Full text of Business Review (Federal Reserve Bank of Philadelphia) : Third Quarter 2005

View original document

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

The Changing Patterns of Payments
in the United States
Based on a speech delivered by President Santomero at the 25th SUERF Colloquium: “Competition and Profitability in European Financial Services: Strategic, Systemic, and Policy Issues,” Marjolin Lecture, Madrid, Spain, October 16, 2004

BY ANTHONY M. SANTOMERO

lthough the origins and evolution of payment
structures in the United States and Europe
are quite different, both systems are moving
toward more electronic forms of payment.
In “The Changing Patterns of Payments in the United
States,” President Santomero highlights the differences
between U.S. and European payments infrastructure;
discusses how the roots and evolution of the U.S.
payments system differs from Europe’s; and outlines the
likely path of the U.S. payment system and the Fed’s role
in it.

As a career academic and current
U.S. central banker, I would like to
offer commentary on some changes
taking place in the financial services
industry in the United States. Specifically, I would like to discuss what
is happening in the U.S. payments
system. The changes occurring in the
U.S. are interesting in their own right
and as a point of comparison and
contrast with what is happening in the
European payments arena.
As anyone who knows the sector
would readily admit, the origins and
evolution of payment structures in the
United States and Europe could not
be more different. Now, however, we
are beginning to see signs that the two
systems are starting to converge. Both
are moving toward more electronic
payment services through a number of
vehicles. In other words, two systems
that started out quite differently are
converging toward similar systems. On
www.philadelphiafed.org

the U.S. side, the pattern of payments
is indeed evolving — some might say
it is experiencing a radical change.
America’s paper-based payments system is giving way to a new realm of
electronic payments vehicles — a transition that has already occurred in Europe. Indeed, there has been quite a bit
of diversity in the forms of payments
used in the U.S. However, as is typical
in this area, change has been, and will
be, greatly affected by our financial
history and its legacy systems.
This presents the Federal Reserve
System with many challenges because,
unlike most central banks in Europe,
the Federal Reserve is not only a regulator but also a service provider. It has
been a vital part of the retail payments
system since its founding more than 90
years ago. From its inception, the Federal Reserve has had a dual role as the
central bank charged with ensuring
the integrity of the payments system

and as a participant in its evolution.
Over time, the Fed’s role in payments and that of European central
banks are likely to converge as well.
The Fed’s role in paper processing will
likely diminish over time as checks
recede in both absolute volume and
relative importance in our retail payments system. As this occurs, it will
further our resemblance to the central
banks of Europe. Over time, both the
Fed and European central banks will
concentrate more of their efforts on
their services on large-dollar gross
settlement, with TARGET2 likely following the evolution of Fedwire.1

TARGET is the system used in the countries
of the European Union for the settlement of
central bank operations, large-value euro interbank transfers, and other euro payments.
TARGET2, the next generation of the system,
is currently under development.
1

Anthony M. Santomero, President,
Federal Reserve Bank of Philadelphia

1 Q3 2005 Business Review

With that prologue, I would like
to share my thoughts on payments,
concentrating on three issues:
• the current status of the U.S.
payments infrastructure vis-à-vis
Europe’s;
• how the roots and evolution of the
U.S. payment system differ from
those of Europe; and
• the likely future path of the U.S.
payment system and the Fed’s role
in it, with an emphasis on how we
are likely to resemble Europe and
how we will be different.

Payment cards account for the
remainder of retail payments, and
there are similarities and differences
between Europe and the United States.
The similarities lie in the use of debit

THE CURRENT STATE OF
PAYMENTS TECHNOLOGY IN
THE U.S.
Historically, Americans and Europeans have long relied on an entirely
different mix of payments vehicles. For
example, Europeans use cash roughly
twice as much as Americans. However,
looking at noncash transactions gives
evidence of where the differences truly
lie. In Europe, half of all noncash retail
payments are made through a Giro
system and only about 15 percent are
made by check. In the United States,
it is almost exactly the reverse. Half of
all noncash retail payments are made
by paper check and less than 10 percent are made through ACH, which
is the American version of a Giro
system.2
The dominance of the Giro in
Europe and of the check in the United
States is a long-standing feature of our
respective payment systems. The history of how this dominance evolved
is interesting and instructive, as I will
elaborate later.

cards. Debit cards, a relatively recent
innovation, have caught on quickly
both in Europe and in the U.S., and
they now account for about a quarter
of noncash retail payments in both
places. The differences are in our use
of credit cards. Credit cards have long
been an important payment vehicle in
the U.S. and, at present, account for
about a quarter of our noncash retail
payments. In Europe, credit cards are
used less frequently — in less than
10 percent of transactions, though
I would note that Europeans’ use of
credit cards has picked up in recent
years.3
The long-standing success of the
credit card in the U.S. and the rapid
rise of the debit card in both Europe
and the U.S. are also interesting and
instructive stories, which I will touch
on as well. First, let’s begin with the
story of the Giro and the check.
The European Structure. To
understand the dominance of the Giro
in Europe and the check in the U.S.
we have to go back about 100 years to
the late 19th and early 20th centuries.

Data from Bank for International Settlements,
cited in Statistics on Payment and Settlement
Systems in Selected Countries, March 2004 (figures for 2002), prepared by the Committee on
Payment and Settlement Systems of the Group
of 10 Countries.

2 Q3 2005 Business Review

At that time, European banks did
not provide routine payment services.
They served primarily as merchant
banks and as private banks for wealthy
individuals.

Credit cards have long been an important
payment vehicle in the U.S. and, at present,
account for about a quarter of our noncash
retail payments.

In the late 1800s, local post offices began establishing postal Giro
systems as a convenient way for common people to deposit savings, and
these systems later evolved to allow
people to remit and receive payments.
The system was successful in that it allowed every post office savings account
holder to make and receive payments
both locally and nationally. This revolutionary achievement rendered noncash payment transactions accessible
to large sectors of the population.
Later, in the 1950s and 1960s,
European banks sought to broaden
their business lines to encompass the
mass market as a way to expand their
deposit base to fund loans. This meant
providing routine payment services to
customers; so bank Giro systems were
created to handle the volume.
This evolution occurred relatively
smoothly and rapidly as a result of Europe’s concentrated banking industry
— a few banks operating nationwide,
cooperating closely with each other.
At the same time, European governments wanted to establish payment
systems that minimized costs and
maximized access. The advent of technological advances created such opportunities through electronification.
When technology made it economical
to replace paper Giros with electronic
Giros, European governments pushed

www.philadelphiafed.org

for the transition, and the concentration of the payments system in the
hands of the postal service and a few
national banks made it relatively easy
to accomplish. Because of its Giro
system, Europe had, or could easily
set up, centralized accounts for credit
transfers. In short, European central
banks encouraged — and in some cases mandated — the use of electronic
Giro systems.
The U.S. Structure. In contrast,
the U.S. payments system evolved
quite differently from Europe’s. Historically, U.S. banks tended to provide
services, including payment services,
to the broad spectrum of people and
businesses. On the loan side, commercial banks focused on commercial and
industrial lending, but they took deposit balances from all economic strata.
In early America, the geographical
expanse of the country encouraged a
fragmented system wherein state banks
issued their own notes. Entry into the
banking business was relatively easy,
but bank branching was very restricted.
Banks were prohibited from branching
outside their home state, and in many
states, branching was restricted still
further. As a consequence, a region
would be served by a relatively large
number of banks, but there were no
banks operating nationwide.
To effect transactions, people paid
one another with paper checks drawn
on their bank or paper currency notes
issued by their bank. The banks would
then clear these checks and notes
among themselves.
With so many individual
banks spread out across such a big
country, and banks clearing paper
instruments among themselves,
effecting transactions outside the
local area was cumbersome. When
someone received a bank check or a
bank note as payment and deposited it
at his bank, the bank would discount
the instrument’s value based on the

www.philadelphiafed.org

cost of presenting it to the “drawn on”
bank for payment and some assessment
of the creditworthiness of the “drawn
on” bank. The farther away the bank,
the less familiar its financial condition
and the greater the transportation cost
associated with clearing the instrument, and so the greater the discount
tended to be. So a merchant in Kansas
City, Missouri, accepting as payment
a check drawn on a bank in Allentown, Pennsylvania, knew he would be
credited with less than the face or par
value of the check and would have to
consult with his bank to find out how
much less. Obviously, this was a payment system inimical to the growth of
national commerce.
By the turn of the 20th century,
it was clear that the U.S. needed a
more well-integrated national payment
system. Indeed, one of the main reasons Congress established the Federal
Reserve System in 1913 was to create
a national clearing system in which
checks could exchange at par value.
To achieve this, the Federal Reserve
offered check-clearing services free of
charge to banks that joined the Fed
System.
However, the Fed did not become
the sole provider of check-clearing
services, despite offering its services for
free. First, not all banks chose to join
the Fed System, primarily because of
some of the regulatory implications. In

addition, large correspondent banks
offered smaller respondent banks
an array of “bankers’ bank services,”
including check clearing, and banks
could take advantage of local and national clearinghouse arrangements.
Nonetheless, the Fed established
a large market presence, providing a
baseline level of national check-clearing services accessible to all banks,
large and small, anywhere in the country. Thus, the Fed contributed to the
viability of both the paper check and
the small community bank.
In the 1960s and 1970s, U.S.
banks and the Fed applied advances
in computing technology to check
processing, increasing the efficiency
of their operations. Banks found the
paper check payments business to be
profitable, and consumers were quite
comfortable and confident in the use
of checks.
In short, checks were the dominant form of noncash payment, and
there was little momentum for change
in the U.S. payments system. One
might argue that bank Giro systems,
which were arising in Europe at the
time, would have increased the efficiency of the payments system even
more. Yet with so many banks in
the U.S. — all serving local markets

Business Review Q3 2005 3

— developing the legal framework, industry standards, and institutional arrangements necessary to establish such
a payments network nationally would
have been a daunting task. In any case,
American banks are forbidden under
antitrust law to work together.
The Fed itself introduced its version of an electronic Giro system in the
early 1970s. We call it the automated
clearinghouse, or Fed ACH. Fed ACH
has met with some success.
However, unlike the European
Giro, ACH has not developed into the
dominant form of electronic payment,
in part, because, traditionally, only
banks — not individuals — could initiate ACH payments. This made ACH
practical only for companies engaged
in batch-processing a large number of
payments, such as payroll disbursement.
In a typical transaction, a firm
would forward to its bank an electronic file containing payments to be
made from the firm’s account. The
bank would then initiate the ACH
transactions by sending the file to the
Fed, which would transfer funds from
the bank’s account to the accounts of
the various payees’ banks, and then
notify them of the account holders to
be credited.
I will add that a relatively recent
variant allows large organizations
to collect regular payments using
the ACH. A typical transaction of
this nature would involve individual
customers’ authorizing their bank to
make ACH payments directly to a firm
— perhaps their utility company or
mortgage company — on a recurring
basis.
CARDS DRIVE CHANGES IN
U.S. PAYMENTS
While Fed ACH saw some success as a means to effect electronic
payments, it was the credit card that
proved most instrumental in moving

4 Q3 2005 Business Review

U.S. payments from paper to electronics. The credit card was the first electronic payments instrument to emerge
in the U.S. Credit cards were introduced in the 1950s, and their use grew
rapidly over the next three decades.
Credit Cards. Not coincidentally,
the U.S. credit card infrastructure
looks a lot like the European banking

While Fed ACH saw
some success as
a means to effect
electronic payments,
it was the credit card
that proved most
instrumental in moving
U.S. payments from
paper to electronics.
system. There are relatively few major
card associations; they operate nationwide; and they are not subject to the
anti-trust laws that prohibited collaboration among U.S. banks. In fact, the
credit card associations benefited from
some early antitrust rulings against
banks.
In the 1990s, when the tech boom
made information processing and
telecommunications more powerful
and less expensive, the credit card
associations were well positioned to
take full advantage of these developments. Low-cost telecom has made
real-time, point-of-service verification
of cardholders and their credit status
widespread, speeding transactions and
curtailing fraud. Of significance for
the future, this technology has made
the credit card a viable means of payment for e-commerce.
Debit Cards. After the credit
card, the debit card is the second most
popular electronic instrument for mak-

ing retail payments in the U.S. today.
The debit card arrived on the scene
relatively recently — during the 1980s
— in both the United States and Europe. But since its arrival, growth in
usage has been dramatic.
In Europe, the debit card emerged
as an evolution of banks’ automated
teller machine (ATM) systems. Instead
of using their card to withdraw cash
from an ATM to pay merchants, bank
customers simply present their card to
the merchants, and their bank account
is debited directly.
This same progression occurred
in the U.S. But in the U.S., the credit
card networks responded with debit
card products of their own. Visa and
MasterCard already had an infrastructure for processing credit card transactions at the point of sale. They leveraged this infrastructure to establish
offline debit card networks. Indeed, in
the U.S., these so-called “signature”
debit cards are proving at least as
popular as ATM, or “PIN-based,” debit
cards.4
Signature debit cards now account
for about two-thirds of the total of
debit transactions. So it could be said
that they are even more popular than
their PIN counterparts. However, PINbased debits are growing a bit faster
than signature.5
In any case, debit cards, in general, seem to be leading the migration
away from cash and checks and toward
electronic payments in the U.S. This
trend is substantiated by the Survey of
Consumer Finances, sponsored by the
Federal Reserve Board of Governors
See the conference summary “Prepaid Cards:
How Do They Function? How Are They Regulated?” produced by the Federal Reserve Bank
of Philadelphia’s Payment Cards Center, June
2004, available at: www.philadelphiafed.org/
pcc/conferences/PrepaidCards_062004.pdf.

See the Retail Payments Research Project: A
Snapshot of the U.S. Payments Landscape, Federal
Reserve System, 2002.

www.philadelphiafed.org

and compiled by the Research Department at the Philadelphia Fed.6
The survey indicates that less
than 18 percent of households used
debit cards in 1995. By 2001, nearly
half of all households were using them.
Not coincidentally, the survey also
divulged a substantial reduction in the
use of cash over the same period.7
The growing popularity of debit
cards in the U.S. seems to be part of a
broader phenomenon. As I mentioned
earlier, debit cards have caught on just
as quickly in Europe. In fact, recently,
for the first time ever, Visa’s global
debit sales volume surpassed its credit
sales volume.8
THE FUTURE OF THE U.S.
RETAIL PAYMENTS SYSTEM
By now, I hope I have given you
some perspective on the current state
of U.S. retail payments and the evolutionary process that brought us there.
Looking ahead, retail payments
in the U.S. will continue moving away
from cash and paper checks and toward electronic instruments, including
credit cards, debit cards, ACH, and
emerging vehicles such as prepaid
cards.
Though roughly half of our noncash payments are still being made by
paper check, the tide has turned. In
fact, recent research by the Federal
Reserve shows check usage peaked in
the mid-1990s and has been declining
steadily ever since. So paper checks are
not only losing market share, they are

See Loretta J. Mester, “Changes in the Use
of Electronic Means of Payment: 1995-2001,”
Federal Reserve Bank of Philadelphia Business
Review, Third Quarter 2003.

actually declining in volume and have
been for about a decade.9
The share of retail transactions
handled by cards will continue to
grow in the U.S., particularly at the
point of sale. Debit cards have made
particularly deep inroads in the realm
of “micropayments” — purchases
under $20. According to a survey by

In the future,
organizations other
than banks will
expand their role in
the payments system,
especially retailers
themselves.
MasterCard International, debit cards
now account for about one-third of all
micropayments, a 61 percent increase
over 2001.10 Visa claims to have authorized 82 percent more payments
at quick-service restaurants between
January and July of 2004 than during
the same period in 2003.11 Here we see
debit transactions replacing cash, since
the survey indicated a substantial drop
in cash micropayments.
Several fast-food chains are promoting greater use of payment cards
at their restaurants. (It undoubtedly
has not escaped their attention that
customers spend, on average, over 50
percent more when they pay with a

See Mester, Federal Reserve Bank of Philadelphia Business Review, Third Quarter 2003.
7

Press release, “Visa Global Debit Card Sales
Volume Surpasses Credit,” Visa International,
April 20, 2004.

www.philadelphiafed.org

See the Federal Reserve System Retail Payments Study, December 2004.

David Breitkopf, “MasterCard, Pulse Report
Wider Use of Debit Cards,” American Banker,
May 17, 2004.

card rather than cash.12) This movement has tremendous upside potential.
Last year, consumers used their cards
to spend $6.5 billion at fast-food restaurants, and that was with only 10
percent of such restaurants accepting
cards.13
In the future, organizations other
than banks will expand their role in
the payments system, especially retailers themselves. As a result of recent
legal action brought by Wal-Mart
against U.S. card companies, retailers
now appreciate the costs and benefits
associated with alternative payment
processing arrangements and will
weigh in to protect their interests. As
you may know, Wal-Mart, the largest
retailer in the U.S., along with other
merchants, balked at the idea of accepting signature debit cards — and
their associated fees — without the
right to negotiation. They sued U.S.
bank credit card associations, prevailing in a good portion of their efforts.
Their settlement eliminated the “honor all cards” rule, effectively allowing
merchants to decline signature debit
products without jeopardizing their
ability to accept credit products or PIN
debit cards.
In short, I expect keen competition among card providers and aggressive marketing by both card providers
and merchants to increase the speed
with which cards replace paper for
point-of-sale transactions in the U.S.
How quickly U.S. consumers
move from paper to electronics, when
it comes to bill paying, is an interesting
question. The speed and scope of that
transition depend on the evolution of
our payments system.

W.A. Lee, “CEO Confident as Visa Posts
More Records,” American Banker, August 5,
2004.

12
Data from W.A. Lee, “CEO Confident as Visa
Posts More Records,” American Banker, August
5, 2004.

“Cards…at participating restaurants,” Electronic Payments International, August 19, 2004.

Business Review Q3 2005 5

As I mentioned earlier, the ACH
system in the U.S. has not been as
successful as Europe’s Giro systems.
But things may be changing. Financial
institutions are finding innovative new
uses for ACH, spanning a broad range
of retail transactions and shifting
substantial volumes to this system, primarily at the expense of check volume.
The most important of these
innovations is accounts receivable
check (ARC) conversion. Large
organizations that receive paper
checks from customers as remittance
for retail payments are now scanning
the checks to digitally capture
their relevant payment information.
The companies can then use this
information to create an electronic
file, which is then transmitted to an
ACH payments provider — usually the
Fed — for processing. In some cases,
even individual merchants who accept
customer checks at the point-of-sale
can use the information on the check
to generate an electronic file. That file
is then sent to the merchant’s bank for
processing through the ACH.
Conversion to ACH is helping
to streamline payments initiated by
check, even when the paper check
would follow. It is also being used to
process one-time payments initiated
via the Internet.
As the owner/operator of the Fed
ACH system, the Federal Reserve
has been working to ensure its ACH
system is equipped to accommodate
changes in volumes and the nature of
payments, even as these applications
proliferate. As in check processing, the
Fed is not the sole provider of ACH.
Though the Federal Reserve network
currently originates about two-thirds
of all ACH payments volume, we are
also seeing growth among privatesector ACH networks. Indeed, as ACH
continues to gain acceptance as a
payment vehicle, its products and
marketing will evolve so as to make

6 Q3 2005 Business Review

it more attractive and accessible to
individuals and businesses.
MANAGING THE TRANSITION
So the private sector is shifting
retail payments in the U.S. away from
paper-based instruments and toward
electronic ones. But history tells us
that people’s payment habits change
only gradually. When people are com-

Control Act of 1980 changed all that.
It required the Fed to offer its payment
services to all banks at prices fully
reflecting the Fed’s costs of production,
including imputed profits. This change
established a marketplace incentive for
the Fed and its private-sector competitors in check processing to maximize
the efficiency of their check processing
operations.

The Fed has been trying to take full
advantage of the efﬁciencies afforded by
electronic processing of payments initiated by
paper check in the interest of maximizing the
efﬁciency of the payment system.
fortable with, and confident in, a payment structure, they are reluctant to
give it up. As a result, the paper check
is likely to be with us for some time.
In the meantime, the Fed has
been trying to take full advantage of
the efficiencies afforded by electronic
processing of payments initiated by
paper check in the interest of maximizing the efficiency of the payment
system. Thus, the Fed is doing what
it can to foster check truncation and
electronification at as early a stage as
possible in the payment process.
The Fed is now well positioned
to pursue this objective. Two pieces of
legislation have set the stage. One is
a law that has been on the books for
25 years: the Monetary Control Act
of 1980. The second was passed in
2003 and went into effect in October
2004: the Check Clearing for the 21st
Century Act, commonly called Check
21. Let me explain the significance of
each.
Recall that when the Fed began
its check processing operations, it
provided the service at no charge to
its member banks. The Monetary

The second piece of legislation,
Check 21, adds an important new
dimension to the competitive drive for
greater efficiency in check processing.
The essence of the new law is that it
makes the facsimile of a check created
from an electronic image serve as the
legal equivalent of the check itself. In
doing so, it eliminates a significant
legal barrier to check truncation and
electronification of check processing.
A collecting bank can soon create an
electronic image of a check, transmit
the image to the paying bank’s
location, and then present the paying
bank with a paper reproduction or
with the electronic image itself. The
hope and expectation is that gradually
more and more paying banks will
prefer the image itself.
Accepting images for both
deposit and presentment eliminates
back office capture of the check as
well as the inconvenience of physical
transportation. Indeed, under the new
Check 21 legislation, it will become
even easier to move toward a more
electronic check process because
banks will be provided with additional

www.philadelphiafed.org

options for processing image-based
payments.
As a provider of financial services,
the Fed has been actively engaged in
bringing a whole array of image products to market to take advantage of
the capability of image clearing. The
Fed has established an image archive
for electronic items; it has enhanced
the ability to produce facsimile checks;
and it has extended clearing times
to encourage the use of the new image technology that the act allows.
In short, the Fed is introducing new
services that will enable banks to take
full advantage of Check 21.
How fast will the transition occur?
Our best guess is that the industry will
be slow to embrace the new capabilities that the law permits. We must also
consider the possibility that making
check processing more efficient will
actually extend the life of the waning
check. In any case, the Federal Reserve Banks’ financial services division
is committed to working with the industry to ensure a smooth transition.
THE CHALLENGE TO THE FED
With the evolution of the payments system in the U.S. accelerating,
the Federal Reserve must make some
major adjustments to its payments services as the changing payments system
alters its role. Nonetheless, the Fed is
committed to working to improve the
reliability and efficiency of the current
generation of payments vehicles, even
as it works to foster innovation and
to support the next generation of payments vehicles. Both commitments are
equally important during this period of
transition.
With this dual commitment in
mind, the Fed continues to fulfill its
traditional role as payments processor
even while it supports the move to the
new electronic clearing environment.
Striking the right balance between
these two seemingly divergent goals is

www.philadelphiafed.org

a challenge. Nonetheless, the Fed has
begun implementing a strategy that
includes key elements to help it successfully meet both commitments.
The Fed has recently announced
a program of “aggressive electronification” of retail payments in the U.S.
This push toward electronics will help
facilitate Check 21 and quicken the
transition to an all-electronic world.
The Fed is also investing heavily in

By setting prices
that reﬂect the low
cost of electronic
check processing
relative to paper,
the Fed will allow,
indeed encourage,
the market to drive
checks toward
electronics.
technologies that enable electronification. In addition, as check volumes
decline, the pressure has been on to
find new processing efficiencies. The
transition will not be easy, particularly
for the Federal Reserve System.
The Fed currently clears about
one-third of all checks written in the
U.S. As check volumes have declined,
the Fed has had to consolidate its
operations, closing down processing
sites where appropriate. Nonetheless, it
has attempted to maintain reasonable
service levels nationally by re-routing
checks to nearby sites.
So that you can see the scale of
this effort, I will note that two years
ago the Fed had 45 check processing
sites. By the end of 2006, we expect
to be down to 22. This downsizing to
match costs and revenues helps the

Fed fulfill its traditional role of payments processor while at the same
time maintaining efficiency in this
new environment.
Such a radical transformation
within the Fed’s financial services
division is made necessary by law. As I
mentioned, the Monetary Control Act
mandated that the Fed set prices on
its services to fully recover its costs. At
the same time, the Fed is required to
adjust its portfolio of services to correspond to the clearing needs of the industry. As such, the aggregate decline
in volume in this volume-based service
creates a substantial challenge to the
System. Achieving full cost recovery
will become more challenging for the
Fed as the volume of check usage continues to decline.
Nonetheless, by setting prices that
reflect the low cost of electronic check
processing relative to paper, the Fed
will allow, indeed encourage, the market to drive checks toward electronics.
In addition, the Fed will continue to
develop its capabilities and expand
its electronics capacity to respond to
the market’s evolution and consumers’
needs. The impact of these changes
and those that follow will ultimately
transform the U.S. payments system
and enable a radical restructuring of
its service capabilities.
A WORD ABOUT WHOLESALE
PAYMENTS
Before closing, let me briefly
discuss the Fed’s wholesale payments
operation. Aside from its role in supporting retail payments, or small-dollar
transactions, the Fed has long had a
role in facilitating wholesale, or largedollar, transactions. Fedwire is the
Fed’s real-time wholesale payments
operation used to transfer both funds
and securities. Fedwire transactions
typically involve large-value, time-critical payments, such as payments for the
settlement of interbank purchases and
Business Review Q3 2005 7

sales of federal funds, or securities or
real estate transactions.
Fedwire first went into operation
back in 1918, and its operations have
evolved with advances in technology
and the integration of financial markets. The Fed has recently centralized
Fedwire operations from all 12 Reserve
Banks to its New York Bank — with
both a hot and a cold backup.
Now, a parallel process seems to
be in motion in Europe. The initiative
known as TARGET2 will likely consolidate European central banks’ wire
transfer operations. As in the case of
Fedwire, this standardized processing
platform will reduce costs through
economies of scale and improve flexibility of wholesale payments.
CONCLUSION
My purpose here was to review
and explain the state of payments
technology in the U.S. vis-à-vis that
of Europe. The roots of these two pay-

8 Q3 2005 Business Review

ment systems lie in the different banking structures of the U.S. and Europe
and different perceptions of appropriate regulation.
Europe’s is a system of a few large
banks that can easily be regulated into
a centralized world — first with nearuniversal Giro accounts and soon with
an electronic world of more centralized
clearing.
In the U.S., markets and consumers led the nation to a multiplicity of
banks and a payments system that
has been paper intensive. This is
changing in the U.S., as cards replace
checks, and electronic clearing truncates the maze of paper that fills U.S.
post offices. Indeed, it seems the U.S.
payments system is moving toward
convergence with the European model.
Our progress, while promising, occurs
largely in fits and starts. The U.S. is
a large nation with many providers,
much complexity, and a philosophy of
market-based solutions.

This has presented challenges for
the Federal Reserve as a provider of
financial services. It has necessitated
restructurings, plant closings, and
difficult decisions that most central
banks in Europe have been spared.
Yet, by law, the Fed is charged with
the dual role of a regulator seeking to
maintain the stability and efficiency of
the payments system and a provider of
payment services. At times, these roles
present different challenges. This is
one of those times.
Nonetheless, as payments technology moves forward in the U.S., our
payments system will continue to
change as evolutionary forces generate
new innovations in payments and new
ways to deliver them. In some ways
we will look more like the European
system even as our two payments
systems move to the next generation
of payments. We will look more alike,
although we will get there from a very
different starting point. BR

www.philadelphiafed.org

The Economic Role of Cities
in the 21ST Century
BY GERALD A. CARLINO

s real income increases, the demand for
a greater variety of goods and services
becomes a more important determinant of
where people choose to live. This implies
that large cities with more choices will attract highincome households that value variety. Members of
these high-income households also tend to be highskill individuals. Their presence supports cities’ new
function as incubators of new ideas and innovation. In
“The Economic Role of Cities in the 21st Century,” Jerry
Carlino focuses on the economic activities that make
firms in cities more productive and that make cities
more attractive to urban households.

What is the role of cities in the
21st century economy? In earlier times,
cities grew near transportation hubs,
such as ports and railroad yards. To
minimize transportation costs, firms
needed to be near these hubs, and
workers needed to live close to their
employers to maintain reasonable
commuting distances. Thus, firms
and households tended to be highly
concentrated in cities. These so-called

Jerry Carlino is a
senior economic
advisor and
economist in
the Research
Department of
the Philadelphia
Fed.

www.philadelphiafed.org

agglomeration economies — the efficiency and cost savings that result
from being close to suppliers, workers,
and customers — were an important
factor in the rise of cities as manufacturing centers.
Agglomeration economies tended
to support mostly the production side
of the economy. That is, proximity to
inputs into the production process led
to gains in output. However, improvements in transportation technology
mean that, today, firms are freer to
locate wherever they want, and, unlike
before, their choice of location will
depend on where their workers choose
to live. This means that an area’s special features, such as its climate, will
be important determinants of where
households, and ultimately firms,
locate.

As a result, agglomeration economies are increasingly concentrated
on the consumption side. Rising real
incomes mean that quality-of-life issues
have become more and more important as determinants of where people
choose to live. For example, growth in
real income increases the demand for
a greater variety of goods and services
(more theaters, varied restaurant
cuisine, and professional sports teams).
Similarly, access to recreational amenities and better public services, such as
good public schools, are also important
quality-of-life issues for households.
This implies that large cities with
more choices will attract high-income
households that put a high value on
variety. Members of these high-income
households also tend to be high-skill
individuals. Their presence supports
cities’ new function as incubators of
new ideas and innovation.
To answer our question about
cities’ role in the 21st century economy,
we will discuss some of the economic
functions of cities, focusing on economic activities that make firms in
cities more productive and that make
cities more attractive to urban households.
AGGLOMERATION ECONOMIES
IN URBAN PRODUCTION
While the discussion in this
article will emphasize agglomeration
economies’ role in urban consumption,
historically, their biggest influence has
been on the production side.
Agglomeration economies constitute an important source of a firm’s
productivity. Increases in productivity due to agglomeration economies
depend not on the size of the firm itself
Business Review Q3 2005 9

(internal economies of scale),1 but
rather on the size of a firm’s industry in
a particular city (localization economies) or on the size of the city itself
(urbanization economies).
Localization. The presence of
an industry in a particular city could
be the result of the available natural
resources or simply historical accident.
But once an industry develops in a city,
other firms in that industry often reap
considerable benefits by also locating
there.
One advantage is sharing inputs.
Consider, for example, the high-tech
industry in Silicon Valley, the TV and
motion picture industry in Los Angeles, and the auto industry in Detroit
— three industries that have concentrated in certain locations. Many production companies in the TV industry,
for example, frequently require the
services of highly specialized workers,
such as people who specialize in writing and editing scripts; workers who
specialize in lighting, sound recording, special effects, and set design and
construction; and talent agencies and
firms that engage in market research.
The need to have quick access
to these types of specialists is particularly important in the production of
TV shows, and consequently, many
of these specialists must be on or near
the production set. A production
company located far from Los Angeles
would need to employ full-time script
editors or sound and lighting personnel
and set designers, for example, or else
spend considerable time and money
bringing them from a distance when
they are needed. But when TV produc-

Economists have long recognized that a
firm’s size can affect its productivity. As a firm
increases its size, it can increase productivity
by having its workers specialize in particular
tasks or by using its capital equipment more
efficiently. In these situations, a firm is said to
enjoy internal economies of scale.

10 Q3 2005 Business Review

tion companies cluster together, their
combined needs for highly specialized
inputs can support at least one firm
that specializes in set design, others that specialize in script analysis,
and so on. Thus, these services are
available at lower cost from a local
firm. All production companies in the
cluster can enjoy a lower average cost
of production by contracting for these
specialized services only when they are
needed.

lays off workers, these unemployed
workers are likely to be hired by one of
the other, more successful firms in that
cluster.2
In addition to reducing the employment risk of workers and firms, labor market pooling also facilitates the
matching of workers and jobs. Having
a large pool of workers in an area
makes it easier for employers to find
people with the set of characteristics
they need. At the same time, workers

A common labor pool allows ﬁrms to more
effectively adjust their demand for labor to
match ﬂuctuations in the demand for their
products.
There also are advantages to
sharing a common labor pool in cities. These advantages arise from the
uncertainty and variability in any one
firm’s demand for workers. If a firm is
uncertain about the number and skill
mix of workers it will hire, the firm has
an incentive to cluster with other firms
in its industry to draw from a common pool of workers. A common labor
pool allows firms to more effectively
adjust their demand for labor to match
fluctuations in the demand for their
products.
Consider our example of the TV
industry once again. Producers of TV
programs are never quite sure if a new
show will be successful. But as economist Arthur O’Sullivan has noted,
“When it becomes clear which programs will be discontinued, actors and
technicians move from the unsuccessful programs to the successful ones.
The concentration of the television
industry in Los Angeles and New York
facilitates the transfer of labor from
one firm to another.”
Common labor pools are also of
value to workers as well. If any one
firm in the cluster is unsuccessful and

are more likely to find jobs that better
match their experience and skills.
Therefore, having a large pool of workers in an area facilitates the number
and quality of matches between firms’
needs and workers’ skills.
Urbanization. Not only does the
size of a firm’s industry in a city matter
but so does the size of the city itself.
Just as some kinds of businesses, such
as a set-design firm, are found only
where specific industries concentrate,
other activities, such as financial and
business services, are generally found
only in urban areas. Often, only a large
city can provide a client base sufficient
for these specialized firms to flourish.
These types of specialized services
give rise to economies of scale, called
urbanization economies, that are external to any one firm and its industry.
Urbanization brings greater efficiency, but it also brings problems that
eventually offset the gains in efficiency.
According to the traditional view, as

See the article by Satyajit Chatterjee for
further discussion of the advantages of labor
market pooling.

www.philadelphiafed.org

cities become more congested, the
increased cost of doing business (for
example, in the form of higher business
rents) will eventually offset any gains
in agglomeration economies from additional growth. At that point, existing
firms have no incentive to expand
production, and new firms will not be
enticed to locate in the city. The city’s
level of population, employment, and
output will have stabilized at a certain
point.
Recently, economists have focused
on a new view: The creation of ideas
in cities can lead to sustained growth
in the output of urban firms even if
population and employment are not
expanding. The basic theory is that
the higher density of population and
employment in cities promotes the
exchange of ideas among individuals, which economists call knowledge
spillovers. The high concentration of
people, especially highly skilled people,
in cities creates an environment in
which ideas move quickly from person
to person. It’s likely that some of these
ideas lead to new goods and to new
ways of producing existing goods.3
To the extent that firms more
readily adopt innovations that are local, they may be able to produce more
output without having to increase the
level of inputs into production. In this
instance, generating ideas has become
an important source of growth, and
proximity to individuals who create knowledge is becoming increasingly important to firms. Thus, urban
locations’ advantages for firms have
shifted from proximity to suppliers
and customers to proximity to highly
skilled workers.

EVIDENCE ON PRODUCTION
BENEFITS OF CITIES
In their 2001 research, economists
Stuart Rosenthal and William Strange
studied the importance of input sharing, labor market pooling, and knowledge spillovers for manufacturing firms

See my 2001 Business Review article and my
paper with Satyajit Chatterjee and Robert Hunt
for further discussion of the role of knowledge
spillovers in cities.

www.philadelphiafed.org

than 2 million people are 8 percent
more productive than metropolitan
areas with less than 2 million people.
In more productive cities, firms can afford to pay higher wages. At the same
time, households and firms are drawn
to relatively high productivity cities.

The decline in the importance of
agglomeration economies to ﬁrms does not
mean that the clustering of people and jobs is
no longer important to cities.
at the state, county, and zip code levels. Among the sources Rosenthal and
Strange considered, labor market pooling has a strong impact on geographic
concentration of manufacturing firms
at all of these levels. They also found
that other types of input sharing, such
as intermediate inputs and natural resources, influence the concentration of
manufacturing firms at the state level
but have no effect on concentration of
manufacturing firms at the county or
zip code levels. The effects of knowledge spillovers on the concentration of
manufacturing firms tend to be more
localized, influencing concentration
only at the zip code level.
While Rosenthal and Strange’s
attempt to identify the relative importance of the various forces that gave
rise to the spatial concentration of
firms, the vast majority of research to
date has tended to analyze the relationship between urban productivity
and city size. In a 1976 study, David
Segal analyzed the change in urban
productivity related to the size of a
metropolitan area.4 He found that, on
average, metropolitan areas with more

The change in urban productivity is the
amount by which output would increase as a
result of increasing population in a city, with all
inputs held constant.

Thus, rents may also rise in these
cities. In sum, if the concentration
of people and jobs in cities is largely
related to urban productivity, both
wages and rents should increase with
city size.
AGGLOMERATION ECONOMIES
IN URBAN CONSUMPTION
Despite agglomeration economies’
historical importance to the production side of urban economies, innovations in transportation, production,
and communication technologies have
weakened the economic advantage of
locating closely related activities near
one another. However, the decline
in the importance of agglomeration
economies to firms does not mean that
the clustering of people and jobs is no
longer important to cities. As we’ll see,
urban locations are still important to
21st century households.
If consumers prefer a large variety
of goods and services and there are
substantial economies of scale in providing them, the number of different
goods and services offered and consumers’ economic welfare will depend
on the size of the local market.
Cultural and leisure activities
offer good examples. As a hypothetical
example, consider professional football,
a good with relatively low per capita
demand. Suppose that to break even,

Business Review Q3 2005 11

the club must sell 30,000 tickets per
game, or 240,000 tickets per season
(based on eight home games per
year). If, on average, 20 percent of a
metropolitan area’s residents attend
a game, a metropolitan area of 1.2
million people is required to support
the football team. But as a metro area’s
population increases, the demand for
variety in professional sports teams
also increases. The greater New York
metropolitan area has a population of
almost 20 million people and is home
to nine professional sports teams in the
four major sports (baseball, football,
basketball, and hockey). Large metropolitan regions such as Los Angeles,
Chicago, and Philadelphia support at
least four teams each. With a population of only about 1 million to 1.5

million, Orlando, Hartford, and Jacksonville support one major professional
sports team each (see the table).
In addition to greater variety,
the quality of a good or service may
improve with the population size of an
area. To continue the sports analogy,
economist Rodney Fort has noted that
large-market teams win much more
frequently than do small-market teams.
The New York Yankees, a large-market
team, are a post-season fixture, whereas the small-market Pittsburgh Pirates
have not made the playoffs since 1992.
Fort points out that teams with a large
fan base earn more revenue for any
given level of quality. Teams in large
markets can outbid small-market teams
for the best players, since large-market
teams can earn more revenue from

TABLE
Big Metro Areas Offer Diversity of Sports
Metro Area
New York
Los Angeles
Chicago
Washington-Baltimore
San Francisco-Oakland-San Jose
Philadelphia
Boston
Detroit
Dallas-Ft. Worth
Houston
Atlanta
Cleveland
Pittsburgh
Cincinnati
Kansas City
Indianapolis
Orlando
Hartford
Jacksonville

No. of Teams
9
5
5
5
6
4
4
4
4
3
3
3
3
2
2
2
1
1
1

Population(Millions)
19.9
15.6
8.6
7.2
6.7
6.0
5.8
5.4
4.7
4.3
3.6
2.9
2.4
1.9
1.7
1.5
1.5
1.1
1.0

these players than do teams in small
markets. The same must be true for
other types of consumer goods and
leisure activities, such as theaters,
orchestras, and restaurants.5
Rising Income. In the 55 years
between 1947 and 2002, per capita
income adjusted for inflation (that is,
real income) almost doubled in the
United States. The rise in real income
has led to more demand for goods and
services, especially luxury goods, such
as meals in gourmet restaurants and
live theater, which are more plentiful in large cities.6 Thus, the greater
variety in consumption found in
large cities is especially attractive to
households as their wealth increases.7
Similarly, rising incomes should increase the value that people (especially
high-skill individuals) place on amenities, such as good weather.
In a 2004 study, Sanghoon Lee
contended that the demand for variety
may increase more than proportionately with income. That is, a 1 percent
increase in income leads to more than
a 1 percent increase in the demand for
variety. Lee went one step further and

Leonard Nakamura discusses how innovation
in retailing (introduction of scanner technology) led to larger supermarkets (superstores)
that offer greater variety to their customers
(bakeries, banking, pharmacies, as well as
greater variety on the shelves). A number of
studies by Joel Waldfogel and co-authors have
shown that larger cities have more and better
newspapers and more and better radio and
television stations.
5

One key feature of goods such as these is that
it’s difficult to transport them; therefore, they
are referred to as nontraded goods and services.
While people can travel to cities offering an
abundance of nontraded goods and services,
there is little substitution for living in the cities,
or their environs, if people value convenient
access to nontraded goods and services.

See, for example, the articles by Jan
Brueckner, Jacques-Francois Thisse, and Yves
Zenou; Edward Glaeser, Jed Kolko, and Albert
Saiz; and Dwight Adamson, David Clark, and
Mark Partridge.

Source: Rodney D. Fort. Sports Economics. New Jersey: Prentice Hall Publishers,
2003, Table 2-2. Used with permission.
12 Q3 2005 Business Review

www.philadelphiafed.org

argued that since high-skill workers
earn more than low-skill workers, highskill workers will account for a larger
share of the work force in large cities
and a smaller share in small cities and
rural areas.8
Other Factors. Economists Ed
Glaeser, Jed Kolko, and Albert Saiz
point out three other ways in which
large cities enhance consumption opportunities. Large cities may provide
a greater variety of public goods, too,
such as more magnet schools per
student (e.g., schools specializing in
fine and performing arts, or those
specializing in science). Furthermore,
large cities make it easier for individuals to make wider social contacts and
to have a more diverse set of friends.
Along this line, large cities appeal to
younger, more highly educated workers
because large cities facilitate better
development of professional and social
connections than small cities and rural
areas. Economists Dora Costa and
Matt Kahn note that “power couples”
(both partners have bachelor’s degrees)
are increasingly locating in large
cities because large cities offer better
employment opportunities for working
couples. Finally, large cities may satisfy
aesthetic preferences, such as the variety of architecture found in many large
cities or the artistic scene in places
such as New York City.
Of course, as with the production
side of the urban economy, urbaniza-

Lee’s discussion ignores the role of the
production side of the economy. If high-skill
workers are relatively more productive than
low-skill workers in cities, high-skill workers
will be disproportionately drawn to large cities.
Put differently, in the extreme case, highly
skilled individuals may be drawn to large cities
not because of the greater variety of goods
and services but because such cities enhance
their productivity. No doubt, both of these
forces (greater productivity and greater variety)
operate in cities. The difficulty is trying to
differentiate the extent to which highly skilled
people locate in cities because of productivity or
because of greater variety.

www.philadelphiafed.org

tion brings not only a greater variety
of goods and services but also problems, such as congestion, that take
the form of long-distance commuting and higher housing costs, which
eventually balance the gains in variety.
The higher cost of housing as cities
become congested reduces households’
purchasing power and limits the inflow
of people.
MORE EVIDENCE ON THE
BENEFITS OF CITIES
The value of a city’s special traits,
such as pleasant weather or the variety
of consumption options, is determined
by what people are willing to pay in order to live there. This amounts to the
sum of what people are willing to pay
for each local characteristic that adds
to the quality of life in an area. The
trick is to determine the prices of these
local traits, since they are not bought
and sold in markets.
Even though there is no explicit
price for local amenities such as nice
weather or greater variety, there is an
implicit price. Suppose you are considering moving either to Metropolis,
which offers its residents great variety
in consumption, or to Smallville,
which has far less variety than Metropolis. Because variety is something
you value, you are willing to pay some
extra amount, say, $1000 a year, to live
in Metropolis.
You could pay your extra $1000
in two ways. One is by bidding up
land prices, and ultimately rents, in
Metropolis relative to Smallville. But
it is not necessarily the case that you
will ultimately pay $1000 more to rent
a house in Metropolis. Part of the cost
of living in a city with more variety
could be paid in the form of wages
lower than you would have accepted
in Smallville. What must be true is
that rent and wage differentials sum
to $1000. Thus, other things equal,
the extent to which rent is higher and

wages are lower (so that wages adjusted
for the cost of living, which economists
call real wages, are lower) is the extent
to which the consumption benefit of
greater variety is absorbed into local
land markets and local labor markets.
This discussion of a city’s special
traits ignores the role of the production
side of the economy. Earlier we saw
that if the concentration of people and
jobs in cities is related to urban productivity, both wages and rents should
increase with city size. But, as we just
saw, if the concentration of people and
jobs in cities is related to urban amenities, higher rents will outweigh higher
wages, so that real wages are lower in
cities offering amenities that people
value.
A number of economists have
looked at the relationship between a
metropolitan area’s size and the level
of local wages and rents to determine
whether productivity or urban amenities better explain the concentration
of people and jobs in cities. The
evidence to date is mixed. In a 2000
article, economists Takatoshi Tabuchi
and Atsushi Yoshida used data for just
over 100 Japanese cities for 1992 and
showed that a doubling of city size is
associated with about a 10 percent
increase in production costs. If firms
are making products for national and
international markets, the only way
firms in relatively high-cost (large) cities can compete with firms in relatively
low-cost (small) cities is if productivity
(that is, agglomeration economies) is
sufficiently higher in high-cost than in
low-cost cities. Thus, according to Tabuchi and Yoshida, firms in large cities
incur higher costs than similar firms
in small cities because large cities offer
firms greater agglomeration economies.
But these authors found that a
similar doubling of city size is associated with a 7 percent to 12 percent
decrease in real wages, which they
attribute to households’ willingness to
Business Review Q3 2005 13

accept lower real wages as a tradeoff
for the greater variety offered in big
cities. On balance, their results suggest
that while productivity is higher in
cities, people’s taste for urban amenities and variety is an important factor
in accounting for the concentration of
population in cities.
In contrast, economists Gianmarco Ottaviano and Giovanni Peri
studied a sample of 160 U.S. metropolitan areas and found no evidence
that cultural diversity (another way
to measure local variety) was important for consumers.9 Instead, cultural
diversity has a net positive impact on
workers’ productivity.
But the interpretation of the
results of these studies assumes workers
have the same level of skill to begin
with; therefore, if higher real wages are
found in large cities, it reflects greater
productivity of similar workers in large
cities. Recently, Sanghoon Lee offered
another reason that real wages may
differ with city size. It could be because
workers with different levels of skill
are attracted to different locales. For
example, if real wages are found to be
higher in large cities, it’s not necessarily the case that agglomeration economies from locating workers together
in a city are making similarly skilled
workers more productive. Rather,
high-skill workers, who tend to earn
more than low-skill workers, may be attracted to large cities in the first place
because of the higher level of amenities
they offer.
As we have noted, we expect
demand for variety to increase with an
individual’s income. Since high-skill
In their study, Ottaviano and Peri measure
cultural diversity in a city as the variety of
languages spoken by city residents.

14 Q3 2005 Business Review

workers also tend to earn more than
low-skill workers, we expect demand
for variety also to increase with a
worker’s skills. Given that variety
increases with city size, we expect to
find that high-skill workers account for
a larger share of the work force in large
cities and a smaller share in small cities
and rural areas.
According to Lee’s theory, then,
it’s the composition of the work force
and not greater productivity that explains why wages tend to rise with city
size. Lee used data from the healthcare industry to test his theory and
found that large cities do, in fact, have
more doctors relative to the number of
nurses than do small cities. No doubt,
both of these forces (greater productivity and greater variety) are at work in
cities. The difficulty lies in trying to
distinguish the extent to which highwage (high-skill) workers locate in
cities because large cities make them
more productive or because large cities
offer greater variety that high-wage
workers value. This is still an open
question.
Although most of the empirical
results focus on the tradeoffs between
wages and consumption amenities
for workers, a recent study by Stuart
Gabriel and Stuart Rosenthal focused on this tradeoff for firms. The
researchers developed quality-of-life
indexes for households and qualityof-business-environment indexes for
firms in 37 cities from 1977 to 1995.
They then considered how much more
in wages and rents a firm is willing to
pay to locate an additional worker in
a city that offers the firm resources for
greater productivity relative to a control city. Gabriel and Rosenthal found
that many cities attractive to households are unattractive to firms (e.g.,

Miami, Tampa, and Albany). Similarly,
they found that some cities that are
attractive to firms are unattractive to
households (e.g., Detroit and Washington, D.C.). Finally, a few cities were
found to be attractive to both households and firms (e.g., New York, San
Francisco, and Los Angeles). If the
views expressed in the current article
are correct, these cities are poised to
do well in the new century.
CONCLUSION
Agglomeration economies will
continue to play a large role in the life
of 21st century cities. But unlike in earlier times, today’s agglomeration economies have turned cities into centers
for consumption, rather than places
for manufacturing goods. In turn, this
shift in focus means that cities now
tend to attract more highly skilled and
highly paid workers—people who want
more consumption options. Consequently, modern cities must offer a
wide choice of amenities to attract the
high-skill workers needed in this new
type of agglomeration economy.
Public policy can play a significant
role in attracting and retaining highly
skilled workers. Even though the productivity advantages that cities offer
to firms may have waned in recent decades, the nation’s largest urban areas
retain many advantages in providing
consumption benefits that people
value. Glaeser and co-authors’ 2001
study suggests that local policymakers
need to focus on life-style issues because they are important in attracting
and retaining high-skill workers. One
such policy is providing good public
schools. Other policies might focus on
reducing urban crime and providing
amenities such as clean streets and
public parks. BR

www.philadelphiafed.org

REFERENCES

Adamson, Dwight W., David Clark,
and Mark Partridge. “Do Agglomeration Effects and Household Amenities
Have a Skill Bias?” Journal of Regional
Science, 44 (2004), pp. 201-23.
Brueckner, Jan K., Jacques-Francois
Thisse, and Yves Zenou. “Why Is Central Paris Rich and Downtown Detroit
Poor? An Amenity-Based Theory,”
European Economic Review, 43 (1999),
pp. 91-107.
Carlino, Gerald A. “From Centralization to Deconcentration: Economic
Activity Spreads Out,” Federal Reserve
Bank of Philadelphia Business Review
(May/June 1982), pp. 3-13.
Carlino, Gerald A. “Knowledge
Spillovers: Cities’ Role in the New
Economy,” Federal Reserve Bank of
Philadelphia Business Review (Fourth
Quarter 2001), pp. 17-26.
Carlino, Gerald A., Satyajit Chatterjee, and Robert Hunt. “Matching
and Learning in Cities: Urban Density
and the Rate of Invention,” Working
Paper 04-16/R, Federal Reserve Bank
of Philadelphia (2004).
Chatterjee, Satyajit. “Agglomeration
Economies: The Spark That Ignites
a City?” Federal Reserve Bank of
Philadelphia Business Review (Fourth
Quarter 2003), pp. 6-13.
Clement, Douglas. “Urban Legends,”
Federal Reserve Bank of Minneapolis:
The Region, 18 (2004).
Costa, Dora L., and Matthew E. Kahn.
“Power Couples,” Quarterly Journal of
Economics, 115 (2000), pp. 1287-1315.
www.philadelphiafed.org

Fort, Rodney D. Sports Economics. Upper Saddle River, New Jersey: Prentice
Hall, 2003.

O’Sullivan, Arthur. Urban Economics. Boston, MA: McGraw-Hill Irwin,
2003.

Gabriel, Stuart A., and Stuart S.
Rosenthal. “Quality of the Business
Environment Versus Quality of Life:
Do Firms and Households Like the
Same Cities?” Review of Economics and
Statistics, 86 (2004), pp. 438-44.

Rauch, James E. “Productivity Gains
from Geographic Concentration of
Human Capital: Evidence from Cities,”
Journal of Urban Economics, 34 (1993),
pp. 380-400.

Glaeser, Edward L., and Albert Saiz,
“The Rise of the Skilled City,” Brookings-Wharton Papers on Urban Affairs
5 (2004), pp. 47-94.
Glaeser, Edward L., Jed Kolko, and
Albert Saiz. “Consumer City,” Journal
of Economic Geography, 1 (2001), pp.
27-50.
Lee, Sanghoon. “Ability Sorting and
Consumer City,” unpublished manuscript, University of Minnesota and
Federal Reserve Bank of Minneapolis,
2004.
Moretti, Enrico. “Estimating the Social
Return to Higher Education: Evidence
from Longitudinal and Repeated
Cross-Sectional Data,” Journal of
Econometrics, 121(2004), pp. 175-212.
Nakamura, Leonard. “Is the U.S.
Economy Really Growing too Slowly?”
Federal Reserve Bank of Philadelphia
Business Review (March/April 1997),
pp. 3-14.
Ottaviano, Gianmarco, and Giovanni
Peri. “The Economic Value of Cultural
Diversity: Evidence from U.S. Cities,”
National Bureau of Economic Research Working Paper 10904 (November 2004).

Rosenthal, Stuart S., and William C.
Strange. “The Determinants of Agglomeration,” Journal of Urban Economics, 50 (2001), pp. 91-229.
Segal, David. “Are There Returns to
Scale in City Size?” Review of Economics and Statistics, 58 (1976), pp. 339-350.
Tabuchi, Takatoshi, and Atsushi Yoshida. “Separating Urban Agglomeration Economies in Consumption and
Production,” Journal of Urban Economics, 48 (2000), pp.70-84.
Waldfogel, Joel. “Who Benefits Whom
in Local Television Markets?” Brookings-Wharton Papers on Urban Affairs, 2003.
Waldfogel, Joel, and Lisa George.
“Who Affects Whom in Daily Newspaper Markets?” Journal of Political
Economy, 2003.
Waldfogel, Joel, and Peter Siegelman.
“Race and Radio: Preference Externalities, Minority Ownership, and the
Provision of Programming to Minorities,” Advances in Applied Microeconomics, 10, 2001.

Business Review Q3 2005 15

The Economics of Asset Securitization
BY RONEL ELUL

sset securitization — transforming illiquid
assets into tradable securities — is a large and
growing market, even rivaling the corporate
debt market in size. While the underlying
assets can be very different — ranging from song royalties
to home mortgages — most asset-backed securities
nevertheless share some distinctive features. In “The
Economics of Asset Securitization,” Ronel Elul explains
why asset-backed securities exist and discusses some
reasons for their common structure.

In 1997 rock star David Bowie
raised $55 million by selling bonds
backed by revenues from his first 25
albums.1 This was the first application of securitization to intellectual
property. Formally speaking, asset
securitization refers to the process
whereby nontraded assets — such as

Despite initial predictions, this has not led to
a wave of such issues, in part because the Bowie
bonds themselves have not performed quite as
well as expected (because online music piracy
has curtailed revenues from music sales).
1

Ronel Elul is a
senior economist
in the Research
Department of
the Philadelphia
Fed.

16 Q3 2005 Business Review

song royalties — are transformed into
tradable securities, called asset-backed
securities, or ABS, through the repackaging of their cash flows. Some more
mainstream examples of asset-backed
securities include mortgage-backed
securities (MBS) and secured credit
card receivables.
Securitization is a large and growing market. Currently, it represents
about 25 percent of new nongovernment borrowing.2 To take just one
of the sectors mentioned above, at
the end of 2003 there was more than
$7 trillion in securitized mortgages,
representing nearly three-quarters of
all outstanding home loans.
While the underlying assets can
be very different (in terms of maturity, collateral, and risk, for example),

Further detail can be found in the Flow of
Funds Accounts tabulated by the Federal Reserve Board.

ABS nevertheless tend to share some
common features. These common
elements, which we discuss in further detail below, include selling the
underlying assets so that they are
moved off the firm’s balance sheet,
grouping individually illiquid assets
into portfolios, taking steps to reduce
the risk of default on the underlying
assets (known as credit enhancement), and subdividing the assets into
several classes of securities (tranching).
Financial economists have attempted
to explain the underlying reasons for
securitization, as well as these common
features.
MORTGAGE-BACKED
SECURITIES: AN EXAMPLE OF
ASSET SECURITIZATION
Consider, for example, a bank
(the originator) that offers a $200,000
mortgage to a home buyer (see Figure)
with an interest rate of 6 percent.
Rather than hold this loan in its
portfolio and receive small monthly
payments for a period of 30 years, the
bank may prefer to move the loan off
its balance sheet by selling it to an
outside investor. In this way the bank
receives funds today from selling the
loan, so that it has the opportunity to
profit further by originating even more
loans; the reason is that the bank typically collects a fee (the origination fee)
for each loan it originates.3 There are
also other motivations for securitization that we will discuss below. But for
now, let’s look at how the bank in our
example might use securitization.

A typical fee is 1 percent of the loan amount.
www.philadelphiafed.org

The problem is that an individual
loan is very illiquid, i.e., hard to sell,
in part because potential buyers know
much less about the homeowner than
does the bank. For example, the bank
probably knows more about its own
underwriting standards than any
potential buyer, or the bank may have
had a prior lending relationship with
the borrower. Instead of selling the
entire loan to an individual buyer, the
bank can agree to sell all or most of its
loans to an issuer — typically a government sponsored enterprise (GSE)
such as Fannie Mae or Freddie Mac
— that pools these loans with ones
made by other lenders (see Figure). For
example, rather than a single $200,000
mortgage, the pool may consist of
$600,000 in mortgages — that is,
three such loans.4 This means that instead of buying 100 percent of a single
mortgage, a potential investor who has
$200,000 to spend may end up with a
claim on one-third of each mortgage.
The GSE will place these mortgages in a trust (also known as a
special-purpose vehicle) (see Figure) and
then insure the pool against default;
this is a form of credit enhancement,
a technique for improving the credit
quality of one or more of the vehicle’s
assets. Credit enhancement can take
several forms: overcollateralization (so
that the dollar value of the assets in
the pool exceeds the value of the securities issued), the use of a GSE or other
outside insurer to guarantee payment,
and tranching, which we discuss later.
In many securitizations more than one
of these may be used.
The trust then issues securities,
known as mortgage-backed securities
(MBS), against this pool. Like other
bonds, these securities promise the

In practice, a typical pool may consist of several hundred loans and have a face value of $50
million.
4

www.philadelphiafed.org

buyer regular interest payments and
the return of principal at maturity, and
they are financed from the cash flows
of the underlying mortgages. Notice
that when the assets are moved off
balance sheet, they are legally separated from the bank that originated
the mortgages, so that creditors of
the bank (such as depositors and its
bondholders) do not have any claims
on these assets, and investors who receive mortgage payments do not have
any claims on the originating bank. A
certain amount is deducted from the
monthly payments on the mortgages
before they are passed through to the
investor; this money covers the servicing of the mortgages (i.e., collecting
the monthly payments, which is often

done by the issuing bank) and also
serves as compensation to the GSE
for its guarantee. For instance, in our
example, although homeowners pay an
interest rate of 6 percent, investors may
receive only 5.5 percent.
Investors will usually find it more
attractive to purchase an MBS than
to purchase an individual mortgage
loan. First, investors are exposed to
much less risk because the pooling
process diversifies away the impact of
an individual mortgage’s performance.
For example, investors do not need
to worry as much about an individual homeowner’s behavior (although
economy-wide disturbances that affect
many homeowners at once will still be
important). Second, the securities are

FIGURE
The Securitization of Mortgages
Homeowner
#1

Homeowner
#2

Loan Originator
(Lender) A

Homeowner
#3

Loan Originator
B

GSE

Trust
Senior Sequential
CMO Tranche

Pension Fund

Junior Sequential
CMO Tranche

Hedge Fund

Business Review Q3 2005 17

also much more liquid than individual
mortgages because the pooling process
makes each MBS much more similar
to its peers; that is, pooling makes
the characteristics of an individual
loan much less important to potential
investors. This reduces the amount of
information potential investors need to
collect before purchasing the security
and thereby makes it easier to trade.
Finally, the issuer of the MBS may
also further manipulate the cash flows
from the pool of mortgages by splitting
them into classes known as tranches
(see Figure).5 The difference between
one tranche and another varies depending on the type of asset securitized. In the case of mortgage-backed
securities, tranches are often structured in terms of principal payments
on the mortgages in the pool. That is,
the structure is used to allocate prepayment risk, the risk that a security will
pay off before its maturity date, thereby
forcing the investor to reinvest his
funds at a (possibly) lower rate. The
simplest structure is known as “sequential pay” (more complex ones are also
used). As the name suggests, in this
case the tranches are retired in sequential order. That is, investors in the first
— senior — tranche receive principal
payments from the underlying assets
first, those in the second tranche next,
and so on. Investors in the last — most
junior — tranche receive principal payments from the mortgages in the pool
only when the tranches ahead of them
in priority have been fully paid.
For instance, suppose that in our
example, the $600,000 pool consisting
of three mortgages was divided into

two tranches: a senior one with a principal balance of $200,000 and a more
junior one with a balance of $400,000.
Then if all mortgages paid according to
schedule, it would take 16.5 years for
the senior tranche to be paid down.6
During this time, the senior tranche
would receive all of the principal payments on the mortgages in the pool, as
well as interest payments of 5.5 percent
on its outstanding balance. The junior

However, in other ABS, the
absence of a GSE guarantee means
that the determining factor in structuring the tranches is typically credit
risk; that is, a senior tranche would
have priority over a junior one in the
event of a default, so that it has first
claim on the securitization’s underlying assets. As a result, tranching can
serve as a form of credit enhancement;
in particular, it enhances the credit

The issuer of the MBS may also further
manipulate the cash ﬂows from the pool of
mortgages by splitting them into classes
known as tranches.
tranche would receive only its interest
payments. After the senior tranche
has been fully paid down, the junior
tranche would then begin to receive
principal payments and would be fully
retired after 30 years.7
Now suppose that shortly after
the mortgages are issued, one of the
homeowners sells his house and pays
off his mortgage. In this case, the
senior tranche is paid off immediately.
The junior tranche would then begin
to receive principal payments as well;
nevertheless, so long as the other mortgages do not pre-pay, it would still take
30 years to fully pay down this tranche.
Notice that the junior tranche is thus
much less sensitive to prepayment risk
than the senior tranche.

quality of the more senior tranches at
the expense of the junior ones (the
senior tranche is typically AAA-rated
in these cases).8
In this example we can see the key
features of asset securitization: a sale of
the underlying assets so that they are
moved off the issuer’s balance sheet,
the pooling of illiquid assets, credit
enhancement, and tranching.

This figure can easily be obtained from any
online mortgage amortization calculator; several
such calculators are available.

Notice that another implication of the sequential pay structure is that the senior tranche has a
shorter maturity than the underlying mortgages;
thus, tranching also facilitates participation in
this market by investors with shorter investment
horizons.
7

In the case of mortgages, a tranched security is
known as a REMIC (real estate mortgage investment conduit) or CMO (collateralized mortgage
obligation).
5

18 Q3 2005 Business Review

WHAT ASYMMETRIC INFORMATION CAN TELL US ABOUT
ASSET SECURITIZATION
When Investors Are Uninformed, Capital Structure Matters.
A firm’s decision about whether — and
how — to securitize assets can be
viewed as a variant of the broader
question of how a firm should finance

Bonds are rated according to their default
risk by ratings agencies, the most prominent
of which are Moody’s and Standard &
Poor’s. Although each agency uses slightly
different classifications, ratings are assigned in
alphabetical order, with AAA being the least
risky (Aaa for Moody’s) and D representing
a bond that is in default. Bonds rated BBB or
above by Standard & Poor’s (Baa for Moody’s)
are termed “investment grade.”

www.philadelphiafed.org

itself. This is known as the capital
structure decision.
In 1958, future Nobel Prize winners Franco Modigliani and Merton
Miller showed that the form of financing a firm uses does not affect the total
value of its assets under a number of
particular assumptions. This is known
as the Modigliani-Miller proposition.
Some key assumptions — which we
will revisit below — are that corporate
bankruptcy is costless, that there are
no applicable government regulations, and that all types of securities
have similar tax treatment. Another
important assumption is that outside
investors are as well informed as the
firm’s insiders (such as management)
about the firm’s prospects. When this
is true, insiders and outsiders are said
to be symmetrically informed.
On the other hand, when insiders know more than outside investors (which is often a more realistic
assumption), the mix of debt securities9
and equity (that is, stock) — and who
holds each — can affect the firm’s
ability to secure funds from outside investors and, ultimately, the value of the
firm itself. Two classic papers examine
these issues, and the ideas in these articles can also be used to explain some
of the key features of ABS.
In their article, economists Hayne
Leland and David Pyle explain why
insiders tend to retain an equity stake
in their firm, rather than selling all of
the firm’s shares to the public. Insiders who believe that a firm’s future
profits are likely to be high would
like to convince skeptical investors.

Corporate and government bonds are common
examples of debt securities. A debt security
represents the issuer’s promise to repay the
loan’s face amount, with interest, in a set
period of time. By contrast, the firm is under
no contractual obligation to pay shareholders
dividends of any set amount.
9

www.philadelphiafed.org

On the other hand, skeptical investors believe that talk is cheap. They
reason that insiders are simply trying
to sell stock in the firm at the highest possible price, whatever the firm’s
true prospects. However, insiders can
credibly signal their information to the
market by holding a larger share of the
firm’s stock. In effect, an insider who

I’ll now show how these ideas can
help to explain some of the distinctive
features of securitization.
Tranching Allows Issuers to Sell
Safe Cash Flows and Retain Risky
Ones. Suppose an issuer (for example,
a bank) has a portfolio of assets such
as credit card receivables, that is,
expected payments on credit card bal-

The problem ﬁrms face when issuing equity is
that outsiders are understandably suspicious
that insiders know something they do not and
that the stock is overvalued.
holds a significant ownership stake is
putting his money where his mouth is.
This allows the firm to sell its stock
at a higher price but will leave insiders exposed to more risk because their
ownership share in the firm keeps
them from holding a well-diversified
portfolio; this increased risk is the cost
insiders must bear to gain credibility.
An article by economists Stewart
Myers and Nicholas Majluf explains
why firms often prefer to sell debt
securities rather than issue equity to
outside investors. The problem firms
face when issuing equity is that outsiders are understandably suspicious that
insiders know something they do not
and that the stock is overvalued. As
a result, the firm can increase the
price investors are willing to pay for
its securities by offering securities that
are informationally insensitive, that is,
securities whose payoffs do not depend
on factors known only to insiders.
For example, since debt payments
are contractually fixed whether the
firm’s profit is high or low, debt is less
informationally sensitive than equity;
therefore, the firm can secure outside
funds at a lower cost by issuing bonds
rather than stock.10

ances. This portfolio is not as liquid as
the issuer would like, and so the issuer
might prefer to sell part of it for cash
through a securitization. However, the
issuer’s information about the quality of its assets is superior to that of
potential investors, perhaps because
the bank has proprietary information
about its customers that it has collected over a long period. Having such
information makes any sale costly and
difficult. The bank’s goal is to structure the security so as to maximize its
revenue from selling the assets.
Economists Peter DeMarzo and
Darrell Duffie show that to maximize
revenue, the issuer should sell a senior
tranche backed by the assets while
retaining the junior tranche. By analogy with the firm’s capital structure

This is known as the Myers-Majluf pecking
order theory because the firm has a “pecking
order” of financing choices. It relies as much as
possible on retained earnings (which bypasses
outside investors completely). If retained
earnings do not suffice to finance its projects, it
issues debt. Only if the firm does not have the
earnings to make debt payments does it issue
equity to outside investors (a start-up firm might
fall into this category).
10

Business Review Q3 2005 19

decision, the most junior tranche is
also often termed the equity stake.
Moreover, they show that the higher
the quality of the assets, the larger this
retained equity stake. This follows the
work of Leland and Pyle in that the
issuer signals that its assets are of high
quality by holding an equity stake;
it is also reminiscent of Myers and
Majluf’s model in that an informationally insensitive security is issued to
uninformed outside investors.
To take a recent example, which
is fairly typical, in a 2002 credit card
securitization by Fleet Bank (now
part of Bank of America), the issuer
retained an equity interest equal to
approximately 10 percent of the total
principal.
Peter DeMarzo further extends
this model to explain why we often
see pooling of assets (recall that this is
a distinctive feature of many securitizations) before tranching occurs.
DeMarzo shows that pooling assets
involves a tradeoff. On the one hand,
by selling different assets as a single
unit, the issuer cannot signal information about the asset by retaining
a specific amount of equity for each
individual asset. On the other hand,
to the extent that pooling diversifies
idiosyncratic risk, it allows the issuer
to sell a larger quantity of informationally insensitive securities.11 When the
benefits from diversification outweigh
the limitations of selling the assets
together (for example, when the issuer
has many similar mortgages available),
then pooling is beneficial.
Tranching Increases Information Production by Investors. While
DeMarzo and Duffie’s model provides

Idiosyncratic risk is risk related to the unique
circumstances of a specific loan or borrower, as
opposed to overall market risk, which affects
many assets at once.
11

20 Q3 2005 Business Review

useful insights, its underlying assumptions do not reflect significant parts
of the ABS market. In many cases,
investors may actually know at least
as much about the assets as the issuer,
and even more significantly, some
potential investors may know more
than others (this is the case for mortgage-backed securities, for example).
Of course, investors do not receive this
information for free. Hedge funds that

In many cases,
investors may actually
know at least as much
about the assets as
the issuer, and even
more signiﬁcantly,
some potential
investors may know
more than others.
specialize in buying mortgage-backed
securities must pay substantial salaries
to Ph.D.s who understand these securities.
Economists Arnoud Boot and
Anjan Thakor develop a model in
which sellers of ABS exploit the fact
that potential investors may choose to
invest in learning about the underlying assets. Boot and Thakor show that
both the pooling and tranching of
assets can encourage investors to learn
about these assets, so that they are
willing to pay more for them.
Their idea is that by separating
the cash flows from the asset into
senior and junior tranches, the issuer
creates a highly informationally sensitive security — the junior tranche.
Since a junior tranche is riskier,
investors need to learn more about the
assets underlying this junior security in
order to determine whether it is worth

buying. By contrast, a high-rated senior
tranche carries less risk, so that even
uninformed investors can safely invest
in it.
This structure maximizes incentives for sophisticated investors to become informed about the value of the
underlying assets, since such investors
can specialize in buying only this most
informationally sensitive portion of the
cash flows. Conversely, uninformed
investors purchase the informationally
insensitive senior tranche. Also note
that unlike in DeMarzo and Duffie’s
model, the issuing firm itself does not
need to retain anything, since it knows
nothing more than investors do.
Boot and Thakor also offer a similar explanation for why securitizations
often involve the pooling of assets. The
reason is that the risks of the assets
pooled in the ABS have two components: a common one (such as interest-rate risk or national price trends in
the case of mortgages) and an idiosyncratic one (e.g., a particular borrower’s
individual default risk). Pooling assets
makes acquiring information more effective because the idiosyncratic risk is
diversified and investors can concentrate their efforts on learning about
the common characteristics of these
assets without worrying that their
efforts will be undone by an individual
homeowner’s unpredictable finances.
Economist Guillaume Plantin
provides evidence that in collateralized
debt obligations,12 it indeed appears
as if sophisticated investors, such as
hedge funds, purchase the more junior
“equity” tranches, whereas relatively
unsophisticated investors specialize in

Collateralized debt obligations, or CDOs, are
securities in which the underlying assets are
themselves loans or bonds, most typically risky
corporate debt (“junk bonds”).
12

www.philadelphiafed.org

the high-rated senior tranches, commonly known as “A” tranches.13
Structures with Many Tranches.
In the models discussed above, the
resulting structure of the securitization is very simple: usually only two
tranches, one senior and one junior. In
practice, most structures are somewhat
more complicated and feature multiple
tranches. For example, in the Fleet
credit card securitization discussed earlier, there were actually three tranches:
a senior AAA-rated “A” tranche, a
more junior “B” tranche (which was A
rated), and the unrated equity tranche.
Plantin’s paper explains why these
multiple tranches might arise; he also
demonstrates that — as in the papers
by Boot and Thakor and DeMarzo
and Duffie — the optimal structure is
a senior-junior securitization in which
the higher-rated senior tranches have
absolute priority over the low-rated
junior ones in the event of a default.
Plantin’s model features multiple
tranches because it includes several
classes of potential investors with
different degrees of sophistication (for
example, hedge funds, pension funds,
and individual investors). For Plantin, a sophisticated investor is more
likely to discover when a given pool of
assets is worth buying, whereas a less
sophisticated investor is more likely to
remain uninformed. Having multiple
investors that differ in their sophistication allows for multiple tranches in the
optimal structure.

13
For example, banks are among the most active
buyers of higher-rated senior tranches. The
reader may find it strange to think of banks as
unsophisticated, but Jianping Mei and Anthony
Saunders have demonstrated that — at least in
the case of real estate loans — banks seem to act
naively in lending on the basis of past returns
rather than expected future performance. Of
course, degrees of sophistication need not
explain why banks favor the senior tranches
— there are regulatory reasons for banks to
invest in less risky securities.

www.philadelphiafed.org

Plantin produces useful insights
by explicitly modeling the sale of ABS
as an auction. Auctions are the common sales method when securities are
privately placed (as opposed to being
publicly issued).14 The auction may
be informal, in which case the issuer
privately consults each potential buyer
before choosing the best offer. Alternatively, if there are many potential
bidders, a formal auction may be used,

winner’s curse. This problem should be
familiar to anyone who has won an
eBay bidding war, only to later discover
that the item is available for retail purchase at a lower price. Note that the
winner’s curse is not the result of bidders’ allowing their emotions to get the
better of their reason. Rather, it arises
because bidders are not equally well
informed about the valuation of the
object (in this example, the price for

In an auction where each bidder has his own
information about the true value of the items
being sold, there is the risk that the buyer who
wins the auction is the one who has overpaid.
typically a first-price sealed-bid auction.15 In either case, economists have
a well-developed set of insights about
the forces at play in an auction.16
In particular, in an auction where
each bidder has his own information
about the true value of the items being
sold, there is the risk that the buyer
who wins the auction is the one who
has overpaid. This is known as the

In a private placement, securities are issued
to “qualified institutional investors” (such as
insurance companies), rather than to the general
public, as in a public offering. The advantage
is that there is much less regulation; the
disadvantage is that since there is a very limited
secondary market, the price received is typically
lower. The “Bowie bonds” discussed earlier were
privately placed; Prudential Insurance Company
purchased the entire issue. More generally,
private placements make up approximately 15
percent of all nonmortgage ABS issued.
14

In a first-price sealed-bid auction, each bidder
submits a sealed bid to the seller (a bid that is
hidden from other bidders). The high bidder
wins and pays his bid for the good. Generally,
a sealed-bid format has two distinct parts: a
bidding period in which participants submit their
bids, and a resolution phase in which the bids
are opened and the winner determined.
15

16
See, for example, the book by Paul Klemperer
and the book by Paul Milgrom.

which it can be bought elsewhere). A
rational bidder takes this into account
when bidding. As a result, instead of
bidding his estimate of the object’s
value, the bidder will shave down his
bid to reflect the fact that he is likelier
to win when he has overestimated the
value of the object.
In Plantin’s model, the issuer
would like to maximize participation
in the auctions for the securities he
offers. The reason is that the more
potential bidders there are, the likelier
it is that some bidder will receive
information that confirms that these
assets are indeed of high quality, in
which case he would be willing to pay
a high price. In particular, the issuer
would like to encourage sophisticated
investors to participate, since they
are likeliest to receive information
concerning the asset. On the other
hand, the more sophisticated investors
there are, the more severe the winner’s
curse. The reason is that those investors who are not informed know they
will win the auction only if none of the
other investors learn that the assets are
of high quality. If many of the other
investors are sophisticated, the absence

Business Review Q3 2005 21

of higher bids suggests that the assets
are indeed of very low quality. Thus,
the uninformed investors are timid
in their bidding, which will reduce
the issuer’s revenue from the auction.
Thus, designing the structure so as to
encourage more sophisticated investors
to participate in the auction creates
a tradeoff: Sophisticated investors
— who are likelier to be well-informed
about the assets — will bid more
aggressively and so will pay a higher
price for high-quality assets. But they
exacerbate the severity of the winner’s
curse for the uninformed investors and
make them timid bidders.
Tranching plays a dual role in
resolving this tradeoff. It draws in sophisticated investors by creating an informationally sensitive junior tranche,
as in Boot and Thakor’s model. Since
Plantin assumes that sophisticated
investors must bear a higher cost to
participate in the auction for any given
tranche, these investors focus their efforts only on the most junior tranche.17
By contrast, unsophisticated investors
participate in the auctions for all of
the tranches. Since the sophisticated
investors bid for only the most junior
tranche, the unsophisticated investors can bid aggressively for the senior
tranches without fear of the winner’s
curse, which increases the issuer’s revenue.18 While these unsophisticated
investors also bid for the most junior
tranches, the winner’s curse means
that they do so very conservatively
and therefore are less likely to end up

Plantin argues that this is because it is difficult
for sophisticated investors to find retail clients
to ultimately hold these securities; for example,
only wealthy “qualified investors” are permitted
to invest in hedge funds.
17

The idea that creating a riskless security can
encourage participation by uninformed investors
was first used by Gary Gorton and George
Pennacchi to explain how insuring bank deposits
protects uninformed investors and thereby
makes them willing to fund banks.
18

22 Q3 2005 Business Review

holding these tranches when the auction closes. This is consistent with the
empirical evidence presented earlier:
junior tranches do indeed seem to be
held by more sophisticated investors.
REGULATION: ANOTHER
DRIVER OF SECURITIZATION
Legal factors and government
regulation are also important drivers of
securitization. Three main regulatory
and legal forces encourage securitization and determine some of its
characteristics.

in the first place, which will raise the
firm’s cost of financing (since they will
obtain a lower price for any securities
they offer).
Economists Gary Gorton and
Nicholas Souleles point out that
moving assets off balance sheet can
be helpful because firms can mitigate
these bankruptcy costs by precluding creditors’ access to these assets.19
For example, when a bank securitizes
mortgages, investors in the mortgagebacked securities are virtually guaranteed that they will be paid in full,

When a bank securitizes mortgages, investors
in the mortgage-backed securities are virtually
guaranteed that they will be paid in full,
regardless of how the bank itself fares in the
future.
Securitization May Reduce
Bankruptcy Costs. As mentioned
above, securitization is typically off
balance sheet in that the underlying
assets are legally separated from the
firm so that the firm’s creditors do
not have any claim on these assets.
Recall that the Modigliani-Miller
proposition assumed that bankruptcy
is costless. In practice, of course, it is
not. Bankruptcy costs take two forms:
direct costs, such as lawyers’ fees and
court costs, and indirect costs, which
include difficulties in raising funds to
make profitable investments, inefficient investments undertaken while in
bankruptcy, and so on. These indirect
costs may also affect a firm when it is
in financial distress, that is, even when
it is close to bankruptcy. Investors
(both shareholders and creditors) will,
of course, ultimately bear these costs
because the value of their securities will be impaired in bankruptcy.
Anticipating these costs, investors
will be more reluctant to offer funds

regardless of how the bank itself fares
in the future. Consequently, they are
willing to offer a high price for these
securities. By contrast, if the bank
retains the mortgages, investors will
share in both the cash flows from the
assets and the costs the issuer incurs
should it find itself in financial distress.
As a result, investors offer a relatively
lower price for these securities. This
is particularly true for risky, low-rated
issuers. A classic example is Chrysler:
It successfully used securitization in a
period of financial distress (1990-91)
when it could neither finance car loans
in the commercial paper market nor
issue long-term debt.20
However, not every type of asset
lends itself to securitization. Economists Kenneth Ayotte and Stav Gaon

Bankruptcy costs are also further reduced
by the credit enhancement that is a feature of
nearly all securitizations.
19

See the article by Dennis Cantwell.

www.philadelphiafed.org

show that if the assets are essential for
the firm’s continuing operations, the
firm’s losing control over them through
a securitization may imperil the firm’s
existence in case of financial distress.
The reason is that the holders of the
securitized assets have little interest
in the firm’s continued survival and
may not be willing to compromise to
help the firm avoid liquidation. Ayotte
and Gaon offer the example of the
bankrupt steel firm LTV, which made
this argument as part of an attempt
to regain control of inventory it had
securitized.21
Securitization Can Lower
Banks’ Regulatory Capital Requirements. Some economists have argued
that bank capital requirements are important drivers of securitization. This
is also known as regulatory arbitrage because securitization might allow banks
to shift assets to lower their minimum
regulatory capital requirements. In
particular, to the extent that minimum
capital requirements do not assign
to each asset the capital that would
be held by an unregulated financial
intermediary, it might be profitable for
banks to sell off low-risk loans (such as
mortgages) and retain high-risk assets.
Note that for this to be an effective
“arbitrage,” the loan’s buyer must have
a lower capital requirement for holding
that loan than the selling bank (for
example, an unregulated hedge fund).
As long as this is true, it is cheaper for
the buyer to hold the loan on its books
than for the bank, and both can profit
from its sale.22

Gorton and Souleles suggest that another
reason firms may not want to securitize all
assets is that interest payments on off-balancesheet debt are not always tax-deductible to
the issuing firm (although in practice lawyers
have developed structures that allow the tax
advantages to flow back to the issuer).
21

In addition to minimum capital requirements,
bank regulators can also limit regulatory
arbitrage through the examination process.
22

www.philadelphiafed.org

Consider the following example.
A bank can make one of two $100,000
loans, both of which require 8 cents
of capital per dollar lent.23 One loan
is an adjustable-rate mortgage with an
80 percent loan-to-value ratio. In 2000
the interest rate on such a mortgage
averaged 7 percent, and the default
rate was approximately 0.5 percent.
The other loan is a small-business
line of credit, with an interest rate of
7.4 percent and a default rate of 1.5
percent.24 Notice that the expected
return on the mortgage can never be
higher than 7 percent. By contrast, the
small-business loan has an expected
return that is at least 7.29 percent.25
Given the regulatory capital requirements, the bank may prefer to hold
the risky small-business loan and
sell the safe mortgage. The reason is
that under current minimum capital
requirements, both loans require the
bank to hold $8,000 of capital, but the
high-risk small-business loan has an
expected return that is nearly 30 basis
points higher.
The evidence as to whether
regulatory arbitrage is an empirically
significant driver of securitization
is mixed. On the one hand, Brent
Ambrose, Michael LaCour-Little, and
Anthony Sanders do find evidence
consistent with regulatory arbitrage
in the mortgage market. By contrast,
however, Bernadette Minton, Anthony

In rough terms, the capital requirement means
that for each dollar lent, the bank must secure at
least 8 cents of funding in the form of retained
earnings, stock, or long-term subordinated debt
(i.e., debt that is junior to deposits).
23

The data on small-business loans are
from the paper by Sumit Agarwal, Souphala
Chomsisengphet, and Chunlin Liu.
24

Because the small-business loan repays 1001.5 percent = 98.5 percent of the time. As a
result, even if this loan returns nothing when it
defaults, its expected return is at least 0.985 ×
7.4 percent = 7.29 percent.

Sanders, and Philip Strahan provide
empirical evidence that casts doubt on
the importance of regulatory arbitrage
and instead supports the hypothesis
that securitization is motivated by a
desire to reduce bankruptcy costs. In
particular, they find that unregulated
issuers (which are not subject to capital
requirements) seem to be more active
securitizers than banks. Moreover, it is
the riskier firms (for which bankruptcy
is obviously more of a concern) that
use securitization the most.
Pension Fund Regulations
Can Explain Credit Enhancement
and Tranching. Finally, regulations
governing pension funds are also
important for securitization. The
most prominent of these regulations
are found in the Employee Retirement Income Security Act (ERISA).
ERISA regulations govern pension
funds’ investment portfolios. Among
these regulations are those that restrict
funds’ holdings of low-rated or very
junior asset-backed securities in certain
circumstances. This clearly encourages
the use of credit enhancement in ABS
structures and, in particular, the creation of high-rated senior tranches.26
In light of the regulations’ obvious
importance, it is somewhat surprising
that economists have yet to examine
their relative weight in the growth of
securitization.
IS SECURITIZATION
EFFICIENT?
One important question we have
not yet discussed is the social implications of securitization. That is, does
it provide a net benefit to society or
perhaps simply lead to a transfer of
wealth from one party to another?
Said differently, many of the models

Many institutional investors also have selfimposed restrictions on the credit quality of
their portfolio.
26

Business Review Q3 2005 23

we have discussed involve the issuer’s
structuring the securitization so as
to maximize his revenues. But is the
issuer’s gain merely the investor’s loss?
Securitization Can Be Socially
Beneficial. Recall that bankruptcy
costs seem to be an important driver of
securitization (explaining its off-balance-sheet feature as well as the credit
enhancement). This ability to mitigate
bankruptcy costs is certainly likely to
be beneficial; we have already seen,
for example, that securitization helped
Chrysler Corporation continue operating during a time of financial distress.
In many of the other models we
examined, securitization is also implicitly beneficial, since it is structured so
as to reduce information asymmetries.
That is, investors may be willing to
pay more for certain tranches either
because they are more confident that
the securities they are buying are of
high quality or because the structure
makes it more profitable for them to
become informed about the assets.
In either case, this lowers the cost of
financing for the firm and could allow
it to fund profitable projects that might
otherwise be infeasible. This is good
for society; everyone can be made
better off if profitable projects are not
forgone.
Securitization May Sometimes
Be Harmful. Having said this, however, securitization could potentially
have social costs for several reasons.
To the extent that securitization
permits firms to circumvent bankruptcy law or to circumvent banks’
minimum capital requirements, it
is unlikely to be socially optimal. In
addition, the recent example of Enron
has shown that securitization can
sometimes be used to facilitate fraud.
By moving assets and liabilities off
its balance sheet, Enron was able to
muddy investors’ picture of the firm.
Enron also implicitly guaranteed some
of the assets it securitized, so that they

24 Q3 2005 Business Review

were not truly off balance sheet. As
a result, the firm was actually much
riskier than it appeared.
Finally, in some of our models,
securities were structured so as to
maximize the investors’ incentives
to become informed. While this may
sometimes facilitate the funding of
projects that otherwise would not be

tion is conducted off balance sheet and
also why it commonly features credit
enhancement.
Another set of explanations we
have explored is based on the existence of differences in information
about the underlying assets — either
between issuers and potential investors or between different classes of

According to some theories, off-balance-sheet
ﬁnancing and, to a limited extent, tranching are
responses to government regulations.
financed, it provides no net social
benefit if the project would have
been financed without securitization.
Moreover, by driving potential buyers
to spend money on acquiring information, the issuer would actually be
encouraging unnecessary investment
in information production.27 To put it
another way, society as a whole would
be better off if the assets were simply
sold without being subdivided into
tranches, and as a result, investors did
not need to invest the resources necessary to purchase these junior tranches.
CONCLUSION
Securitization is a large and
growing area of corporate finance. Its
key features are that it is typically off
balance sheet, combines many small
assets into a pool, and often divides
this pool of cash flows into tranches.
According to some theories,
off-balance-sheet financing and, to a
limited extent, tranching are responses
to government regulations. Bankruptcy
costs also help explain why securitiza-

27
This is similar to the argument often made
against advertising.

investors. These theories show that
securities may be designed to alleviate
these differences in information, so
that outside investors are comfortable
purchasing them, and they may also
be designed to encourage investors
to become better informed about the
underlying assets. This is manifested in
the pooling of assets and the subsequent division of these cash flows into
tranches.28
While there is a well-developed
body of theoretical work that explores
the determinants and structure of
securitization, the empirical significance of these models, and in
particular the impact of government
regulation and bankruptcy law on
securitization, remains a ripe area for
future research. BR

28
Information asymmetries and regulations are
not the only explanations for why new securities are introduced. There is an interesting
literature in which securities are designed to fill
unmet needs for risk-sharing, that is, to complete
markets. For example, a futures contract allows
farmers to lock in a price for wheat so that they
are not exposed to the risk that prices will collapse. For a model in which completing markets
drives financial innovation, see Franklin Allen
and Douglas Gale.

www.philadelphiafed.org

REFERENCES

Agarwal, Sumit, Souphala Chomsisengphet, and Chunlin Liu. “Determinants of Small Business Default,”
Working Paper, University of Nevada,
Reno, 2004.
Allen, Franklin, and Douglas Gale.
“Optimal Security Design,” Review of
Financial Studies, 1, 1988, pp. 229-63.
Ambrose, Brent, Michael LaCourLittle, and Anthony Sanders. “Does
Regulatory Capital Arbitrage or Asymmetric Information Drive Securitization?” Working Paper, Ohio State
University, November 2003.
Ayotte, Kenneth, and Stav Gaon.
“Asset-Backed Securities: Costs and
Benefits of ‘Bankruptcy Remoteness’,”
manuscript, 2004.
Boot, Arnoud, and Anjan Thakor.
“Security Design,” Journal of Finance,
48, 1993, pp. 1349-78.
Cantwell, Dennis. “How Public Corporations Use Securitization in Meeting
Financial Needs: The Case of Chrysler
Corporation,” in Kendall, Leon T., and
Michael J. Fishman (eds.): A Primer on
Securitization. Cambridge, MA: MIT
Press, 1996.

DeMarzo, Peter. “The Pooling and
Tranching of Securities: A Model of
Informed Intermediation,” Review of
Financial Studies, 18, 2005, pp. 1-35.
Gorton, Gary, and Nicholas Souleles.
“Special Purpose Vehicles and Securitization,” NBER Working Paper 11190,
March 2005.
Gorton, Gary, and George Pennacchi.
“Financial Intermediaries and Liquidity Creation,” Journal of Finance, 45,
1990, pp. 49-71.
Klemperer, Paul. Auctions: Theory
and Practice. Princeton, NJ: Princeton
University Press, 2004.
Leland, Hayne, and David Pyle. “Informational Asymmetries, Financial
Structure, and Financial Intermediation,” Journal of Finance, 32, 1977, pp.
371-87.
Mei, Jianping and Anthony Saunders.
“Have U.S. Financial Institutions’
Real Estate Investments Exhibited
‘Trend-Chasing’ Behavior?” Review of
Economics and Statistics, 79, 1997, pp.
248-58.

Milgrom, Paul. Putting Auction Theory
to Work. Cambridge University Press,
2003.
Minton, Bernadette, Anthony Sanders,
and Philip Strahan. “Securitization by
Banks and Finance Companies: Efficient Financial Contracting or Regulatory Arbitrage?” Working Paper, Ohio
State University, September 2004.
Modigliani, Franco, and Merton
Miller. “The Cost of Capital, Corporation Finance and the Theory of Investment,” American Economic Review, 48,
1958, pp. 261–97.
Myers, Stewart, and Nicholas Majluf.
“Corporate Financing and Investment
Decisions When Firms Have Information That Investors Do Not Have,”
Journal of Financial Economics, 13,
1984, pp. 187-221.
Plantin, Guillaume. “Tranching,”
Financial Markets Group Discussion
Paper DP-449, Revised December
2004.

DeMarzo, Peter, and Darrell Duffie.
“A Liquidity-Based Model of Security
Design,” Econometrica, 1999, 67, pp.
65-99.

www.philadelphiafed.org

Business Review Q3 2005 25

Do Budget Deﬁcits Cause Inﬂation?
BY KEITH SILL

s there a relationship between government
budget deficits and inflation? The data show
that some countries—usually less developed
nations—with high inflation also have large
budget deficits. Developed countries, however, show little
evidence of a tie between deficit spending and inflation.
In “Do Budget Deficits Cause Inflation?,” Keith Sill
states that the extent to which monetary policy is used
to help balance the government’s budget is the key to
determining the effect of budget deficits on inflation. He
examines the theory and evidence on the link between
fiscal and monetary policy and, thus, between deficits
and inflation.
In 2004, the federal budget deficit
stood at $412 billion and reached 4.5
percent of gross domestic product
(GDP). Though not at a record level,
the deficit as a fraction of GDP is now
the largest since the early 1980s. Moreover, the recent swing from surplus to
deficit is the largest since the end of
World War II (Figure 1). The flip side
of deficit spending is that the amount
of government debt outstanding rises:
The government must borrow to
finance the excess of its spending over
its receipts. For the U.S. economy, the

Keith Sill is a
senior economist
in the Research
Department of
the Philadelphia
Fed.

26 Q3 2005 Business Review

amount of federal debt held by the
public as a fraction of GDP has been
rising since the early 1970s. It now
stands at a little over 37 percent of
GDP (Figure 2).
For a long time, economists and
policymakers have worried about the
relationship between government
budget deficits and inflation. These
worries stem from the possibility that
the government will finance its deficits
by borrowing or by printing money.
Should deficit spending and a large
public debt be worrisome for monetary
policymakers who are concerned about
the economy’s level of inflation? Do
government budget deficits lead to
higher inflation? When looking at
data across countries, the answer is:
it depends. Some countries with high
inflation also have large government
budget deficits. This suggests a link
between budget deficits and inflation.

Yet for developed countries, such as
the U.S., which tend to have relatively
low inflation, there is little evidence
of a tie between deficit spending and
inflation. Why is it that budget deficits
are associated with high inflation in
some countries but not in others?
The key to understanding the relationship between government budget
deficits and inflation is the recognition
that government deficit spending is
linked to the quantity of money circulating in the economy through the
government budget constraint, which is
the relationship between resources and
spending. At its most basic level, the
budget constraint shows that money
spent has to come from somewhere: in
the case of local and national governments, from taxes or borrowing. But
national governments can also use
monetary policy to help finance the
government’s deficit.
The extent to which monetary
policy is used to help balance the
government's budget is the key to
determining the effect of budget
deficits on inflation. In this article, we
will examine theory and evidence on
the link between fiscal and monetary
policy and, thus, between deficits and
inflation.
BUDGETS AND ACCOUNTING
Budget constraints are a fact of
life we all face. We’re told we can’t
spend more than we have or more than
we can borrow. In that sense, budget
constraints always hold: They reflect
the fact that when we make decisions,
we must recognize we have limited
resources.
An example can help fix the idea.
Imagine a household that gets income
www.philadelphiafed.org

FIGURE 1
Federal Surplus/Deﬁcit Relative to GDP
0.06
0.04
0.02
0
-0.02
-0.04
-0.06
-0.08

1950 1954 1958 1962 1966 1970 1974 1978 1982 1986 1990 1994 1998 2002

Source: Haver Analytics

FIGURE 2
Federal Public Debt Outstanding
as a Fraction of GDP
0.6
0.5
0.4
0.3
0.2
0.1

67
19
69
19
71
19
73
19
75
19
77
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
20
01
20
03

Source: Ofﬁce of Management and Budget, Flow of Funds Accounts

from working and from past investments in financial assets. The household can also borrow, perhaps by using
a credit card or getting a home-equity
loan. The household can then spend
the funds obtained from these sources
to buy goods and services, such as
www.philadelphiafed.org

food, clothing, and haircuts. It can also
use the funds to pay back some of its
past borrowing and to invest in finan-

cial assets such as stocks and bonds.1
The household’s budget constraint
says that the sum of its income from
working, from financial assets, and
from what it borrows must equal its
spending plus debt repayment, plus
new investment in financial assets.
There are no financial leaks in the
budget constraint: The household’s
sources of funds are all accounted for,
its spending is all accounted for, and
the two must be equal. The household
may use borrowing to spend more than
it earns, but that source of funding is
accounted for in the budget constraint.
If the household has hit its borrowing
limit, fully drawn down its assets, and
spent its work wages, it has nowhere
else to turn for funds and would therefore be unable to finance additional
spending.
Just like households, governments
face constraints that relate spending
to sources of funds. Governments can
raise revenue by taxing their citizens,
and they can borrow by issuing bonds
to citizens and foreigners. In addition,
governments may receive revenue from
their central banks when new currency is issued. Governments spend
their resources on such things as goods
and services, transfer payments such
as Social Security to its citizens, and
repayment of existing debt. Central
banks are a potential source of financing for government spending, since the
revenue the government gets from the
central bank can be used to finance
spending in lieu of imposing taxes or
issuing new bonds. For example, the
U.S. Treasury received a little more
than $22 billion from the Federal
Reserve in 2003.2
Much of a central bank’s revenue
comes from its monetary policy opera-

Recent detail on Federal Reserve payments to
the Treasury can be found in the 90th Annual
Report, Board of Governors of the Federal Reserve System, 2003, Table 5, page 270.

The household can also sell some of its assets
to finance consumption. This is tantamount to
negative investment in assets.

Business Review Q3 2005 27

tions. An important aspect of modern
monetary policymaking is controlling
the short-term interest rate. Central
banks do this by purchasing and selling
interest-earning government bonds.
If the central bank wants to raise the
interest rate, it sells government bonds.
If it wants to lower the interest rate, it
buys government bonds. As a consequence of these open market operations,
central banks have government bonds
in their portfolios, and these bonds
earn interest. Thus, one component of
central bank revenue is interest earned
on the government bonds it holds.
The second component of central
bank revenue is also related to open
market operations. Central banks
are able to create and issue money to
pay for the government bonds they
purchase. The money that central
banks create is called high-powered
money, and it takes the form of currency held by the nonbank public plus
the reserves banks are required to
hold against certain types of deposits. Since the central bank can issue
high-powered money to pay for things
like government bonds, an increase
in high-powered money represents a
source of central bank revenue.
Revenues are one side of the
central bank’s budget constraint.
What does the central bank spend its
revenue on? As mentioned, a major
use of funds is to purchase government
debt in the conduct of open market
operations. The other component of
central bank spending is residual: what
is left over after the central bank pays
its expenses. In the U.S., this residual
gets turned over to the Treasury each
year.
We can get a consolidated government budget constraint by combining
the budget constraints of the treasury
and the central bank. The government spends its revenue on:
• Goods and services;
• Transfer payments; and

28 Q3 2005 Business Review

• Interest payments on government
debt held by the public.3
This spending is funded by:
• Tax receipts;
• The increase in debt held by the
public; and
• The increase in high-powered
money.
Note that if the government
increases the quantity of high-powered
money it can reduce other taxes or
borrowing.
The revenue the government
gets from the increase in high-powered money is called seigniorage.4
The extent to which governments
use seigniorage as a means for financing budget deficits plays a key role in
the link between budget deficits and
Recall that interest paid on government debt
held by the central bank goes back to the treasury.

More technically, seigniorage is the real
increase in the stock of high-powered money
(currency held by the nonbank public plus bank
reserves), i.e., the increase in the stock of highpowered money adjusted for the level of prices
in the economy. As shown in Figure 3, for the
U.S., this measure of seigniorage has been small.
See the book by Frederic Mishkin.

inflation. Since the creation of highpowered money, and thus seigniorage,
is undertaken by the central bank, the
consolidated budget constraint shows
the link between fiscal policy and
monetary policy. Money creation is a
source of revenue for the government.
The amount of revenue the government gets from seigniorage has implications for the government’s choices
about taxes, borrowing, and spending.5
HOW MUCH CAN THE
GOVERNMENT BORROW?
The consolidated budget constraint shows the link between the
There is also a subtle way in which governments can use monetary policy to help finance
spending. If the government can generate surprise inflation, the real value of the payments
it makes to holders of its debt falls below what
investors expected to receive when they bought
the debt. Surprise inflation erodes the value
of government debt, which means that a lesser
amount of real tax revenue must be raised to
pay off bondholders. However, generating surprise inflation to finance spending is ultimately
a losing game for the government. Eventually,
investors will catch on to what the government
is doing and demand a high enough interest
payment to compensate them adequately for the
government’s inflation policy.
5

FIGURE 3
Seigniorage Relative to Government Spending
0.06
0.05
0.04
0.03
0.02
0.01
0
-0.01
-0.02

1962
Q4

1966
Q4

1970
Q4

1974
Q4

1978
Q4

1983
Q4

1986
Q4

1990
Q4

1994
Q4

1998
Q4

2002
Q4

Source: Haver Analytics

www.philadelphiafed.org

government’s choices about spending,
taxing, borrowing, and seigniorage.
This relationship is a constraint only
in the sense that there may be limits
on the government’s ability to borrow
or raise taxes. Obviously, if there were
no such limits, there would be no constraint on how much the government
could spend at any point in time.
Certainly governments are limited
in their ability to tax citizens. (That is,
the government can’t tax more than
100 percent of income.) But are governments constrained in their ability
to borrow? Indeed they are. Informally,
the value of government debt outstanding today cannot be more than
the value of the resources the government has to pay off the debt.6
How do governments pay their
current debt obligations? One way is
for the government to collect more
tax revenue than it spends. In this
case, the surplus can be used to pay
bond holders. Another way to finance
existing debt is to collect seigniorage revenue and use that to pay bond
holders. Finally, the government can
borrow more from the public to pay existing debt holders. If the government
chooses this last option, any new debt
it issues would, in turn, have to be paid
off using future surpluses, future seigniorage, or future borrowing. As long
as the amount of debt the government
issues to pay its obligations does not
grow too fast over time, we can think
of the current value of outstanding
government debt as being ultimately
backed by a stream of future surpluses
and future seigniorage.7 Since investors generally prefer to receive payouts
sooner rather than later, the future
stream of surpluses and seigniorage
A formal derivation of this relationship can be
found in the Technical Appendix.

7
We have assumed that in the long run, government debt does not grow at a rate faster than
the interest rate.

www.philadelphiafed.org

that backs government debt must be
discounted to take account of the time
value of money. That is, the current
value of debt must equal the present
discounted value of future surpluses
and future seigniorage.8

Monetary policy
does not necessarily
have to adjust money
growth in response
to deﬁcit spending...
provided that deﬁcit
spending is expected
to be offset by future
surpluses.
We call this relationship the
government’s intertemporal budget constraint.9 It indicates that the government must plan to raise enough revenue (in present value terms) through
taxation and seigniorage to pay off its
existing debt and to pay for its planned
expenditures on goods, services, and
transfer payments.
The intertemporal budget constraint has some interesting implications for monetary and fiscal policy.
Suppose the government decides that,
for a set path of future spending, it will
lower current and future taxes permanently. This policy would lower the
present discounted value of future surpluses. So, to fund the path of future
spending, the government would need
to increase the present discounted

value of seigniorage. Since seigniorage is related to high-powered money
growth, the implication is that money
growth must increase in the future.
Similarly, if the government decides to
permanently increase future surpluses
— for example, a permanent increase
in taxes or a permanent reduction in
borrowing — so that the present discounted value of future surpluses rises,
the present discounted value of future
seigniorage must fall; therefore, future
money growth must fall.10
Note that the constraint does not
say that an increase in deficits must
be accompanied by a rise in seigniorage. An increase in the deficit could
be temporary in the sense that it will
be offset by future surpluses. In other
words, a deficit today could be negated
by a future surplus, so that the present
discounted value of future surpluses
remains unchanged. In that case, no
offsetting adjustment in the value of
discounted future seigniorage would
be necessary. Monetary policy does
not necessarily have to adjust money
growth in response to deficit spending by the government, provided that
deficit spending is expected to be offset
by future surpluses. But if the present
discounted value of future surpluses
changes, there must be an offsetting
change in the present discounted value
of seigniorage, and vice versa.

Present value refers to an amount of money today that will become a given amount at a stated
point in the future, depending on the interest
rate. For example, if the interest rate is 10 percent, $100 today will be worth $110 in one year.
So the present value of $110 one year from now
(with an interest rate of 10 percent) is $100.

POLICY, DEFICITS, AND
INFLATION
Suppose that whenever there is a
change in the present discounted value
of seigniorage, fiscal policy adjusts so
that the intertemporal budget constraint holds. In this case, monetary
policy is independent in the sense that
monetary policymakers take action
without regard to fiscal policy, and
then fiscal policy adjusts to maintain

An “intertemporal constraint” shows how government resources and spending are linked over
time.

10
The government can permanently increase future surpluses by raising taxes or borrowing less.

Business Review Q3 2005 29

a balanced budget.11 With monetary
independence, policymakers are free to
pursue goals such as low and stable inflation and not have to worry about using money growth to finance treasury
budget deficits. In this case, we would
not expect a tight link between government budget deficits and inflation
because current government budget
deficits are expected to be largely offset
by future government budget surpluses.
In addition, the path of government
budget surpluses is expected to offset
changes in seigniorage, so that the
intertemporal budget constraint holds.
This does not mean that we may
not observe some correlation between
deficits and spending. For example,
if the economy is hit by a recession,
the deficit is likely to rise because tax
revenues fall. At the same time, monetary policymakers may lower interest
rates to combat the recession, an act
that may subsequently lead to higher
inflation. In this case, though, deficits
are not, per se, the cause of inflation.
Rather, deficits and inflation are both
consequences of the recession.
The alternative case is one in
which monetary policy is dependent.
When monetary policy is dependent,
the central bank adjusts seigniorage
so that the budget constraint holds.
Monetary policy responds to fiscal
policy, so that seigniorage revenue
becomes an important component of
government finance. An independent
treasury might decide to run permanent deficits, a situation that requires
seigniorage to make up the gap between the value of the public debt and
the present discounted value of budget
surpluses. In this case, we could expect
to see a link between deficits and
inflation, since monetary policymakers respond directly to a fiscal policy
of deficit spending. Whether monetary
See Michael Dotsey’s article for more on independent and dependent monetary policy.

30 Q3 2005 Business Review

policy is independent and fiscal policy
is dependent or vice versa is the key
to answering the question of whether
budget deficits imply higher inflation.
Dependent Monetary Policy May
Result in Unexpected Outcomes. In
a 1981 article, Thomas Sargent and
Neil Wallace offer a famous example
of how dependent monetary policy can
lead to unexpected outcomes. Suppose
fiscal policy is independent, monetary
policy is dependent, monetary policy
responds to fiscal policy, and the

It seems safe to say
that, for the U.S.
economy, there
is little, if any, link
between deﬁcits and
inﬂation.
intertemporal budget constraint holds.
In this case, an attempt by monetary
policymakers to rein in inflation today
by lowering money growth can result
in higher inflation in the future:
Policymakers are ultimately defeated
in their efforts to lower inflation. How
could this happen?
Suppose monetary policymakers lower current money growth in an
effort to bring down inflation. Lower
money growth means lower seigniorage. If government spending and taxes
do not change, the government will
have to borrow more from the public
in order to make up for the lost revenue from seigniorage. If the outstanding public debt increases, the intertemporal budget constraint implies that
there must be a corresponding increase
in the present discounted value of future budget surpluses and seigniorage.
In a regime of fiscal independence,
fiscal policy does not adjust, so the
present discounted value of budget
surpluses does not change. But that
means that the present discounted

value of seigniorage must rise to match
the increase in the value of public
debt outstanding. That is, the central
bank will be required to increase the
rate of money growth (seigniorage), an
action that ultimately leads to higher
inflation.12 In this case, efforts to use
monetary policy to lower inflation are
self-defeating.
EMPIRICAL EVIDENCE ON
INFLATION AND DEFICITS
Economic theory suggests that the
strength of the relationship between
government budget deficits and inflation depends on whether monetary
policy is independent or dependent
relative to fiscal policy. In countries
where seigniorage is an important
component of government finance,
we are likely to find that government budget deficits and inflation are
empirically linked. In countries with
independent monetary authorities, the
link between deficits and inflation is
likely to be weaker.
Evidence for the U.S. Economy.
As we can see from a plot of deficits
and inflation for the U.S. economy
since the end of World War II, there
does not appear to be much of a relationship between government budget
deficits and inflation (Figure 4). The
contemporaneous correlation between
federal budget deficits and inflation
(GDP deflator inflation) is essentially
zero. It is possible that deficits today
are more highly correlated with future
inflation than with current inflation
— it may take some time for deficits to
be felt in the form of higher inflation.
But even if we look for the largest correlation between current deficits and
future inflation, we find that it is still
rather low at 10 percent, when current
deficits are correlated against inflation
12
There is a strong empirical link between money growth and inflation for a wide range of
countries over a long span of time. See the article by George McCandless and Warren Weber.

www.philadelphiafed.org

six quarters ahead. It seems to be the
case that for the U.S. economy, deficits
and inflation are largely unrelated.
It seems safe to say that, for the
U.S. economy, there is little, if any,
link between deficits and inflation.
The reason is that the Federal Reserve
largely sets monetary policy independently of what the Treasury is doing to
finance the federal government budget
deficit. The Fed turns over its profit
to the Treasury each year, but the Fed
does not conduct monetary policy to
raise revenue for the Treasury. Rather,
the Fed focuses on stabilizing inflation and unemployment and does not
conduct monetary policy with an eye
toward financing fiscal deficits.
More thorough evidence than
simple correlations bears out the
finding that deficits and inflation are
weakly linked, if at all, in the U.S. and,
for that matter, in most of the world’s
advanced economies.13 However, there
does seem to be a link between deficits
and inflation in the world’s less-developed economies. For those countries,
high inflation is often associated
with high average government budget
deficits.
Evidence for the Rest of the
World. A recent study by Stanley
Fischer, Ratna Sahay, and Carlos Vegh
classified a sample of 94 countries
into high-inflation and low-inflation
countries. High-inflation countries, of
which there were 24 in their sample,
are those that experienced at least one
episode of 12-month inflation exceeding 100 percent over the span 1960 to
1995. On average, inflation in those
countries was a bit over 150 percent
per year. Seigniorage as a fraction of

An older set of empirical studies tended
to find that there was at best a tenuous link
between deficits and inflation for the U.S.
economy. See the papers by G. Demopoulos,
G. Katsimbris, and S. Miller; K. Grier and
H. Neiman; D. Joines; and Robert King and
Charles Plosser.

GDP averaged about 4 percent in highinflation countries versus an average of
1.5 percent in low-inflation countries.
High-inflation countries rely more on
seigniorage to help finance government spending. The authors find that
for high-inflation countries, a worsening fiscal balance is much more likely
to be accompanied by an increase in
seigniorage than is the case in low-inflation countries.
What triggers inflation? The authors use standard techniques to show
that fiscal deficits lead to high inflation when the government depends on
revenue from seigniorage to finance
debt. They find that for high-inflation countries, a 10-percentage-point
reduction in the fiscal balance (i.e.,
deficit) as a fraction of GDP is associated with, on average, a 4.2 percent
increase in seigniorage. For low-inflation countries, there is no significant
link between deficits and seigniorage.
Also, when high-inflation countries
experience episodes of low inflation,
the link between deficits and inflation
weakens dramatically.

A 2003 study by Luis Catao and
Marco Terrones uses a broader sample
of 107 countries over the period 1960
to 2001 to look for a link between fiscal deficits and inflation. They find a
strong link between fiscal deficits and
inflation in developing countries. For
example, a 1 percent reduction in the
ratio of the budget deficit to GDP is
associated with an 8.75 percent lower
inflation rate. Catao and Terrones also
find results similar to those of Fischer,
Sahay, and Vegh when the sample is
broken into high-inflation and low-inflation countries using the 100 percent
annual inflation rule. But they also
find a statistically significant relationship between deficits and inflation in
countries with moderate inflation as
well, though the link is weaker. For
low-inflation and advanced countries,
Catao and Terrones find no link between fiscal deficits and inflation.
For developing countries, seigniorage is a significant source of revenue,
and fiscal policy appears to be an
important ingredient for the amount
of inflation. Indeed, over the period

FIGURE 4
Federal Deﬁcit and Inﬂation

www.philadelphiafed.org

Source: Haver Analytics

Business Review Q3 2005 31

1980 to 1995, seigniorage as a fraction
of GDP averaged about 2.2 percent,
compared with only 0.64 percent in
advanced economies such as the U.S.,
Germany, and Japan.14 One possible
reason for the greater reliance on seigniorage revenue in developing economies is that, for them, seigniorage may
be a relatively efficient method to raise
revenue compared with other forms
of taxes. In developing countries, it
may be difficult to collect tax revenue,
since the tax base tends to be small
and difficult to identify, especially
when the government does not have a

lot of resources to devote to building
an efficient tax-collection system.
SUMMARY
Monetary policy and fiscal policy
are linked because money growth,
in the form of seigniorage, provides
revenue to the fiscal branch of the
government. But whether deficits lead
to inflation depends on the extent to
which monetary policy is independent,
that is, the extent to which monetary
policymakers must react to fiscal
financing developments when setting

We have focused on the possible inflation
consequences of government budget deficits.
Other questions of interest we have not explored include the impact of budget deficits on
real interest rates and exchange rates.

For more detail on seigniorage revenue in
developing and advanced economies, see the
article by Paul Masson, Miguel Savastano, and
Sunil Sharma.
14

policy goals and implementing them.15
For the U.S. economy, there is
little evidence of a link between fiscal
deficits and inflation, precisely because
monetary policymakers have been free
to pursue goals such as low and stable
inflation. They are able to do this
because fiscal policy is seen as sustainable, in the sense that deficit spending
today is not expected to continue to
the extent that monetary policy will
have to provide major funding for the
Treasury. This is largely the case for
the developed countries of the world.
Developing countries, however, often
require revenue from seigniorage to
meet their fiscal financing needs.
Thus, these countries tend to show a
strong link between fiscal deficits and
subsequent inflation. BR

Technical Appendix
The Government's Intertemporal Budget Constraint
We can express the consolidated budget constraint in the symbolic form:

i t -1 B t -1+G t =

Tt+(B t – B t -1 )+( H t – H t -1 )

where G t is government spending at time t, i t - 1 B t - 1 is interest payments on publicly held government debt outstanding, Tt is tax receipts, and H t is high-powered money. The left-hand side of the expression is total spending by the
government and the right-hand side is total sources of revenue. It is convenient to put the budget constraint in inflation-adjusted, or real, terms by dividing through by the price level Pt . Define the real interest factor as

(1+rt ) =

1+it
(Pt / Pt-1)

We’ll use lower case to denote real values. Then re-arranging terms, we can write the consolidated budget constraint as:

( 1 + r ) b t- 1 + g t = t t + b t + s t
In this expression, tt is the real value of taxes collected, and st is the real value of the increase in money, or
seigniorage. Finally, ( 1+ r ) is the real interest factor on government debt, which we assume (for simplicity) is constant
over time. If we iterate the budget constraint forward T times into the future, we get:
T

(1+ r ) b t -1=∑
i=0

t t+i - g t+i+∑ s t+i
T

(1+ r ) i

i=0

(1+ r ) i

bt+T

(1+ r ) T

As long as the real amount of debt outstanding grows no faster than the real interest rate, which is a condition that
says enough economic resources will be available to fully pay off any debt outstanding, then as T gets larger, the last
term in the expression should get closer and closer to zero.
The first term on the left-hand side of the equal sign is the present discounted value of future budget surpluses. The
second term is the present discounted value of future seigniorage. The equation shows that the real value of debt held
by the public (principal and interest) is constrained by the government’s ability to raise revenue to pay it off.
32 Q3 2005 Business Review

www.philadelphiafed.org

REFERENCES

Board of Governors of the Federal
Reserve System. 90th Annual Report,
2003.

Grier, K., and H. Neiman. “Deficits,
Politics and Money Growth,” Economic
Inquiry, 25(2), April 1987, pp. 201-14.

Catao, Luis, and Marco Terrones.
“Fiscal Deficits and Inflation,” IMF
Working Paper WP/03/65 (2003).

Joines, D. “Deficits and Money
Growth in the United States 18721983,” Journal of Monetary Economics,
16(3), November 1985, pp. 329-51.

Demopoulos, G, G. Katsimbris, and S.
Miller. “Monetary Policy and CentralBank Financing of Government
Budget Deficits,” European Economic
Review, 31(5), July 1987, pp. 1023-50.
Dotsey, Michael. “Some Not-SoUnpleasant Arithmetic,” Federal
Reserve Bank of Richmond Economic
Quarterly, 82, 4 Fall 1996.
Dwyer, Gerald P Jr., and R.W. Hafer.
.,
“The Federal Government’s Budget
Surplus: Cause for Celebration,”
Federal Reserve Bank of Atlanta
Economic Review, July 1998.
Fischer, Stanley, Ratna Sahay, and
Carlos Vegh. “Modern Hyper- and
High Inflations,” Journal of Economic
Literature 60 (2002), pp. 837-80.

www.philadelphiafed.org

King, Robert, and Charles Plosser.
“Money, Deficits, and Inflation,”
Carnegie-Rochester Conference Series
on Public Policy, 22, Spring 1985, pp.
147-96.
Klein, Martin, and Manfred Neumann.
“Seigniorage: What Is It and Who
Gets It?” Weltwirtschaftliches Archiv.,
126(2), 1990, pp. 205-21.
Masson, Paul R., Miguel A. Savastano,
and Sunil Sharma. “Can Inflation
Targetting Be a Framework for
Monetary Policy in Developing
Countries?” IMF Finance and
Development, March 1998, pp. 34-37.

McCandless, George, and Warren
Weber. “Some Monetary Facts,”
Federal Reserve Bank of Minneapolis
Quarterly Review, 19(3), Summer
1995, pp. 2-11.
Mishkin, Frederic A. The Economics of
Money, Banking and Financial Markets,
6th edition. Reading, MA: AddisonWesley, 2001.
Sargent, Thomas, and Neil
Wallace. “Some Unpleasant
Monetarist Arithmetic,” Federal
Reserve Bank of Minneapolis
Quarterly Review, 5(3), Winter 1981,
pp. 1-17.
Walsh, Carl. Monetary Theory and
Policy. Cambridge, MA: MIT Press,
2003.

Business Review Q3 2005 33

Challenges and Opportunities in a Global Economy:
Perspectives on Outsourcing, Exchange Rates, and Free Trade
A Summary of the 2004 Philadelphia Fed Policy Forum
BY LORETTA J. MESTER

“C

hallenges and Opportunities in a Global
Economy: Perspectives on Outsourcing,
Exchange Rates, and Free Trade” was the
topic of our fourth annual Philadelphia Fed
Policy Forum held on December 3, 2004. This event,
sponsored by the Bank’s Research Department, brought
together a group of highly respected academics, policymakers, and market economists, for discussion and debate
about the macroeconomic impact of developments in
the global economy. Our hope is that the 2004 Policy
Forum serves as a catalyst for both greater understanding
and further research on policymaking in an increasingly
global economy.

Over the past couple of years,
the widening U.S. trade deficit and
rising oil prices became front page
news in discussions of U.S. economic
performance. The longer-term impact
of globalization on our labor markets
and economic well-being became a
discussion topic at cocktail parties and
around dinner tables. The feeling
that globalization was leading to the
loss of U.S. jobs made some people
even question whether free trade was

Loretta J. Mester
is a senior vice
president and
director of
research at the
Federal Reserve
Bank of
Philadelphia.

36 Q3 2005 Business Review

as positive for the U.S. economy as
economists know it to be. As world
economies become more integrated,
topics such as the macroeconomic
effects of outsourcing, exchange rate
policies and the flow of financial capital, and free trade and the cross-border
flow of goods and services are garnering increased attention from policymakers and researchers. How best to
seize the opportunities and meet the
challenges of the global economy was
the focus of the 2004 Philadelphia Fed
Policy Forum.
Anthony M. Santomero, president of the Federal Reserve Bank of
Philadelphia, began the day discussing
the breadth and depth of the global
economy’s influence. The international
marketplace is widening geographically, and the U.S.’s relationships with its
traditional trading partners in North

America and Europe, with Japan, and
with the emerging markets of Asia are
evolving.
In Santomero’s view, developments
in the global economy are transforming the basic structure of the economy,
the issues policymakers need to address, and the questions researchers are
studying. The revolution in information technology and the emergence of
new market economies are opening up
opportunities to reallocate production
and distribution around the globe.
Yet, so far, the potential effects of
this outsourcing on the U.S. economy have been difficult to quantify.
Similarly, there is still much to learn
about the distribution of the costs and
benefits of free trade. An examination
of the sharp decline in the value of the
dollar during the mid-1980s suggests
that a substantial relative price change
causes an expansion or contraction of
economic activity in well-established
sectors but does not open up brand
new areas of international trade.
Declining trade barriers, however,
bring more fundamental change to
the economies affected. For example,
as Timothy Kehoe discussed later in
the day, the North American Free
Trade Agreement (NAFTA) led to an
increase in trade in goods and services that were traded only in limited
quantities previously and accelerated the transfer of new technologies
across borders. Santomero conjectures
that one possible explanation for the
difference in effects is that a change
in tariffs is perceived as being more
permanent than a change in exchange
rates; hence, it elicits a larger response.
He also posits another possible explanation: that changes in exchange rates
www.philadelphiafed.org

affect relative prices across a broader
array of goods and services and so
evoke smaller adjustments across that
broad array, while changes in tariffs
affect a smaller number of goods and
services and so have narrower but
larger effects.
While opening up free trade
brings participants an improved standard of living, it also creates dislocations and imposes cost on individual
sectors within nations. As Santomero
points out, free trade is beneficial
provided the people and firms who
gain from it are able to compensate the
losers. Policymakers need to grapple
with the political problem of how to
redistribute the benefits of free trade
in order to build and maintain support
for free-trade policy. Countries are
approaching free trade along various
paths. Some are pursuing global trade
arrangements, others are pursuing
free trade areas, and some are pursuing bilateral trade agreements. In
Santomero’s view, the success of each
of these strategies in building the
necessary support for free trade is an
open question.
OUTSOURCING1
The Policy Forum’s first session
considered the issue of outsourcing.
Was it a reason for employment’s slow
recovery in this expansion? What has
it meant for the industrial sector? And
what determines whether a firm will
choose to outsource its operations?
Labor markets have been weaker
for longer in this recovery than in any
of the other postwar recoveries, even
the one in 1991, which has been called
the jobless recovery. The 1990 and
2001 recessions were about the same
length – eight months – but it took

1
Many of the presentations reviewed
here are available on our web site at www.
philadelphiafed.org/econ/conf/ policyforum2004.
html.

www.philadelphiafed.org

almost four years for U.S.
employment to recover back
to the level of its previous
peak in March 2001. During
the 1991 recovery, it took
about two and a half years.
Cathy Minehan, president of the Federal Reserve
Bank of Boston, elaborated
on the behavior of labor
markets during this business
cycle. In her view, foreign
outsourcing has not played
a major role in the relatively
slow rate of job growth
during this recovery. The
U.S. economy in the third
quarter of 2004 looked quite healthy,
growing at a sustainable pace, with the

Cathy Minehan

assumed about population growth, the
labor-force participation rate, and the
noninflationary unemployment rate.

Foreign outsourcing has not played a major
role in the relatively slow rate of job growth
during this recovery.
unemployment rate trending down,
inflation well contained, and productivity growth strong. Still, sluggish job
growth had been a concern during the
recovery. Labor-force growth had outpaced job growth during and after the
recession and opened an employment
gap. Unemployment had been longer
in duration than typical, and Minehan posited that this was because job
losses during the recession had been
of a more permanent than temporary
nature. Highly educated middle-aged
workers lost jobs this time, but the less
educated, younger workers made up
more of the long-term unemployed.
Also unique to this recovery is that
labor-force participation continued to
decline as the recovery unfolded.
Minehan presented a range of
estimates of how much job growth
would be needed to close the gap
between actual and full employment.
These estimates depend on what is

To meet demographic growth in the
labor force, which includes population
growth and changing patterns of work
and aging, Minehan estimates the
economy needs to add about 120,000
jobs per month. If labor force participation continues on the low side, then
the economy needs to create fewer jobs
to absorb labor supply. But if laborforce participation reverts to its more
normal level, the economy would need
to add more workers. Also, the lower
one believes unemployment can go
without inflation becoming a problem,
the more jobs can be created. Depending on the assumptions about laborforce participation and the natural rate
of unemployment, Minehan estimates
that somewhere between 125,000 and
225,000 jobs per month would have
to be created to absorb the increase in
labor supply.
Minehan evaluated two factors
that the media have often mentioned
Business Review Q3 2005 37

as factors for the recent unusually slow
job growth. First, the loss in manufacturing jobs has continued, and it has
become steeper in recent years. But in
Minehan’s view, while this is part of
the recent story, it cannot fully explain
sluggish job growth, since the economy
has been losing manufacturing jobs
for most of the last 30 years. Second,
foreign outsourcing has expanded. Not
only goods-producing industries but
also service-producing industries have
begun to outsource. But, again, this
cannot be the full explanation. While
U.S. firms are outsourcing to India and
China, Minehan points out that those
countries appear to be buying more
services from the U.S. than the U.S. is
from them, and this creates an offset in
terms of jobs. The fact that U.S. firms
do not point to imports or outsoucing
as the main cause of extended layoffs
is taken by Minehan
as evidence against the
outsourcing explanation
of slow job growth.
Then what is the explanation? Why are U.S.
firms demanding less
labor? Partly, this may be
due to structural change
as the economy shifts its
mix of products and services; partly it might be
a reaction to increasing
labor costs, especially the
cost of benefits; partly it
might be firms’ response
to higher uncertainty,
perhaps over the staying
power of the recovery
because of high oil prices
and geopolitical concerns; and perhaps
it’s because firms are driven to become
ever more productive. Minehan concludes that the latter two factors – uncertainty and the drive for increased
productivity – might be the best
explanations of the sluggish job growth
that characterized this recovery.

38 Q3 2005 Business Review

Robert Lawrence of Harvard
University extended the discussion
of the relatively weak employment
growth the U.S. experienced during
the recovery. The media have focused

stronger productivity growth because
productivity growth was rapid not only
in manufacturing but also in other sectors, which experienced fewer losses.
Looking deeper at the data shows that

Devising reliable measures to determine
trade’s impact on employment is not easy.
on the role of international trade,
particularly with China and India, and
the effects of outsourcing on the U.S.
economy were discussed during the recent presidential campaign. Lawrence
described some of his recent research
with Martin Baily of the Institute for
International Economics that attempts
to quantify the role of trade on the
employment losses between 2000 and
2003. Like Minehan, he pointed out

Robert Lawrence

the sharp drop in manufacturing employment during the recent recession.
In fact, while the share of employment in manufacturing declined
throughout the 1990s, the number of
workers employed in manufacturing
didn’t begin to decline until 2001, the
beginning of the recession. In his view,
one cannot simply attribute this to

managers and production workers suffered the largest job losses, but many of
those managers were in the manufacturing sector. Another factor during
this recovery that Lawrence highlighted was the abnormally slow recovery
in investment, which he feels is an important part of the story. Indeed, the
largest manufacturing employment decline was in computers and electronic
products, which lost about 30 percent
of its jobs. In effect, it was the capital
goods part of the manufacturing sector
that experienced the highest job losses.
In addition, exports during this cycle
were quite a bit weaker than they were
over other cycles, while imports were
somewhat weaker. Since manufacturing productivity growth was much
higher over the 2000-2003 period than
either manufacturing export or import
growth, jobs attributable to exports
declined over this period, as did jobs
embodied in imports.2
But devising reliable measures to
determine trade’s impact on employment is not easy. Lawrence and Baily
take an input-output approach to
determine the sectors in which exports
create jobs and the sectors in which
imports subtract from jobs in the sense
that jobs in those sectors would have
been higher had we produced those
imports domestically rather than buy-

2
Over the 2000-2003 period, manufacturing
productivity growth rose 15.2 percent,
manufacturing exports declined 8.8 percent, and
manufacturing imports rose 2.3 percent.

www.philadelphiafed.org

ing them from abroad. Since total output equals production for domestic use
plus exports minus imports, after jobs
attributable to exports and imports
are determined, jobs attributable to
domestic use can be calculated as the
residual. The results of their analysis suggest that weak U.S. domestic
demand and trade both contributed to
the loss in employment from 2000 to
2003, but that domestic demand had
a larger effect than trade. Moreover,
most of the job losses due to trade were
due to weak exports and not to increased imports. Merchandise imports
as a share of goods GDP were stable,
31.8 percent in 2000 and 31.4 percent
in 2003, while merchandise exports
as a share of goods GDP fell from 22.7
percent in 2000 to 20.1 percent in
2003. Based on data available as of December 2004, Lawrence and Baily estimate that of the 2.85 million jobs lost
between 2000 and 2003, 2.54 million
were due to weak domestic demand,
0.74 million were due to weak exports,
and imports actually contributed 0.43
million jobs. Lawrence concludes that
the job losses during the recession and
first part of the recovery were “made
in America.” His analysis also reveals
that the decline in U.S. exports is a
market-share story rather than a weakforeign-demand story. The U.S. lost
competitiveness against other suppliers
to the world market. If the U.S. had
held its share in world markets, exports
would have risen by 23.5 percent rather than declined. The lagged effects
of the rise in the value of the dollar in
the late 1990s played an important role
in limiting U.S. exports as well.
Finally, Lawrence turned his focus
to the future of manufacturing employment. Here there are two countervailing effects. If the U.S. closes the trade
deficit by 2015, this will create jobs
as U.S. exports increase and imports
decrease. But if at the same time productivity growth in manufacturing sta-

www.philadelphiafed.org

bilizes at its average 3.9 percent pace
seen over the past decade, then net
employment creation will be much less.
Lawrence concluded that contrary to
the discussion in the popular press,
trade was not a large part of the story
of the employment losses during the
recession and it isn’t likely to be a large
part of the manufacturing employment
story of the future.
While the session’s first two
speakers concentrated on the macroeconomic effects of trade and
outsourcing, the next speaker, Gene
Grossman of Princeton University,
refocused the discussion, taking a
microeconomics perspective on how

are traded in outsourcing relationships
are often customized for a particular
user. This is different from the types
of products that trade theory usually
considers, which are homogeneous
goods that can be bought in multiple
markets. This customization requires
relationship-specific investments,
which enhance the value of the relationship. Outsourcing also requires
contracts to govern the relationship.
Offshoring also has distinctive features. One aspect is the cost of transportation and communication. These
fixed costs can create complementarities between offshoring activities.
Once one activity is moved offshore, it

It is often the largest and most productive ﬁrms
that ﬁnd it cost effective to move production
offshore.
multinational firms decide to organize
their production activities. Grossman
explained that trade theory is concerned with the allocation of resources
over the longer run rather than the
shorter-run dynamics discussed by
our first two speakers. He began by
explaining the difference between
“outsourcing” and “offshoring,” terms
that are often used synonymously in
popular discussions but that trade
economists view as distinct. Outsourcing pertains to how a firm chooses to
organize itself. Does the firm perform
an activity in-house, or does it subcontract the activity to another producer?
A decision to outsource is a decision to
go outside the boundaries of the firm.
Offshoring pertains to the location of
an activity, either at home or abroad.
A firm that subcontracts, say, its call
center, to another firm that sets up the
center in India, would be offshoring
and outsourcing.
Outsourcing has several distinctive features. The types of goods that

is cheaper to offshore another activity.
This can lead to an increase in the volume of activity that is moved offshore.
Thus, there’s a positive feedback.
Once a firm has paid the fixed costs
of moving an activity offshore, say,
to a low-wage country, the firm’s unit
costs of production will be lower and it
will gain sales. But the increased sales
give the firm the incentive to lower its
unit costs in other ways, so the firm
may consider paying the fixed costs
to move another production activity
offshore to achieve further reductions
in unit costs. Also, if transportation
costs are high, firms might move
several parts of the production process
offshore at the same time to economize
on these costs. Thus, the economy can
go from exhibiting a small amount of
offshoring to exhibiting a large amount
in a short period of time. Hence, the
fact that U.S. firms aren’t offshoring
that much production yet does not
imply that they won’t in the future.
Another aspect of offshoring is that it

Business Review Q3 2005 39

is often the largest and most productive firms that find it cost effective to
move production offshore, since these
firms are better able to bear the fixed
costs needed to obtain savings on the
variable costs of production and to
bear the increased cost of monitoring
performance across a longer distance.
The new literature on trade is
drawing on the theory of the firm
to address some of the interesting
questions regarding outsourcing and
offshoring. What accounts for the
increasing fragmentation of the production process? What determines the
form of offshoring? Does it differ by
country? By industry? What characteristics of the firm or its activities help
us understand the organizational mode
it would choose?
With his co-author Elhanan
Helpman of Harvard University,
Grossman has studied some of the
tradeoffs between outsourcing production versus producing in an integrated
firm. On the one hand, specialized
suppliers of inputs can usually produce
more efficiently, especially if they
provide those inputs to more than
one customer. On the other hand,
because not every contingency can be
written into the contract, the supplier
and the final producer may be subject
to potential “hold-up” problems. The
final producer may end up having to
pay more than expected for the inputs.
Or the supplier, after having made
the relationship-specific investments
needed to produce the specialized
input, might find it difficult to get the
final purchasers to share in the cost
of those investments. This creates an
incentive for underinvestment relative to what an integrated firm would
do. Also, the supplier might do less
customization of its input so that it
could sell to other buyers if it has to.
Once the input has been fully customized, it’s harder to sell to any other
buyer, and this puts the input supplier

40 Q3 2005 Business Review

in a weak bargaining position relative
to the buyer. Thus, the theory predicts
that there would be a tendency toward
less firm-specific investment and
customization in industries with more
outsourcing. In industries where these
types of investment and customization
are very important to the production
process, integrated production rather
than outsourcing would predominate.

The new literature
on trade is drawing
on the theory of the
ﬁrm to address some
of the interesting
questions regarding
outsourcing and
offshoring.
Other research suggests that one
mechanism for getting around the
potential underinvestment and undercustomization problem between supplier and final producer is cost-sharing for
the investments. We often see firms
providing their suppliers with specialized equipment or lending them funds
to purchase such equipment or raw
materials. Cost-sharing on the labor
side is much less common. But this
cost-sharing means that the supplier
has more bargaining power in any ex
post renegotiations with the producer.
This hold-up problem will be worse the
more capital-intensive the production
process is. Hence, this theory predicts
that we would see more outsourcing in
industries that are more labor intensive
and less in industries that are capital
intensive – and, analogously, more outsourcing to countries with abundant
labor and less outsourcing to countries
with abundant capital. This seems to
fit reality.

Another feature of countries
that firms would outsource to is what
Grossman calls thick-market externalities. A firm is looking for a producer
to customize its input, so it wants to
find partners with the proper expertise
to make what it wants. This could
differ from what another producer is
looking for. If we think of potential
suppliers arrayed along a spectrum
according to their type of expertise,
finding someone with expertise close
to what the producer is looking for is
important. The denser or thicker the
market of suppliers, the more likely
the producer will find one with the
expertise close to what he is looking
for. There is a positive feedback. If
more U.S. producers outsource business services to India, it will be more
profitable for Indian firms to develop
the expertise to provide those services.
And as more Indian firms enter the
market and develop the expertise, the
easier it will be for a U.S. firm to find
a suitable supplier in India. On the
other hand, if no firms are outsourcing to a particular country, then a
firm might not want to be the first to
outsource there, since it might not find
the expertise it is looking for.
Grossman’s research also suggests
that a country’s legal environment
is an important determinant of the
volume of outsourcing the country
can expect to obtain. An improved
contracting environment, all else
equal, makes the country more attractive to outsourcers. However, all else
is not equal – eventually wages rise as
the contracting environment improves, and this may lead firms to look
elsewhere, especially if the original
motivation for outsourcing was to save
on labor costs.
Another tradeoff when considering outsourcing versus integration
concerns the incentives the firm can
give to managers for good performance. Since an external supplier has

www.philadelphiafed.org

to put up the cost of the inputs and
the labor for producing the inputs, it
typically has more at stake than an
internal manager does, and this would
provide a better incentive for good
performance. On the other hand, it
is probably easier for a firm to monitor
the performance of one of its own internal divisions than an external supplier. These considerations imply that
outsourcing will more likely be chosen
by firms with very high or very low potential productivity and that firms with
intermediate productivity will choose
integration. In addition, for those
firms that remain integrated, offshoring is chosen most often by the more
productive of these firms. A look at
the data suggests this seems to accord
well with actual experience. However, economists are just beginning to
empirically test the theories explaining
firms’ choices of outsourcing versus
integration and home versus offshore
production. According to Grossman,
this empirical work shows promise, and
it, along with new theoretical models,
is helping us understand which types of
firms in an industry are the ones that
go offshore or engage in outsourcing,
which types of industries are prone to
these types of trade relationships, and
in which types of countries we should
expect to see one form of production
versus another.
EXCHANGE RATES
The Policy Forum’s next session
looked at implications of exchange
rate policies and trade deficits on
the macroeconomy. Jeremy Siegel
of the Wharton School, University
of Pennsylvania, began the session
emphasizing the demographic component of structural trade deficits or
surpluses across countries, which in
his view is often neglected. Over the
past 50 years, life expectancy has risen
and retirement age has fallen. In 1950
in the U.S., the difference between

www.philadelphiafed.org

the two was only 1.6 years
and today it is 14.4 years
– a large change. However, these trends cannot
continue. In 1950 in the
U.S., the number of workers
per retiree was seven to one.
Now it is five to one, but it
is slated to decline to two
and a half to one by 2050.
And other countries, including Japan, Italy, Spain,
and Greece, are aging more
quickly than the U.S. In
Japan, the number of workers per retiree approaches
one to one by 2050, which
means the workers have to produce
not only for themselves but also
transfer goods to the retirees. These
trends imply that retirement age has

Jeremy Siegel

people have to work longer because
they are living longer – the retirement
age has to increase almost twice as fast
as projected life expectancy. Things
are worse if life expectancy rises more

Economists are just beginning to empirically
test the theories explaining ﬁrms’ choices
of outsourcing versus integration and home
versus offshore production.
to increase. To investigate the effects
of these demographic trends, Siegel
has built an economic model to study
who in the world is going to produce
the goods and who is going to buy the
assets in the economy. In the model,
income grows at the rate of productivity growth until a person retires and
then it is zero, and consumption grows
at the rate of productivity growth
until a person retires and then it is
flat. The outcome of the model is the
equilibrium retirement age, assuming
that Social Security taxes are fixed.
The model suggests that by 2050
the retirement age in the U.S. has to
increase to 73, which implies that the
difference between life expectancy and
retirement age narrows to 9.2 years.
As Siegel points out, it isn’t merely that

than the conservative estimates Siegel
uses in his model simulation.
What can help solve this “age
wave” problem? Faster productivity
growth can help the situation, but only
modestly. That’s because when productivity growth accelerates, wages go
up, and when wages go up, retirement
benefits go up. So there’s not much
help there. Immigration might help.
But a half billion immigrants into the
U.S. over the next 45 years would be
needed to keep the retirement age in
the mid 60s; that number is far higher
than the current U.S. population of
294 million.
Siegel says the hope comes from
the developing world, where 85 percent
of the world’s population lives and
where the population is much younger

Business Review Q3 2005 41

than the developed countries’. The
developing world’s age profile is about
50 years behind that of the developed
world – for example, the distribution
of population by age group in India
today looks like that of Japan or the
U.S. in 1950. The number of workers
per retiree is projected to decline in
India but only to four to one by 2050.
According to Siegel’s model, if the
developing world can grow at 6 percent
per year into the future, which is optimistic but not overly so given current
experience, then the retirement age in
the U.S. and other developed countries
can stay roughly where it is today. If
growth in the developing world is less,
then retirement age in the U.S. and
other developed countries will have
to rise. But assuming that growth
in the developing world is 6 percent,
then it is the developing countries
that produce the goods and buy the
developed world’s assets. Today, the
developing countries own less than
10 percent of the world’s capital, but
the model simulations suggest that by
2050, they will own most of the world’s
capital and they will be producing
most of its goods. The model implies
that the developing countries will be
running large trade surpluses, while
the developed countries will be running increasingly large trade deficits.
Because most of their populations will
be retired, the developed countries will
need to import goods for consumption, and they will sell off the assets
they have been accumulating for many
decades. These trade flows come out
of the demographics; they are not
structural imbalances.
What are the implications for
exchange rates? In Siegel’s model, the
trade deficits in the U.S. and the other
countries of the developed world are
sustainable at current exchange rates
– they are driven by the demographics.
Thus, even though the U.S. trade deficits are very large, they do not cause a

42 Q3 2005 Business Review

depreciation of the dollar. As long as
foreigners want to acquire U.S. assets
and Americans want to acquire foreign
goods, the trade deficits won’t put
pressure on the dollar exchange rate.
Given this, Siegel suggests that when
we are trying to determine whether a
particular trade deficit is sustainable,
we shouldn’t use a zero deficit as the

tinues at about 5 percent and the current account deficit remains at about 6
percent of GDP, net external liabilities
as a percent of nominal GDP would
rise to 120 percent of GDP, double
what any industrial country has been
able to achieve and sustain. While it’s
possible the U.S. could sustain such
a high level, he thinks it is unlikely.3

Because most of their populations will be
retired, the developed countries will need
to import goods for consumption, and
they will sell off the assets they have been
accumulating for many decades.
basis of comparison but the structural
deficit that will obtain in the long run
because of large differences in demographics across countries.
Michael Mussa of the Institute
for International Economics followed
Siegel with an opposing view of the
sustainability of the U.S. current
account deficit and the path of the
exchange value of the dollar. Acknowledging that a wide range of
outcomes for both exchange rates and
the deficit have been observed in the
past, Mussa made a case for why, in his
view, the dollar remained overvalued.
In his view, it is difficult for the U.S. to
borrow against many of its assets on a
world market. For example, borrowing
against our domestic human capital is
not really feasible. In the U.S., U.S.owned assets abroad used to exceed
foreign-owned assets in the U.S. by
about 25 percent of GDP. Now, it is
the opposite – foreign-owned assets
in the U.S. exceed U.S.-owned assets
abroad by about $2.5 trillion, or 25
percent of GDP. Mussa points out that
no industrial country has ever seen
that ratio go above about 60 percent of
GDP, but if nominal GDP growth con-

Sustaining a current account deficit of
2 to 3 percent of GDP over the next
decade or longer would be feasible in
Mussa’s view, since there are many
reasons that foreigners want to invest
in the U.S.
What’s needed to bring the
current account deficit down from 6
percent of GDP to what, in his view, is
a more sustainable level? Mussa cites
two things: first, a switch in the pattern of world demand toward purchases of U.S.-produced goods and services
and therefore away from rest-of-world
goods and services; second, an adjustment in the level of spending relative
to income both in the U.S. and abroad.
For the U.S. that means reducing our
spending; for the rest of the world that
means increasing their demand relative
to their income.
Mussa says that both the private
sector and the government sector in
the U.S. will need to change their
behavior to effect these changes. The

In contrast, Jeremy Siegel said in the session’s
question and answer period that he thinks it is
quite likely that the U.S. will break the historical
maximum of 60 percent of GDP
.
3

www.philadelphiafed.org

years would be a welcome
development.
Mussa concluded
with his perspective on
whether a strong dollar is
good or bad for the U.S.
When the dollar is strong,
the U.S. gets paid high
prices for the goods and
services it produces and
sells abroad, and it pays
relatively low prices for
the goods and services it
purchases from the rest
Gertrude Tumpel-Gugerell
of the world. All else
equal, that is a good thing.
But, again, all else is not
private sector needs to save more; the
necessarily equal. If the value of the
government needs to put its Social Sedollar is so high that demand for U.S.
curity and Medicare budgets in order.
goods and services by the rest of the
If the dollar depreciates, it will help
world falls so that the U.S. doesn’t
reduce the drag from lower governearn enough on what it sells abroad
ment spending by improving the U.S.’s
to afford what it buys from the rest of
net export position. If public-sector
expenditures are not controlled, then if
the dollar depreciates substantially, in
Mussa’s view, the Fed will need to raise
interest rates to curb overly expansionary effects of higher net exports on
U.S. economic growth.
Mussa believes the rest of the
world faces a more difficult situation
than the U.S., since they need to get
demand up. In Europe, he looks to
the world, the U.S. will have to borrow
the European Central Bank to use
more to finance the gap. And we may
monetary policy as much as it can.
want the value of the dollar to fall in
In Japan, Mussa thinks there is not a
order to restore equilibrium. As Mussa
good deal more that monetary policy
puts it, the goal is to have the strongest
can do in the short run. The developdollar consistent with maintaining a
ing countries of Asia, which have been
sustainable equilibrium position in our
resisting an exchange rate correction,
external payments position over time.
will need to allow that to happen. In
In Mussa’s view, that’s a dollar that
Mussa’s view these countries’ massive
is a fair bit weaker than seen in 2001
interventions to buy dollars in order to
through early 2002, and significantly
keep their currencies from appreciating
weaker than the current value on a
must slow down. Their purchases of
trade-weighted basis – perhaps not
U.S. Treasury securities as investment
much weaker against the currencies of
need not stop, but Mussa feels that
most of the other industrial countries
$100 billion or $200 billion fewer purbut much weaker against the currenchases per year over the next couple of

cies of a number of emerging market
economies.
Gertrude Tumpel-Gugerell, a
member of the Executive Board of
the European Central Bank, brought
an international perspective to the
discussion. She discussed the question
of whether large swings in exchange
rates matter for the real economy and
what the appropriate monetary policy
response to exchange rate swings is.
The consensus of both academic
economists and policymakers is that
exchange rate movements are difficult
to predict and that random walk models generally predict as well as standard
macroeconomic models. Since exchange rates are asset prices, they are
strongly influenced by expectations,
which are difficult to measure and to
include in formal models. TumpelGugerell points out the irony, then,
in the many calls for policy responses

The consensus of both academic economists
and policymakers is that exchange rate
movements are difﬁcult to predict and that
random walk models generally predict as well
as standard macroeconomic models.

www.philadelphiafed.org

every time the value of the currency
moves markedly. She views this as a
sign that exchange rate movements are
seen as important, despite all the difficulties in understanding them.
The international monetary
system has generally evolved toward
greater exchange rate flexibility
between major currency pairs. She
distinguishes several major phases in
which this evolution has occurred.
The period of the gold standard was
one of fixed exchange rates and convertibility of currencies into gold. It
ended with the advent of World War I.
During the interwar periods, counBusiness Review Q3 2005 43

tries went progressively back to the
gold standard with the goal of restoring fiscal discipline. But this broke
down again after the Great Depression
convinced many that the system was
faulty, and there was a brief period of
flexible exchange rates. The Bretton
Woods Agreement in 1944 led to all
the major industrialized countries pegging their currencies to the U.S. dollar.
This lasted until 1971. Tumpel-Gugerell calls the 1970s the trial-and-error
system, which led to more flexibility in
the 1980s and the 1990s, and the creation of the single European currency.
She characterizes the current system as
one of flexible exchange rates among
major currencies accompanied by
international cooperation.
Within the current framework,
how much do exchange rate movements matter for the economy? Tumpel-Gugerell distinguishes between
effects taking place through the
price-competitiveness channel and
those associated with market uncertainty. Regarding the former channel,
theory suggests that exchange rate
movements will have less of an effect
on closed economies than on small
open economies. Research suggests
that a persistent exchange rate movement can have a significant effect on
prices and GDP in the euro area, but
that the effect usually is seen with a
lag. For example, firms can squeeze
their profit margins or can attempt to
hedge against adverse exchange rate
movements, thus delaying the effect
of a persistent move. The European
integration process, including the
introduction of the euro, has reduced
instability generated by shocks to the
exchange rate. In Tumpel-Gugerell’s
view this reduction in volatility should
help to boost trade across the countries
in Europe.
But according to Tumpel-Gugerell, the main way to limit undesirable
exchange rate instability is for poli-

44 Q3 2005 Business Review

cymakers to focus on achieving and
maintaining sound macroeconomic
fundamentals. She believes that if
monetary policymakers are committed
to price stability, this will lead to exchange rate stability over the long run.

Kehoe focused his talk on the lessons
we’ve learned over the past 25 years
from the economic integration that
has taken place – from our empirical
experience with integration and from
economic models, from where the

While opening up to free trade and investment
may be an important ingredient for generating
economic growth, it is not sufﬁcient.
Sound and sustainable fiscal policy will
also play a role in achieving economic
balance among the world’s economies.
FREE TRADE
I had the pleasure of moderating our final session, which looked at
free trade. For economists, free trade
is not very controversial – it offers
participants the benefit of an improved
standard of living. But the recent
negative discourse in the popular press
has led to a more nuanced discussion
of the benefits – more documentation
of those benefits – as well as discussion
of the dislocations and other costs of
the transition to free trade.
Timothy Kehoe of the University of Minnesota discussed how free
trade agreements have affected trade
and capital flows across
countries. There has
been an expansion of
regional trade agreements in both Europe
and the Americas. The
U.S. signed a trade
agreement with Chile
in 2003, negotiated the
Central American Free
Trade Agreement with
a number of countries,
and has been in negotiations with several South
American countries.
The European Union has
been expanding as well.

economic models have worked well in
predicting the effects of integration,
and from where they have failed and
need improvement.
Kehoe’s first lesson is that while
opening up to free trade and investment may be an important ingredient
for generating economic growth, it is
not sufficient. The Mexican Apertura,
or opening up of the country, which
began in the late 1980s and led to the
North America Free Trade Agreement (NAFTA), had a large impact on
Mexico, generating large increases in
foreign trade and investment. Mexico
now exports almost twice as much as
the rest of Latin America combined.
But while it generated significant
growth in exports, it did not generate
much overall economic growth – at

Timothy Kehoe

www.philadelphiafed.org

least not until after the 1994-1995
crisis there.
Lesson two is that a free trade
area such as NAFTA or the European Union is neither necessary nor
sufficient for generating foreign trade
and foreign investment. Chile has just
negotiated a free trade agreement with
the U.S.; yet after its economic crisis in
1981-1982, its exports surged and are
now about 25 percent of its GDP. Its
GDP growth also accelerated sharply,
and the increase was not only export
driven. In contrast, Greece joined
the European Economic Community
in 1980, yet its exports as a percent of
GDP are still under 10 percent, and
foreign investment in Greece is also
very small.
Lesson three is that to get foreign
investment, domestic institutions such
as banks are important; protections of
investors’ rights are important; property rights – like bankruptcy laws – are
important. Although the Mexican
banking system was opened to foreign
participation in 1995, it still is not
functioning well in financing private
investment, which is still low compared to other countries like Chile.
Thus, signing a free trade agreement is
not a guarantee of direct investment.
Kehoe’s fourth lesson reiterates
Siegel’s point, namely, that demographic differences can be important
determinants of international capital
flows. Mexico’s baby boom was much
stronger than the U.S.’s, and Mexico
today has many young people. The
median age in Mexico is 20 compared
to 34 in the U.S. Similarly, the other
countries in Latin America are young.
In contrast, the European integration is between rich, old, and aging
countries and poor, old, and aging
countries. These demographics will
affect both trade and capital flows
across countries.
But Kehoe’s fifth lesson is that
capital flows may be substitutes rather

www.philadelphiafed.org

than complements of trade flows.
When we look at the U.S., we see that
the volume of trade flows between the
U.S. and our NAFTA partners is much
higher than between the U.S. and the
European Union, while the volume of
investment flows is much higher with

are accompanied by trade deficits and
depreciations of the real exchange
rate. These inflows eventually stop,
the trade deficit becomes a surplus,
and the currency appreciates. Kehoe
points out that this happened after
Spain joined the European Com-

Despite the dislocations and reallocations
that have to be borne, the steady march of
technology and economic adjustments have
allowed us to reap higher per capita income
across the decades.
Europe. Kehoe posits that this might
be because the U.S. is afraid of further
trade restrictions and protectionism if
trade volumes increase in Europe.
Lesson six is that applied general equilibrium economic models of
NAFTA’s impact did a poor job of
capturing the very significant increase
in trade volumes in North America,
and they did a poor job of identifying
the sectors in which trade increased.
For example, if we compare one of
the best model’s predictions of U.S.
exports to Mexico in different industry
sectors over 1988-1999 to the actual
data on exports, we find a correlation
of less than 1 percent. One reason the
models performed poorly is that they
were unable to capture a fact shown
in Kehoe’s research: that much of the
expansion of trade took place in sectors where there was little or no trade
before trade liberalization. Models
that focus on the exchange rate will
not capture this new-goods effect; it
happens with changes in trade policy.
Lesson seven is that dynamic
applied general equilibrium models
can do a good job capturing the path
of capital flows when a country opens
itself up to foreign investment. Flows
of capital into a relatively poor country
that opens itself to foreign investment

munity in 1986 and capital started
flowing into the country. In 1992,
the process reversed, and while the
Spanish government was caught off
guard and called the outflow of capital
a crisis, this is exactly what the model
predicted would happen.
Kehoe’s eighth and final lesson
is that signing a free trade agreement
does not always mean an increase in
free trade. It depends on the level of
trade barriers and tariffs the country
operated under to begin with. In
Kehoe’s view, Ecuador’s signing a
free trade agreement with the U.S. is
a large step toward free trade, since
there’s a high level of tariffs and trade
barriers there. For Latvia and Slovenia, joining the European Union will
give them access to European markets,
but it will increase the level of tariffs
under which they currently operate
and so will be a step away from free
trade. Kehoe predicts they will find it
difficult to import to non-EU countries.
Douglas Irwin of Dartmouth College elaborated on the evolving debate
over free trade. He pointed out that
the first debates over U.S. trade policy
took place when the new Congress
met at Congress Hall, just a few steps
away from the Philadelphia Fed. James
Business Review Q3 2005 45

Madison of Virginia introduced the
first tariff bill on the floor of the U.S.
House when the first Congress met in
April 1789. It passed in July, but only
after a lively debate. Indeed, trade
policy has always been a controversial
aspect of U.S. economic policy. Perhaps the main reason is that trade is
associated with economic change and
it affects the distribution of income
within the country. This means that
trade is likely to always elicit various
opinions. Irwin points out that the
same arguments against trade tend to
recur time and time again and that the
current complaints that the U.S. can’t
compete because of low wages abroad,
that foreign countries are unfair traders, and that trade will damage the
economy have all been heard before.
Nonetheless, the debate on trade has
shifted over time. In the 1970s one of
the issues was that multinationals were
draining America of capital, investing in foreign countries rather than
at home. In the 1980s, the debate
focused on Japan and its high-tech
development. In the 1990s, NAFTA
was the issue. Currently, outsourcing to China and India has moved to
the forefront. Irwin’s study of history
suggests that these issues will pass and,
by 2010, a new country or issue will
emerge as the focus of the debate.
When economists are asked if
trade is good for the U.S. economy,
the answer is yes. Despite the dislocations and reallocations that have to be
borne, the steady march of technology and economic adjustments have
allowed us to reap higher per capita
income across the decades. Irwin acknowledges that going through the adjustments can be painful, but stopping
the dislocation and economic change
would create many more problems.
And even though fear of trade has
been constant through our history, the
U.S. has consistently over the past 30
or 40 years pursued an agenda of open-

46 Q3 2005 Business Review

ing up markets and keeping the U.S.
market open. Irwin points out that the
U.S. has done this in two ways. It has
negotiated with foreign trade partners
in the context of the World Trade
Organization, and it has negotiated a
number of regional and bilateral trade
agreements. There is some debate
among economists about whether the
bilateral agreements are better or worse
than multilateral negotiations, but
both are proceeding with increased
momentum.
This raises the question: if there
is so much fear of globalization, why
is it proceeding apace and why have
markets remained open? Irwin points

Canadian softwood lumber and steel.
The users of Canadian wood have
made it much more difficult for the
U.S. government to give protection to
domestic producers. Steel consumers
put pressure on the Bush administration against steel tariffs. A third factor
that has worked against protectionism
is the macroeconomic stability the
U.S. has enjoyed over most of the postwar period. Economic growth helps
ameliorate the pain associated with the
economic dislocations that accompany
increased trade and the opening of
markets.
Irwin pointed to an example
that illustrates that protectionism is

Recent research has shown that countries that
are more open to trade have higher ratios of
private-sector credit to GDP.
to three factors that help explain why
there hasn’t been a great backlash
against globalization. First, domestic
industries that compete with our imports, such as shoes and apparel, have
been losing their political importance.
They have shrunk in size or, in some
cases, have been totally wiped out. For
example, in the mid 1960s we imported a third to a half of shoes consumed
in the U.S.; now we import over 95
percent. Also, a number of industries
that faced foreign competition, such
as semiconductors and automobiles,
have gone global. In the past, they
have argued for trade protection. Now,
they’ve undertaken foreign investments, have diversified their production across many countries, and import
many goods themselves. A second
factor is that many U.S. imports are
intermediate goods. Their consumers
are businesses, not households, and
they are dependent on getting these
imports to carry out their own production. Irwin points to two examples:

increasingly being viewed as a poor
policy option. The state of Indiana has
considered legislation to ban state contracts going to firms that outsourced
to other countries. Not outsourcing
the processing of state unemployment
claims would cost the taxpayers of Indiana $16 million that could otherwise
be spent on public works such as roads
or schools, tax cuts, or servicing the
debt. This cost has been publicized,
and this, plus the fact that these jobs
are not currently in Indiana anyway,
has led many to question the proposed
legislation.
In Irwin’s view it will be difficult
for trade opponents to move the U.S.
away from its current very low tariff
position and its open market. Irwin
ended his presentation saying he believes there will always be critics of free
trade and they will need to be rebutted
by those who have a stake in and support the system of open world trade.
In recent testimony before the Finance Committee of the U.S. Senate,

www.philadelphiafed.org

Chairman Greenspan expressed the
view that it is essential that we not put
“our future at risk with a step back into
protectionism.”4
How can we ensure that
U.S. markets remain open? Or, as
Raghuram Rajan, economic counselor
and director of research at the International Monetary Fund asks, how can
we build constituencies for free trade?
First, as was pointed out earlier in the
day by Kehoe, it’s important to have
well-functioning institutions and welldefined property rights to realize the
benefits of free trade. Those benefits
include stronger economic growth.
But also, over the 20th century,
countries that have become more open
for trade have tended to have better
developed financial markets, which in
itself helps to foster growth. This is an
example of a positive feedback – better
institutions allow the benefits of free
trade and free trade allows development of better institutions.
Why do such correlations exist
between openness and financial development? One possibility is that free
trade strengthens the domestic constituencies for financial sector reform. For
example, industries that want to begin
trading more will need to finance
that trade and will exert pressure on
financial markets to develop to meet
their needs. Or industries that feel
the competition from foreigners could
push for improved financial markets
to aid them in remaining competitive. Recent research has shown that
countries that are more open to trade
have higher ratios of private-sector
credit to GDP, and that seems to come
about because the constituencies that
are pro-finance become more powerful
after trade liberalization.
See Testimony of Chairman Alan Greenspan
on China, before the Committee on Finance,
U.S. Senate, June 23, 2005, www.federalreserve.
gov/boarddocs/testimony/2005/20050623/
default.htm.
4

www.philadelphiafed.org

But Rajan argues that the direction of causality may run the other way
as well. The development of financial
markets may increase the power of
constituencies in favor of free trade
relative to those opposed. Trade liberalization creates winners and losers;
it does not make everyone uniformly
better off. So to understand how
constituencies in favor of free trade
are developed, one must identify the
winners and losers. Economic theory
suggests that those who have the

The development of
ﬁnancial markets may
increase the power of
constituencies in favor
of free trade relative
to those opposed.
endowments in which the country is
rich will be more pro-trade, since those
who are relatively higher endowed will
benefit from trade. For example, the
U.S has more highly educated people
than other countries. So opening up
U.S. markets to trade will tend to benefit these people, since the U.S. is the
country that can supply this type of
worker. Thus, they are the ones that
are pro-trade in the U.S. The low-skill
workers in the U.S. will be hurt by free
trade, since other countries can supply
low-skill workers. Thus, in the U.S.,
the low-skill workers will tend to be
against free trade. In poorer countries,
where low-skill workers predominate,
the more highly educated tend to be
against free trade.
But if free trade is beneficial
overall, why can’t the winners compensate the losers? Rajan conjectures
it is because many of the required side
payments would need to be enormous,
and they would have to take place over

such a long period of time that they
would be hard to commit to.
If this is the case, then how does a
country go about changing the political balance in favor of free trade? Rajan sees three broad possibilities. The
first is through committing to external
agreements like those of the World
Trade Organization or setting a date
far in the future when the trade and
capital markets will open up, for example, the United Kingdom’s big bang.
The second possibility is through a
crisis, as happened in India. The crisis
exposes the fact that the country’s
policy of closed markets creates very
bad outcomes, or the crisis reduces the
relative political power of the status
quo, who are against open markets.
The third possibility is through building constituencies. In developing
countries this entails showing them
that there is more opportunity. The
more trade that is occurring outside
a country’s borders, the more its own
firms want to partake. Also, when
the rest of the world is enjoying more
flows of goods and capital, there can
be more leakages across a country’s
borders, and the country may find it
more advantageous to open itself up
and control the flows, rather than
have them go on without any control.
Consumers also see the benefits of
free trade in the form of lower prices
and can create a pro-trade constituency. Firms that are more efficient
are less likely to fear the increased
competitiveness that comes from opening up markets. Hence, increasing
entry into their industry can create
more efficient firms that then emerge
as a free-trade constituency. Similarly,
individuals may fear free trade because
they don’t have access to education
or the resources that will enable them
to handle the changes that free trade
will bring. Creating a safety net for
these individuals will help shift their
opinions regarding free trade.
Business Review Q3 2005 47

Rajan concluded his presentation with some data from the World
Value Survey, a survey of over 150,000
individuals in 66 countries between
1981 and 2000, which shows that
preferences for competition, a proxy for
free trade, do vary with factors such
as education, income, age, and type of
occupation. It turns out that younger
people are more against competition
than older people. This might reflect
the fact that younger people tend to
be producers and fear the job loss and
older people tend to be consumers and
value the lower cost of goods. Those
with higher wealth, higher social status, and higher education tend to favor
competition. Unskilled workers are
more against competition than moderately or higher skilled workers. An interesting finding is that small business
owners’ attitudes toward competition
are influenced by their access to credit,
while managers’ and employees’ attitudes are not. Small-business owners

48 Q3 2005 Business Review

in countries with strong credit markets
are much more likely to be pro-competition than those in countries with
weak credit markets. That is, if they
have access to resources and feel they
can get the resources to run their businesses, they favor competition. This
is evidence that institutions matter
and that financial development and
well-functioning institutions that allow
access to resources can foster freer
trade – the reverse causality mentioned
earlier. It also suggests that a country
that finds itself with dysfunctional
institutions might find it very hard
to build support for changing those
institutions and to build a constituency
for free trade. Can an institution like
the International Monetary Fund help?
In Rajan’s view the answer is yes, but
only at the margin. There needs to be
momentum within the country itself
for change. Large, developed economies can help develop that internal
momentum by helping to ensure that

trade spreads to the poor, developing
countries. Freer trade offers outside
opportunities to the people in those
countries, who can then develop into
a constituency within the country in
favor of even more openness and freer
trade.
SUMMARY
The 2004 Policy Forum generated
lively discussion among the program
speakers and audience participants on
a number of the challenges and opportunities brought by an increasingly
global economy. Our hope is that the
ideas raised will spur further research
and foster a greater understanding of
today’s economy.
We will hold our fifth annual
Philadelphia Fed Policy Forum, “Fiscal
Imbalance: Problems, Solutions, and
Implications,” on Friday, December 2,
2005. You will find the agenda on page
35. BR

Full text of Business Review (Federal Reserve Bank of Philadelphia) : Third Quarter 2005

FRASER