although all of its efficiency scores fluctu-

Table 3
DEA Window Analysis
Efficiency Scores
Bank
48

YR84

YR85

YR86

1.00

1.00
1.00

1.00
1.00
1.00

0.84

0.76

0.85
0.86

0.68
0.70

0.89
0.90
0.90

0.60
0.60
0.59

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

YR87

Summary Measures

0.67
0.70
0.71

0.96
0.98

0.75
0.76

TGD
0.00

0.05

0.14

0.68
0.63
0.63
0.65

0.91
0.94
0.96

1.00
1.00

YR90

0.00

0.92
0.94
0.94
0.96

1.00
1.00
1.00

YR89

MEAN
1.00

1.00
1.00
1.00

YR88

0.04

0.18

1.00

0.98

0.77

Table 4
Distribution of Average DEA Scores
(1984-1990)
Model
CCR

Five-year average
DEA score

Number
of banks

1.00
0.98 — 0.99
0.96 — 0.97
0.93 — 0.95
0.91 — 0.92
0.90
0.88 — 0.89
0.86 — 0.87
0.83 — 0.85
0.80 — 0.82
0.79
0.68

1
8
4
13
7
3
4
10
5
3
1
1

ated slightly in the other th ree w indows, they
tended to increase. W ith a gradual im provem ent
in its DEA efficiency over the seven years, bank 41
w as alm ost fully efficient in the last year, w ith a
DEA score of 0.98. However, its average-efficiency
score of 0.92 does not put it am ong the top 13
banks for the period.
In contrast to the banks previously discussed,
bank 59 displayed relatively erratic and inefficient
behavior over the entire seven-year period. Its
average DEA score of 0.68 w as the lowest of the
60 M issouri banks analyzed.
The w indow analysis enables us to identify the
best and the w orst banks in a relative sense, as
well as the m ost stable and m ost variable banks in
term s of th eir seven-year average DEA scores.

CONCLUDING REMARKS
The DEA m ethodology discussed in this article
has the potential to provide crucial inform ation
about banks’ financial conditions and m anagem ent
perform ance for the benefit of bank regulators,
m anagers and bank stock investors. The DEA
fram ew ork is extrem ely general, perm itting
multiple criteria for evaluation purposes.
M oreover, DEA requires only data on the quantity
of inputs and outputs; no price data are necessary.
This is especially appealing in the analysis of
banking because of the difficulties inherent in
defining and m easuring the prices of banks’ inputs
and outputs.
In addition, the DEA m ethod is highly flexible. In
particular, the selection of inputs and outputs has

considerably few er limitations th an alternative
econom etric approaches. Nevertheless, if the anal
ysis is to be useful, care m ust be exercised in the
selection of inputs and outputs.

REFERENCES
Ahn, T., A. Charnes, and W. W. Cooper. “ Some Statistical and
DEA Evaluations of Relative Efficiencies of Public and
Private Institutions of Higher Learning,” Socio-Economic
Planning Sciences, Vol. 22, No. 6,1988, pp. 259-69.
________“ Efficiency Characterizations in Different DEA
Models,” Socio-Economic Planning Sciences, Vol.22, No. 6,
1988, pp. 253-57.
Aly, Hassan Y., Richard Grabowski, Carl Pasurka, and Nanda
Rangan. “ Technical, Scale, and Allocative Efficiencies in
U.S. Banking: An Empirical Investigation,” Review of
Economics and Statistics (May 1990), pp. 211 -18.
Amel, D., and L. Froeb. “ Do Firms Differ Much?” Finance &
Economics Discussion Series, Federal Reserve Board, #87
August 1989.
Banker, Rajiv D. “ Estimating Most Productive Scale Size Using
Data Envelopment Analysis,” European Journal of Opera
tional Research 217(1984), pp. 35-40.
Banker, Rajiv D., A. Charnes and W. W. Cooper. “ Models for
Estimating Technical and Scale Efficiencies,” Management
Science, Vol. 30, (1984), pp. 1078-92.
Banker, Rajiv D., R. F. Conrad and R. P. Strauss.“ A Compara
tive Application of DEA and Translog Methods: An Illustrative
Study of Hospital Production,” Management Science Vol. 36
(1986), pp. 30-34.
Banker, Rajiv D., and Ajay Maindiratta. “ Nonparametric Anal
ysis of Technical and Allocative Efficiencies in Production,”
Econometrica (November 1988), pp. 1315-32.
Berg, S. A., F. R. Forsund, and E. S. Jansen. “ Deregulation and
Productivity Growth in Norwegian Banking 1980-1988: A
Non-parametric Frontier Approach,” (Bank of Norway, 1990).
Booker, Irene O. “ Tracking Banks from Afar: A Risk Monitoring
System,” Federal Reserve Bank of Atlanta Economic Review
(November 1983), pp. 36-41.
Bovenzi, John F., James A. Marino, and Frank E. McFadden.
“ Commercial Bank Failure Prediction Models,” Federal
Reserve Bank of Atlanta Economic Review (November 1983),
pp. 14-26.
Charnes, A., W. W. Cooper, B. Golany, L. Seiford and J. Stutz.
“ Foundations of Data Envelopment Analysis for ParetoKoopmans Efficient Empirical Production Functions,”
Journal of Econometrics (November 1985), pp. 91-107.
Charnes, A., W. W. Cooper, Z. M. Huang and D.B. Sun. “ Poly
hedral Cone-Ratio DEA Models with An Illustrative Applica
tion To Large Commercial Banks,” Journal of Econometrics
(October/November 1990), pp. 73-91.
Charnes, A., W. W. Cooper and E. Rhodes. “ Measuring Effi
ciency of Decision Making Units,” European Journal of Oper
ational Research Vol. 1 (1978), pp. 429-44.
Day, D. L., A. Y. Lewin, R. J. Salazar, and H. Li. “ Strategic
Leaders in the U.S. Brewing Industry: A Longitudinal Anal
ysis of Outliers,” presented at the conference on New Uses
of DEA, Austin, Texas, September 27-29,1989.
Ehlen, James G. Jr. “ A Review of Bank Capital and its
Adequacy,” Federal Reserve Bank of Atlanta Economic
Review (November 1983), pp. 54-60.

JANUARY/FEBRUARY 1992

Elyasiani, Elyas, and Seyed M. Mehdian. “ A Nonparametric
Approach to Measurement of Efficiency and Technological
Change: The Case of Large U.S. Commercial Banks,”
Journal of Financial Services Research (J u ly 1990),
pp. 157-68.

Congress of the Econometric Society, Barcelona, Spain,
August 21-28,1990.

Fare, Rolf, Shawna Grosskopf, and C. A. K. Lovell. The Meas
urement of Efficiency of Production (Kluwer-Nijhoff, 1985).

Putnam, Barron H. “ Concepts of Financial Monitoring,”
Federal Reserve Bank of Atlanta Economic Review
(November 1983), pp. 6-13.

Fare, Rolf, and W. Hunsaker. “ Notions of Efficiency and Their
Reference Sets,” Management Science Vol. 32 (February
1986), pp. 237-43.
Gilbert, R. Alton. ‘‘Bank Market Structure and Competition, A
Survey,” Journal of Money, Credit, and Banking (November
1984, Part 2), pp. 617-45.
Grosskopf, Shawna. ‘‘The Role of the Reference Technology in
Measuring Productive Efficiency,” The Economic Journal
(June 1986), pp. 499-513.
Hancock, Diana. “ Bank Profitability, Deregulation, and the
Production of Financial Services,” Research Working Paper
89-16, Federal Reserve Bank of Kansas City (December
1989).
Koopmans, T. C. “ An Analysis of Production as an Efficient
Combination of Activities,” in T. C. Koopmans, ed., Activity
Analysis of Production and Allocation, Cowles Commission
for Research in Economics, Monograph No. 13 (John Wiley
and Sons, Inc., 1951).
Korobow, Leon, and David P. Stuhr. “ The Relevance of Peer
Groups in Early Warning Analysis,” Federal Reserve Bank of
Atlanta Economic Review (November 1983), pp. 27-34.
Lovell, C. A. K „ and K. D. Zieschang. “ A DEA Approach to the
Problem of New and Disappearing Commodities in the
Construction of Price Indexes,” presented at the Sixth World

Noonan, John H., and Susan Kay Fetner. “ Capital and Capital
Standards,” Federal Reserve Bank of Atlanta Economic
Review (November 1983), pp.50-53.

Rangan, Nanda, Richard Grabowski, Hassan Y. Aly, and Carl
Pasurka. “ The Technical Efficiency of U.S. Banks,”
Economics Letters Vol. 28, No. 2 (1988), pp. 169-75.
Sherman, H. David, and Franklin Gold. “ Bank Branch Oper
ating Efficiency: Evaluation with Data Envelopment Anal
ysis,” Journal of Banking and Finance (June 1985), pp.
297-315.
Thrall, R. M. “ Overview and Recent Development in DEA: The
Mathematical Programming Approach,” paper presented at
IC2 Institute, Conference Proceedings, University of Texas at
Austin, October 1989.
Wall, L. “ Why Are Some Banks More Profitable Than Others?”
Working Paper Series No. 12, Federal Reserve Bank of
Atlanta (November 1983).
Watro, Paul R. “ Have the Characteristics of High-Earning
Banks Changed? Evidence From Ohio,” Economic Commen
tary, Federal Reserve Bank of Cleveland (September 1,
1989).
Whalen, Gary. “ Concentration and Profitability in Non-MSA
Banking Markets,” Federal Reserve Bank of Cleveland
Economic Review (1:1987), pp. 2-9.
Zukhovitskiy, S. I., and L. I. Avdeyeva. Linear and Convex
Programming (W.B. Saunders Company, 1966).

Appendix A
A Comparison of the CCR

The CCR R atio M odel
The m ost im portant characteristics of the DEA
m ethodology can be presented w ith the CCR Ratio
Model. Consider a general situation w here n deci
sion m aking units, DMUs, convert the same m
inputs into the same s outputs. The quantities of
these outputs can be different for each DMU. In
m ore precise notation, the j-th DMU uses a
m-dimensional input vector, Xj, (i = 1 ,2 ,...,m), to
produce an s-dimensional output vector, yrj
(r = 1,2,..., s). The particular DMU being evaluated
is identified by subscript 0; all others are denoted
by subscript j. The following optim ization problem
is form ed for each DMU:
s

r= 1

i= 1

Max h 0 = I u ry r0 / I V;Xi0
subject to the constraints:
s

1
r = 1

uryrj I I VjXji < 1 , u r > 0, Vj >

i= 1

for i = 1 ,2 ,..., m; r = 1 ,2 ,..., s; j = 1 ,2 ,..., n.

FEDERAL RESERVE BANK OF ST. LOUIS

w here the output weights denoted by u r
(r = l , 2 ,...,s) and the input w eights denoted by V;
(i = 1 ,2 ,...,m) are required to be non-negative
(i.e., u r, v, > 0 for r = 1 , 2 , ..., s; i = 1 , 2 , ..., m).
The “virtual output” is the sum ( 1 u ryri) and the
s

r= 1

"virtual input” is the sum ( 1. VjxJ. The objective
function is defined by h„, that is, the ratio of
virtual output to virtual input. The solution is a set
of optimal input and output w eights. The
maximum of the objective function is the DEA effi
ciency score assigned to DMU0. The first set of
inequality constraints guarantees th at the effi
ciency ratios of other DMUs (com puted by using
the same w eights ur and Vj) are not greater than
unity. The rem aining inequality constraints simply
require all input and output w eights to be positive.
Since every DMU can be DMU0, this optimization
problem is well-defined for every DMU. Because
the weights (vi; u r) and the observations of inputs
i= 1

and outputs (xii; yrj) are all positive and the con
straints m ust be satisfied by D M U 0, the maximum
value of h 0can only be a positive num ber less than
or equal to unity. If the efficiency score h 0 = 1,
D M U 0 satisfies the necessary condition to be D EA
efficient; otherw ise, it is D EA inefficient.
The above problem cannot be solved as stated
because of difficulties associated w ith nonlinear
(fractional) m athem atical program m ing. Charnes
and Cooper, how ever, have developed a m athe
matical transform ation (the so-called “CC transfor
m ation”) w hich converts the above nonlinear
program m ing problem into a linear one. Existing
duality theory and simplex algorithm s in linear
program m ing are used to solve the transform ed
problem .1
For a linear program m ing problem , th ere exists
a pair of expressions w hich are "dual” to each
other. The CCR ratio m odel is form ed by problem
1 and problem 2 below:
Problem 1:
M inh 0 = 0O- e( iZ1 s ] + rI=1 s+
r)
=
subject to
m

0 0 ^io —
n

j= i

^ij

j= l

^ i

yrj A, - s+ = yr0, A, > 0 ,
r

i-i

yri A, and

0oxio >

i-i

X„ Aj;

for r = 1 ,..., s; i = 1 ,..., m.
2. Technical efficiency will be achieved if, and
only if, all of the following conditions are satisfied:
60 = 1 and s+ = 0, = 0
r
for i = 1 ,.., m; r = 1 ,.., s.
The condition 90 1 ensures that D M U 0 is located
on the production frontier; the conditions s+ = 0
r
and s] = 0 exclude situations such as F6 in figure 1
of the text.
3. The constant retu rn s to scale condition for
=

s] > 0,

s+ >
r

DMU0 occurs if 1 A = 1, otherwise; X A > 1
:
j
0;

r= 1

y rtl <

—

for i = 1,.., m; r = 1,.., s; j = 1,.., n.
Problem 2:
Max Y0 = S MrYro
subject to
m

to ensure th at all of the observed inputs and
outputs have positive values or shadow prices and
th at the optimal value h 0is not affected by the
values assigned to the so-called "slack variables"
( s + or S7). 2
r
The main conclusions from the CCR model are
sum m arized as follows:
1. The optim al values of s+, S7, and A, via
problem 1 m ust be positive. The following inequal
ities should then be satisfied:

j= i

implies decreasing retu rn s to scale; iI- i A ,< 1
implies increasing retu rn s to scale.
4. An adjustm ent can be m ade in order to move
(or project) inefficient D M U 0 onto the efficiency
frontier. The projection (x', y ‘) in the CC R m odel is
form ed by the following form ulas:

I V xi0 = 1 , 1 M - 2 ^x,, < 0,
;
1=
1
r =1 .-yr, 1=1

Mr >£. Vj > £
for i = 1 ,.., m; r = 1 ,.., s; j = 1 ,.., n.
As before, the subscript 0 represents the D M U
being evaluated, x(j denotes input i, yrj denotes
output r of DM U j, and /ur and represent the
weights for outputs and inputs, respectively. An
arbitrarily small positive num ber, £, is introduced

x i0‘

0 n X io

yro‘ = y ro +

“

S i

r r = 1.

•- m

The differences (xi0 - xi0'), i = 1 ,.., m, represent
am ounts of inputs to be reduced; (yro' - yr0),
r = l,..,s, represen t th e am ounts of outputs to be
increased in order to move D M U 0 onto the effi
ciency frontier. Hence, these differences can
provide diagnostic inform ation about the ineffi
ciency of D M U 0.

1This also opens the way for many different DEA models
which are refined, more flexible or more convenient for
computations. These DEA models (BCC model, additive
DEA model, cone ratio DEA model, CCW model) and their
mathematical characteristics are beyond this paper.
2For the e-Method, see Zukhovitskiy et al. (1966), pp. 46-51.

JANUARY/FEBRUARY 1992

5. Problem 1 is defined as the "prim al” problem
while problem 2 is the "dual.” The dual variables
have the economic interpretation of “shadow
prices.” The value of vi indicates the m arginal
effect of input x i0on the DEA efficiency score. The
value of fu, indicates the m arginal effect of output
yr on the DEA efficiency score. A com parison of
these dual variables provides inform ation on the
relative im portance of inputs and outputs in the
DEA evaluation.
6 . In the CCR model, problem 1 (or problem 2) is
solved for each DMU. Theoretically, th ere is no
limitation on how m any DMUs can enter the DEA
model. Hence, the DEA model can perform an effi
ciency diagnosis for m any DMUs.
W hy is this approach referred to as data
envelopm ent analysis? The tw o inequalities in
conclusion 1 ,
n

yro < i-i yrjAj and 0oxio > j-i x,^,
Z
I
for r = 1 ,..., s; i = 1 ,..., m
are constraints to be satisfied for the optimal solu
tion. The first inequality implies th at the output of
DMU 0should not exceed the linear com bination of
all observed output yrj; thus, the optimal solutions
will create a hyperplane to envelop the output of
DMU 0from above. Similarly, the second constraint
can be in terpreted such th at the optimal solutions
create another hyperplane w hich envelops the
input of DMU 0from below. Since both outputs
and inputs of the DMU evaluated are enveloped
from above and below, the nam e DEA exactly
m atches the geom etric interpretation of the
procedure.
To see how this w orks, assum e that th ere is a
group of DMUs th at produces the same outputs
using the same inputs, bu t in varying am ounts. In
ranking th eir efficiencies of DMUs, DEA assigns
weights to the outputs and inputs of each DMU.
These w eights are neither predeterm ined nor
based on p rior inform ation or preferences of the
decision m akers. Instead, each DMU receives a set
of "optim al” weights th at are determ ined by
solving the above m athem atical program m ing
problem . This procedure generates a DEA effi
ciency score for the DMU evaluated based on the
solution value for the input and output weights.
A set of constraints guarantees th at no DMU,
including the one evaluated, can obtain an effi
3See Charnes et al. (1985).

FEDERAL RESERVE BANK OF ST. LOUIS

ciency score th at exceeds unity. In this way, DEA
derives a m easure of the relative efficiency rating
for each DMU in the cases of m ultiple input and
output.

The A dditive M odel

Among DEA models, the additive m odel has
been im portant in applications. The additive
m odel can be form alized as the following tw o
problem s, w hich are dual to each o th er .3
Problem 3:
m

i= 1

r= 1

Max I st / |x j + I s \ l |y j
subject to
n

xi0 - i-i Xjj Aj — s 7 = 0, i-i yri Aj - s+ = y*,,
I
I
r
2

A, =

1 ,

A, >0, sj >0, s+ > 0,

for i = 1 ,.., m; r = 1,.., s; j = 1,.., n.
Problem 4:
s

r=1

i~l

Min I /iry r0 + I VjX.o + u 0
subject to:
s

r= 1

M + 1
,-yn

1= 1

VjXjj + u 0 > 0,

V > 1 / |xi0|, H < i / l y j ,
i
r

for i = 1 ,.., m; r = 1 ,.., s; j = 1 ,.., n.
Com pared w ith the CCR model, the additive
m odel has introduced another constraint
I I = 1 and a new variable u 0. The new
i-i
constraint in problem 3 ensures th at the efficiency
frontier is constructed by the convex com bina
tions of original data points rath er than a convex
cone as in the CCR model. The new variable u 0in
problem 4 is used to identify retu rn s to scale. The
other variables in the additive m odel have
interpretations similar to the CCR model.
In addition, th ere is a difference in the w ay the
additive model and the CCR ratio model locate the
efficient reference point on the facet. In figure
A .l, an output isoquant consists of input com bina
tions for five firm s (F„ F2, F3, F4and F.) in the case
of one-output (y) and tw o-input (xt and x2). Point F5
represents an inefficient DMU w hich uses m ore of
x, and x 2to produce the same am ount of output as
n

Figure A.1
The Difference Between CCR and Additive DEA Models
Inpu t X 2

to the intersection point B divided by the length
from the origin to Fs. In the additive model,
how ever, the reference efficient point on facet
F2-F3is denoted by A, w hich is determ ined by
maximizing the sum of the slacks, s, + s2. Geom etri
cally, the slack variables are expressed by the
horizontal line starting from Fs and the vertical
line extending to the facet F 2-F3. Point A is selected
such th at the sum of the lengths of the horizontal
and vertical lines are maximized. The DEA effi
ciency score in th e additive m odel th at we used is
com puted by the following form ula:
m

1*1

r - 1

i«1

r= 1

r = 1

( I x(o + I y'J /( I x l0 + I y r t + I

2 s;).

w here x -0and y r0are corresponding inputs and
'
outputs of the efficient reference point, such as
point A.
The DEA scale efficiency in the additive model is
identified by a variable u 0in problem 4 in accor
dance w ith the following criteria:
O

V .________________________________________________

In pu t X j

its efficient reference DMUs, F2and F3. By the CCR
ratio model, the efficiency score is determ ined via
a value h0, w hich can be interpreted in term s of
the ray from the origin to F5. That is, h 0 is ex
pressed by the length of the ray from the origin

If u 0 = 0, DMU 0has constant retu rn s to scale;
otherw ise,
u 0 > 0 implies decreasing retu rn s to scale;
u 0 < 0 implies increasing retu rn s to scale.
The value of variable u 0is part of an optimal solu
tion of the additive model and is produced by the
com puter code such th at facet rate = - u 0.

Appendix B
Data Envelopment Analysis: An Alternative Approach
In m easuring and evaluating technical and scale
efficiencies th ere are tw o basic approaches: the
DEA technique developed by Charnes, Cooper and
others in operations research and the approach
developed by Farrell, Fare and Grosskopf, am ong
others, in econom ics .1 The latter approach is
based upon a set of axioms on production tech
nology to define the concept of efficiency. Some
connections of the tw o approaches have been
investigated by Banker, Charnes and Cooper
(1984) and by Fare and H unsaker (1986).

Both approaches share the characteristics that
th ere is no need to specify a production function
or cost function and to estim ate the param eters.
T herefore, they are nonparam etric, nonstochastic
techniques th at can be used to construct a
m ultiproduct frontier relative to w hich the effi
ciency m easures of the entities in the sam ple are
calculated. Because the frontier in these
approaches is generated by data and all observa
tions are enveloped by the frontier, both
approaches can be viewed as Data Envelopm ent

'See Fare and Hunsaker (1986); Fare, Grosskopf and
Lovell (1985).

JANUARY/FEBRUARY 1992

Analysis. In this appendix, some of the differences
and similarities am ong the CCR and the additive
models and the Farrell or Russell models are
discussed.
The choice of efficiency reference on the rele
vant frontier is a m ajor difference am ong these
DEA models. In the Farrell or Russell models,
th ree m easures of technical efficiency can be
defined: input, output and graph efficiency
m easures.
Using the input efficiency m easure, the ob
served output vector is fixed and the search for
efficient reference is constrained to proportion
ally reducing inputs until the efficient frontier is
reached. The “ratio of contraction,” as it is called,
is the ratio of the particular input to be efficient to
the cu rren t level of inputs (in the Farrell input
model).
Using the output efficiency m easure, the ob
served input vector is fixed and the outputs p ro
portionally expanded until th e efficient frontier is
reached. The "stretch ratio” of the output, as it is
called, is th e ratio of efficient output to the cu rren t
level of output (in the Farrell output model).
For the graph efficiency m easure, both input
and output vectors are varied. Inputs are reduced
and outputs are expanded, both proportionally,
w ith the input ratio reciprocal to the output ratio.
In the case of figure 1 in the text, A is the refer
ence point for the input efficiency m easure, B is
the reference point for the output efficiency
m easure and C m ight be the reference point for
the graph efficiency m easure. These th ree effi
ciency m easures can be classified as radial
because proportional changes of inputs and/or
outputs are used in defining them .
To illustrate the input efficiency m easure, ray
OF3in figure 1 of the text is used to represen t the
optimal scale that would be generated by long-run
com petitive equilibrium . The overall input effi
ciency m easure is defined w ith respect to the ray
OF3, while the input pure technical efficiency is
defined w ith respect to the line segm ent connect
ing F1( F 3 and F5. The m easure of input overall
technical efficiency, KD/KF,, can be decom posed
into the m easure of pure technical input efficiency
given by the ratio KA/KF2 and the m easure of input
scale efficiency given by the ratio KD/KA. W hen
2See Grosskopf (1986).

FEDERAL
http://fraser.stlouisfed.org/ RESERVE BANK OF ST. LOUIS
Federal Reserve Bank of St. Louis

the scale efficiency equals unity, the constant re
tu rn s to scale occur; otherw ise non-increasing or
varying retu rn s to scale hold.
It is clear from these examples that, in general,
these radial efficiency m easures will be different.
M oreover, th ere is nothing to guarantee th at a
firm th at is output efficient by this m easure is also
input efficient or vice versa. For example, the firm
denoted by F6in figure 1 of the text is output effi
cient by the output efficiency m easure, but is not
input efficient (see Fare, Grosskopf and Lovell
(1985)). However, the Farrell input efficiency
m easure is reciprocal to the Farrell output effi
ciency m easure, if and only if, the technology is
hom ogeneous degree one. Because this condition
is satisfied by constant retu rn s to scale tech
nology, the Farrell input and output efficiency
m easures are "identical” in this case. For models
w ith other technologies, simple relationships
betw een input and output efficiency m easures do
not hold.
An im provem ent of the Farrell or Russell models
over the others is the use of non-radial efficiency
m easures. The use of proportional changes of
inputs and/or outputs in searching for efficient
reference is abandoned.
M oreover, different piecewise linear technology
can be accom m odated in both Farrell and Russell
models to m eet the needs of various users. For
example, to m easure scale efficiency w e can use
constant retu rn s to scale, non-increasing retu rn s
to scale or varying retu rn s to scale technologies.
These technology constraints can be easily im posed
by corresponding restrictions on the "intensity
param eters” in the Farrell or Russell models.
In the CCR or additive DEA m odel discussed in
appendix A, how ever, only one efficiency m easure
is defined: the CCR m odel uses the radial m easure
of efficiency while the additive m odel uses the
non-radial m easure.
Geometrically, the efficiency frontier w ith cons
tant retu rn s to scale technology is a convex cone,
bu t it is a convex hull in cases of both non-increasing and varying retu rn s to scale. In general, these
constraints on technology form a chain such that
one efficiency frontier is enveloped by another.
Consequently, the associated efficiency m easures
are com patible and nested .2

As is presented in appendix A, the CCR model
has a convex cone efficiency frontier that implies
technology w ith constant retu rn s to scale. The
additive m odel uses a convex hull as its efficiency
frontier th at is associated w ith the varying returns
to scale. Even though the efficiency frontier of the
additive model is enveloped by the efficiency
frontier of the CCR model, the efficiency scores
given by both models are not com patible because
one uses a radial m easure while the other uses a
non-radial m easure. The efficiency ratio of the
CCR m odel is identical to the Farrell input effi
ciency m easure (or reciprocal output efficiency
m easure) w ith constant retu rn s to scale technol
ogy. Although both additive and Russell models
define non-radial efficiency m easures, the defini
tions are not identical. Hence, the efficiency m ea
sures given by these m odels are not compatible.

W ith our 1984 data of 60 M issouri com m ercial
banks, we used the Farrell m odel w ith input and
output efficiency m easures and different tech
nology constraints. The overall technical efficien
cies and scale efficiencies are presented in table
B.l. The reported results are based upon the input
m easure of efficiency.
Com paring table B.l w ith table 1 in the text, we
can see that the CCR model and the Farrell input
m odel give identical technical efficiency m easures
and classification of retu rn s to scale. Farrell input
scale efficiency m easures in table B.l indicate that
the scale inefficiency was not a m ajor source of
technical inefficiency in this group of banks. For a
few of the banks in the sample, how ever, the scale
inefficiency might be a problem .

Table B.1
Farrell Technical and Scale Efficiencies of 60 Missouri
Commercial Banks (Input Efficiency Measure)_______
Bank
no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

Efficiency
measure

Scale
measure

Type of
scale1

Bank
no.

Efficiency
measure

Scale
measure

Type of
scale1

.8545
.9228
.9033
.8588
1.0000
.8766
.8709
.8841
.8735
.8115
.9086
.7852
.8338
.9739
.8937
.8292
.8705
.9684
.8439
.9527
.9746
.8681
.9744
.9003
1.0000
.8714
1.0000
1.0000
.8753
.9003

.8556
.9228
.9177
.9852
1.0000
.9983
.9308
.9980
.9731
.9943
.9962
.9740
.9572
.9994
.9550
.9938
.9714
.9939
.8439
.9867
.9747
.9306
.9843
.9877
1.0000
.9930
1.0000
1.0000
.9622
.9538

D
I
I
I
C
I
I
I
I
I
I
I
I
I
I
I
I
I
D
I
I
I
I
I
C
I
c
c
I
D

31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

.8568
.9305
.8509
.8392
.8596
1.0000
.8712
.8707
1.0000
1.0000
.8500
.8867
.8220
.8254
1.0000
.9124
1.0000
1.0000
.9507
1.0000
1.0000
1.0000
.8992
.9443
.9303
.8889
.8434
1.0000
.7600
.8614

.9813
.9964
.8777
.9738
.9849
1.0000
.9870
.9316
1.0000
1.0000
.9853
.9637
.9836
.9887
1.0000
.9769
1.0000
1.0000
.9983
1.0000
1.0000
1.0000
.9758
.9443
.9762
.8889
.9427
1.0000
.9565
.9830

D
D
D
I
I
C
I
I
C
C
I
I
I
I
C
I
C
C
I
C
C
C
I
I
I
D
I
C
I
I

Where C = constant returns to scale
I = increasing returns to scale
D = decreasing returns to scale
1Determined by the Farrell input measure.

JANUARY/FEBRUARY 1992

The Federal Reserve Bank of St. Louis

1991 J Research

W orking Paper Series

Number

Author(s)

Title

91-001A

Mark D. Flood

Market Structure and Inefficiency
in the Foreign Exchange Market

91-002A

James Bullard and
Alison Butler

Nonlinearity and Chaos in Economic
Models: Implications for Policy
Decisions

91-003A

James Bullard

Collapsing Exchange Rate Regimes:
A Re-interpretation

91-004A

James Bullard

Learning Equilibria

91-005A

David C. Wheelock and
Subal C. Kumbhaker

Which Banks Choose Deposit
Insurance? Evidence of Adverse
Selection and Moral Hazard in a
Voluntary Insurance System

91-006A

David C. Wheelock

Regulation and Bank Failures: New
Evidence from the Agricultural Col
lapse of the 1920s

Note:

Single copies of Research papers are
available upon request by writing to:
Research and Public Information Dept.
Federal Reserve Bank of St. Louis
P.O. Box 442
St. Louis, Missouri 63166-0442

F e d e ra l R e s e rv e B an k o f St. L o u is

Post Office Box 442
St. Louis, Missouri 63166

The R e v ie w is pu b lish ed six
tim es p e r y e a r b y the R esearch
an d P ublic In form ation
D epartm en t o f the F ederal
R eserve R ank o f St. Louis.
Single-copy su b scrip tio n s are
available to the pu b lic f r e e o f
charge. Mail requ ests f o r
su bscrip tio n s, b a ck issues, o r
a d dress changes to: R esearch
an d P ublic In form ation
D epartm ent, F ederal R eserve
Rank o f St. Louis, P.O. Rox 442,
St. Louis, M issouri 63166.
The view s ex p ressed are th ose
o f the individual au thors an d do
n ot n ecessarily reflect officia l
p o sition s o f the F ederal R eserve
R ank o f St. Louis o r the F ederal
R eserve System . A rticles herein
m ay b e reprin ted p ro v id e d the
sou rce is credited . P lease p ro vid e
the R an k’s R esearch an d P ublic
In form ation D epartm en t w ith a
co p y o f reprin ted m aterial.

Full text of Review (Federal Reserve Bank of St. Louis) : January/February 1992

FRASER