View original document

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

Call for Papers
2008 Conference on
Bank Structure a«(d
Competition

Federal Reserve Ban
of Chicago

Fourth Quarter 2007

RESEARCH LIBRARY
Federal Reserve Bank
of St. Louis

DEC I 2 2007

perspectives
2

Understanding the evolution of trade deficits:
Trade elasticities of industrialized countries
Leland Crane, Meredith A. Crowley, and Saad Quayyum

18

Evidence on entrepreneurs in the United States:
Data from the 1989-2004 Survey of Consumer Finances
Mariacristina De Nardi, Phil Doctor, and Spencer D. Krane

37

A bank by any other name ...
Christian Johnson and George G. Kaufman

52

Index for 2007

Economic

perspectives

President
Charles L. Evans
Senior Vice President and Director of Research
Daniel G. Sullivan
Research Department
Financial Studies
Douglas Evanoff, Vice President

Macroeconomic Policy Research
Jonas Fisher, Economic Advisor and Team Leader

Microeconomic Policy
Daniel Aaronson, Economic Advisor and Team Leader
Payment Studies
Richard Porter, Vice President
Regional Programs
William A. Testa, Vice President
Economics Editor
Anna L. Paulson, Senior Financial Economist

Editor
Helen O'D. Koshy

Associate Editors
Kathryn Moran
Han Y. Choi
Graphics
Rita Molloy
Production
Julia Baker
Economic Perspectives is published by the Research
Department of the Federal Reserve Bank of Chicago. The
views expressed are the authors’ and do not necessarily
reflect the views of the Federal Reserve Bank of Chicago
or the Federal Reserve System.

© 2007 Federal Reserve Bank of Chicago
Economic Perspectives articles may be reproduced in
whole or in part, provided the articles are not reproduced
or distributed for commercial gain and provided the source
is appropriately credited. Prior written permission must
be obtained for any other reproduction, distribution,
republication, or creation of derivative works of Economic
Perspectives articles. To request permission, please
contact Helen Koshy, senior editor, at 312-322-5830 or email
Helen.Koshy@chi.frb.org.
Economic Perspectives and other Bank
publications are available on the World Wide Web
at www.chicagofed.org.

fit chicagofed. org
ISSN 0164-0682

Contents
Fourth Quarter 2007, Volume XXXI, Issue 4

2

Understanding the evolution of trade deficits:
Trade elasticities of industrialized countries
Leland Crane, Meredith A. Crowley, and Saad Quayyum

In this article, the authors present updated trade elasticities—measures of how much imports and
exports change in response to income and price changes—for the U.S. and six other industrialized
countries, collectively known as the Group of Seven. They find that the imports and exports of
these countries are slightly more responsive to changes in a country’s total income over a period
that ends in 2006, compared with a period that ends in 1994.

18

Evidence on entrepreneurs in the United States:
Data from the 1989-2004 Survey of Consumer Finances
Mariacristina De Nardi, Phil Doctor, and Spencer D. Krane
Using data from the Federal Reserve Board’s Survey of Consumer Finances, the authors examine
characteristics of entrepreneurs and the businesses they run. Their analysis confirms that business
owners are important sources of saving and wealth creation in the U.S. and that they are less risk
averse than other wealthy households. This discounts the notion that the wealth of entrepreneurs
disproportionately reflects a buildup of precautionary balances to guard against financial risk.

37

A bank by any other name ...
Christian Johnson and George G. Kaufman

“Banks” are regulated by the government. However, because the generic term bank applies
to a number of different types of financial institutions that provide different services, different
types of regulation are required. This issue has attracted much attention in recent years as
nonfinancial firms—including Wal-Mart—attempted to obtain a “bank.” This article traces
changes in the definition of the term commercial bank for the purposes of regulation and
discusses the implications for one type of “bank”: industrial loan companies.

Conference on Bank Structure and Competition: Call for papers

52

Index for 2007

Understanding the evolution of trade deficits:
Trade elasticities of industrialized countries
Leland Crane, Meredith A. Crowley, and Saad Quayyum

Introduction and summary
In 2006, Americans bought $1,928 billion of goods and
services produced in foreign countries. In the same year,
American sales of goods and services to foreigners
amounted to only $1,304 billion.1 The difference between American exports and imports of $624 billion
is known as the trade deficit. When this number is
positive—that is, when the U.S. sells more to foreign
countries than it buys—it is called a trade surplus. More
generally, the difference between the value of a country’s exports and imports is known as the trade balance.
Over the past several years, the U.S. trade deficit has
attracted a great deal of attention in the popular press
and among policymakers in Washington, DC. Why is
there so much interest in the trade deficit?
To place things in a historical context, the U.S.’s
trade balance has been negative—that is, the U.S. has
been importing more than it has been exporting—for
much of the post-World War II period. Over the past
25 years, the trade deficit has tended to become a larger
and larger fraction of the total output of the U.S. economy, or U.S. gross domestic product (GDP). In the first
quarter of 2005, the U.S. trade deficit peaked at a postWorld War II high of 5.7 percent of GDP. Most recently, in response to a weakening U.S. dollar, the trade
deficit has shrunk a bit, standing at 5.2 percent of GDP
in the first quarter of 2007. However, the U.S.’s trade
deficit of more than 5 percent of GDP is large both by
historical standards and in comparison with other industrialized countries’.
Why does the size of the U.S. trade deficit matter?
Whenever the U.S. trade balance is in deficit, this means
that the U.S. is borrowing from foreigners to finance
its consumption of imports. Borrowing in and of itself
is not necessarily a bad thing. In fact, one way to view
the trade deficit is to recognize that foreigners are choosing to invest in the United States. One can imagine
that they choose to do so because they think that the



returns they will earn on an investment here are better
than the returns they will earn in other countries.
However, debts need to be repaid, and the large
and persistent U.S. trade deficit means that the U.S.
has been borrowing more and more from foreigners
for a long time. This raises the concern that at some
point in the future, foreigners may begin to doubt the
ability of the U.S. to repay its debts and will cease lending. Or, more likely, that the cost of borrowing from
foreigners will increase significantly. Trouble could
also arise if the dollar experiences a “hard landing,”
a sudden large fall in its value relative to other currencies. A large fall in the value of the dollar would
eventually result in much higher prices for imports,
which could leave U.S. consumers much worse off
than they otherwise would be. So the size of the trade
deficit matters because it could be a contributing factor to a sudden change in the value of the dollar. In turn,
an abrupt change in the value of the dollar could, by
raising the prices of imported goods and services, reduce the consumption of Americans (see box 1).
Given the importance of the trade balance, economists want to understand how it will evolve in the
future. Two important factors in determining how much
Americans import from other countries are U.S. national income and the relative price of imported goods
to domestically produced ones. Similarly, important
determinants of U.S. exports are the income of our
trading partners and the relative price of U.S. exports
Leland Crane is an associate economist and Meredith A.
Crowley is a senior economist in the Economic Research
Department at the Federal Reserve Bank of Chicago.
Saad Quayyum is a graduate student in economics at the
University of Wisconsin–Madison. The authors thank
Jaime Marquez, Craig Furfine, Yukako Ono, and seminar
participants at the Federal Reserve Bank of Chicago for
valuable comments and insights. Bertrand Pluyaud generously shared his data on French imports and exports.

4Q/2007, Economic Perspectives

BOX 1

The trade balance, the current account balance, and the macroeconomy
A second perspective on the current account deficit focuses on its international financial foundations.
The current account measures international transactions as changes in current trade flows, income payments, and unilateral transfers. It is roughly equivalent
to the financial account, which measures international
transactions arising from changes in the stocks of real
and financial assets, including foreign direct investment,
as well as the public and private holdings of stocks,
bonds, bank accounts, and currency. Thus, the financial account balance is equal to the difference between
foreign spending for U.S. assets (U.S. capital inflows)
and U.S. spending on foreign assets (U.S. capital outflows). The financial account surplus is loosely understood to reflect U.S. net borrowing from the rest
of the world. That is, when U.S. expenditures exceed
U.S. income, the shortfall is made up in a net inflow
of capital to the U.S. or, equivalently, the purchase
of U.S. assets by foreigners. One interpretation of the
large U.S. current account in recent years is that it has
been driven by foreign investors’ strong desire for
relatively secure U.S. assets (Bernanke, 2005).
A final perspective on the current account deficit
is one that emphasizes the international trade flows
that underlie it. The current account is equal to the
sum of the trade balance (exports less imports), net
foreign income, and unilateral transfers. Net foreign
income is the difference between the overseas earnings
of U.S. investors that are sent to the U.S. and the domestic U.S. earnings of foreign investors that are sent
to their countries. Unilateral transfers are payments
by governments or individuals to foreign governments
or residents of foreign countries. This includes foreign
aid and remittances from U.S. resifigure b1
dents to family and friends overseas.
Figure B1 presents the components of
Components of the U.S. current account
the U.S. current account balance as a
percent of gross domestic product
fraction of U.S. GDP from 1980 through
2
2007. The trade balance and current
account move together closely throughout this period. Thus, one way to un0
derstand the evolution of the current
account is to study the evolution of its
–2
largest component, the trade deficit.
The international trade perspective
examines how flows of imports and
–4
exports respond to changes in the
Trade balance of goods and services
national incomes of the importing
Net foreign income
–6
countries and the relative prices of
Unilateral transfers (net)
imported and exported goods and
Current account
services.
–8

The trade balance represents the overwhelming share
of another measure of a country’s external balance—
the current account. To gain some insight into the
macroeconomic significance of the U.S. trade deficit,
we review the three different, yet equivalent, definitions of the current account. Mann (2002) describes
these as three perspectives on the current account.
Ferguson (2005) refers to them as three different lenses
through which to view the current account. Ultimately,
the current account changes in response to the forces
underlying all of its three different definitions. Each
perspective highlights different forces that drive the
current account and can be analyzed empirically by
economists who wish to quantify the importance of
different forces that change its value.
A first perspective on the current account emphasizes its domestic macroeconomic foundations. The
current account balance is defined as the difference
between a nation’s total income and its expenditures
on consumption and investment or, equivalently, as
the difference between the nation’s total (public and
private) saving and its total (public and private) investment. From this perspective, we can see that the
U.S. current account deficit reflects the fact that investment expenditures exceed saving in the U.S. Thus,
many critics of the U.S. current account deficit argue
that Americans should save more. It is true that an
increase in the level of saving, holding the level of
investment constant, will by definition lead to a smaller current account deficit. However, this argument tends
to oversimplify the complexity of a world in which
individuals make consumption and saving decisions
in response to relative prices and investment returns.

1980 ’82 ’84

’86 ’88 ’90 ’92

’94 ’96 ’98 2000 ’02 ’04 ’06

Sources: Authors’ calculations based on data from the U.S. Bureau of Economic
Analysis and Haver Analytics.

Federal Reserve Bank of Chicago



to the price of other goods that are available to consumers in foreign countries. In order to predict how
exports or imports will change in the future, economists
estimate trade elasticities. Trade elasticities measure
how much a country’s imports or exports will change
in response to changes in national incomes or the relative price of imported goods and services to domestically produced ones. For example, the import elasticity
with respect to income is a number that specifies how
much imports will increase in response to a 1 percent
increase in the total income of a country.
In this article, we present updated trade elasticities
for the United States and six other industrialized
economies—Canada, France, Germany, Italy, Japan,
and the United Kingdom. These countries are collectively known as the Group of Seven (G-7) industrialized
countries. We find that the imports of the G-7 countries
are slightly more responsive to changes in a country’s
total income over a period that ends in 2006 compared
with a sample period that ends in 1994.2 Similarly,
the exports of the G-7 countries appear to be as responsive or more responsive to changes in the tradeweighted average income of the country’s trading
partners over the period 1981–2006 compared with
the period 1981–94. With respect to prices, we find
that the imports of several G-7 countries are more
responsive to import price changes than other studies
have indicated. Our estimates of the responsiveness
of G-7 exports to price changes for most countries
differ from those in previous studies, but not in any
systematic way.
What do our findings imply about the magnitude
of the U.S. trade deficit? We can use the elasticities
estimated in this article to make predictions about how
large or small the U.S. trade deficit would have been
if prices or incomes had been different. For example,
suppose that the relative price of U.S. imports had been
10 percent higher and the relative price of U.S. exports
had been 10 percent lower than they actually were in
2006. Our estimates of the price elasticity of demand
for imports of –0.63 and of the price elasticity of demand for exports of –0.61 imply that the U.S. trade
deficit in 2006 would have been $424 billion instead
of the actual $624 billion.3
In addition to our national level estimates, we
estimate elasticities for different U.S. industry sectors.
We find that the U.S. export elasticity for services with
respect to foreign income exceeds the U.S. import elasticity for services with respect to income over the period
1988–2006. This means that if the U.S. were to grow at
the same rate as its trading partners and prices remained
constant, over time the U.S. trade balance in services
would move toward larger and larger trade surpluses.



In the following section, we discuss what trade
elasticities are. Then, we review previous empirical
studies of trade elasticities. Next, we present our
econometric model and discuss the data we used in
estimating trade elasticities for the G-7 countries.
Finally, we present our results and draw some conclusions from them.
What are trade elasticities?
The theoretical model underlying the estimation
of trade elasticities is an imperfect substitutes model—
that is, a model in which it is assumed that exports
and imports are imperfect substitutes for domestically
produced goods. Goldstein and Khan (1985) provide
a detailed discussion of this model.
In an imperfect substitutes model, the foreign demand for U.S. goods and services is determined by
three main factors: foreign income, the prices of U.S.
goods and services, and the prices of goods and services
that compete with U.S. goods and services in the foreign market. Similarly, U.S. demand for foreign goods
and services is determined by U.S. income, the prices
of foreign goods and services, and the prices of goods
and services that compete with foreign goods and services in the U.S. market.
The income elasticity of demand for imports measures to what extent changes in an importing country’s
income affect changes in its imports. Similarly the income elasticity of demand for exports measures to
what extent changes in foreign countries’ incomes affect the exporting country’s exports.
Theoretically, the import and export elasticities
with respect to income are positive. That is, an increase in a country’s income leads it to buy more from
foreign countries. An income elasticity of imports or
exports that is equal to one implies that imports or exports increase proportionately with income. Deviations
from this imply long-term imbalances in the global
economy. Specifically, an income elasticity for imports
of more than one implies that domestic consumers
have a stronger preference for foreign goods than for
domestic goods. If prices do not adjust, having imports increase more than proportionately to income
growth means that imports would eventually exceed
GDP. Because estimates of the elasticity with respect
to income that are greater than one yield this kind of
implausible prediction, they are hard to reconcile with
a view of long-term balance in the global economy.
Turning to the relationship between prices and
imports, we estimate how imports respond to the
price of imported goods and services relative to the
price of domestically produced goods and services
that compete with imported goods. As imported

4Q/2007, Economic Perspectives

goods become more expensive relative to domestic
goods, economic theory predicts that the volume of
imported goods will fall. In other words, the import
elasticity with respect to the relative price of imports
is negative. A similar relationship holds for prices and
exports. As the price of exported goods increases relative to the price of domestically produced goods in
the importing countries, the volume of exports falls.
Thus, the export elasticity with respect to the relative
price of exports is also negative.
Figure 1 presents the U.S. trade balance from 1955
through 2006 as a fraction of U.S. GDP, along with two
measures of international prices—the relative price of
imports and the relative price of exports. In this figure,
we see that an increase in the relative price of imports
leads to a fall in the volume of goods and services
imported, which, in turn, leads the U.S.’s trade deficit
to shrink. In a similar manner, an increase in the relative price of exports (which often coincides with an
appreciation of a country’s currency) makes exports less
attractive to foreign consumers and, thus, leads to a
fall in the volume of goods and services that a country
exports. This, in turn, leads to an increase in a country’s trade deficit.
From figure 1, we can see that a relationship between relative prices and the trade balance is apparent,
but it is not strikingly clear. The econometric analysis
we conduct in this article will allow us to precisely

quantify the relationship between relative prices and
trade flows.
Previous research on trade elasticities
A highly influential paper by Houthakker and
Magee (1969) estimated the income elasticity of demand for imports and exports with ordinary least
squares for 15 industrialized countries, using annual
data from 1951 through 1966. They identified an important robust empirical relationship that has become
known as the Houthakker–Magee asymmetry. Specifically, the U.S.’s income elasticity with respect to imports was higher than its income elasticity with respect
to exports by a factor of roughly 1.5. See table 1.
Houthakker and Magee’s estimate of the U.S. income
elasticity with respect to imports of 1.51 implies that
for every 1 percent increase in U.S. GDP, Americans
increase their purchases of imports by 1.51 percent.
In contrast, Houthakker and Magee’s estimate of the
U.S.’s income elasticity with respect to exports of
0.99 means that for every 1 percent increase in the
GDP of the U.S.’s trading partners, the U.S.’s exports
increase by only about 1 percent. The asymmetry between these two estimates has important implications
for the U.S. trade deficit.
Johnson (1958) noted that how the trade balance
evolves over time depends crucially on each economy’s
income elasticity of exports and imports. Johnson

figure 1

U.S. trade balance
percent of gross domestic product

price index, 2000 = 100

8

180

6

160
140

4

120

2

100
0
80
–2

60

–4

Trade balance (LHS)
Relative price of exports (RHS)
Relative price of imports (RHS)

–6
–8
1955

40
20
0

’60

’65

’70

’75

’80

’85

’90

’95

2000

’05

Notes: LHS means left-hand scale. RHS means right-hand scale.
Sources: Authors’ calculations using data from the U.S. Bureau of Economic Analysis; International Monetary Fund, International Financial
Statistics database; and Haver Analytics.

Federal Reserve Bank of Chicago



Table 1

Selected estimates of U.S. long-run trade elasticities
	
	
	
	
	
Houthakker and
Magee’s estimates	

Imports	
	
	
Real	
	
	
effective	
	
Relative	 exchange	
Income	 price	
rate	
1.51	

	
	
	
Income	

–0.54		

0.99	

Exports
	
Real
	
effective	
Relative	 exchange	
price	
rate	
–1.51		

	
Methods	

Sample
period

OLS	

1951–66

Hooper, Johnson, and
Marquez’s estimates	
1.79	
–0.31		
0.83	
–1.47		
Johansen 	 1961–94, imports
							
ML	
1976–94, exports
								
Chinn’s estimates	
2.29		
–0.12	
1.62		
–0.73	
Johansen	
1975–2003
							
ML
Cardarelli and
Rebucci’s estimates 	

2.03	

–0.69		

1.85	

0.02		

OLS	

1973–2006

Cardarelli and
Rebucci’s estimates—
correcting for
aggregation bias	

1.68	

–1.45		

1.60	

–0.26		

OLS	

1972–2006

Authors’ estimates	
1.93	
–0.63		
2.34		
–0.61	
							

Johansen	 1960–2006, imports
ML	
1981–2006, exports

Notes: OLS means ordinary least squares. Johansen ML means Johansen’s (1988) maximum likelihood estimator.
Sources: Authors’ calculations; Haver Analytics; Houthakker and Magee (1969); Hooper, Johnson, and Marquez (2000); Chinn (2004); and
Cardarelli and Rebucci (2007).

showed that if trade between two countries is initially
balanced, relative prices are constant, and income growth
is constant in both economies, and if an economy’s
income elasticity of demand for imports is not the same
as the foreign income elasticity of demand for its exports, then the trade balance in each economy can
change over time. Johnson’s model and the Houthakker–
Magee asymmetry together imply that if the U.S. and
the rest of the world grew at the same rate, the U.S.
trade deficit would widen over time if relative prices
remained constant.
Various reasons have been put forward to explain
the Houthakker–Magee asymmetry in the literature.
Mann (2002) lists demographic differences between
the U.S. population and its major trading partners as
possible reasons for the asymmetry. Citing Gould (1994)
and Marquez (2002), Mann notes that there is a relatively high share of immigrants in the U.S. population
and that these immigrants have a strong preference for
goods from their respective home countries. Furthermore, a relatively young population in the U.S. has a
greater demand for goods, especially imported goods,
compared with the relatively older populations in
Europe and Japan that spend proportionally more on
domestic services, such as health care.



A misspecified model could also be behind the
asymmetry. Krugman (1989) developed a model in
which countries grow by producing new goods that
can be exported. His theory implied that empirical
models used to estimate elasticities should include a
supply term in the import demand equation. Gagnon
(2004) includes a supply term—the GDP of the exporting country—in the import demand equation, and finds
strong evidence of a supply effect. The inclusion of
this variable in the equation leads to a reduction in
the estimate of the U.S.’s income elasticity for imports.
The U.S.’s trade elasticities have been estimated
in numerous papers since Houthakker and Magee’s
findings were published in 1969, but the asymmetry
has proven to be robust across time periods and econometric methods.4 See table 1.
Table 1 summarizes the results of several recent
papers. An important contribution by Hooper, Johnson,
and Marquez (2000) uses Johansen’s (1988) cointegration technique to estimate the long-term trade elasticities for the G-7 countries, and finds that the U.S. income
elasticity of imports is roughly twice as high as the income elasticity of exports. Chinn (2004), using different measures of relative import and export prices and
several additional years of more recent data, estimates

4Q/2007, Economic Perspectives

income elasticities of imports and exports that are larger
in magnitude than both Houthakker and Magee (1969)
and Hooper, Johnson, and Marquez (2000). While the
absolute difference between the import and export
elasticities is a substantial 0.7, the relative asymmetry
appears more modest than in earlier studies, with the
income elasticity for imports only 1.4 times larger
than the income elasticity for exports.
Continuing down the rows of table 1, Cardarelli
and Rebucci (2007) estimate large income elasticities
for imports and exports of 2.0 and 1.9, respectively.
Their study notes that trade elasticities are affected by
aggregation bias. Specifically, aggregate price elasticities might be understated relative to a trade-weighted
average of sector elasticities if goods with relatively
low price elasticities face stronger price variation than
goods with relatively high price elasticities. Cardarelli
and Rebucci correct for aggregation bias by estimating separate trade elasticities for 17 categories of imports and 16 categories of exports and then taking the
simple average of these separate estimates. Income
elasticities are lower and price elasticities are higher
when the aggregate estimate is constructed from sector-level estimates. It is interesting to note that the
Houthakker–Magee asymmetry almost disappears
in Cardarelli and Rebucci’s estimates that correct for
aggregation bias.
Finally, one important feature of U.S. imports and
exports that has been changing recently is that services
trade is becoming more prominent. To the extent that
trade elasticities for services differ from those for goods,
this could have important implications for the evolution
of the U.S. trade balance. For example, Wren-Lewis
and Driver (1998) find that the elasticity of U.S. exports
of services to foreign income of 1.95 is much higher
than that of manufactured goods (1.21), and they find
that the elasticity of U.S. imports of services with respect to U.S. income (1.72) is much lower than that
of manufactured goods (2.36).5 In other words, the
Houthakker–Magee asymmetry is reversed for U.S. trade
in services. Mann (2002) argues that the Houthakker–
Magee asymmetry between the import and export elasticities of trade in goods and services might gradually
attenuate as the world’s economies mature and spend
more on services and less on manufactured goods.
Econometric model
The empirical model relating imports to national
income and relative import prices and the model relating exports to foreign national income and relative
export prices come from Hooper, Johnson, and Marquez
(1998, 2000). These models assume that income and
price elasticities of demand for imports and exports

Federal Reserve Bank of Chicago

are constant over time. We estimate each system of
equations using quarterly data for each of the G-7
countries.
The system for real imports is
1)

2)

n

n

n

j =1

j =1

j =1

n

n

n

j =1

j =1

j =1

mit = ∑ α ij mit − j + ∑ γ ij yit − j + ∑ λ ij rpmit − j + εit ,

yit = ∑ τij mit − j + ∑ υij yit − j + ∑ φij rpmit − j + ςit ,

			
n

n

n

j =1

j =1

j =1

		

3) rpmit = ∑ θij mit − j + ∑ ν ij yit − j + ∑ χij rpmit − j + ϕit ,
where mit is the log of real imports of country i at
time t, yit is the log of real GDP of country i at time t,
and rpmit is the log of the relative price of imports to
domestic goods and services, or more precisely, the
log of import prices relative to the GDP deflator for
country i at time t.
Similarly, the system for real exports is given by
4)

5)
	
6)

n

n

n

j =1

j =1

j =1

n

n

n

j =1

j =1

j =1

xit = ∑ δij xit − j + ∑ κij fyit − j + ∑ ξij rpxit − j + µit ,
fyit = ∑ ρij xit − j + ∑ ηij fyit − j + ∑ ιij rpxit − j + σit ,
n

n

n

j =1

j =1

j =1

rpxit = ∑ ωij xit − j + ∑ ςij fyit − j + ∑ ψ ij rpxit − j + eit ,

where xit is the log of real exports of country i at time
t, fyit is the log of real trade-weighted foreign GDP of
country i’s export partners at time t, and rpxit is the log
of a measure of the relative price of exports for country i at time t. Construction of all variables is detailed
in the next section.
In brief, we construct a trade-weighted foreign GDP
series for each exporting country i, using data on all
of its trading partners for which we could obtain a quarterly real seasonally adjusted GDP series over a sufficiently long time horizon. Thus, our confidence in the
estimated export elasticities with respect to prices and income rises with the coverage of our foreign GDP series.
For the U.S., we also estimate the model using real annual data because although this reduces the number of observations in our data sample considerably, it increases
the country coverage of our foreign GDP variable.



We use two different price series as the relative
price of exports. The first measure follows Hooper,
Johnson, and Marquez (1998, 2000) and is essentially
the log of the ratio of export prices to the trade-weighted
GDP deflators in the importing countries. Although
this measure is theoretically preferred, our relatively
low level of country coverage led us to use also the
International Monetary Fund’s (IMF) real effective
exchange rate, or REER (the ratio of unit labor costs
in the exporting country divided by export-weighted
unit labor costs in the destination markets), as a proxy
for the relative price of exports.
We estimate both the import system and the export system by using Johansen’s (1988) full information maximum likelihood estimator, and report estimates
for which test statistics support our assumption that at
least one cointegrating relationship is present in the
data. This estimator essentially assumes that each variable in the system is stationary in first differences, but
information on the level of a variable or variables also
helps to describe the system.

where PXi is the export price of country i and GPFi is
the geometric mean of the domestic prices of country i’s
export partners adjusted by the nominal exchange
rate index. Note that GDPFj is the GDP deflator for
country j.11 Also, note that E$/j is an exchange rate index for country j that normalizes the amount of dollars
that can be bought with a unit of local currency to a
value of 1 in the year 2000. The weight wij is the proportion of country i’s exports going to country j. The
weights are constructed using bilateral exports of
goods in 2000, which were obtained from the OECD’s
STAN Bilateral Trade Database.12 For estimates reported in table 2 on a data sample that ends in 1994,
we construct foreign GDP, using weights on bilateral
exports of goods in 1995.
The foreign income used in the export equation
is constructed as the geometric mean of the real GDP
of each of country i’s export partners weighted by the
export shares of trade in goods. It is calculated as follows:

Data

FYi = Π (Y j × E$/j ) wij , Σwij = 1,

Quarterly data on real seasonally adjusted imports
and exports of goods and services and chained price
indexes of imports and exports of goods and services
for each of the G-7 countries come from Haver Analytics’ Group of Ten (G-10) database.6 Data on quarterly real GDP and GDP deflators for the G-7 countries
and their trading partners come from Haver Analytics’
G-10 database and the Organization for Economic
Cooperation and Development’s (OECD) Main Economic Indicators.7 The GDP data for six additional
G-7 trading partners—Argentina, Brazil, China, Hong
Kong, Singapore, and Taiwan—come from internal
estimates of Federal Reserve Board staff. The periods
for which data are available vary by country. The periods used in estimating the import and export systems
of each G-7 country are reported in tables 3 and 4.
While we have tried to use data from the OECD
whenever possible, estimates for France, Germany,
and Japan utilize additional data sources.8
The construction of all variables follows Hooper,
Johnson, and Marquez (2000). The relative price of
imports (rpmi) is constructed as:
rpmi = log (PMi /PYi),
where PMi is the chained price index of imports for
country i and PYi is the GDP deflator for country i.9
The relative price of exports (rpxi) is constructed
as follows:10

GPFi = Π (GDPFj × E$/j ) wij , Σwij = 1,

where FYi is the aggregate foreign income for country i,
Yj is the real GDP of trading partner j for country i,
and the weight wij is the proportion of country i’s exports going to country j.
Lastly, we also estimate the export system using
the IMF’s real effective exchange rate as a measure of
the relative price of exports. The appendix contains a
description of the REER and compares it with the relative
price of exports (rpxi) measure described previously.
Results: Long-run elasticities
Estimates of long-run trade elasticities are presented in table 2 through table 6. Our results are generally in line with previous studies, but some interesting
differences exist. Notably, our estimates of the import
elasticities with respect to income for the G-7 countries
over a period of time that ends in 2006 are generally
higher than those reported by Hooper, Johnson, and
Marquez (1998, 2000), whose sample period ends in
1994. Further, many of our import price elasticities
are larger and more negative than those reported by
Hooper, Johnson, and Marquez. On the export side,
our estimates of the export elasticities with respect to
income are as large as or larger than Hooper, Johnson,
and Marquez’s estimates, which cover an earlier period.
The export price elasticities we report differ markedly
from those in Hooper, Johnson, and Marquez’s research,
but not in any systematic way.

rpxi = log (PX i × E$/i /GPFi ),



4Q/2007, Economic Perspectives

Table 2

Long-run elasticities of industrialized countries through 1994
A. Estimates
	
	
	
	
	
Canada	
France 	
Germany	
Italy	
Japan	
UK	
U.S.	
U.S., annual	

Income	
Hooper, Johnson,	
and Marquez’s 	
estimates	
Exports	
Imports	
1.1*	
1.5*	
1.4*	
1.6*	
1.1*	
1.1*	
0.8*	
—	

1.4*	
1.6*	
1.5*	
1.4*	
0.9*	
2.2*	
1.8*	
—	

Price	
	

	
Authors’ estimates	
Exports	
Imports	
1.56*	
—	
2.06*	
1.64*	
0.99*	
0.97*	
2.33*	
1.06*	

1.50*	
1.30*	
2.26*	
1.63*	
1.65*	
1.70*	
1.92*	
1.78*	

Hooper, Johnson,
and Marquez’s
estimates	
Exports	
Imports	
–0.9*	
–0.2	
–0.3	
–0.9*	
–1.0*	
–1.6*	
–1.5*	
—	

–0.9*	
–0.4*	
–0.06*	
–0.4*	
–0.3*	
–0.6	
–0.3*	
—	

Authors’ estimates
Exports	
Imports
–0.61*	
—	
–0.79*	
–0.57*	
–0.74*	
–1.31*	
–0.24*	
–0.97*	

–1.14*
–0.50*
–0.42*
–0.33*
–0.15*
–0.38*
–0.25*
–0.19*

							
B. Lags and sample periods
	
Number of lags	
Sample period start dates	
	
Hooper, Johnson,	
	
Hooper, Johnson,
	
and Marquez’s 	
	
and Marquez’s	
	
estimates	
Authors’ estimates	
estimates	
Authors’ estimates
	
Exports	
Imports	
Exports	
Imports	
Exports	
Imports	
Exports	
Imports
	
Canada	
9	
8	
4	
2	
1976:Q1	 1961:Q1	
1981:Q1	 1961:Q1
France 	
2	
3	
—	
5	
1975:Q4	 1971:Q3	
—	
1978:Q1
Germany	
2	
2	
3	
3	
1977:Q4	 1968:Q1	
1981:Q1	 1979:Q4
Italy	
2	
4	
3	
5	
1976:Q1	 1971:Q2	
1981:Q1	 1981:Q1
Japan	
5	
6	
3	
8	
1976:Q1	 1955:Q2	
1981:Q1	 1980:Q1
UK	
4	
5	
6	
4	
1976:Q1	 1955:Q1	
1981:Q1	 1955:Q1
U.S.	
2	
9	
3	
4	
1976:Q1	 1959:Q3	
1981:Q1	 1955:Q1
U.S., annual	
—	
—	
3	
3	
—	
—	
1981	
1955
	

	

	

	

	

*Significant at the 5 percent level.
Note: All sample periods end in 1994:Q4.
Sources: Authors’ calculations; Haver Analytics; and Hooper, Johnson, and Marquez (2000).

To summarize, most of our estimates appear reasonable; however, some estimates, discussed in detail
later, appear questionable. Interestingly, our estimates
suggest that trade elasticities with respect to income
are increasing over time. This is consistent with a globalizing world economy in which trade is becoming
more important.
Comparing our results with previous research
Table 2 presents our estimates of the long-run trade
elasticities of the G-7 countries for the period through
the fourth quarter of 1994.13 The objective of this exercise is to replicate as closely as possible the long-run
elasticities of Hooper, Johnson, and Marquez (2000).
This is a useful starting point because before we can
draw any conclusions about whether elasticities have
changed over time, we would like to understand how
closely our data and econometric techniques are able
to reproduce previous work.
Overall, our import elasticities with respect to income and prices are close to those of Hooper, Johnson,

Federal Reserve Bank of Chicago

	

and Marquez (2000), but they do not match perfectly.
We suppose that the differences are due to three causes:
1) Hooper, Johnson, and Marquez have a longer time
series of data extending further back into history than
ours; 2) they likely made adjustments to some data
series;14 and 3) we report estimates using a different
number of lags than Hooper, Johnson, and Marquez.15
As for the estimates of export elasticities with respect to income and prices, the differences between our
estimates and those of Hooper, Johnson, and Marquez
(2000) using data through 1994 are small for most
countries. However, our estimate for the U.S. income
elasticity using quarterly data seems implausibly large.
In addition to the differences in sample periods,16 there
are likely two additional causes for the discrepancies
between our estimates and Hooper, Johnson, and
Marquez’s estimates of export elasticities: 1) We use
the IMF’s REER as a measure of the relative export
prices (see the discussion in the appendix) and 2) Hooper,
Johnson, and Marquez likely have a measure of foreign GDP that covers a larger share of each country’s



trading partners. Because Johansen’s (1988) estimator
uses information on the level of foreign GDP as well
as the growth rate, the estimated elasticities are sensitive to the construction of the foreign GDP variable.
Turning to columns 2 and 4 of panel A of table 2,
we see the estimates of the import elasticities with respect to income of Hooper, Johnson, and Marquez
(2000) and our own, respectively. The major discrepancies between the Hooper, Johnson, and Marquez’s
estimates and our estimates occur for the UK, Germany,
and Japan. Our estimated income elasticity of 1.7 for
the UK is closer to the long-run equilibrium elasticity
of 1 suggested by economic theory. Further, we estimate a statistically significant negative price elasticity
of –0.38 in line with the negative elasticity predicted
by theory, whereas Hooper, Johnson, and Marquez’s
estimate was not statistically different from zero.
For both Japan and Germany, our estimated income elasticities with respect to imports diverge from
Hooper, Johnson, and Marquez’s by more than a factor
of 1.5. We attribute these differences to the longer time
span of their data and possible differences in their
handling of German reunification. They use German
data series that begin in 1968, while ours start in 1979.
Moreover, we constructed long time series of German
imports, import prices, and GDP from series on West
Germany and reunified Germany. Also, Hooper, Johnson,
and Marquez’s Japan data series begins in 1955, while
ours begins in 1981.
In columns 1 and 3 of panel A of table 2, we show
Hooper, Johnson, and Marquez’s (2000) estimates for
export elasticities with respect to income, as well as
our own. There are small differences between the two
sets of estimates for most countries. For Italy, Japan,
and the UK, our export elasticities with regard to income are quite close to Hooper, Johnson, and Marquez’s.
However, we estimate substantially larger income
elasticities for Canada, Germany, and the U.S. We do
not report export coefficients for France because no
specification gave sensible results. On the price side,
our estimated elasticities are uniformly smaller, with
the exception of Germany.
Lastly, we estimated the import and export systems
using annual data for the U.S. through 1994 (last row of
panel A of table 2). Our annual estimates of trade
elasticities are close to or the same as Hooper, Johnson,
and Marquez’s estimates using quarterly data. On the
export side, by estimating the system on annual data,
we were able to incorporate a measure of foreign GDP
that includes several of the U.S.’s smaller trading partners for which we only have annual data on real GDP.
The annual model’s generally better agreement with
Hooper, Johnson, and Marquez’s results suggests that the
construction of foreign GDP is of primary importance.

10

Import elasticities for industrialized countries
through 2006
In table 3, we present our estimates of the import
elasticity with regard to income and prices through
2006 alongside our estimates for the period through
1994. For all countries except the UK, the estimated
import elasticities with respect to income are higher
over the sample period through 2006, suggesting that income elasticities might be increasing over time.17
Income and price elasticity estimates for Canada,
France, Italy, the UK, and the U.S. are of the expected
sign, and they are statistically significant at a 5 percent significance level. Estimates for Germany and
Japan, which are unreasonably high for income and
no different from zero for price, may be attributed to
the data issues outlined earlier.
The import income elasticities are substantially
greater than one for all countries over both sample
periods. As mentioned earlier, this implies a long-run
imbalance in that, as a nation grows, if relative prices
are constant, imports will eventually exceed GDP.
Second, the income elasticities have increased in all
of the G-7 countries except the UK and possibly the
U.S. Estimates of the income elasticity for the U.S.
using quarterly data show no change between the earlier and later sample periods. However, estimating the
import system on annual U.S. data suggests that the
income elasticity has increased over time. Estimates
of the U.S. price elasticity appear to have increased
over time using either quarterly or annual data.
The higher price elasticities that we estimate for
France, the UK, and the U.S. could be the result of increasing global price competition. As tariffs and other
trade barriers fall, consumers might be able to switch
their purchases to lower-cost producers more easily,
resulting in an increased sensitivity of imports to prices.
Two different phenomena could be behind the apparently higher income elasticities in the sample that includes data through 2006. First, import price indexes tend
to overstate the true price of imports. It is well known
that much trade growth comes from new products, which
could be an old product that is coming from a new, cheaper market (for example, China) or a truly new product
that is likely to be of higher quality or a relatively lower
price than the previously existing product. Because statistical agencies tend to treat imports of new products
as having the same price as old products, this creates
an upward bias in import prices. As increasing shares
of imports come from developing countries, this bias
could be increasing. In estimating the import elasticity
with respect to income and prices, an upwardly biased
price measure would lead to a larger income elasticity
and a smaller (in absolute value) price elasticity.

4Q/2007, Economic Perspectives

Table 3

Long-run import elasticities for industrialized
countries through 2006
A. Estimates
	
	

Start through 2006	
Income	
Price	

Start through 1994
Income 	
Price

Canada	
1.67*	
–1.17*	
1.50*	
–1.14*
France 	
1.62*	
–0.61*	
1.30*	
–0.50*
Germany	
3.28*	
0.08	
2.26*	
–0.42*
Italy	
2.48*	
–0.23*	
1.63*	
–0.33*
Japan	
1.94*	
0.05	
1.65*	
–0.15*
UK	
1.65*	
–0.60*	
1.70*	
–0.38*
U.S.	
1.93*	
–0.63*	
1.92*	
–0.25*
U.S., annual	
1.95*	
–0.47*	
1.78*	
–0.19*
			
		
B. Sample periods
	
Start through 2006	
Start through 1994
Canada	
1961:Q1–2006:Q4 	
1961:Q1–1994:Q4
France 	
1978:Q1–2005:Q2	
1978:Q1–1994:Q2
Germany	
1979:Q4–2006:Q4	
1979:Q4–1994:Q4
Italy	
1981:Q1–2006:Q4	
1981:Q1–1994:Q4
Japan	
1980:Q1–2006:Q3	
1980:Q1–1994:Q3
UK	
1955:Q1–2006:Q4	
1955:Q1–1994:Q4
U.S.	
1955:Q1–2006:Q4	
1955:Q1–1994:Q4
U.S., annual 	
1955–2006 	
1955–1994
			
		
C. Number of lags
	
Start through 2006	
Start through 1994
Canada	
France 	
Germany	
Italy	
Japan	
UK	
U.S.	
U.S., annual 	

4	
3	
2	
7	
3	
4	
3	
3	

*Significant at the 5 percent level.
Sources: Authors’ calculations and Haver Analytics.

Second, vertical integration is thought to be behind much of the recent rapid expansion of trade. Vertical integration is the process by which firms have
spread their production processes across several countries so that production processes requiring lower-skilled
labor (such as manual assembly) happen in less developed countries, whereas production processes that
are more capital intensive and require higher-skilled
labor (such as building an engine) happen in more developed countries. An example of a vertically integrated
production process would be a car manufacturer that
previously produced an entire car from start to finish
in the U.S. Under a vertically integrated production process, the engine and other higher-tech components would
be made in the U.S. and then exported to Mexico, where
the assembly of the car would take place. When the

Federal Reserve Bank of Chicago

2
5
3
5
8
4
4
3

finished car is imported into the U.S., the
total value of the car (including not only
the value created in Mexico when the car
was assembled, but also the value of the
parts exported from the U.S.) is recorded
in U.S. import statistics. Thus, because
the total value of imports increases with
vertical integration, this leads estimates
of the import elasticity with respect to income to be overstated.18
Lastly, returning to our estimates in
table 3, we note that Germany has the
largest increase in income elasticity across
the two periods as well as the largest income elasticity in both periods. The sizable
increase may be attributable to the inclusion of both preunification and postunification Germany in the sample. Preunification
(West) Germany had relatively higher
import growth and GDP growth than
postunification Germany. The sample of
data through 1994 includes only a few
years of unified German data; thus, the
model essentially estimates elasticities
for preunification Germany. In contrast,
the sample through 2006 includes several
years of both preunification and postunification data. The differences in GDP
and import growth rates between the
preunification and postunification periods
might make it appear that there is a particularly strong relationship between income and imports.

Export elasticities for industrialized
countries in 1981–2006
Table 4 compares our estimates of
export elasticities on data from 1981 through 1994
with our estimates on data from 1981 through 2006.
The last column reports the share of each country’s exports that each foreign GDP series covers. With the
exception of Canada, estimates of the income elasticity
for the G-7 countries over the period 1981–2006 are
as large as or larger than those for the period 1981–94.
This could be interpreted as evidence that export elasticities with respect to income are increasing over time.
For Canada, France, Germany, Italy, Japan, and
the U.S., estimates of the export elasticities with regard to income and prices are of the expected sign
and are statistically significant. The estimate of the
UK’s export elasticity with respect to income is of the
correct sign, but its elasticity with respect to relative
export prices is positive. This implies that the UK

11

Table 4

Long-run export elasticities for industrialized
countries through 2006
A. Estimates
	
	

  1981–2006	
Income	
Price	

1981–94
Income 	
Price	

Canada	
1.06*	
–0.18*	
1.56*	
–0.61*	
France 	
1.22*	
–2.86*	
—	
—	
Germany	
2.67*	
–1.15*	
2.06*	
–0.79*	
Italy	
1.74*	
–0.74*	
1.64*	
–0.57*	
Japan	
1.70*	
–0.34*	
0.99*	
–0.74*	
UK	
1.28*	
1.17*	
0.97*	
–1.31*	
U.S.	
2.34*	
–0.61*	
2.33*	
–0.13*	
U.S., annual	
—	
—	
1.06*	
–0.97*	
			
B. Number of lags
	
Canada	
France 	
Germany	
Italy	
Japan	
UK	
U.S.	
U.S., annual	

1981–2006	

1981–94

8	
4	
3	
4	
4	
4	
2	
—	

4
—	
3	
3	
3	
6	
3	
3	

Export

share
95
57
55
70
72
61
77
83

through 2006 and on the sample through
1994. This diverges from previous estimates that found that Japan’s export elasticity for income exceeded its import
elasticity for income (Houthakker and
Magee, 1969; and Hooper, Johnson, and
Marquez, 2000).
There is no obvious time trend in the
Houthakker–Magee asymmetry across
countries. The asymmetry appears to have
increased in Canada, Germany, and Italy,
while it moderated in Japan and the UK.
Returning to table 1 (p. 6), we see that
studies of U.S. elasticities incorporating
the most recent data have tended to find
a more moderate relative asymmetry in
the U.S.

Estimates of import elasticities for the
U.S. by sector
Table 5 presents disaggregated import elasticities with respect to income
and prices for three periods: 1967–2006,
	
	
*Significant at the 5 percent level.
1967–87, and 1988–2006.19 In choosing
Note: Estimates use a 1981:Q1–2006:Q4 sample and a 1981:Q1–1994:Q4 sample.
1988 as a somewhat arbitrary breakpoint,
Sources: Authors’ calculations and Haver Analytics.
we hoped to split the sample into an early
period of relatively high trade barriers
and high inflation and a later period of
exports more when its products are more expensive
lower trade barriers and more stable prices. Moreover,
than its competitors’ and, consequently, is difficult to
by 1988, much of the U.S. dollar depreciation formalreconcile with an imperfect substitutes model of trade.
ized in the Plaza Accord of 1985 and the Louvre
The export share of a country’s trading partners
Accord of 1987 is likely to have fully passed through
included in its foreign GDP measure (column 5) apinto import prices.20 Comparing the early and later
pears to be highly correlated with the quality of the
sample periods across disaggregated imported goods
estimates. The countries that have the most implausiand services, we generally observe higher income
ble income (Germany) and price (France, Germany,
elasticities and, with some exceptions, higher price
and the UK) elasticities also have the smallest share
elasticities in the later period.
of their trading partners included in their respective
The first two columns present the estimates on
foreign GDP measures. For example, we cover only
the 1967–2006 sample. The next two present the esti57 percent of French exports in constructing the foreign
mates on the 1967–87 sample. The following two colGDP measure for France, and estimate an extremely
umns present the estimates on the 1988–2006 sample.
large price elasticity over the period 1981–2006. As
To give the reader a sense of how important each catstated previously, we estimate a positive and significant
egory is, the final column of table 5 shows each endprice elasticity for the UK, where our foreign GDP
use category’s share of year 2000 imports.
series covers only 61 percent of exports. These results
Beginning in the top row of table 5, the income
contrast sharply with the estimates for Canada, where
elasticity for total imports appears to have increased
we cover 95 percent of Canada’s trade-weighted tradover time. The price elasticities for total imports show
ing partners in our foreign GDP variable and the cothe same upward trend over time as the estimates in
efficients are much closer to those implied by theory.
table 3.21 We turn next to the estimates for industrial
Comparing the import and export elasticities on
durables (row 4) and industrial nondurables excluding
the samples through 2006, we see that the Houthakker–
oil (row 5). Imports of industrial durables are primarMagee asymmetry holds for all the G-7 nations exily composed of iron, steel, other metals, and building
cept the U.S. Of particular interest, in our estimates
materials. In the 1988–2006 period, the estimated
the asymmetry holds for Japan, both on the sample
price elasticity of industrial durables is not significantly

12

4Q/2007, Economic Perspectives

Table 5

Long-run U.S. import elasticities, by sector
	

	
	
	
	
Total imports	
Goods	
Industrial goods except oil	
Industrial durables 	
Industrial nondurables	
Petroleum	
Capital goods except autos	
Autos	
Consumer goods except autos	
Durable consumer goods	
Nondurable consumer goods	
Services	
Nonpetroleum goods	

	

1967–2006	
	

1967–87	
	

1988–2006
	

2000

Income	

Price	

Income	

Price	

Income	

Price	

import
share

1.98*	
2.10*	
1.33*	
1.14*	
1.63*	
1.05*	
2.54*	
1.64*	
2.42*	
2.21*	
2.41*	
1.58*	
2.20*	

–0.47*	
–0.42*	
–0.43*	
–0.89*	
–0.32*	
1.00*	
–1.04*	
–0.38	
–0.84*	
–1.05*	
–1.02*	
–1.32*	
–0.63*	

1.94*	
1.98*	
1.12*	
0.62*	
1.71*	
0.30	
4.08*	
3.07*	
2.83*	
2.68*	
3.05*	
1.80*	
2.41*	

–0.37*	
–0.22*	
–0.32*	
–0.21*	
–0.41*	
0.82*	
–0.87*	
–1.10*	
–0.97*	
–0.90*	
–1.04*	
–1.55*	
–0.81*	

2.11*	
2.18*	
1.82*	
2.11*	
1.56*	
1.23*	
–1.20	
2.03*	
1.76*	
2.56*	
3.68*	
1.64*	
1.82*	

–0.62*	
–0.69*	
–0.41*	
–0.04	
–0.79*	
–0.03	
–2.39*	
0.11	
–1.78*	
–0.87*	
1.34	
0.06	
–1.07*	

100
84
12
6
6
8
24
13
19
10
9
16
76

								
*Significant at the 5 percent level.
Notes: All estimates were calculated using three lags. The 2000 import share column presents that sector’s or subsector’s import value as a
percent of total imports. Because we present the import shares of both aggregated sectors (for example, goods and services) and some of the
finely disaggregated subsectors (for example, industrial durables and nondurables), the shares do not necessarily add up to 100.

different from zero. The high income elasticity and
the low price elasticity may be indicative of price
mismeasurement or vertical integration. Alternatively,
since steel and other metals form a large share of industrial durables, government intervention and the threat
of trade policy restrictions may play a role. The prospect of government action to protect the domestic
industry may discourage price competition among exporters. In contrast, imported industrial nondurables
are mostly chemicals and paper products. The increasing price elasticity we observe for industrial nondurables might suggest international price competition is
increasing in this sector.
Continuing down to row 7, the income and price
elasticities of demand for imports of capital goods excluding automobiles are of particular interest, since
capital goods represent almost a quarter of total imports.
Unfortunately, our estimates for this sector are not
easily interpretable. The large negative price elasticity for the period 1988–2006 is probably an artifact of
rapidly falling, yet difficult to measure, computer prices.
Because computers make up a large share of this category, we might expect there would be significant
difficulties in the construction of a price index for this
sector. On both the 1967–2006 sample and the 1967–87
sample, the income elasticity of capital goods is quite
high, higher than any other category. This is consistent with investment or purchases of capital goods being strongly pro-cyclical. These results are consistent
with Chinn (2004), who found high income elasticities on capital goods.

Federal Reserve Bank of Chicago

Next, we turn to the estimates of the elasticities
for consumer durables (row 10) and consumer nondurables (row 11). As with capital goods, income elasticities of consumer goods are higher than those of most
other categories. This suggests that luxury goods may
be playing an important role in imports of consumer
goods. The elasticities for consumer durable imports
are remarkably stable over both the 1967–87 and
1988–2006 samples. The pattern of change in the consumer nondurables coefficients suggests price mismeasurement. Given that this category is primarily apparel,
one might not expect price measurement problems to
be present (as opposed to a sector including computers). However, price varies widely across import
source, and the end of textile quotas has led to a great
deal of change in the source of apparel imports.
Finally, elasticities for imported services are presented in row 12 of table 5. We estimate the income
elasticity for services imports to be 1.64 over the period 1988–2006, considerably lower than our estimated
elasticity for imports of goods of 2.18. These estimates
are close to those reported by Wren-Lewis and Driver
(1998), who found the income elasticity for services
imports to be 1.72 and for goods to be 2.36 by using
the same methodology over an earlier period (1980–95).
Estimates of export elasticities for the U.S. by sector
Table 6 reports the disaggregated export elasticities by end-use category. Because quarterly real foreign
GDP data are only available since 1981, we only estimate on the 1981–2006 and 1988–2006 samples. We
present estimates that use the real effective exchange

13

Table 6

Long-run U.S. export elasticities, by sector
	
	
	
Total exports	
Goods	
Industrial goods except oil	
Industrial durables 	
Industrial nondurables 	
Agriculture	
Capital goods except autos	
Autos	
Consumer goods except autos	
Durable consumer goods	
Nondurable consumer goods	
Services	
Nonagricultural goods	

Real effective exchange rate	
1981–2006	
1988–2006	
Income	
Price	
Income	 Price	
2.34*	
2.51*	
1.62*	
1.85*	
1.48*	
0.98*	
3.33*	
2.42*	
2.79*	
3.00*	
2.59*	
2.04*	
2.70*	

–0.61*	
–0.63*	
0.03	
–0.16	
0.04	
0.19	
–1.79*	
–0.01	
–0.83*	
–1.11*	
–0.44*	
–0.25*	
–0.77*	

1.86*	 –5.07*	
1.91	
–8.56*	
1.65*	 –0.07	
1.78*	
0.30	
1.57*	 –0.18*	
1.10*	
0.07	
–5.94	 –63.07*	
2.53*	 –0.82*	
2.76*	 –0.49*	
2.91*	 –0.59*	
2.59*	 –0.41*	
1.87*	 –0.61*	
1.96	 –10.14*	

Relative price of exports
1981–2006	
1988–2006	
Income	
Price	
Income	 Price	
2.76*	
3.04*	
1.62*	
2.16*	
1.48*	
1.41*	
7.12*	
2.83*	
2.77*	
2.79*	
2.78*	
2.38*	
3.32*	

0.12	
0.20	
0.07	
0.23	
0.06	
0.55*	
1.28*	
0.35*	
–0.75*	
–1.09*	
–0.26	
0.10	
0.23	

3.83*	 1.78*	
4.90*	 2.21*	
1.58*	 0.26*	
8.70	 –76.98*	
1.54*	 –0.09	
1.27*	 0.30*	
–21.51*	 –11.47*	
2.68*	 0.19	
2.53*	 –0.39	
2.53*	 –0.56*	
2.58*	 –0.10	
2.00*	 0.31	
5.60*	 2.60*	

2000
export

share
100
72
15
6
9
5
24
7
8
4
4
28
67

*Significant at the 5 percent level.
Notes: All estimates were calculated using two lags and year 2000 weights. The 2000 export share column presents that sector’s or subsector’s
export value as a percent of total exports. Because we present the export shares of both aggregated sectors (for example, goods and services)
and some of the finely disaggregated subsectors (for example, industrial durables and nondurables), the shares do not necessarily add up to 100.
Sources: Authors’ calculations and Haver Analytics.

rate in the first four columns and, for comparison, estimates using the relative price of exports in columns
5 through 8. The final column shows each end-use
category’s share of total U.S. exports in the year
2000.
While both price measures produce some problematic estimates, the REER estimates in general appear
more reasonable. Using the REER as our price variable,
most of the disaggregated categories have sensible estimates. These include the consumer goods categories,
industrial goods, services, and automobiles. Some of
these sectors have price elasticities that are not significantly different from zero. The very small price elasticities might be attributable to high-quality or unique
U.S. exports, for which few substitutes exist. The major problem is the extreme capital goods price elasticity, which further appears to dominate any aggregate
measure that includes capital goods. Given that U.S.
exports include not only a large amount of high tech
but also airplanes, we should expect problems.
Lastly, row 12 of table 6 presents the income and
price elasticities of demand for exported services. Using
the REER as the foreign relative price measure, we find
that the income elasticity for services over the 1988–
2006 period is a relatively large 1.87. Consistent with
the findings of Wren-Lewis and Driver (1998), we
find that the Houthakker–Magee asymmetry is reversed
for services trade. Our income elasticity for services
exports is considerably larger than the elasticity of
services imports reported in table 5 of 1.64.

14

Conclusion
In this article, we present new estimates of trade
elasticities for seven industrialized countries using
data through 2006. We find that the Houthakker–
Magee asymmetry, which implies an increasing trade
deficit if relative prices are held constant, is present in
all countries, with the exception of the U.S. Our high
estimate of the U.S. income elasticity of demand for
exports is found to be highly sensitive to the construction of foreign GDP. Thus, we do not think that our
estimate is definitive.
While the Houthakker–Magee asymmetry has
been present for most of the G-7 countries for a long
time, our article is the first that we know of to find
Japan’s estimated import elasticity with respect to income is larger than its export elasticity with respect
to income.
Finally, in estimating elasticities for disaggregated
sectors, we find that our estimate of the U.S. export
elasticity for services with respect to foreign income
of 1.87 exceeds the U.S. import elasticity for services
with respect to income of 1.64. This means that if the
U.S. were to grow at the same rate as its trading partners,
over time the U.S. trade balance in services would
move toward larger and larger trade surpluses. This is
consistent with previous research (Wren-Lewis and
Driver, 1998) and suggests that the Houthakker–
Magee asymmetry for aggregate trade in goods and
services could gradually attenuate as services trade
increases as a share of total trade.

4Q/2007, Economic Perspectives

notes
These numbers are provided by the U.S. Bureau of Economic
Analysis, and they are in real 2000 chain-weighted dollars. In current dollars, which do not adjust for inflation, 2006 imports were
$2,229 billion and exports were $1,467 billion.
1

The start dates for the estimation sample for each country vary
according to data availability. We use the longest period available
for each country. Sample start dates are reported in panel B of
table 2 (p. 9).
2

These numbers are in real 2000 chain-weighted dollars.

3

Goldstein and Khan (1985) contain references to the early literature.

4

Wren-Lewis and Driver’s (1998) sample period is 1980–95. The
cited numbers are those estimated using Johansen’s (1988) maximum likelihood estimator.
5

The Group of Ten actually comprises 11 nations: namely, Belgium,
Canada, France, Germany, Italy, Japan, the Netherlands, Sweden,
Switzerland, the UK, and the U.S. Luxembourg is an associate member of the G-10. Haver Analytics’ G-10 database provides statistics
from each country’s national accounts in an easy-to-use format.
6

Trading partners available from the OECD for sufficiently long
periods include Australia, Canada, France, Germany, Italy, Japan,
South Korea, Mexico, the Netherlands, Switzerland, Taiwan, the
UK, and the U.S.

Countries included in the calculation of the relative price of exports are: Argentina, Australia, Brazil, Canada, France, Germany,
Hong Kong, Italy, Japan, South Korea, Mexico, the Netherlands,
Singapore, Switzerland, Taiwan, the UK, and the U.S. For Argentina,
Brazil, Hong Kong, Singapore, and Taiwan, the GDP deflator was
not available. We used the Consumer Price Index (CPI) in place of
a GDP deflator for these countries.
11

Table A1 in the appendix lists the cumulative share of exports
accounted for by those countries listed in note 11 for each of the
G-7 countries.
12

As noted previously, the sample start dates vary by country
according to data availability, and they are reported in panel B of
table 2 (p. 9).
13

For example, a long time series for German data must be built
from historical data on West Germany and reunified Germany. In
the process of constructing this series, different researchers are
likely to make different adjustments to the raw data.
14

We estimate the model using between two and nine lags and then
select the shortest lag length that corresponds to at least one cointegration vector and produces plausible results.
15

7

For France, we follow Pluyaud (2006) and use the import price
deflator from Eurostat, rather than the OECD, beginning in 1995.
The two price indexes for French imports (Eurostat versus OECD)
diverge after the introduction of the euro, apparently reflecting a
difference in methodology. For Germany, we construct a time series
for German real GDP growth by splicing the GDP series for the
unified Germany to the (appropriately scaled) West German GDP
series in 1991. Data on real German imports come from the IMF’s
International Financial Statistics (IFS) database. For Japan, real
chain-weighted GDP is available from 1994 to the present. To obtain a longer time series of Japanese GDP, we splice a fixed-weighted GDP series for the period 1981–93 onto the chain-weighted
series. For Japanese imports, we use real imports from the IMF’s
IFS database.

Hooper, Johnson, and Marquez’s (2000) export series start in the
mid to late 1970s, while ours start in 1981. The omission of the
second period of oil shock years from our sample could be an important source of differences.
16

8

We do not conduct any formal tests for structural breaks or parameter stability. Hooper, Johnson, and Marquez (2000) conduct
parameter stability tests and find that export elasticities are generally more instable than import elasticities.
17

Cardarelli and Rebucci (2007) estimate trade elasticities for the
U.S. after making an adjustment for the value of U.S. exports in
categories of goods that are likely to be parts in a vertically integrated production process. They find that this correction for vertical integration lowers the income elasticity with respect to imports.
18

Data on quarterly imports and exports disaggregated by end-use
category from the U.S. Bureau of Economic Analysis begin in 1967.
19

9

The relative price measure has the same product coverage as our aggregate measure of trade. One disadvantage of this price measure is
that it includes commodities such as oil, which should be perfectly substitutable across locations of production, and thus, the measure is
somewhat inconsistent with the imperfect substitutes model of trade.

20

The product coverage in the export price index of country i coincides with that country’s aggregate measure of exports. This price
measure is designed to incorporate price differences between exported goods from country i and the domestic goods at the export
destination. It fails to incorporate the price of exports of other countries that compete with country i’s exports.

21

10

Federal Reserve Bank of Chicago

The Plaza Accord of 1985 was an agreement among the central
banks of France, Germany, Japan, the UK, and the U.S. to reduce
the value of the dollar through coordinated intervention in currency
markets. The Louvre Accord of 1987 was a similar agreement in
which the same central banks agreed to stop the dollar’s decline.
Differences between table 3 (p. 11) and table 5 (p. 13) are due to
the use of slightly different sample periods.

15

appendix

In this appendix, we discuss the IMF’s real effective exchange rate and then compare the REER with the relative
price of exports to foreign GDP deflators.
The IMF’s real effective exchange rate
The REER for each of the G-7 countries is taken from
the International Monetary Fund, Statistics Department
(2007). It is calculated as:

REERi = log (ULCi × E$/i /ULCFi ),
ULCFi = Π (ULC j /E$/j ) wij , Σwij = 1,
where ULCi is the unit labor cost of the ith country expressed in U.S. dollars and ULCFi is the unit labor cost
for rest of the world expressed in U.S. dollars. Note that
ULCFi is calculated as the geometric mean of the unit
labor cost in country i’s trading partners, adjusted by the
exchange rate. The 20 trading partners, denoted by j, are
selected from the 21 industrialized countries.1 The weights
wij are based on aggregate trade flows for manufactured
goods. An increase in the REER represents a real appreciation of the domestic currency.
According to the International Monetary Fund,
Statistics Department (2007), unit labor costs are compensation of employees per unit of real output (or value
added) in the manufacturing sector. It takes into account
employer-paid social insurance premiums and other employment taxes in addition to wages and salaries. However, the International Monetary Fund, Statistics
Department (2007) also notes that for the most recent quarters, indexes typically refer more narrowly to wages or wages and salaries per unit of total output of manufactured
goods (rather than that of value added in the manufacturing sector).
The total trade weights were chosen to make the
REER index sensitive to movements in costs affecting
Table A1

Percentage of trade accounted for
in the relative price of exports and
the real effective exchange rate
	
	
Canada	
France 	
Germany	
Italy	
Japan	
UK	
U.S.	

Relative	
price	

Real effective	
exchange rate

95	
57	
55	
70	
72	
61	
77	

94
77
73
71
49
79
54

exports and imports of manufactured goods. The
weights, which are built up from aggregate trade flows
for manufactured goods (Standard International Trade
Classification 5–8, or SITC 5–8) and are averaged over
the period 1999–2001, take into account the relative importance of a country’s trading partners in its direct bilateral relations with them, in both the home and foreign
markets; they also take into account the relative importance of the competitive relations with third countries in
particular markets.
Chinn (2004) calls this measure an empirical proxy
for “cost competitiveness” and points out that one of the
drawbacks of this measure as a proxy for cost competitiveness is that it reflects competitiveness in terms of labor cost, and not total cost. Given that we are estimating
trade elasticities of goods and services, the facts that the
weights (wij ) are based on only manufactured goods and
the compensations are from the manufacturing sector are
further drawbacks of this measure.
Comparing the REER with the relative price of
exports to foreign GDP deflators
The REER has more extensive country coverage
than the relative price term for France, Germany, Italy,
and the UK. Table A1 reports the percentage of exports
accounted for by the countries included in the calculation of the REER and the relative price of exports. The
countries included in the REER account for 77 percent
of France’s exports, 73 percent of Germany’s exports,
71 percent of Italy’s exports, and 79 percent of the UK’s
exports.2 The relative price term has a better coverage of
export partners for Canada, Japan, and the U.S.
The weights in the relative price of exports are
based on exports of all goods, while the weights in the
REER are based on aggregate manufacturing trade flows.
This makes the relative price measure a more appropriate price measure in the export equation.
Important emerging markets are left out of the REER.
Countries such as Argentina, Brazil, Hong Kong, South
Korea, Mexico, Singapore, and Taiwan are included in
the relative price of exports, but not in the REER. Hooper,
Johnson, and Marquez (2000) report that including developing countries in the relative price and foreign GDP
measure affects the estimated trade elasticities.
The 21 countries included in the calculation of the REER are
Australia, Austria, Belgium, Canada, Denmark, Finland, France,
Germany, Greece, Ireland, Italy, Japan, the Netherlands, New
Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the UK,
and the U.S.
1

The trade shares are the authors’ own estimates using information
from the IMF on countries included in the measure. Trade shares
are calculated using 2000 bilateral trade of goods from the OECD’s
STAN Bilateral Trade Database.
2

Sources: Authors’ calculations and Haver Analytics.

16

4Q/2007, Economic Perspectives

REFERENCES

Bernanke, Ben S., 2005, “The global saving glut
and the U.S. current account deficit,” remarks at the
Sandridge Lecture, Virginia Association of Economics,
Richmond, VA, March 10.
Cardarelli, R., and A. Rebucci, 2007, “Exchange
rates and the adjustment of external imbalances,”
in World Economic Outlook, Washington, DC:
International Monetary Fund, April, pp. 81–120.
Chinn, M. D., 2004, “Incomes, exchange rates, and
the U.S. trade deficit, once again,” International
Finance, Vol. 7, No. 3, December, pp. 451–469.
Ferguson, Roger W., Jr., 2005, “U.S. current account
deficit: Causes and consequences,” remarks to the
Economics Club of the University of North Carolina
at Chapel Hill, Chapel Hill, NC, April 20.
Gagnon, J. E., 2004, “Productive capacity, product
varieties, and the elasticities approach to the trade
balance,” International Finance Discussion Papers,
Board of Governors of the Federal Reserve System,
working paper, No. 781.
Goldstein, M., and M. S. Khan, 1985, “Income
and price effects in foreign trade,” in Handbook of
International Economics, Vol. 2, Ronald W. Jones
and Peter B. Kenen (eds.), Amsterdam: Elsevier,
pp. 1041–1105.
Gould, D., 1994, “Immigrant link to the home country: Empirical implications for U.S. bilateral trade
flows,” Review of Economics and Statistics, Vol. 76,
No. 2, May, pp. 302–316.
Hooper, P., K. Johnson, and J. Marquez, 2000,
“Trade elasticities for the G-7 countries,” Princeton
Studies in International Economics, Princeton
University, Department of Economics, monograph,
No. 87, August.

Houthakker, H. S., and S. P. Magee, 1969, “Income
and price elasticities in world trade,” Review
of Economics and Statistics, Vol. 51, No. 2, May.
International Monetary Fund, Statistics Department,
2007, International Financial Statistics, Washington,
DC, April, CD-ROM.
Johansen, S., 1988, “Statistical analysis of cointegration vectors,” Journal of Economic Dynamics and
Control, Vol. 12, No. 2–3, pp. 231–254.
Johnson, H. G., 1958, International Trade and
Economic Growth, Cambridge, MA: Harvard
University Press.
Krugman, P., 1989, “Differences in income elasticities
and trends in real exchange rates,” European Economic
Review, Vol. 33, No. 5, May, pp. 1031–1046.
Mann, C. L., 2002, “Perspectives on the U.S. current
account deficit and sustainability,” Journal of Economic
Perspectives, Vol. 16, No. 3, Summer, pp. 131–152.
Marquez, J., 2002, Estimating Trade Elasticities, Boston:
Kluwer Academic Publishers.
Pluyaud, B., 2006, “Modeling imports and exports
of goods in France, distinguishing between intra and
extra euro area trade,” in Convergence or Divergence
in Europe? Growth and Business Cycles in France,
Germany, and Italy, Olivier de Bandt, Heinz Herrmann,
and Giuseppe Parigi (eds.), Berlin and Heidelberg,
Germany: Springer, pp. 325–359.
Wren-Lewis, S., and R. L. Driver, 1998, Real
Exchange Rates for the Year 2000, Policy Analyses
in International Economics, No. 54, Washington, DC:
Institute for International Economics.

__________, 1998, “Trade elasticities for G-7 countries,” International Finance Discussion Papers,
Board of Governors of the Federal Reserve System,
working paper, No. 609, April.

Federal Reserve Bank of Chicago

17

Evidence on entrepreneurs in the United States:
Data from the 1989–2004 Survey of Consumer Finances
Mariacristina De Nardi, Phil Doctor, and Spencer D. Krane

Introduction and summary
A country’s national saving rate is a crucial determinant
of its ability to accumulate capital and generate growth;
hence, it is an important determinant of the country’s
future prosperity. Another important determinant of a
country’s prosperity is innovation—the ability to generate new goods and services and provide existing ones
in a more efficient manner. Accordingly, it is vital to
study the households that are savers, as well as the
managers of the businesses that are innovators. In
this article, we consider the behavior of a group of individuals who play both roles in the U.S. economy—
entrepreneurs.
First, entrepreneurs accumulate capital. As noted
by Quadrini (1999, 2000), on average, entrepreneurs
save a good deal more than other households: Even
though households headed by entrepreneurs make up
only 7–8 percent of the population, they own nearly
one-third of the wealth in the United States.
Second, entrepreneurial risk-taking is thought to
be an important way that individuals with skills, ideas,
and business savvy introduce new products, technologies, and business strategies into the economy. This is
the entrepreneur described by economists such as
Schumpeter (1934) and Knight (1921). Such individuals are willing to put their financial well-being on the
line in risky business ventures, with the expectation
of earning large returns and expanding their wealth.
A third important feature of entrepreneurship is
that the business owner’s personal skills and financial
resources are much more closely linked to the operations and performance of the firm than is the case with
owners of widely held publicly traded corporations.
Most shareholders of public firms have little say in
the management of the business; and because their risk
exposure is limited to the value of their shares, their
personal portfolios are irrelevant to the assessment of
the firm’s creditworthiness. In contrast, entrepreneurs’

18

relationships with their businesses are anything but at
arm’s length. As managers, they make all of the dayto-day decisions about the firms’ operations. As owners,
they reap most of the rewards of success, but in many
cases, their personal assets help finance the business and
are at risk if the business fails. This means that there
is a fundamental and bi-directional link between entrepreneurial households’ portfolios and the performance
of their businesses.
In this article, we present a number of stylized
facts that can help us understand the roles that entrepreneurs play in the U.S. economy. First, we construct
an empirical counterpart to Schumpeter and Knight’s
notion of the entrepreneur in the context of the information collected by the Federal Reserve Board’s
Survey of Consumer Finances (SCF).
Second, we use the SCF data to document a number of facts about entrepreneurs and the businesses
that they run. We show that entrepreneurs, as a group,
are very rich. They account for about 30 percent of
the households in the top decile of the wealth distribution in the United States. They also earn more income than others, though the disparity is not as great
as it is for wealth. They hold about as much total net
wealth relative to their income as other rich people.
Entrepreneurs also fall into two demographic categories that have more wealth than the population as a
whole; that is, they are more educated and less likely
to be a minority than the general population. Looking
at their businesses, we see entrepreneurs operating in

Mariacristina De Nardi is a senior economist, Phil Doctor
is a senior associate economist, and Spencer D. Krane is
a vice president and economic advisor in the Economic
Research Department at the Federal Reserve Bank of
Chicago. The authors thank Gene Amromin, Marco
Bassetto, Craig Furfine, and Kevin Moore for helpful
comments.

4Q/2007, Economic Perspectives

a wide range of industries. We also find large changes
in the legal organization of their firms over time, with
more of them being organized as less risky limited liability entities.
Third, we shed light on how entrepreneurs’ success
in business affects their personal wealth. The vast majority of entrepreneurs start their own businesses as opposed
to buying or inheriting them—an indication that the
businesses’ performance reflects the entrepreneurs’
personal skills. We also find a great deal of dispersion
in measures of the businesses’ success, such as firm
size, income, and rates of return, and we find that rich
entrepreneurs with large businesses have owned their
firms for longer than others. In combination, these
facts indicate that entrepreneurs can face very large
risks in their business ventures, but successful ones
can earn huge returns and become very wealthy.
Fourth, we turn the tables and look at the role of
entrepreneurial wealth in supporting the operations of
the business. Some households and firms are liquidityconstrained—that is, they cannot borrow as much as
they would like to or can only borrow at higher interest
rates than other market participants. Liquidity constraints
cause households to save more in order to finance large
expected expenditures or to have a precautionary buffer to insure against unexpected shocks. We find evidence that entrepreneurs face liquidity constraints:
Many report having been turned down for credit, and
many provide personal loans to their businesses or
pledge personal assets as collateral to secure loans for
their firms. But we also present some evidence that
these borrowing constraints may have fallen over time.
Finally, we document that entrepreneurial households appear to be less risk averse than other comparable households. Despite large variation in the returns
to their business ventures, relative to other rich households, they do not have higher net worth (relative to
income), they tie up a large portion of their wealth in
their business ventures, and they carry much more
debt. In addition, entrepreneurs’ responses to attitudinal questions suggest that they are more willing to
take risks in order to achieve high financial returns.
Who are the entrepreneurs and
how do we measure them?
It is difficult to define what an entrepreneur is and
to determine the best empirical counterpart to that definition (see Gentry and Hubbard, 2004). We think of
entrepreneurs as people who actively manage their
own businesses and invest their own wealth in them.
Our entrepreneurs are not simply managers, because
managers may not have risked a personal investment
stake in the firm. Nor are our entrepreneurs simply

Federal Reserve Bank of Chicago

investors, because investors may not have a key active
role in the decision-making of the firm. Finally, our
entrepreneurs are not people simply working on their own
because they can’t find a suitable job at another firm.
Our unit of observation is the household. We classify entrepreneurs as those households in which the head
declares being self-employed as a primary job, owning a business (or a share of one), and having an active
management role in the firm. We refer to these households as self-employed business owners, or SEBs.1
By requiring the respondent to be self-employed,
we exclude people who have a full-time wage-earning job and are running a business as a hobby. By requiring that the entrepreneur manages the firm, we help
reduce the reverse causation between business ownership and wealth—that is, we likely eliminate many
people who simply are rich and acquire a business as
a passive investment. Finally, by requiring that the
entrepreneur has an investment stake in the business,
we also likely exclude those who are self-employed only
because their outside wage opportunities are very poor.
We investigate the characteristics of the SEBs
using six waves of the Survey of Consumer Finances:
1989, 1992, 1995, 1998, 2001, and 2004.2 The SCF is
well suited for this exercise. First, one of the primary
purposes of the SCF is to measure household balance
sheets: Accordingly, the survey has been designed to
capture comprehensive information about the value
of assets and liabilities as accurately as possible. Second, since the distribution of wealth is highly skewed
in the U.S., with a small fraction of the population holding much of the wealth, the SCF oversamples wealthy
households in order to better measure economy-wide
aggregates. Because SEBs are disproportionately wealthy,
the SCF will sample more SEBs than other surveys
that sample according to standard demographic probability weights. Each of the six SCF waves we consider contains between 565 and 930 SEBs; this is a large
number when compared with other data sets, such as
the University of Michigan’s Panel Study of Income
Dynamics, or PSID. (See the appendix for details.)
Third, in addition to wealth, the survey contains
a wide range of detailed information on income, employment, borrowing activity, business history, and
demographic characteristics.3 It also records respondents’ subjective attitudes toward saving and risk.
One drawback of the SCF is that it is a series of
cross-sectional surveys as opposed to a panel that follows individual households over time. Accordingly,
we do not have data on the flow of savings. Instead,
following standard practice, we study the wealth of
the household, which is the accumulated flow of its
past savings (where savings are broadly measured to

19

include capital gains and losses) plus any inherited
wealth. The lack of a panel also prevents us from analyzing linkages between entrepreneurial behavior,
business failure, and the process of taking successful
firms public. This means our analysis is susceptible to
two types of “survivor bias.” The first bias is that our
sample does not allow us to identify individuals who
are no longer entrepreneurs because their past businesses ventures failed. The second bias is that we do
not capture those entrepreneurs who have successfully taken their businesses public. This first bias would
leave our SEB sample with households that are more
successful than the entire population of people who
have undertaken entrepreneurial activity; the second
would push us toward underestimating entrepreneurial
performance. While there is no way to quantify these
biases with our data, the first is likely more relevant
for younger and smaller businesses, while the second
is likely more relevant for older and larger firms.
Entrepreneurs are rich and what else?
Table 1 displays data for total household net worth
and total pretax household income of SEBs and compares them with other households in the U.S. We find
that SEBs are wealthier than other households.4 They
own about one-third of the household wealth in the
U.S., and the median net worth of SEB households
varies between four times to six and a half times greater

than the median net worth of non-SEBs. The median
wealth of SEBs runs between $260,000 and $540,000
(in 2004 dollars). On average, about 55 percent of
SEBs fall in the top wealth quintile, and SEBs make
up about 20 percent of the total number of households
in the top 20 percent of the wealth distribution in the
U.S.5 There is no systematic cyclical variation or time
trend in the allocation of SEBs over the aggregate
wealth distribution during the 1989–2004 period.
The SEBs also earn more income than other households, but the difference is not at dramatic as it is for
wealth. The median income of SEBs is about two times
greater than the median income of other households,
and they earn between 15 percent and 20 percent of
total pretax household income as measured by the SCF.
As with wealth, there is no clear trend or cyclical pattern to the share of income earned by SEBs.
As noted by Quadrini (1999, 2000), higher income
does not account for the higher share of wealth held
by the SEBs. As seen in the bottom half of table 1,
the wealth-to-income ratios of SEBs are a good deal
higher than the ratios for other households. For the
non-SEB households, the median wealth-to-income
ratio was about 1.5 in the late 1980s and early 1990s,
and then rose to 1.9 in 2004. For the SEBs, the ratio
varied between 4.7 and 5.7 over the first four SCF
waves we consider and then rose to 6.5 in 2004. Clearly,
because SEBs own businesses, a good deal of the

Table 1

Wealth and income of SEBs and others
	
	

SEB percentage of total
  Wealth 	
  Income	
	
Median wealth
(2004 dollars)	
  SEBs	
  Others	
	
Median income
(2004 dollars)	
  SEBs	
  Others	
	
Median ratio of
wealth to income	
  SEBs	
  Others	
	
Median ratio of wealth
(excluding business
net equity) to income	
  SEBs	
  Others	

	

	

	

	

1989	

1992	

1995	

1998	

2001	

2004

33.1	
21.3	
	

31.3	
17.3	
	

29.2	
14.0	
	

30.6	
17.9	
	

30.8	
17.9	
	

31.9
18.3

	
388,704	
60,139	
	

	
274,726	
56,399	
	

	
262,421	
63,606	
	

	
384,235	
72,318	
	

	
487,850	
79,728	
	

536,000
80,600

	
68,841	
36,715	
	

	
67,475	
33,738	
	

	
52,927	
35,285	
	

	
70,492	
37,596	
	

	
85,403	
40,511	
	

79,069
41,075

	
5.7	
1.5	
	

	
4.8	
1.5	
	

	
5.1	
1.6	
	

	
4.7	
1.8	
	

	
6.2	
1.8	
	

6.5
1.9

	
2.6	
1.5	

	
2.3	
1.5	

	
2.7	
1.5	

2.6	
1.7	

3.4	
1.7	

3.5
1.8

Note: SEB means self-employed business owner.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

20

4Q/2007, Economic Perspectives

difference between wealth-to-income ratios reflects
business assets. Overall, the median SEB household
holds about one-third of its wealth as net equity in its
business ventures. Non-SEB households hold very little
business wealth. Even excluding business net equity,
however, SEB households are wealthier than non-SEB
households. The last two rows of table 1 show that
excluding business net equity, SEBs still hold one and
a half to two times more wealth relative to income
than other households.
Are the self-employed business owners different
from the other rich households?
Comparing median income, wealth, and wealthto-income ratios of all SEBs with all other households
is of limited value because SEBs fall disproportionately in the upper wealth categories. For a more useful comparison, therefore, table 2 compares the wealth
and income of SEBs with those of only the households
in the top decile of the wealth distribution. Thirty-six
percent to 43 percent of SEBs are in this category
(row 1), and they account for between one-quarter
and one-third of the total number of households in
this portion of the wealth distribution (row 3). Even
in this narrower category, SEBs are concentrated in
the upper ends of the wealth and income distributions—
their median wealth is 30–47 percent more than
the others, and their median income ranges between
18 percent and 66 percent more.

Although they are wealthier and earn more income,
as seen in the bottom two rows of table 2, there is no
systematic difference between wealthy SEBs and other
rich households in the ratio of total household wealth
to income. Note, though, that while the wealth-toincome ratio of non-SEBs has shown a fairly steady
uptrend since 1989, there has been more noticeable variation in the ratios for SEBs. This reflects volatility in
the business net equity portions of the SEBs’ portfolios.
Demographics
Table 3 describes the demographic characteristics
of SEBs. As we noted previously, SEBs account for
between 7 percent and 8 percent of the households in
the U.S. There are no trends in the share of SEBs in
the population over the 1989–2004 period. Some commentators and researchers have postulated that the share
of self-employed should be countercyclical, as some
workers who cannot find work for pay when job prospects are poor turn to self-employment. We do not find
much evidence for such behavior among SEBs.6
There is no systematic difference in the ages of
the heads of SEB households and other households
in the SCF. This means that there are no life-cycle
reasons for SEBs to have accumulated more wealth
than other households. However, SEBs do fall disproportionately into two other demographic groups that
have higher-than-average saving rates. First, SEBs
are more educated than the rest of the population.

Table 2

Wealth and income for the top decile of the wealth distribution
	
	
Percentage of SEBs
in top decile 	
Percentage of others
in top decile 	
	
Percentage of top
decile who are SEBs	
	
Median wealth
(2004 dollars)	
  SEBs	
  Others	
	
Median income
(2004 dollars)	
  SEBs	
  Others	
	
Median ratio of
wealth to income	
  SEBs	
  Others	

	

1989	

1992	
	

1995	

2001	
	

2004

42.8	

37.9	

35.9	

39.2	

37.1	

37.6

7.2	
	

7.5	
	

8.1	
	

7.7	
	

7.7	
	

7.8

32.7	
	

30.9	
	

24.0	
	

29.2	
	

28.8	
	

28.1

	
1,215,984	
923,723	
	

	
1,038,488	
752,902	
	

	
1,100,371	
765,240	
	

	
1,277,309	
980,487	
	

	
1,848,665	
1,279,169	
	

1,956,500
1,331,200

	
140,742	
110,146	
	

	
125,504	
94,465	
	

	
114,676	
97,033	
	

	
140,984	
92,815	
	

	
206,937	
125,914	
	

164,300
133,494

	
10.5	
9.7	

	
8.8	
9.6	

	
10.4	
9.0	

	
9.9	
11.0	

	
11.8	
11.9	

	

1998	

	

13.9
12.3

Note: SEB means self-employed business owner.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

Federal Reserve Bank of Chicago

21

Table 3

Demographic characteristics of SEBs and others
	

1989	

1992	

1995	

1998	

2001	

2004

SEB percentage
of households	
	
Average age	
  SEBs	
  Others	
	
Average education levela	
  SEBs	
  Others	
	
Percentage of minorityb	
  SEBs	
  Others	

7.6	
	
	
45.4	
48.1	
	
	
2.8	
2.4	
	
	
9.5	
26.5	

8.1	
	
	
46.9	
48.6	
	
	
3.1	
2.6	
	
	
13.3	
25.7	

6.7	
	
	
47.7	
48.5	
	
	
3.0	
2.6	
	
	
10.6	
23.3	

7.4	
	
	
47.6	
48.8	
	
	
3.1	
2.7	
	
	
7.7	
23.4	

7.8	
	
	
49.7	
48.9	
	
	
3.1	
2.7	
	
	
8.0	
25.1	

7.5
50.0
49.5
3.1
2.7
10.8
27.7

Index: 1 corresponds to no high school degree; 2 to high school degree only; 3 to some college; and 4 to college degree.
Percentage other than white.
Note: SEB means self-employed business owner.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.
a
b

The SCF records education according to four categories: no high school degree, high school degree but no
college, some college, and college degree or more.
Table 3 shows an index from these responses.7 On average, SEBs score about 0.4 points higher than non-SEBs.
The fraction of SEBs with a college degree is 15–20
percent higher than the fraction for the rest of the population. Similar percentages of SEBs and others have some
college education, and the fraction of SEBs with a high
school degree or less is 15–20 percent lower than the
rest of the population. Second, the share of SEBs who
are minorities is well below the share of non-SEB
households headed by a minority. As documented by
Altonji, Doraszelski, and Segal (2000) and Altonji
and Doraszelski (2005), higher educated households
have higher-than-average saving rates, while minority
households tend to save at less-than-average rates.
What kinds of businesses do self-employed
business owners run?
Self-employed business owners are found in many
different industries. The SCF asks detailed questions
regarding the type of business the SEB manages; since
1995, however, the public use data sets have aggregated
the responses into categories that are too broad to discern much about the industrial composition of the SEBs.
Still, the detailed data for 1989 and 1992 reveal some
interesting facts. As seen in table 4, on average in those
surveys, the largest single category of SEBs was in
professional practices, such as law, medicine, and accounting; these represented a little over 16 percent of
the SEB households. The next largest categories were
contracting and construction, farming and agricultural

22

services, and general retail and wholesaling—each with
between 12 percent and 15 percent. Real estate and
insurance, manufacturing, business services, and personal services have shares from 4.5 percent to 7 percent.
While some of these occupations and industries might
not be the first that would come to mind when we think
about entrepreneurship, any of them could include individuals testing the market with some combination
of innovative technologies, new products, or novel
business practices.
Legal structure
The SEBs can choose a variety of legal structures
for their businesses. The firms can be sole proprietorships or partnerships, in which there is little legal distinction between personal assets and liabilities and
those of the business operation. They also can be limited
liability companies (LLCs), closely held C corporations,
or S corporations.8 These entities are subject to more
legal restrictions on governance and management than
proprietorships and partnerships, but have the advantage of sheltering personal assets from the firm’s
creditors. They may also provide certain fringe benefits tax free to owners. Furthermore, distributions of
LLCs and S corporations are treated as personal income and thus avoid double taxation of dividends.
As seen in table 5, the legal structure of SEBs
has changed over time. The percentages of respondents whose businesses are sole proprietorships or
partnerships have trended down, while the percentages of LLCs, C corporations, and S corporations have
moved up. These trends reflect a variety of factors.
Limited liability companies were first introduced in

4Q/2007, Economic Perspectives

SEBs acquired their businesses and how
long they have owned and managed their
firms. In order to link the business operation to the creation of wealth, we then
consider the relationships between ownership tenure, the size of the firm,
and the income and rates of return that it
generates. These results will also be useful for our discussion of credit constraints
and risk aversion later in this article.

Table 4

SEBs, by type of business
	

Percent

Professional practice (law, medicine, etc.)	
Contracting/construction	
Farm, agricultural services, and landscaping	
General retail/wholesale	
Real estate/insurance	
Manufacturing	
Other business services	
Personal services	
Business management/consulting	
Restaurant/bar	
Food/liquor	

16.2
15.2
13.3
12.1
6.7
6.1
5.6
4.5
2.2
2.1
1.4

Business acquisition and
ownership tenure
Notes: SEB means self-employed business owner. The values here are derived 	
Table 6 presents some more facts
from the pooled 1989 and 1992 waves of the Survey of Consumer Finances. 	
The column does not total because there are other types of businesses owned 	
about SEB businesses. One indication
by SEBs that are not reported here.
that our SEB classification is capturing
Source: Authors’ calculations based on data from the Board of Governors 	
of the Federal Reserve System, Survey of Consumer Finances.
entrepreneurial activity is the fact that between 65 percent and 79 percent of the
SEBs started their own business as opposed to buying an existing operation, being promotthe U.S. by Wyoming in 1977, and it took some time
ed to ownership/management status, or inheriting the
for other states to offer this option. The favorable tax
firm or receiving it as a gift. Personally starting an
treatment for distributions is relatively new—1986 for
operation likely takes more initiative and is more
S corporations and 1988 for LLCs. Furthermore, some
amenable to the incorporation of new ideas than taklegal restrictions on ownership and operation have
ing over an existing operation. Furthermore, the share
changed over time; notably, they were relaxed for
of SEBs starting their own businesses has risen over
S corporations in 1997. Finally, there have been several
time—a fact that we will return to later when discusschanges in the relative tax rates on corporations, indiing credit constraints on entrepreneurial borrowing.
viduals, and self-employment over the 1989–2004
The SCF asks SEBs how long they have owned
period. On net, however, the increased attractiveness
and operated their businesses. We will refer to the reof the limited liability structures appears to have only
sponse to this question as ownership tenure. If the SEB
shifted the share of SEBs taking on these legal strucalso started the firm, then ownership tenure would cortures and not produced an uptrend in the overall share
respond to firm age. If instead the SEB acquired an
of SEBs in the population.
existing firm through purchase, promotion, inheritance,
Are self-employed business owners wealth
or gift, then the tenure would underestimate the firm’s
creators?
age. Of course, since most of the businesses we observe
were started by the SEB, ownership tenure equals
Do SEBs create wealth by operating successful
firm age for most of our sample.
business ventures? Or are they primarily rich people
Between 9 percent and 15 percent of the businesses
who buy existing successful business operations? In
we observe were formed or acquired within the current or
order to address this issue, we first look at how the
Table 5

SEBs, by legal structure of business
	

1989	
1992	
			
Sole proprietorship	
Partnership	
LLC, C corporation, or other	
S corporation	

54.6	
22.4	
12.9	
10.1	

58.8	
15.2	
14.7	
11.3	

1995	

1998	

2001	

2004

53.9	
14.8	
15.8	
15.5	

51.4	
11.6	
17.4	
19.7	

47.3
12.1
22.4
18.2

(percent)
56.5	
17.7	
11.2	
14.6	

Notes: SEB means self-employed business owner. LLC means limited liability company. Columns may not total because of rounding. 	
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

Federal Reserve Bank of Chicago

23

previous year of the SCF sampling date, about onequarter two years to five years prior to the sampling
date, and about 60 percent at least six years prior to
the date. Accordingly, our data set likely contains a
fairly good mix of start-ups and established firms. There
is no trend in these tenure shares. However, there is a
noticeable increase in the share of businesses that were
started or acquired one year ago or less in 1992 and a
marked drop in this share in 2001. Both of the 1992
and 2001 SCF waves are capturing entrepreneurial
activity during periods of macroeconomic weakness.
Thus, the data provide conflicting readings on whether small business formation serves as a substitute for
work for pay during periods of economic weakness.
A somewhat larger share—between 23 percent and
27 percent—of wealthy SEBs (those in the top decile
of the wealth distribution) purchased their businesses
(panel B of table 6). Still, in every year we consider,
about two-thirds of the rich SEBs started their businesses themselves. So it is likely that much of the business wealth accumulated by the rich SEBs reflects their
efforts to start and nurture a successful business plan.
Since rich SEBs hold a good deal more business
wealth and older firms have more net equity, it is likely
the case that wealthy SEBs own older operations. This is
indeed true. Panel B of table 6 presents the tenure of
the businesses managed by SEBs in the top decile of
the wealth distribution. Only between 6 percent and

8 percent of these firms fall in the one year or less category—somewhat less than the 9 percent to 15 percent
for the SEB population as a whole. And between
73 percent and 80 percent are in the six years or more
year classification, about 10–20 percentage points
more than for the overall SEB population.
Business net equity
The SCF calculates business net worth by taking
the respondent’s assessment of the net value that the
household would receive if the business were sold
today plus the net of any lending or collateral provision from the personal accounts of the household to
the business ledger. (See the appendix for more details.9) Figure 1 shows histograms of the net equity
in the businesses that entrepreneurs actively manage.
We have combined all six waves of the SCF, recorded
net equity in 2004 dollars, and combined all businesses
valued at $1 million or more in the rightmost bar of
the histograms. For reference, table 7 shows the businesses’ median net equity by SCF year.
Panel A of figure 1 plots net equity for all firms
and shows that there is large variation and skewness
in business net equity. Most firms are small. The leftmost bar of the histogram corresponds with business
net worth between –$2,200 and $22,000, and indicates
that 27 percent of firms fall in this category.10 Median
business net equity is a bit under $100,000. However,

Table 6

Characteristics of businesses
	
1989	
1992	
1995	
1998	
2001	
			
(percent)
A. All SEBs						
Acquisition method	
	
	
	
	
	
  Started	
65.3	
72.4	
71.4	
73.9	
78.5	
  Purchased	
24.7	
23.0	
21.3	
20.4	
17.5	
  Joined/promoted	
0.7	
0.0	
1.4	
1.9	
1.5	
  Inherited/gift 	
9.4	
4.6	
5.9	
3.8	
2.5	
	
	
	
	
	
	
Ownership tenure	
	
	
	
	
	
  0–1 year	
13.2	
15.2	
13.7	
12.6	
9.2	
  2–5 years	
27.3	
21.0	
24.8	
26.9	
23.4	
  6 or more years	
59.5	
63.8	
61.5	
60.6	
67.4	
	
	
	
	
	
	
B. Wealthy SEBs						
Acquisition method	
	
	
	
	
	
  Started	
59.2	
65.1	
63.1	
69.7	
68.8	
  Purchased	
25.9	
26.5	
26.7	
23.3	
27.4	
  Joined/promoted	
0.3	
0.0	
1.3	
2.8	
1.0	
  Inherited/gift 	
14.6	
8.4	
9.0	
4.2	
2.9	
	
	
	
	
	
	
Ownership tenure	
	
	
	
	
	
  0–1 year	
8.2	
5.5	
5.7	
6.0	
6.2	
  2–5 years	
19.1	
16.2	
15.2	
14.4	
17.4	
  6 or more years	
72.7	
78.3	
79.1	
79.7	
76.4	

2004

77.9
17.3
2.8
2.0
12.6
25.6
61.8

67.1
25.2
3.2
4.5
6.2
18.0
75.8

Notes: SEB means self-employed business owner. Wealthy SEBs are those in the top decile of the wealth distribution. Columns may not total 	
because of rounding.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

24

4Q/2007, Economic Perspectives

many entrepreneurs own firms worth a good deal more;
about one-third own firms valued at greater than
$275,000, and 11 percent—the rightmost bar in the histogram—own operations valued at $1 million or more.
The net equity in the business varies substantially
by how long the SEB has owned the business. Panels B,
C, and D of figure 1 plot net equity histograms by length
of ownership tenure. As one would expect, young firms
are disproportionately small: 16 percent of firms with
tenure of one year or less have zero net equity, and
39 percent have net equity of less than $22,000 (the
leftmost hand bar in the histogram). Still, there are a
few medium-sized young firms, with 10 percent of
them having net equity above $365,000. Older operations are much bigger; the median net equity in businesses with ownership tenure of six years or more is
about $135,000, and nearly 15 percent have net equity greater than or equal to $1 million.

Business income
We study two measures of pretax business income.
(See the appendix for details.) The first reflects profits
from operations or net equity extracted during a given
SCF year. We call it “business income cash flow,” or
BCF, and compute it as wage and salary income and
other distributions drawn from the business by the
entrepreneur and his or her spouse.
The second income measure accounts for unrealized capital gains. The SEB may extract only a fraction of the profits earned by the firm, leaving the rest
in the business to support and expand its operations.
In addition, some of the SEB’s efforts may not increase
current cash flow, but still raise the value of the firm
by enhancing its ability to generate cash flow in the
future. Following Moore (2004), we compute our second
measure of business income by adding an estimate of
the average annual unrealized capital gains earned by

figure 1

Distribution of business net equity, by ownership tenure
A. All tenures

B. 0–1 year

percentage of SEBs

percentage of SEBs

40

40

30

30

20

20

10

10
0

0
–76

47

169 291 413 536 658 780 902 1,024

–76

47

169 291 413 536 658 780 902 1,024

thousands of dollars

thousands of dollars

C. 2–5 years

D. 6 or more years

percentage of SEBs

percentage of SEBs

40

40

30

30

20

20

10

10
0

0
–76

47

169 291 413 536 658 780 902 1,024

thousands of dollars

–76

47

169 291 413 536 658 780 902 1,024

thousands of dollars

Notes: SEB means self-employed business owner. All dollar values are in 2004 dollars.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

Federal Reserve Bank of Chicago

25

the SEB since acquiring the firm to business income
cash flow.11 We compute average unrealized capital
gains for the SEB by taking the difference between
current firm value and the acquisition costs of the firm
divided by business tenure. The acquisition cost is the
tax cost basis the owner would declare if he sold the
business today. We call this measure BCF–CG for the
ratio of business income to cash flow including unrealized capital gains. The difference between the two
income measures is hence an estimate of how much
the SEB has been able to boost the net value of the firm
through retained earnings, debt reduction, and successful entrepreneurial effort that increases the value of
the firm.
Figure 2 plots histograms of BCF that combine
all six waves of the SCF that we use. The histograms
combine all firms with earnings greater than or equal
to $513,000 and those with a loss in excess of $10,000
in the rightmost and leftmost bars, respectively. Table 7
presents firms’ median income by SCF year.
As with net equity, there is a large variation in
income, and the distribution is skewed: 30 percent of
the firms make less than $25,000, but 30 percent make
more than $81,000, and 10 percent more than $200,000.
While there are some differences in the distribution
of income by ownership tenure, they are not as great
as the disparities in net equity. For example, the median income of firms with ownership tenure of one
year or less is $36,000, while the median income of
businesses with tenure of six years or more is $50,000.
There are some differences in the tails of the distributions. About 10 percent of the newer firms experience
losses, while fewer than 5 percent of the longest-held
ones do; 10 percent of the firms with the shortest tenure

earn $130,000 or more, while 20 percent of the firms
with the longest tenure have such income.
Figure 3 and table 7 show BCF–CG. Unrealized
capital gains significantly boost the business income
for SEBs. Over all years, the median BCF–CG measure of income is $56,800, compared with $47,000
for the median BCF measure. As noted by Moore
(2004), the increases are sizable across the entire income distribution. Still, the BCF–CG distribution is
more skewed than the BCF distribution. For example,
the 30th percentile of the BCF–CG distribution is
$30,450, compared with $25,000 for that of BCF; the
70th BCF–CG percentile is $104,900 versus $81,000
for that of BCF; and the 90th BCF–CG percentile is
$287,000—almost one and a half times greater than
that of BCF. The increase in the median income of
firms with ownership tenure of one year or less is
large: It rises from $36,000 to $50,000.12 The gains
(in percentage terms) are smaller for more established
operations—from $45,000 to $59,400 for firms with
tenure of two years to five years and from $50,000 to
$56,700 for firms with tenure of six years or more.
Rates of return
We also compute two measures of the rate of return
from running one’s business. The first measure, which
we call RBCF, divides BCF by business net equity. It
measures the return extracted from the business by the
entrepreneur in the year covered by the SCF. The second measure, which we call RBCF–CG, replaces BCF
in the numerator with BCF–CG. Thus, it is a more comprehensive measure that includes both extracted returns
and the unrealized increase in value due to retained
earnings, debt reduction, and entrepreneurial labor.

Table 7

Financial returns from the business
	
Median business wealth
net equity (2004 dollars)	
	
Median income (2004 dollars)	
  BCF a 	
  BCF–CGb	
	
Median rate of return (percent) 	
  RBCF c	
  RBCF–CGd	

1989	

1992	

1995	

1998	

95,343	
	
	
35,000	
40,500	
	
	
21.1	
32.3	

78,990	
	
	
42,031	
49,663	
	
	
28.6	
37.6	

61,514	
	
	
37,018	
47,104	
	
	
36.2	
46.4	

92,685	
	
	
52,999	
60,200	
	
	
39.0	
50.7	

2001	

2004

106,518	
	
	
55,999	
70,752	
	
	
33.5	
45.5	

144,000
59,028
79,000
30.0
45.8

Business income plus wages and salaries of household heads and spouses working for the business as a percent of total household income.
Business income cash flow including unrealized capital gains.
c
Business income divided by business net equity.
d
Business income cash flow including unrealized capital gains divided by business net equity.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.
a
b

26

4Q/2007, Economic Perspectives

figure 2

Distribution of business income cash flow (BCF), by ownership tenure
A. All tenures

B. 0–1 year

percentage of SEBs

percentage of SEBs

16

16

12

12

8

8

4

4
0

0
3

67

130

194

258

322

385

449

513

3

67

130

194

258

322

385

449

513

385

449

513

thousands of dollars

thousands of dollars

C. 2–5 years

D. 6 or more years

percentage of SEBs

percentage of SEBs

16

16

12

12

8

8

4

4

0

0
3

67

130

194

258

322

385

449

513

thousands of dollars

3

67

130

194

258

322

thousands of dollars

Notes: SEB means self-employed business owner. All dollar values are in 2004 dollars. Because of the scaling used in this figure, 	
the leftmost bar, which represents businesses with a loss in excess of $10,000, is not clearly visible in each panel.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

Figure 4 reports data on RBCF, while figure 5
(on p. 30) reports data on RBCF–CG. These figures
display histograms of the rate of return on the businesses by ownership tenure for the pooled surveys.
In both figures, we have combined the rates of return
of 250 percent or more and all negative rates of return
in the rightmost and leftmost bars, respectively.13
Table 7 shows the businesses’ medians of RBCF and
RBCF–CG for each SCF year.
With regard to the cash flow measure, the median
RBCF over all years and SEBs is 33 percent. If this
were simply a one-year return to capital, then it would
be a very high rate—after all, this measure excludes
capital gains, and for comparison, the annual nominal
rate of return on the Standard & Poor’s 500 stock index was about 8.5 percent over the 1989–2004 period.

Federal Reserve Bank of Chicago

However, the SEBs’ return includes the return to the
entrepreneurs’ labor efforts, and so it should exceed
the return to capital by a good deal.
The median RBCF masks a wide range of outcomes. About 5 percent of the SEBs lose money, and
over one-third of them earn less than a 15 percent return. But many of them earn very large rates of return—
30 percent of them earn more than an 80 percent rate
of return. So clearly there is a good deal of financial
risk involved with the business ventures of SEBs.
Much of this risk-taking appears to be undertaken by
smaller, younger businesses. As seen in figure 4, the
dispersion of returns for firms with the shortest ownership tenure is much greater than that for older firms.14
Among firms with ownership tenure of one year or less,
9.4 percent experience losses and another 8.7 percent

27

figure 3

Distribution of business income including unrealized capital gains (BCF–CG), by ownership tenure
A. All tenures

B. 0–1 year

percentage of SEBs

percentage of SEBs

16

16

12

12

8

8

4

4

0

0
3

67

130

194

258

322

385

449

513

3

67

130

194

258

322

385

449

513

385

449

513

thousands of dollars

thousands of dollars

C. 2–5 years

D. 6 or more years

percentage of SEBs

percentage of SEBs

16

16

12

12

8

8

4

4

0

0
3

67

130

194

258

322

385

449

513

thousands of dollars

3

67

130

194

258

322

thousands of dollars

Notes: SEB means self-employed business owner. All dollar values are in 2004 dollars. Because of the scaling used in this figure, 	
the leftmost bar, which represents businesses with a loss in excess of $10,000, is not clearly visible in panels C and D.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

have positive returns of less than 5 percent. But the
median RBCF of the businesses is about 50 percent,
and about 30 percent of SEBs earn rates of return
greater than 125 percent. As firms mature, they are
less likely to suffer very poor annual cash flows, but
they are also less likely to experience very large returns.
Only about 4 percent of the businesses with ownership
tenure of six years or more experience losses. But their
median RBCF is more than 20 percentage points less
than the median for the new businesses, and only 20 percent of the mature ventures earn cash flow returns
greater than 100 percent.
As seen in figure 5, including an estimate of unrealized capital gains in the income measure changes
the rate of return histograms a good deal. But it does
not change the basic story that entrepreneurs take on

28

a great deal of risk and that much of this risk-taking
in search of large returns appears to be done by
smaller, younger businesses.
First, RBCF–CG is substantially larger than RBCF:
The median return for all SEBs and all years rises from
33 percent to 43 percent. Second, unrealized capital
gains boost the returns across all categories of business
ownership tenure. But the gains are larger for ventures
with short ownership tenure. For the firms with ownership tenure of one year or less, the median for RBCF
is 49 percent, while the median RBCF–CG is 102 percent. For firms with tenure of two to five years, the median RBCF and RBCF–CG rates are 40 percent and 58
percent, respectively, and for businesses with tenure
of six years or more, the median RBCF and RBCF–
CG rates are 28 percent and 32 percent, respectively.

4Q/2007, Economic Perspectives

figure 4

Cash flow rate of return (RBCF), by ownership tenure
A. All tenures

B. 0–1 year

percentage of SEBs

percentage of SEBs

18

18

15

15

12

12

9

9

6

6

3

3

0

0
0

25
.

50

0

7
. 5 100 125 150 175 200 225 250

25
.

50

C. 2–5 years

D. 6 or more years

percentage of SEBs

percentage of SEBs

18

18

15

15

12

12

9

9

6

6

3

3

0

7
. 5 100 125 150 175 200 225 250

RBCF, percent

RBCF, percent

0
0

25
.

50

7
. 5 100 125 150 175 200 225 250

RBCF, percent

0

25
.

50

7
. 5 100 125 150 175 200 225 250

RBCF, percent

Note: SEB means self-employed business owner.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

Third, according to RBCF–CG, few surviving firms
experience very low returns. For example, even among
the firms in the shortest ownership tenure category,
only 7.5 percent experience losses and 90 percent of
the firms earn returns greater than 16 percent.
Survivorship and buyout biases
In this section, we discuss some important sources
of bias in the rate of return calculations. The first is
an upward survivorship bias. We are only measuring
the ex post returns of the successful ventures that have
survived. We do not have any information on how many
firms lost enough money that they had to cease operations. Accounting for such losses would lower our
overall rate of return calculations.
The second set of biases relates to buyouts
and divestitures. We do not observe the returns of

Federal Reserve Bank of Chicago

entrepreneurs who have taken their firms public or
have been bought out by other private firms. This would
tend to bias down our return measures. However, the
results of Moskowitz and Vissing-Jorgensen (2002)
suggest that this bias might be small. They compute
an aggregate, value-weighted return to a portfolio of
private equity that includes an adjustment for realized
capital gains in the form of initial public offerings
and takeovers. They find the adjustment has a small
effect on the rate of return, on the order of 50 basis
points. Another potential source of downward bias in
our rate of return measures is that they do not account
for income resulting from the divestiture of assets
that occurs between the time the business was acquired
and the SCF year (see the appendix). This underestimate may be more important for older firms.

29

figure 5

Total rate of return including unrealized capital gains (RBCF–CG), by ownership tenure
A. All tenures

B. 0–1 year

percentage of SEBs

percentage of SEBs

20

20

15

15

10

10

5

5

0

0
0

25
.

50

7
. 5 100 125 150 175 200 225 250

0

25
.

50

7
. 5 100 125 150 175 200 225 250

RBCF–CG, percent

RBCF–CG, percent

C. 2–5 years

D. 6 or more years

percentage of SEBs

percentage of SEBs

20

20

15

15

10

10

5

5
0

0
0

25
.

50

7
. 5 100 125 150 175 200 225 250

RBCF–CG, percent

0

25
.

50

7
. 5 100 125 150 175 200 225 250

RBCF–CG, percent

Note: SEB means self-employed business owner.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

Evidence of entrepreneurial wealth creation
Because we do not have a panel, we cannot make
any definitive statements about the process in which
successful, young firms evolve into rich, old firms. Still,
the results relating rates of return and net equity to the
tenure of business ownership provide some insight
into what that process looks like.
Many entrepreneurs bring new business ideas to
the market. Many of these fail. Those that succeed
have the potential to generate some large cash flow
returns, even early on in the operating life of the firm.
A firm with short ownership tenure can also generate
unrealized capital gains, for example, by paying down
debt. And such a firm’s survival can also signal an increase in the franchise value of the business. As a successful business ages and grows, the rates of return,
as measured by the cash flowing from the firm to the

30

household, fall a good deal. Retained earnings, debt
reduction, and entrepreneurial effort are still successful in increasing the equity value of the firm. But even
including unrealized capital gains, the typical returns
to older firms fall short of those to operations with a
shorter tenure. This suggests that as the businesses
age and grow, larger rates of return become less
common because of diminishing returns to scale and
because imitators enter the entrepreneur’s niche and
erode market share. The successful entrepreneurial
venture ends up with more wealth, a larger business,
and smaller, but more stable, returns.
In sum, entrepreneurs face substantial risk–return
trade-offs. But those who have good ideas, learn, and
survive can persistently generate substantial returns
for their businesses. In the end, the successful businesses make the entrepreneurs wealthy.

4Q/2007, Economic Perspectives

Credit constraints, risk preferences, and
precautionary saving
Ideally, the entrepreneur would be able to borrow
as needed at a cost that is contingent on the particular
state of the world that eventually transpires. Such ideal conditions do not exist because, in reality, certain
information is available to the entrepreneur, but not
the lender; there is limited enforceability of contracts;
and there is a risk that entrepreneurs might reduce their
work effort if they do not bear all of the risks and reap
all of the rewards from their business ventures. These
features generate less-than-perfect risk sharing between lenders and entrepreneurs. And in light of the
large variability in returns, the importance of survivorship, and the relevance of unrealized capital gains for
entrepreneurs, this lack of risk sharing likely has an
important influence on the economic environment
faced by entrepreneurs and their creditors.
One potentially important implication is that entrepreneurs can face credit constraints. Such constraints
can take many forms—differences between borrowing
and lending rates, collateral requirements, or outright
denial of credit at any price.15 Credit constraints can
deter entrepreneurs from investing as much as they
think is necessary to make their businesses profitable
and can expose them to consumption fluctuations due
to unforeseen shocks. The more the household is averse
to risk, the larger the cost of these consumption fluctuations. And so there is an incentive for entrepreneurs
to self-insure against these costs by building wealth for
precautionary balances and to finance their business
operations. However, the degree to which entrepreneurs
will do so also depends on their attitudes toward risk
and the severity of the credit constraints they face.
We find qualitative evidence that entrepreneurs
face credit constraints, though there are indications
that these have eased over time. We also find evidence
that SEB households are less risk averse than other
wealthy households. This discounts the notion that
the wealth of entrepreneurs disproportionately reflects
a buildup of precautionary balances.
Credit constraints
The data seem consistent with the proposition
that SEBs do face important credit constraints, although
there is no way to determine how much of SEB wealth
is associated with such constraints. Furthermore, some of
our other results suggest that credit constraints have
moderated over the 1989–2004 period.
The SCF calculates business net worth by taking
the net value the household would receive if the business were sold today plus the net of any lending or
collateral provision from the personal accounts of the

Federal Reserve Bank of Chicago

household to the business ledger (see the appendix).
For example, if an SEB puts up a house as collateral
to secure a line of bank credit to be used by the business, this amount is included in the value of business
net worth. The greater is the degree of credit constraints on business lending, the greater the need for
such lending or collateralization.
As seen in table 8, a substantial share of SEBs
make such commitments. In every year we consider,
more than 15 percent of SEBs have a loan or a guarantee or they have pledged collateral to their business.
Furthermore, the size of these commitments is not trivial. The median value as a share of net equity of the
firm varies between 12 percent and 25 percent; in
terms of the household’s net worth, the commitments
vary between 4 percent and 15 percent. In general,
there are more cases of lending between households
and businesses for the limited liability ventures. This
may simply reflect the fact that for such entities there
are more legal reasons to distinguish the balance sheet
of the owner from that of the business. The size of the
commitment, however, does not vary systematically
between the various legal forms of business structure.
Further insight into credit constraints can be
gleaned from a question in the SCF that asks if over
the past five years the respondent has been turned down
for credit or not been given as much credit as requested.
As seen in the bottom portion of table 8, over the first
four survey years in our sample, SEBs were slightly
more likely than others in the population to have experienced such a constraint. However, the relative share
of SEBs experiencing such limits seems large given
the disproportionate representation of SEBs in the upper ranges of the wealth and income distributions. When
we consider only those households in the top 10 percent of the wealth distribution, SEBs experienced a
much higher rate of being turned down for credit than
other households. As we document, SEBs have much
higher debt-to-income ratios than other households
(see table 9). They also may try to borrow more often
than non-SEB households in order to finance business
operations. Accordingly, SEBs may experience more
instances of credit denial by lenders that are concerned
about their ability to service debt.
Even though SEBs apparently face credit constraints, such restrictions appear to have relaxed over
time. The share of SEBs making loan commitments
to their businesses declined from nearly 30 percent in
1989 and 1992 to the 15–20 percent range in the late
1990s and early 2000s.16 A decline in personal lending
to the business is evident for all forms of legal structure.
Furthermore, the percentages of all households in the
top decile of the wealth distribution that have been

31

turned down for credit are substantially smaller in the
2001 and 2004 surveys, and the declines are much larger
for SEBs than for other wealthy households. Indeed, the
gap between the two sets of households is quite small
in the last two surveys. Finally, recall that table 6 (p. 24)
showed a steady uptrend—from 65 percent in 1989 to
78 percent in 2004—in the share of SEBs that started
their own business. The trend could be consistent
with the view that reductions in liquidity constraints
make it easier for entrepreneurs to finance new business ventures as opposed to purchasing established
operations that might be more easily collateralized
than a new operation.
Risk aversion and precautionary saving
All else being equal, one might expect SEBs to
engage in more precautionary saving to compensate
for the extra risk associated with the large variation in
returns to their businesses. However, entrepreneurs appear to be less risk averse than other wealthy households, which mitigates the degree of precautionary
saving we would expect to observe.
The SCF asks respondents about the amount of
financial risk they are willing to take in order to receive
financial returns when they save or make investments.
The survey asks respondents if they are willing to:

1) Take substantial financial risks expecting to earn
substantial returns, 2) Take above average financial
risks expecting to earn above average returns, or 3) Take
average financial risks expecting to earn average returns,
or if they are 4) Not willing to take any financial risks.
As seen in table 9, on balance, SEBs respond that their
willingness to take risks for high returns is somewhat
more than average, while non-SEBs respond that
their tolerance for risk is somewhat less than average
(also see Moore, 2004, and Herranz, Krasa, and
Villamil, 2007).
In addition, entrepreneurs do not seem to diversify
very much away from their businesses. The median
share of household income SEBs earn from their businesses is well over one-half (see table 9). Furthermore,
this share increased steadily over the sample period;
from just under 60 percent in 1989 to nearly 85 percent in 2004.17 By comparison, the median non-SEB
household (not shown in table 9) earns no business
income, and the average share of business income in
total income for such households varies between just
2.5 percent and 6.25 percent. Also, SEBs leave a large
portion—about one-third—of their wealth in their
businesses, meaning that a substantial portion of their
ability to consume is unhedged against bad business
outcomes. This lack of diversification stands out

Table 8

Credit constraints evidence
	
Percentage of SEBs with a
loan, guarantee, or collateral
commitment	
  Sole proprietorship	
  Partnership	
  LLC, C corporation, other	
  S corporation	
  SEB total	
	
Value as percentage of
business net wortha	
Value as percentage of
household net worthb	
	
Percentage of households
ever turned down for credit	
  All SEBs	
  All others	
  SEBs in the top decile of
   wealth distribution	
  Others in the top decile of
   wealth distribution	

1989	

1992	

1995	

1998	

2001	

2004

	
25.2	
33.3	
25.8	
46.3	
29.2	
	

	
23.5	
27.0	
39.8	
30.3	
27.2	
	

	
18.9	
15.2	
27.7	
36.1	
21.8	
	

	
10.1	
18.3	
33.8	
38.9	
19.5	
	

	
11.1	
20.0	
22.8	
22.7	
16.4	
	

17.1
17.8
24.6
26.4
20.6

12.7	

19.8	

12.2	

22.9	

25.0	

24.7

4.9	
	

6.8	
	

5.7	
	

8.4	
	

9.3	
	

14.5

	
18.2	
17.3	

	
21.6	
20.0	

	
20.1	
18.4	

	
21.1	
19.1	

	
15.1	
17.1	

13.9
17.8

16.4	

13.8	

7.4	

12.9	

3.7	

5.2

3.9	

4.7	

4.5	

3.9	

2.1	

2.2

For those households reporting a loan, guarantee, or collateral commitment to their businesses, this is the value of that transaction relative 	
to the net equity in the business. Median share.
b
For those households reporting a loan, guarantee, or collateral commitment to their businesses, this is the value of that transaction relative to
household net worth. Median share.
Notes: SEB means self-employed business owner. LLC means limited liability company.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.
a

32

4Q/2007, Economic Perspectives

starkly when looking at all the wealthy households
(those in the top 10 percent of the wealth distribution). As seen in the bottom portion of table 9, the
overall wealth-to-income ratio of wealthy SEBs is
about the same as that of other rich households. However, the ratio of wealth (excluding business net equity) to income for SEBs is 30 percent to 50 percent
below that of non-SEB households.18
An interesting, complementary finding by Heaton
and Lucas (2000) suggests that entrepreneurs try to
limit their risk exposure by holding less wealth in stock,
compared with other similarly wealthy households.
Furthermore, SEBs raise both sides of their balance
sheets to a much greater degree than other households.
Notably, SEBs carry substantially more debt than other
households: Their debt-to-income ratios are between
one and a half and three times higher than those of
wealthy non-SEBs.19 This means that relative to other
households, a much greater percentage of SEBs’ income stream is precommitted to servicing debt.
The combination of similar total wealth-to-income
ratios, high concentrations of wealth in business equity,
and high debt-to-income ratios suggest a greater tolerance for risk by entrepreneurs than by others, and
hence, a lower predilection for precautionary saving.

Conclusion
Our findings support the view that entrepreneurs
are important sources of saving and wealth creation
in the U.S. economy. They start new businesses, introduce new ideas or business concepts, invest large
amounts of their net worth in their businesses, and
take very large risks for potentially very large returns.
Our findings also support the view that “market frictions” prevent the entrepreneurs from completely diversifying away risks and, perhaps, from investing in
their firms at an economically efficient level. We also
find, however, indicators pointing to some easing of
the borrowing constraints faced by entrepreneurs after
the early 1990s. Our results on portfolio and income
diversification and from direct questions regarding
risk attitudes also indicate that entrepreneurs are less
risk averse than the U.S. population at large.

Table 9

Risk tolerance of SEBs and others
	
Subjective risk intolerancea	
  SEBs	
  Others 	
	
Percentage of SEB wealth
invested in the business	
Percentage of SEB income
coming from the business	
	
Wealthy SEBs vs.
other wealthy households	
  Median ratio of wealth to income	
    SEBs	
    Others	
  
  Median ratio of wealth
   (excluding business
   net equity) to income	
    SEBs	
    Others	
  Median ratio of debt to income	
    SEBs	
    Others	

1989	

1992	

1995	

1998	

2001	

2004

	
3.0	
3.3	
	

	
2.9	
3.4	
	

	
2.9	
3.3	
	

	
2.8	
3.1	
	

	
2.9	
3.1	
	

2.8
3.2

34.4	

33.3	

30.6	

32.8	

33.3	

34.3

58.3	
	

65.2	
	

70.9	
	

75.7	
	

78.3	
	

83.5

	
	
10.49	
9.51	

	
	
8.82	
9.59	

	
	
10.44	
9.05	

	
	
9.86	
10.82	

	
	
11.81	
11.90	

13.88
12.29

	
4.98	
8.38	

	
4.46	
8.76	

	
5.55	
8.39	

	
5.42	
10.28	

6.51	
11.07	

7.84
11.11

	
0.29	
0.17	

	
0.64	
0.15	

	
0.50	
0.29	

	
0.90	
0.31	

	
0.61	
0.20	

0.91
0.43

a
Index: 1 corresponds to willing to take substantial risk; 2 to take above average risk; 3 to take average risk; and 4 to take no risk.
Notes: SEB means self-employed business owner. Wealthy SEBs and other wealthy households are those in the top decile of the wealth distribution.
Source: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

Federal Reserve Bank of Chicago

33

notes
Although we require that SEBs own at least one business, we do
not restrict our analysis to households whose business net worth is
above a given threshold. Therefore, we have not necessarily excluded
owners of tiny businesses that may enter or exit operation quickly.
1

We chose to not use the 1983 and 1986 SCF waves because these
surveys appear to be of lower quality, and they did not ask all of the
questions regarding the households’ business interests found in the
six SCF waves that we use.
2

The SCF generally takes place in the second half of the calendar
year. Flow variables, notably income, refer to the previous calendar
year; stock variables and demographic or business ownership identifiers cover current values.
3

The summary statistics we present in this article are calculated by
first multiplying the relevant observations by their SCF weight.
These are demographic weights that indicate how representative
the observed household is of the U.S. population as a whole.
4

The SEBs are overrepresented in the other portions of the
upper ends of the wealth distribution as well. Nearly 40 percent
of SEBs fall in the top wealth decile, and nearly one-quarter are
in the top 5 percent of the wealth distribution. The SEBs make up
25–30 percent of the top 10 percent of the wealth distribution and
30–40 percent of the top 5 percent.
5

There is some evidence of a countercyclical pattern for self-employed
workers as a whole in the Current Population Survey (CPS), which
is conducted by the U.S. Census Bureau for the U.S. Bureau of
Labor Statistics. (See Aaronson, Rissman, and Sullivan, 2004, and
Rissman, 2006.) The CPS does not allow one to identify who among
the self-employed own and manage a business operation. With regard
to our results, many of those making cyclical transitions between
self-employment and work for pay may not consider themselves as
actively managing a business, and so they would not be included in
our SEB sample. The movements in and out of self-employment of
such individuals may indeed be countercyclical.
6

The education index is constructed by coding household heads as
a 1 if they do not have a high school degree; a 2 if they have a high
school degree only; a 3 if they have some college; and a 4 if they
have a college degree. The index is the population-weighted average across respondents.
7

Note that C corporations may or may not be publicly traded, but
the SCF questions regarding active business ownership instruct
respondents to give only information on corporations that are not
publicly traded that they own or share ownership of and in which
they have an active management role.
8

Clearly, measurement error may be an important issue with business net worth. The value of the firm is a self-reported number. So
unless the SEB has recently received a credible buyout offer or
other external assessment of the firm’s valuation, the answer to this
question will be subjective. Valuation might be particularly difficult for SEBs running start-ups, since there is little track record
available to judge the long-run potential of the firm.
9

A small percentage of firms—0.05 percent—have a net worth less
than –$2,200.
10

34

We differ from Moore (2004) in that we do not include an adjustment for the difference between the unrealized capital gain and the
opportunity cost of capital.
11

This result may seem surprising for firms with short ownership
tenure, since they have had little time to reap the gains from capital
investments made using retained earnings. However, other factors
may be boosting their net equity. First, the SEB may have paid
down a substantial amount of debt incurred when acquiring the
business. Second, the survival of the SEB’s firm may provide a positive signal of the ongoing viability of the business venture, causing
the SEB to give a more optimistic assessment of the value of the
firm. Note, though, that we likely are underestimating annualized
capital gains to the firms with short ownership tenure. This is because
we do not know the exact month the firm was acquired; hence, we
divided the change in value by one, even if the firm was created
less than a year ago.
12

In addition, to avoid skewing the rates of return because of the
earnings of businesses with little or negative net equity, we perform
rate of return computations only for businesses with a net worth
greater than $1,000. Even with this adjustment, however, some
firms’ high rates of return may reflect relatively modest income
flows against a base of very little net equity. Unfortunately, the
SCF only records net equity in the business and does not separate
assets and liabilities. Accordingly, we cannot calculate an alternative rate of return on assets.
13

Some of the higher dispersion among businesses with ownership
tenure of one year or less might be affected by sampling error, since
the number of SEBs in this category is a good deal smaller than
those in the other two tenure classifications.
14

Some researchers, for example, Hurst and Lusardi (2004) question the importance of credit constraints to entrepreneurial activity
documented by much other literature on the subject.
15

That said, conditioned on making a loan or pledging collateral,
the size of such lending has trended up over time. The median
commitment rose from between 12 percent and 20 percent of the
net worth of the business in the first three surveys in our sample
to about 25 percent in the 2001 and 2004 surveys.
16

The trends in legal structure do not appear to account for the
trend in the business income share of total SEB income. All legal
forms of business show an uptrend in the share of income derived
from their business activity, and the differences in business shares
among the legal forms are small.
17

For comparison, among non-SEB households in the top decile
of the wealth distribution, the median amount of wealth held in
private business equity is zero, and the average amount varies between 9 percent and 12 percent. Moskowitz and Vissing-Jorgensen
(2002) also document the lack of portfolio diversifications for
entrepreneurs.
18

Undoubtedly, this is a natural consequence of borrowing to operate their businesses, although we cannot tell for sure because the
SCF provides information on only the net value of the businesses,
without separating their assets and liabilities.
19

4Q/2007, Economic Perspectives

appendix

The number of observations in each wave of the Survey
of Consumer Finances we use in our article appears in
table A1.
Business net worth

Business net worth equals the net equity if the business were sold today, plus loans from the household to
the business, minus the loans from the business to the
household not reported as personal debt by the respondent, plus the value of personal assets used as collateral
for business loans not reported as an asset by the household earlier. Net equity is the self-reported answer to the
question: “What is the net worth of the business? What
could you sell it for today?”
Self-employed business owners’ business income

Isolating the income SEB households earn from their
businesses is not straightforward. We do so by combining the SCF questions that directly ask respondents how
much salary or wages they earn from their main jobs and,
in addition to regular salary, how much of the net earnings
or other income they received from their businesses. We
include income received by the spouses from the businesses.
These data are not without problems, and measurement
error is still a concern. For example, they may suffer

from recollection error, and while salary data are for the
current year, other business income refers to the calendar year prior to the SCF. See Moore (2004) for a discussion of these and other issues. Still, we feel this approach is superior to using tax data, which the SCF also
records. Importantly, partnership and S corporation income
are found in form 1040, line 17, but line 17 also includes
income from rental real estate, royalties, and trusts that
might not be actively managed businesses by SEB households. Furthermore, S corporations and other closely
held corporations can pay salaries to their owners. Such
salary income is included in the regular wage and salary
reporting in form 1040, line 7, but they also include wage
earnings of family members from outside of the businesses.
Computing business income net of taxes would be
very interesting but requires work that goes beyond the
scope of this article. It would be relatively easy to do so
for businesses organized as C corporations because they
face a flat tax rate on business income. Such firms, however, make up only a small fraction of the SEBs in our
sample. For other types of SEBs, business income is included in the households’ total income, and hence, it is
tax-based on all of the household financial and demographic
characteristics. Computing business income net of taxes
does require a complicated and detailed algorithm.

Table A1

Number of observations in each SCF wave
	

1989	

SEBs	
Others	

1992	

1995	

1998	

2001	

2004

566	

837	

838	

856	

878	

931

2,577	

3,068	

3,461	

3,449	

3,564	

3,588

Notes: SEB means self-employed business owner. SCF means Survey of Consumer Finances.
Source: Board of Governors of the Federal Reserve System, Survey of Consumer Finances.

Federal Reserve Bank of Chicago

35

REFERENCES

Aaronson, Daniel, Ellen R. Rissman, and Daniel
G. Sullivan, 2004, “Can sectoral reallocation explain
the jobless recovery?,” Economic Perspectives, Federal
Reserve Bank of Chicago, Vol. 28, No. 2, Second
Quarter, pp. 36–49.
Altonji, Joseph G., and Ulrich Doraszelski, 2005,
“The role of permanent income and demographics on
black/white differences in wealth,” Journal of Human
Resources, Vol. 40, No. 1, Winter, pp. 1–30.
Altonji, Joseph G., Ulrich Doraszelski, and Lewis
M. Segal, 2000, “Black/white differences in wealth,”
Economic Perspectives, Federal Reserve Bank of
Chicago, Vol. 24, No. 1, First Quarter, pp. 38–50.
Gentry, William M., and R. Glenn Hubbard, 2004,
“Entrepreneurship and household savings,” Advances
in Economic Analysis and Policy, Vol. 4, No. 1, article 8,
available at www.bepress.com/bejeap/advances/
vol4/iss1/art8.
Heaton, John, and Deborah Lucas, 2000, “Portfolio choice and asset prices: The importance of entrepreneurial risk,” Journal of Finance, Vol. 55, No. 3,
pp. 1163–1198.
Herranz, Neus, Stefan Krasa, and Anne P. Villamil,
2007, “The impact of entrepreneur characteristics and
bankruptcy rules on firm performance,” University of
Illinois at Urbana–Champaign, mimeo, May.
Hurst, Erik, and Anna Lusardi, 2004, “Liquidity
constraints, wealth accumulation, and entrepreneurship,” Journal of Political Economy, Vol. 112, No. 2,
pp. 319–347.

36

Knight, Frank H., 1921, Risk, Uncertainty, and Profit,
Boston: Hart, Schaffner, and Marx and Houghton
Mifflin.
Moore, Kevin, 2004, “Comparing the earnings of
employees and the self-employed,” Board of Governors
of the Federal Reserve System, mimeo, October 19.
Moskowitz, Tobias, and Annette Vissing-Jorgensen,
2002, “The returns to entrepreneurial investment: A
private equity premium puzzle?,” American Economic
Review, Vol. 92, No. 4, September, pp. 745–778.
Quadrini, Vincenzo, 2000, “Entrepreneurship, saving,
and social mobility,” Review of Economic Dynamics,
Vol. 3, No. 1, January, pp. 1–40.
­­
__________,
1999, “The importance of entrepreneurship for wealth concentration and mobility,” Review
of Income and Wealth, Vol. 45, No. 1, March, pp. 1–19.
Rissman, Ellen R., 2006, “The self-employment
duration of younger men over the business cycle,”
Economic Perspectives, Federal Reserve Bank of
Chicago, Vol. 30, No. 3, Third Quarter, pp. 14–26.
Schumpeter, Joseph A., 1934, The Theory of Economic
Development: An Inquiry Into Profits, Capital, Credit,
Interest, and the Business Cycle, Redvers Opie (trans.),
Cambridge, MA: Harvard University Press, originally
published as Theorie der Wirtschaftlichen Entwicklung,
1911, Leipzig, Germany: Duncker und Humblot.

4Q/2007, Economic Perspectives

A bank by any other name …
Christian Johnson and George G. Kaufman

Introduction and summary
Banks come in a wide variety of forms. These include
commercial banks, savings banks, savings and loans,
and credit unions. But, all banks are not perceived as
equally vital to the economy so as to require the same
degree of government regulation to promote their safe
and efficient operation. To regulate efficiently, it is
necessary to carefully define the entity to be regulated.
The issue of what constitutes a bank for regulatory
purposes emerged in 2005 from being an arcane subject of interest primarily to a small number of regulatory attorneys to being of interest to a much larger and
broader group. This interest was sparked when the
large retailer Wal-Mart applied to the Federal Deposit
Insurance Corporation (FDIC) to obtain federal deposit insurance for a newly chartered “bank” in Utah
that was not subject to the ownership restrictions applicable to most other “banks.” This article examines the
definition of “bank” for financial regulatory purposes,
traces and explains the evolution of the definition through
time, and explores the controversy surrounding the
recent attempt by Wal-Mart to establish its own bank.
Wal-Mart has since withdrawn its application.
All depository institutions, including commercial
and savings banks, need to obtain a special charter
from either the federal government or their home state
government rather than a general corporate charter. The
charter identifies the activities in which the institutions
are permitted to engage. Each chartering and regulatory agency specifies a definition of “bank” to which
its authority applies. Restrictions on permissible activities may be imposed by the FDIC on insured banks
and by the Board of Governors of the Federal Reserve
System on holding companies that own bank subsidiaries.
The definition of bank need not be the same across
agencies nor for any one agency through time. Differences and changes in definition may occur for a number
of reasons, including differences in regulatory objectives

Federal Reserve Bank of Chicago

among agencies, changes in legislation, changes in
the demand for different types of financial services,
changes in the supply of particular financial services,
innovations in financial products and institutions, and
changes in the operations of financial institutions.
In recent months, controversy about the definition
of a bank has been ignited by an attempt, since abandoned, by Wal-Mart to obtain FDIC insurance for an
industrial loan company (ILC) to be chartered in Utah.1
An ILC is a “bank” chartered in a limited number of
states that is granted the same or slightly fewer product powers than are commercial banks chartered in that
state. Importantly, ILCs are currently explicitly exempted
from the definition of “bank” in the Bank Holding
Company Act (BHCA) if, among other characteristics,
they do not accept demand deposits when their assets
exceed $100 million. As long as the proposed ILC had
satisfied these conditions, the parent holding company
Wal-Mart would not have been legally classified as a
bank holding company—a holding company that owns
one or more institutions legally defined as a “bank”—
and would have been subject neither to regulation by
the Federal Reserve nor to the restrictions of the Bank
Holding Company Act. If it had been, the nonfinancial activities of the parent company Wal-Mart would
have prohibited its ownership of a bank subsidiary.
Christian Johnson is a professor of law at the Law School
of Loyola University Chicago. George G. Kaufman is the
John F. Smith Professor of Economics and Finance at
Loyola University Chicago and a consultant to the Federal
Reserve Bank of Chicago. The authors are indebted to
Christine Blair and Donald Hamm of the Federal Deposit
Insurance Corporation, Tara Rice of the Federal Reserve
Bank of Chicago, and the participants at presentations at
the Western Economic Association and the Federal Reserve
Bank of Chicago for their valuable assistance and suggestions and to the staff of the Knowledge Center of the
Federal Reserve Bank of Chicago for collecting much
of the underlying documentation.

37

To some, this “loophole” in the legal definition of
a bank permits the piercing of the separation of banking (financial) and commerce (nonfinancial) that the
BHCA was designed to maintain and is perceived as
providing holding companies owning an ILC an unfair
advantage over holding companies that own legally
defined banks, such as commercial banks. This generated opposition to the Wal-Mart application for FDIC
insurance, which was necessary for it to be an ILC
that is exempt from the restrictions of the BHCA. In
response to this opposition, the FDIC imposed a sixmonth moratorium in July 2006 on this and all other
pending applications for federal insurance either for
a new ILC or for an existing ILC undergoing a change
in control through January 31, 2007. The FDIC then
extended the moratorium in January 2007 for another
year on new and pending applications from commercial (nonfinancial) firms for the operation of federally
insured ILCs. This moratorium is due to expire on
January 31, 2008.2 In March 2007, Wal-Mart withdrew
its application.
Evolution of the definition of “bank” and
“bank holding company”
A bank is a type of financial institution. A financial institution is an entity that deals primarily in financial instruments and derives most of its revenues from
interest and fees charged on its loans, investments, and
deposits, or from trading in these securities. A popular dictionary of banking terms defines a bank as
[an organization,] usually a corporation, that
accepts deposits, makes loans, pays checks,
and performs related services for the public.3
What differentiates a bank from most other financial institutions is that a bank can accept deposits of
funds that the bank may re-lend but that need to be
repaid to the depositor at full value at a future specified or unspecified date. As such, banks belong to the
broader class of depository institutions, which includes
other institutions that are chartered to accept deposits
and make loans but traditionally have provided a narrower and more specialized range of services, such as
savings and loan associations and credit unions.
As noted, unlike most other business corporations,
banks require a special corporate bank charter from a
government entity; in the United States this is either
from the federal government (national bank) or the home
state government (state bank).4 Their powers are
defined in the charter. For example, national banks
chartered by the Comptroller of the Currency may:

38

exercise … all such incidental powers as shall
be necessary to carry on the business of banking
by discounting and negotiating promissory notes,
drafts, bills of exchange, and other evidences of
debt; by receiving deposits; by buying and selling exchange, coin and bullion; by loaning
money on personal security. …5
The National Bank Act, as currently amended, specifies individually the permissible powers in addition
to deposit taking and loan making.
The charter imposes both advantages and disadvantages on a bank. The institution can offer various
types of deposits, such as demand, time, and savings.
These deposits are currently insured up to a maximum amount of $100,000 per eligible account by the
Federal Deposit Insurance Corporation, which is an
agency of the federal government. The bank is also
provided direct access to the national payments system through the Federal Reserve’s check and electronic clearing facilities. To the extent that bank charters
are not granted freely, the chartering agencies may
restrict entry and reduce competition.
In return for these advantages, the charter subjects
the bank to a number of disadvantages in the form of
costly regulation and supervision for reasons of safety,
fairness, efficiency, and monetary policy. In the words
of former Federal Reserve Chairman Paul Volcker:
Handling other people’s money, which is what
banking is all about, connotes a fiduciary responsibility. …To that end, banking systems in
virtually all countries are regulated.6
Types of regulation and supervision that have
been frequently imposed on chartered banks include:
n	 Restrictions on types of products and services that
may be offered;
n 	 Restrictions on the number and location of offices;
n 	 Minimum capital requirements;
n	 Restrictions on ownership by holding companies;
n	 Restrictions on mergers with other banks;
n 	 Restrictions on interest paid on deposits and
charged on loans;
n 	 Examination by bank regulatory agencies for
financial soundness and compliance with other
regulations;
n	 Frequent reporting of financial condition to the
regulatory agencies; and
n 	 Special nondiscrimination lending and reporting
requirements.
Until relatively recently, the term bank was often
defined only loosely in federal legislation.7 For example, the Federal Reserve Act of 1913 defines bank

4Q/2007, Economic Perspectives

Table 1

Changes in definition of bank in Bank Holding Company Act
1956	
	

Any national or state-chartered commercial, savings,
or trust bank

1966	

Any institution that accepts demand deposits

1970	
	

Any institution that both accepts demand deposits
and makes business loans

1987	
	

All banks insured by the FDIC except thrifts, credit card
banks, and industrial loan companies and banks

Table 2

Changes in the definition of bank (savings and loan)
holding company for purposes of Holding Company Act
1956 	
	

Bank Holding Company Act (BHCA) applied to holding
companies (HC) owning two or more chartered banks

1967 	
	
	

Saving and Loan Association Holding Company Act (SLHCA)
applies provisions similar to BHCA to S&Ls owning two or
more institutions

1970 	
	

BHCA expands definition of covered HC to owning only one
bank or more

1987	
	

BHCA expands covered HCs to any owning one or more
FDIC insured banks but lists specific exemptions

1999	
	

Gramm–Leach–Bliley Act expands definitions of SLHCA
to an S&L owning one or more institutions

to include state bank, banking association, and
trust company, except where national banks or
Federal Reserve banks are specifically referred to.8
The important Banking Act of 1933 (Glass–Steagall)
refers to the definition used in the Federal Reserve Act.
However, the term bank came to be more precisely
defined with the Bank Holding Company Act of 1956.
The definition of the term bank reflects the primary
purpose of the act, which was to prevent both excessive economic concentration in banking and conflicts
of interest that could arise if banks and nonbanks were
under common ownership, enabling banks to provide
preferential treatment to customers of their affiliates.9
(The major changes in the legislated definitions of
“bank” and “bank holding company” since 1956 are
summarized in tables 1 and 2.)
Thus, the act restricted the nonfinancial activities
of bank holding companies (BHC), prohibited bank
holding companies from owning subsidiaries that engage in nonfinancial activities or in financial activities that were defined by the Federal Reserve as not
being so closely related to banking as to be incidental

Federal Reserve Bank of Chicago

to it, and restricted the ability of bank
holding companies to acquire banks in
other states.10 The Fed developed a “laundry list” of financial activities that it considered sufficiently incidental to banking
to be offered by nonbank subsidiaries of
BHCs. Although commercial banks were
generally prohibited from engaging in
nonfinancial (commerce) activities by
their charters, there were no previous restrictions on the activities of subsidiaries
of holding companies that also owned
one or more chartered banks or of the
nonfinancial activities of such a holding
company.
To achieve its objective, the BHCA
needed to define “bank holding company.”
Because the major concern with both excessive economic concentration and conflicts of interest was with respect to banking
firms, the act defined bank holding company with respect to the type of bank that
it owned or controlled. The definition in
the 1956 act defined “bank” to include:
any national banking association
or any State bank, savings bank, or
trust company…11

and “bank holding company” as any corporate firm that owned two or more banks
so defined.12 In addition, bank holding
companies had to register with the Federal Reserve
and receive permission from the Fed for further
acquisitions.
In time, the BHCA’s definition of a bank was viewed
as broader than necessary to achieve its objectives, as
the definition included many types of financial institutions that were unlikely to produce excessive economic concentration or meaningful conflicts of interest
if owned by a holding company that also owned nonbank subsidiaries. Thus, in 1966, the Bank Holding
Company Act was amended to define a bank more
narrowly as:
Any institution that accepts deposits that
the depositor has a legal right to withdraw
on demand. ...13
This amendment changed the definition of bank
from a chartering test to an activities test. Because
deposits subject to withdrawal on demand (demand
deposits) were at the time generally restricted to
commercial banks, this definition effectively defined
a bank holding company only as a company that owned
two or more commercial banks.

39

The Senate report that accompanied this and other
amendments at the time to the BHCA explained the
reason for the change as follows:
Section 2(c) of the [1956 BHCA] defines “bank”
to include savings banks and trust companies,
as well as commercial banks. The purpose of
the [BHCA] was to restrain undue concentration of control of commercial bank credit, and
to prevent abuse by a holding company of its
control over this type of credit for the benefit
of its nonbanking subsidiaries. This objective
can be achieved without applying the [BHCA]
to savings banks, and there are at least a few
instances in which the reference to “savings
bank” in the present definition may result in
covering companies that control two or more
industrial banks. To avoid this result, the bill
redefines “bank” as an institution that accepts
deposits payable on demand (checking accounts),
the commonly accepted test of whether an institution is a commercial bank so as to exclude industrial banks and nondeposit trust companies.14
Note the express exclusion of industrial banks in the
legislative history from the definition of “bank” for
purposes of the act.
In 1970, the definition of “bank” for purposes of
the act was narrowed further to:
any institution organized under the laws of the
United States, any State of the United States …
which 1) accepts deposits that the depositor has
a legal right to withdraw on demand, and 2) engages in the business of making commercial
loans.15
This definition excluded a few institutions that accepted demand deposits but did not make business
loans. Lending for noncommercial purposes was considered less likely to cause the problems that the act
was designed to prevent. In addition, in response to a
sharp increase in the number of holding companies
owning only one bank and engaging in activities not
permitted for holding companies owning two or more
banks, the 1970 amendments also broadened the definition of a bank holding company to cover ownership
of only one bank so defined.
In the early 1980s, however, an increasing number of bank holding companies organized or purchased
banks that either accepted demand deposits but did
not make commercial (business) loans or made commercial loans but did not accept demand deposits.
Thus, they were not defined as “banks” for purposes
of the act at that time. These institutions became known
as “nonbank banks.” Holding companies that owned

40

such nonbank banks were not subject to the restrictions of the act that were imposed on holding companies that owned banks that met the definition of the
act, particularly the prohibition against banks being
owned by companies that were nonfinancial firms or
owned them. Indeed, most but not all of the newly
chartered nonbank banks were owned by holding
companies that also owned nonfinancial firms.
To restrict this type of holding company going
forward, the act was amended in 1987 by the Competitive Equality Banking Act (CEBA) to broaden the
definition of bank from institutions that both accept
demand deposits and make business loans to all banks
insured by the FDIC.16 (Existing nonbank banks were
grandfathered, but subject to asset growth restrictions.)
However, this definition captured some banks and
other financial institutions that were generally considered unlikely to cause either excessive economic concentration or conflicts of interest if they were owned
by a nonfinancial holding company or by a holding
company that owned financial companies that were
not on the Federal Reserve’s permissible list.
To address this problem, the CEBA amendments
for the first time specifically excluded from the definition of “bank” foreign banks, federally insured savings
and loan associations, credit unions, credit card banks,
and most federally insured industrial loan companies.
However, as seen earlier, ILCs were already noted as
not being a target of the BHCA in the Senate report
accompanying the 1966 amendments. What most of
these exempted institutions had in common is that, at
the time, while they generally accepted deposits and
made loans, they did not offer demand deposits and
did little, if any, commercial lending. Companies that
owned such excluded institutions were not subject to
the act’s restrictions. In explaining his support for the
new definition, Paul Volcker, Chairman of the Board
of Governors of the Federal Reserve System at the
time, testified before the Senate Banking Committee:
Essentially, the nonbank bank has become a device for tearing down the separation of commerce and banking by permitting a commercial
firm to enter traditional banking business without abiding by the provisions of the Bank Holding Company Act. … Fundamentally at stake is
not a few in-house consumer banking offices of
some retail chains. … We want to protect
against instability, excessive concentration of
power, and undue conflicts of interest, while
preserving the institutional framework for monetary policy. In seeking these goals, the separation of banking and commerce has been a basic
part of the American tradition for what seems
to me sound reasons.17

4Q/2007, Economic Perspectives

The specific exemption for industrial loan companies
and industrial banks in CEBA was introduced in the
final drafting of the act by then Senators Alan Cranston
of California and Jake Garn of Utah, who served on
the Senate Banking Committee and represented the
two states with the largest number of such institutions.18
In 1999, Congress effectively reaffirmed the ILC
exemption from the definition of “bank” and thereby
also the restrictions of the BHCA, when it included a
provision in the Gramm–Leach–Bliley Act (GLBA)
that slightly expanded the permissible activities of eligible ILCs but did not otherwise change the exemption.
It is evident from this chronology of the evolution of the definition of both “bank” and “bank holding company” for regulatory purposes that the legal
definition at any moment in time reflects the pressing
public concerns of the time. As the concerns changed,
so frequently did the definitions.
Industrial loan companies
Partially as a result of the broadening of the definition of bank in the BHCA through time, both nonfinancial (commercial) firms that wished to own a bank
and were prohibited from doing so by the BHCA and
nonbank financial companies that wished to own banks
but did not wish to be legally classified as a bank
holding company, and therefore be subject to Federal
Reserve regulation, became more restricted in their
options. ILCs were a remaining available option.19
CEBA explicitly exempted ILCs from the definition
of bank in the BHCA if:
1.	 In 1987, the state in which they were chartered required them to be insured by the FDIC, and either
2.	 They have less than $100 million in assets or, if
greater, they do not offer demand deposits,20 or
3.	 There has been no change in control since 1987.
In addition, in 1999, some firms that could have
owned a single (unitary) thrift institution were brought
under the restrictions of the Savings and Loan Holding
Company Act (SLHCA) by the Gramm–Leach–Bliley
Act. However, such firms may have preferred an ILC
because, unlike a thrift institution, an ILC is not subject to the qualified lender provision, which effectively requires thrifts to hold a minimum percentage of
mortgage loans in their portfolios.21
Seven states that charter ILCs satisfy the federal
deposit insurance requirement of CEBA. They are
California, Colorado, Hawaii, Indiana, Minnesota,
Nevada, and Utah. A number of companies that wanted
to escape the restrictions of the BHCA or SLHCA
chose to purchase or organize ILCs in these states,
primarily in Utah, California, and Nevada, or to grow
existing ILCs faster than they would have otherwise.

Federal Reserve Bank of Chicago

ILCs originated in the early 1900s as small depository institutions, aimed primarily at the financial
needs of low- and moderate-income households that
were not being well served by existing larger financial institutions. They differed little either in mission
or in operation from other consumer-oriented smaller
financial institutions of the day, such as Morris Plan
banks and credit unions.22 They were chartered only
at the state level, but could generally branch across
state lines. ILCs remained relatively small until the end
of the 1990s when their aggregate asset size jumped
dramatically, even though they declined in number.
Although the FDIC has insured Morris Plan banks
since the FDIC’s establishment in 1934, ILCs became
eligible for FDIC insurance only in 1982, after the
enactment of the Garn–St Germain Act.
Since the enactment of CEBA in 1987, when the
ability of firms to avoid the BHCA restrictions by owning banks that either did not take demand deposits or
did not make business loans was terminated, aggregate
assets at federally insured ILCs increased from less
than $5 billion to more than $150 billion by year-end
2006. All but $15 billion of this increase occurred since
1998, when the ability of additional firms to avoid the
restrictions of the SLHCA by owning only one thrift
institution (unitary thrift holding companies) was terminated by the Gramm–Leach–Bliley Act of 1999.
Despite their rapid growth, ILCs account for less than
2 percent of total assets at FDIC insured institutions.23
At the same time, the number of federally insured
ILCs declined sharply from 105 to 59.24 Only three of
the largest 15 ILCs in 1987 remained active in 2006.
By far, the largest increase in ILC assets in this period occurred in Utah, which increased its market share
of national ILC assets from 11 percent to 82 percent
by 2004.25 Both the rapid growth of ILCs in total and
the particularly rapid growth in Utah can be explained
in part by changes in Utah’s legislation and the state’s
supportive regulatory environment for ILCs.26 In 1986,
Utah put a moratorium on new ILC charters after a
number of ILCs had experienced significant financial
difficulties that required some $45 million of state assistance to meet their depositor claims. The moratorium was lifted in 1997 after the industry regained its
financial health, and the number of charters grew from
18 to 33 by June 30, 2006. Total assets also grew from
$18 billion in 1997 to $133.8 billion in 2006.27 Over
the same period, the size of the individual institutions
has also changed greatly. In 1987, the largest Utah
chartered ILC had $290 million in assets.28 At yearend 2006, the largest ILC in Utah reported assets of
$67 billion.29

41

While most ILCs are relatively small, seven had
assets in excess of $10 billion at year-end 2006 and
ranked among the largest 125 FDIC insured depository
institutions of the nearly 9,000 such institutions in the
country. (A listing of the largest 15 ILCs by asset size
at year-end 2006 is shown in table 3.) All but three 	
of these were chartered in Utah. The industry is also
highly concentrated. In mid-2006, the largest ILC 	
accounted for 40 percent of all assets in the industry
and the five largest accounted for about 75 percent of
the industry’s total assets.30
Contrary to their earlier days, few of today’s
larger ILCs are independent community-oriented 	
institutions. Although large ILCs are prohibited from
taking demand deposits, the current powers of ILCs
are not greatly different in most states from those of
commercial banks; many ILCs operate as limited 	
service or specialized lending institutions.
ILC parent holding companies represent a wide
range of financial and nonfinancial firms, and the 	
activities of their subsidiary ILCs are directed at an
equally broad range of economic sectors that may or
may not be associated with the primary activities of
the parent. The largest four ILCs are owned by major
financial firms, including one of the largest commercial banks in the world. The largest ILC, Merrill Lynch
Bank USA, is owned by the investment banking firm
	

of Merrill Lynch. It focuses on securities-based consumer loan products as well as consumer and business
loans. The bank also makes first and second mortgage
loans, as well as community development loans and
investments to satisfy its Community Reinvestment
Act (CRA) responsibilities.31 The next largest ILC 	
focuses on loans to high wealth households, and the
third on loans generated through general credit cards
originated by its parent firm.
Some ILCs are owned by financial firms or by
firms that are not otherwise generally prohibited from
owning a bank. Other ILCs are owned by nonfinancial
firms that use their ILCs to finance the sales of goods
they either manufacture or sell or to finance unrelated
activities. These firms could not own commercial banks
under the current provisions of the BHCA. According
to their websites and Community Reinvestment Act
reports, Volkswagen owns an ILC that finances primarily indirect automotive, home equity, and credit
card loans. Until recently General Motors (GM) owned
General Motors Acceptance Corporation (GMAC),
which in turn owned two Utah ILCs, one of which focuses on commercial mortgage loans and the other on
automotive loans. The GMAC Automotive Bank was
the fifth largest ILC in 2006. In November 2006, in
an exception to its moratorium, the FDIC permitted 	
a change in ownership of the larger of the two ILCs

Table 3

Fifteen largest industrial loan companies, by asset size, 2006
	
	
Rank	
	

	
	
ILC	
	

	 1.	

Merrill Lynch Bank USA	

	 2.	

UBS Bank USA	

	 3.	

American Express Centurion Bank	

	 4.	

Parent	
holding	
company	
	

	
State	
chartered	
	

Year
	
	
	
Federally	
Chartered	
insured	
	
	

Total
assets
2006
($ billion)

Merrill Lynch	

Utah	

1988	

1988	

67.2

UBS	

Utah	

2003	

2003	

22.0

American Express	

Utah	

1989	

1989	

21.1

Morgan Stanley Bank	

Morgan Stanley	

Utah	

1990	

1990	

21.0

	 5.	

GMAC Automotive Bank	

General Motors	

Utah	

2004	

2004	

19.9

	 6.	

Fremont Investment and Loan	

Fremont General Corp.	

California	

1937a 	

1984	

12.9

	 7.	

Goldman Sachs Bank	

Goldman Sachs	

Utah	

2004	

2004	

12.6

	 8.	
	 9.	
	 	
	 	

USAA Saving Bank	
Capmark Bank	
(formerly GMAC Commercial 	
Mortgage Bank)	

USAA Life Co.	
Cerberus Capital	
Management
Consortium

Nevada	
Utah	

1996	
2003	

1996	
2003	

5.8
3.8

	10.	

Lehman Brothers Commercial Bank	

Lehman Brothers	

Utah	

2005	

2005	

3.2

	11.	

CIT Bank	

CITGroup	

Utah	

2000	

2000	

2.8

	12.	

BMW Bank of North America	

BMW Group	

Utah	

1999	

1999	

2.2

	13.	

GE Capital Financial Inc.	

General Electric	

Utah	

1993	

1993	

2.0

	14.	

Advanta Bank Corp.	

Advanta	

Utah	

1991	

1991	

2.0

	15.	

Beal Saving Bank	

Beal Financial Group	

Nevada	

2004	

2004	

1.9

a
Originally chartered ILC was purchased by Fremont General in 1990.
Sources: iBanknet and Federal Deposit Insurance Corporation.

42

4Q/2007, Economic Perspectives

owned by GM, which was undergoing major restructuring, to a consortium of four financial institutions.
BMW uses its Utah ILC to finance sales of BMW
automobiles and motorcycles, and the retailer Target
uses its Utah ILC to finance its in-house credit card
sales for small business customers.
The wide variety of both ownership and business
lines of ILCs is reflected in the eight types of business
models into which the two principal ILC trade groups
divide the industry: 1) ILCs owned by securities companies, 2) ILCs owned by commercial finance companies, 3) ILCs owned by consumer finance companies,
4) ILCs owned by a commercial company conducting
an independent core financial services business,
5) commercially owned ILCs offering financial services to customers of the corporate group that are not
affiliate transactions, 6) ILCs owned by a commercial
company that finance transactions with affiliates subject to the restrictions in Sections 23A and 23B of the
	

Federal Reserve Act and the anti-tying provisions of
the Bank Holding Company Act, 7) ILCs owned by
title insurance holding companies, and 8) independently owned ILCs.32 A brief description of each business
model and an ILC example are shown in table 4.
Primarily because of the rapid growth of ILCs in
recent years and the ongoing controversy surrounding
Wal-Mart itself, its application for required FDIC insurance for its proposed ILC in Utah attracted immediate
attention and widespread opposition from many bankers,
retailers, and policymakers, including members of
Congress. The opposition arose despite Wal-Mart’s
stated intentions in the application of not engaging in
full-service banking, but only in credit and debit card
and fund transfer (payments system) operations. At
its filing, the application raised at least two important
public policy issues:
1.	 Should a decision to increase the mix between
banking and commerce be made administratively

Table 4

ILC business models
Business model	

Description	

ILC example

Banks owned by securities 	
companies	
	

Provide commercial and consumer	
credit to customers of securities
companies	

Merrill Lynch Bank USA

Banks owned by commercial 	
finance companies	

Provide commercial loans to customers	
that are not customers of an affiliate	

Advanta Bank

Banks owned by consumer 	
finance companies	
	
	
	
Banks owned by a commercial 	
company conducting an independent	
core financial services business	

Provide credit cards and other forms	
of consumer credit and services to
customers that are not customers of
affiliates

American Express Centurion Bank

Provide traditional banking services	
to customers that are not customers
of affiliates

GE Capital Financial

Commercially owned banks offering 	
Provide credit and financial services	
financial services to customers of the	
to customers of owner
corporate group that are not affiliate
transactions		

BMW Bank of North America

Banks owned by a commercial 	
Provide credit to customers of affiliates	
company that finances transactions 	
(credit and services are subject to the
with affiliates subject to the restrictions	
covered transaction rules)
in Sections 23A and 23B of the Federal
Reserve Act and the anti-tying provisions
of the Bank Holding Company Act		

Target Bank

Banks owned by title insurance 	
Provide financial services	
holding companies		

First Security Thrift

Independently owned banks	
	
	
	

Celtic Bank

Provide financial services (owners 	
not engaging in commercial activities
prohibited by bank holding company
rules)

Source: Utah Association of Financial Services and California Association of Industrial Banks (2006).	

Federal Reserve Bank of Chicago

43

by a regulatory agency within the authority
Congress granted it, or should it be made legislatively by Congress in the light of the changed
circumstances described earlier?, and
2.	 Are the current regulatory prudential powers of
the FDIC sufficient for consolidated supervision
of ILC holding companies relative to the prudential powers of the Federal Reserve for bank
(financial) holding companies under the BHCA?
Because Wal-Mart was not the first large nonbank
firm to have received or applied for FDIC insurance
for an ILC or even the first large commercial firm—
only the most controversial—these two issues were
not necessarily muted by the withdrawal of the application. As discussed earlier, large firms, such as Merrill
Lynch, General Motors (until recently), BMW, and
Target all own ILCs. Home Depot has an insurance
application pending, but action on it has been delayed
by the moratorium.
Public policy issues
The mixing of banking and commerce
The mixing of banking and commerce in “universal” banks, as exists in many countries, has long
been controversial in U.S. banking history. Most state
charters for banks and the federal charter for national
banks limit the activities of banks to accepting deposits
and making loans, but permit other services viewed
as incidental to banking. This was generally interpreted
by regulators as prohibiting the banks from engaging
in some financial activities, such as insurance underwriting and real estate brokerage, and all nonfinancial
activities. Until the enactment of the BHCA in 1956,
these limitations were not generally applied to bank
holding companies, so that commercial firms could
own banks. Thus, Ford Motors and Sears, among other
large nonfinancial firms, operated banks. But, as discussed earlier, growing fears in the 1950s that such
combinations could lead both to excessive economic
and social power and to potential conflicts of interest
favoring sellers resulted in the enactment of the BHCA
in 1956 and its expansion in 1970. Since then, the thrust
of legislation, which often is preceded by changes in
the marketplace, has reversed. The financial powers of
BHCs have been expanded significantly, most recently in
the Gramm–Leach–Bliley (Bank Modernization) Act of
1999, and the nonfinancial powers have been expanded moderately. However, unlike ILCs, commercial
banks may still not be owned by commercial firms.
Two questions appear to arise going forward. First,
the ILC industry has changed dramatically since 1987,
when ILCs were first specifically exempted from the
restrictions of the BHCA primarily because they were

44

small and insignificant on a national scale. Thus, it
may reasonably be asked whether this issue has now
become sufficiently important that further piercing of
the separation of banking and commerce is too important to leave to the regulatory agencies by default.33
Rather, does it now deserve a review by Congress?34
Indeed, in her explanation for the one-year extension
of the moratorium on granting insurance to additional
ILCs owned by commercial firms in January 2007, the
FDIC Chairman, Sheila Bair, noted that “The moratorium will provide Congress with an opportunity to
address the issue legislatively.”35,36
In particular, it may be asked if Congress would
have specifically exempted ILCs from the BHCA in
earlier years had some of the institutions been as large
then as they are today? For example, the largest ILC
in 1987 had total assets of only some $400 million.
Indeed, only one of the current largest 15 ILCs was
chartered and federally insured before 1987. It is effectively a new industry. In testimony at the FDIC’s
open hearing on the Wal-Mart application, former
Senator Garn, who sponsored the exemption in 1987,
stated that he had not intended for ILCs to move into
the retail banking business and now opposes such expansion.37 Moreover, if after review, Congress determined that increased mixing of banking and commerce
is desirable, should this be limited to ILCs, or should
it be extended to all bank and financial holding companies to level the playing field?38
Second, by 1999, when Congress last retained
the ILC exemption by broadening it slightly, the ILC
industry had already begun a rapid expansion. The
largest ILC, owned by American Express, had assets
in excess of $15 billion, and four other ILCs had assets in excess of $2 billion each; one of these was
owned by a commercial firm. Thus, if Congress was
not sufficiently concerned at the time, and has taken
no action since, some may question whether it is appropriate for a regulatory agency to delay approval of
applications that are not in conflict with existing law
until Congress acts. Indeed, some have suggested
that, in this instance, the issue goes beyond whether
the mixing of banking and commerce is appropriate
and may be an issue with Wal-Mart per se.39 Wal-Mart
is the world’s largest retailer with an extensive distribution network and a perception as utilizing aggressive marketing and labor practices.40
Indeed, an application for a Utah chartered
ILC by large retailer Target in 2004 was viewed as
sufficiently routine by the FDIC to be approved at the
staff level rather than by the FDIC’s board of directors.41
Nor did the approval of the application ignite much
public opposition. In contrast, Wal-Mart’s application

4Q/2007, Economic Perspectives

to the FDIC attracted nearly 14,000 written letters, including 150 from members of Congress, almost all
opposed to the application, and caused the FDIC to
schedule three days of open hearings that attracted
some 70 witnesses, again almost all opposed.42
Although Wal-Mart has withdrawn its application,
there is some concern that, in the absence of congressional action, it may reapply in the future, after the
expiration of the moratorium. Wal-Mart has recently
established a full-service bank in Mexico and has announced its intentions to offer a wide range of nonbank financial services at its U.S. stores.
The FDIC’s prudential authority over ILCs
Because ILCs are state-chartered FDIC insured
institutions and none have chosen to be members of
the Federal Reserve System, their primary federal
regulator is the FDIC. In addition, they are regulated
by the banking agency in the state in which they are
chartered. All three federal regulators of commercial
banks—the Comptroller of the Currency, the Federal
Reserve, and the FDIC—have effectively the same
statutory prudential authority for the banks they supervise. But this is not necessarily true for their authority
over parent holding companies of these banks. The
Federal Reserve has clear authority under the BHCA
to supervise and examine bank holding companies, as
defined in the act, on a consolidated basis.43 This would
include the operation of the parent holding company,
subsidiary banks, and any subsidiary nonbank firms.
The underlying justification for such consolidated supervision is that these entities are usually managed in
terms of risk exposures on a centralized or consolidated
basis, so that full understanding of the risk exposure of
any one component of the entity requires knowledge of
all components combined.
Consolidated top-down supervision is widely
viewed as necessary despite the fact that Federal Reserve regulations 23 A and B limit the amount of transactions between the bank and the other affiliates of
the holding company and require that permissible transactions be priced on an “arm’s length” basis. These
regulations attempt to isolate the bank subsidiary from
the other components of the holding company, so that
the bank more closely resembles an independent, freestanding institution. A recent study (table 5) by the federal government’s Government Accountability Office
(GAO) compared the current statutory consolidated supervision powers of the FDIC and Federal Reserve (as
well as the Office of Thrift Supervision for parent
holding companies of savings and loan associations)
and found the FDIC’s weaker.44
For example, with limited exceptions, the FDIC
focuses on the ILC itself rather than the parent on a

Federal Reserve Bank of Chicago

consolidated basis—a bottom-up approach. The FDIC
generally examines or imposes sanctions and enforcement actions on the parent company or its non-ILC
affiliates only if it is concerned about the financial condition of the insured ILC. Thus, for example, the FDIC
recently issued a cease and desist order against the
Fremont Investment and Loan (an ILC) in California
and its parent holding companies for problems at the
ILC related to its underwriting of subprime mortgage
loans without noting either the large losses simultaneously experienced for the same reason by the parents
or requiring similar changes to be made by them as at
the subsidiary ILC.45 Major differences in the explicit
supervisory powers of the federal agencies over parent
holding companies of insured depository institutions
according to the GAO are shown in table 5.
To some, the more limiting powers over parent
holding companies may hamper the FDIC’s ability to
evaluate and protect the safety and soundness of ILCs.
Partially in recognition of this concern, the FDIC announced in its extension of the moratorium that it had
proposed a regulation that would provide for enhanced
supervision of ILC parent holding companies that engage only in financial activities to ensure their ability
to provide financial support to their institutions and
require them to maintain the capital of the ILC at a
specified minimum level.46 This proposal is still pending. The proposal did not include parent holding companies that engage in nonfinancial activities, pending
additional study by both the FDIC and Congress.
Recent developments
In May 2007, the House of Representatives
passed the Industrial Bank Holding Company Act of
2007 that would prohibit any firm that receives more
than 15 percent of its annual gross revenues on a consolidated basis from nonfinancial activities from owning or controlling an ILC. On October 4, 2007, the
Senate Banking Committee held hearings on Senate
Bill 1356, which is identical to the House bill. Firms
that owned an ILC before January 28, 2007, were
generally grandfathered. But, an ILC subsidiary of
a commercial firm that did not own the subsidiary
before 2003 cannot engage in activities in which it
did not engage in on January 28, 2007, or operate
branches in states in which it did not operate branches
on that date. The act would also broaden the FDIC’s
authority to examine and require reports from the ILC
parent holding company and affiliates and to enforce
sanctions and capital standards on the ILC parent
holding company and affiliates. This change would
bring the regulatory environment for ILC holding
companies into greater conformity with that for

45

	

Table 5

Comparison of explicit supervisory powers of the FDIC, Federal Reserve Board, and OTS
Description of explicit supervisory authority	

FDICa	

Board	

OTS

Examine the relationships, including specific transactions, if any,	
between the insured institution and its parent or affiliates.	

b	

b	

b

Examine beyond specific transactions when necessary to disclose the	
nature and effect of relationship between the insured institution and
the parent or affiliate.

b	

b	

b

Examine the parent or any affiliate of an insured institution,	
including a parent or affiliate that does not have any relationships
with the insured institution or concerning matters that go beyond
the scope of any such relationships and their effect on depository
institution.

	

b	

b

Take enforcement actions against the parent of an insured institution.	

b,c	

b	

b

Take enforcement actions against affiliates of the insured institution	
that participate in the conduct of affairs of, or act as agents for,
the insured institution.

b	

b	

b

Take enforcement actions against any affiliate of the insured institution,	
even if the affiliate does not act as agent for, or participate in the
conduct of, the affairs of the insured institution.

	

b	

b

Compel the parent and affiliates to provide various reports such as reports	
of operations, financial condition, and systems for monitoring risk.

b,d	

b	

b

Impose consolidated or parent-only capital requirements on the parent	
and require that it serve as source of strength to the insured depository
institution.

d	

b	

b

Compel the parent to divest of an affiliate posing a serious risk to the	
safety and soundness of the insured institution.

e	

b	

b

Explicit authority.

Less extensive authority.
No authority.
a
FDIC may examine an insured institution for interaffiliate transactions at any time and can examine the affiliate when necessary to disclose the
transaction and its effect on the insured institution.
b
The authority that each agency may have regarding functionally regulated affiliates of an insured depository institution is limited in some respects.
For example, each agency, to the extent it has the authority to examine or obtain from a functionally regulated affiliate, is generally required to
accept examinations and reports by the affiliates’ primary supervisors unless the affiliate poses a material risk to the depository institution or the
examination or report is necessary to assess the affiliate’s compliance with a law the agency has specific jurisdiction for enforcing with respect to
the affiliate (for example, the Bank Holding Company Act in the case of the Board). These limits do not apply to the Board with respect to a company
that is itself a bank holding company. These restrictions also do not limit the FDIC’s authority to examine the relationships between an institution and
an affiliate if the FDIC determines that the examination is necessary to determine the condition of the insured institution for insurance purposes.
c
FDIC may take enforcement actions against institution-affiliated parties of an ILC. A typical ILC holding company qualifies as an institution-affiliated
party. FDIC’s ability to require an ILC holding company to provide a capital infusion to the ILC is limited. In addition FDIC may take enforcement action
against the holding company of an ILC to address unsafe or unsound practices only if the holding company engages in an unsafe and unsound
practice in conducting the affairs of the depository institution.
d
FDIC maintains that it can achieve this result by imposing an obligation on an ILC holding company as a condition of insuring the ILC. FDIC also
maintains it can achieve this result as an alternative to terminating insurance. FDIC officials also stated that the prospect of terminating insurance
may compel the holding company to take affirmative action to correct violations in order to protect the insured institution. According to FDIC officials,
there are no examples where FDIC has imposed this condition on a holding company as a condition of insurance.
e
In addition to an enforcement action against the holding company of an ILC in certain circumstances (see note b), as part of prompt corrective
action the FDIC may require any company having control over the ILC to 1) divest itself of the ILC if divestiture would improve the institution’s financial
condition and future prospects, or 2) divest a nonbank affiliate if the affiliate is in danger of becoming insolvent and poses a significant risk to the
institution or is likely to cause a significant dissipation of the institution’s assets or earnings. However, the FDIC generally may take such actions only
if the ILC is already significantly undercapitalized.
Notes: FDIC is the Federal Deposit Insurance Corporation. OTS is the Office of Thrift Supervision.
Source: Hillman (2006), pp. 15–16.

BHCs and give the FDIC powers over ILC holding
companies more similar to those the Federal Reserve
has over BHCs.
Wal-Mart withdrew its application to operate an
ILC, but not its intention to engage in a wide range
of bank-like activities for which a bank charter is not

46

required. It has announced its intention to open “money
centers” in its stores that will offer, among other financial products, low-cost prepaid stored-value cards as
well as check cashing and money transfer (remittance)
services. In addition, it will offer a Wal-Mart branded
Visa debit card through a third party bank vendor.

4Q/2007, Economic Perspectives

Payroll and social security checks could be directly
transmitted by customers to Wal-Mart to be added to
the stored-value card or to support the debit card.
This is intended to increase both safety and convenience over currency transfers. Through time,

Wal-Mart has expressed intentions to add additional
financial services directed largely at low-income
“unbanked” customers.47

notes
In some states, Utah, for example, industrial loan companies are
referred to as industrial banks. The Wal-Mart application was initially filed in Utah for a charter in July 2005 and simultaneously with
the FDIC for insurance. The FDIC application was withdrawn in
March 2007. See Wal-Mart Stores, Inc. (2005).
1

FDIC (2007b).

2

Fitch (2000), p. 40.

3

Depository institutions are one of the few types of corporations
that may be chartered by either the federal government or the home
state.
4

National Bank Act, Chapter 106, Section 8, June 3, 1864, 13 Stat. 99,
codified at 12 USC §24.

If the parent holding company also owns a thrift institution, the
company is subject to regulation by the Office of Thrift Supervision
as a savings and loan holding company.
19

This may not be overly restrictive since large ILCs may offer
consumer NOW accounts, which resemble demand deposits.
20

12 USC § 1467(a)(m)(1).

21

For additional information about and the history of Morris Plan
banks, see http://eh.net/encyclopedia/article/philips.banking.
morris_plan.
22

23

5

Volcker (1987), p. 200.

6

Hillman (2006), pp. 5–7. Jones (2006).

There are apparently many more small ILCs that are not federally
insured, not included in the federal statistics, and not exempt from
the restrictions of the BHCA. Weiss (2007).
24

7

25

Government Accountability Office (2005), p. 20.

Federal Reserve Act, 63rd Cong. Chapter 6, Section 1, December
23, 1913, 38 Stat. 251.

26

Sutton (2002).

27

State of Utah, Commissioner of Financial Institutions (2006).

This section draws on Di Clemente (1983).

8

Bank Holding Company Act of 1956, Senate report, No. 84-1095,
July 25, 1955, pp. 1–4.
9

The separation of banking and commerce was not complete. BHCs
were permitted limited investment in nonfinancial firms. A review
of the permissible nonfinancial activities of banks appears in Haubrich
and Santos (2003).
10

Bank Holding Company Act of 1956, Ch. 240, 70 Stat 133,
Section 2(c). May 9, 1956.

State of Utah, Commissioner of Financial Institutions (2006).
State of Utah, Commissioner of Financial Institutions (1987).
28

See www.ibanknet.com (financial reports of industrial loan
companies).
29

30

11

Ibid. Companies that owned or controlled savings and loan associations and other thrift institutions insured first by the Federal Savings
and Loan Insurance Corporation (FSLIC) (and then the FDIC) were
not defined as bank holding companies and were initially not subject to any restrictions. After the enactment of the Savings and Loan
Holding Company Act (SLHCA) in 1967, those companies, for a
time, were subject to lesser restrictions until 1999, when the BHCA
and SLHCA became more comparable.
12

13

Public Law 89-485, Section 3(c), July 1, 1966, 80 Stat. 236.

14

S. Rep. No. 1179, 89th Cong., 2d Sess. 2391 (1966).

Bank Holding Company Act of 1970 (Public Law 91-607), Sect. 2(c),
December 31, 1970, 84 Stat. 1760.
15

Competitive Equality Banking Act of 1987, PL100-86, Sect. 101,
August 10, 1987, 101 Stat 552.
16

17

Volcker (1987), p. 200.

Comment submitted by Wal-Mart to the FDIC, October 10, 2006,
Appendix 1, p. 40, available at www.fdic.gov. Wilmarth (2007,
p. 1572), however, argues that Senator Garn’s cosponsor was
Senator William Proxmire of Wisconsin rather than Senator Cranston.
18

Federal Reserve Bank of Chicago

Hillman (2006).

Public Disclosure, January 10, 2006, Community Reinvestment
Act Performance Evaluation, Merrill Lynch Bank USA, available
at www.FDIC2.gov/crapes.
31

Utah Association of Financial Services and the California Association
of Industrial Banks (2006), pp. 11–13. See also Weiss (2007).
32

For a summary of the public policy issues in mixing banking and
commerce see Haubrich and Santos (2003), Blair (2004, 2007),
and Ergungor and Thomson (2006).
33

An analogous situation may be the demise of the controversial restrictions on underwriting and trading in private securities by banks
and bank holding companies introduced in the Banking (Glass–Steagall)
Act of 1933. In response to changing economic conditions and in
the absence of congressional action, the Board of Governors and the
other bank regulatory agencies slowly started to permit bank holding
companies into these activities in 1982 through administratively
liberalizing the interpretation of the restrictive language in the act
for subsidiaries authorized in Section 20 of the Federal Reserve
Act. Congress ultimately enacted liberalizing legislation in the
Gramm–Leach–Bliley Act of 1999. For a history of these issues
see Kaufman and Mote (1989, 1990).
34

35

FDIC (2007c).

47

The FDIC has approved a number of applications for insurance
since the adoption of the moratorium from firms that it considers as
financial or that propose activities by ILCs that are complementary
to financial activities and thus are not covered by the moratorium.
The extension of the moratorium applies only to ILCs to be owned
by commercial firms and not by nonbank financial firms, which do
not involve a mixing of banking and commerce.
36

Wilmarth (2007), p. 1572

37

Since the initial adoption by the FDIC in July 2006 of the moratorium on new and pending applications for federal deposit insurance
for both new and existing ILCs undergoing a proposed change in
control, assets at ILCs as a whole have increased sharply. In the six
months before the moratorium, assets at the 25 largest ILCs at yearend 2006 increased by some $12 billion from $145 billion at yearend 2005 to $157 billion at midyear 2006, or 8 percent. In the six
months following the moratorium, assets at these ILCs jumped by
$51 billion, or fully 32 percent.
Most of this unusual spurt in asset size can be attributed to three
ILCs—two are owned by nonbank financial firms and the third received special permission from the FDIC for a change in control
from GMAC to a consortium of four financial firms in anticipation
of a major restructuring of General Motors. The asset jump at these
ILCs may have been precautionary, in case Congress limited the
ILC exemption to the ownership restrictions of the BHCA. If so,
these ILCs may have anticipated that, as frequently is the case, existing ILCs would be grandfathered but their future growth would
be restricted.

Gramm–Leach–Bliley Act in 1999, which ended the unitary thrift
exemption, and then by enactment of restrictions on commercial
firm ownership of California chartered ILCs by the California state
legislature. It currently leases space to branch offices of some 300
independent banks in more than 1,000 of its stores. But an earlier
attempt in 2001 to have its own employees man such branch offices
and share in the proceeds with a chartered thrift institution was
denied by the Office of Thrift Supervision (Nolan, 2006).
41

38

Featherstone (2005).

39

Jorde (2003, 2006). This was not Wal-Mart’s first attempt to establish and operate a bank or thrift institution. It had previously attempted
to obtain a thrift institution in Oklahoma in 1998 and an ILC charter
in California in 2002, but was denied first by the enactment of the
40

Adler (2007a).

Wilmarth (2007), pp. 1545–1546. In addition, as of January 2007,
five states had enacted legislation since the Wal-Mart application in
Utah to prevent Utah chartered ILCs from branching further into
their states, and another five were considering such legislation.
(Adler, 2007b).
42

The Office of Thrift Supervision (OTS) has similar consolidated
supervisory authority for savings and loan holding companies. As
of year-end 2006, eight of the 15 largest ILCs holding 71 percent
of the assets of these ILCs were owned by parent companies that
also owned a thrift institution and thus were classified as savings
and loan holding companies and subject to OTS consolidated supervision (Reich, 2007).
43

Hillman (2006). This has also been argued by Federal Reserve
officials (Kohn, 2007).
44

45

FDIC (2007a).

FDIC (2007c). However, this still leaves them with weaker consolidated supervisory powers relative to the Federal Reserve. Equating
the two would require congressional action.
46

47

McWilliams (2007) and Barbaro and Dash (2007).

REFERENCES

Adler, Joe, 2007a, “FDIC Board sends staff a policy
reminder,” American Banker, July 5, pp. 1, 3.
, 2007b, “More states put ILC curbs on
agenda,” American Banker, January 18, p. 1.
Alvarez, Scott G., 2006, “Testimony before the Committee on Financial Services, U.S. House,” Washington, DC: Federal Reserve Board, July 12.
Barbaro, Michael, and Eric Dash, 2007, “At WalMart, a back door into banking,” Wall Street Journal,
June 21, p. C1.
Blair, Christine E., 2007, “Banking and commerce:
What difference does Wal-Mart make,” paper presented at the 2007 Western Economic Association
Meeting, Seattle, WA, June 21.
, 2004, “The mixing of banking and
commerce: Current policy issues,” FDIC Banking
Review, Vol. 16, No. 4, January, pp. 97–120.

48

Di Clemente, John, J., 1983, “The meeting of passion and intellect: A history of the term ‘bank’ in the
Bank Holding Company Act,” Federal Reserve Bank
of Chicago, staff memoranda, No. 83-1.
Ergungor, O. Emre, and James B. Thomson, 2006,
“Industrial loan companies,” Economic Commentary,
Federal Reserve Bank of Cleveland, October 1.
Featherstone, Liza, 2005, “The Bank of Wal-Mart?,”
The Nation, September 12.
Federal Deposit Insurance Corporation, 2007a, “FDIC
issues cease and desist order against Fremont Investment and Loan, Brea, California, and its parents,”
Washington, DC, press release, No. PR22-2007, March 7.
, 2007b, “Moratorium on certain industrial bank applications and notices,” Federal Register,
Vol. 72, No. 23, February 5, available at www.fdic.gov.
, 2007c, “FDIC extends moratorium on industrial loan company (ILC) applications by commercial

4Q/2007, Economic Perspectives

companies for one year; Will move forward on applications from financial companies,” press release,
No. PR-7-2007, January 31.
Fitch, Thomas P., 2000, Dictionary of Banking Terms,
Hauppauge, NY: Barron’s Educational Services.
Government Accountability Office, 2005, “Industrial
loan corporations: Recent asset growth and commercial interest highlight differences in regulatory authority,” report to the U.S. House, Committee on Financial
Services, Washington, DC, No. 05-621, September.
Haubrich, Joseph G., and João A. C. Santos, 2003,
“Alternative forms of mixing banking with commerce:
Evidence from American history,” Financial Markets,
Institutions, and Instruments, Vol. 12, No. 2, May,
pp. 121–164.
Hillman, Richard J., 2006, “Industrial loan corporations: Recent asset growth and commercial interest
highlight differences in regulatory authority,” testimony before the U.S. House, Committee on Financial
Services, Washington, DC: Government Accountability Office, No. 06-961T, July 12.
Jones, Douglas H., 2006, “Industrial loan companies: A review of charter, ownership, and supervision
issues,” statement before the U.S. House, Committee
on Financial Services, Washington, DC: FDIC, July 12.
Jorde, Terry J., 2006, “Keys to the kingdom,” Independent Banker, June, p. 11.
, 2003, “Banking and commerce in the financial marketplace,” paper presented at The Future
of Banking Symposium held by the FDIC, July 16.
Kaufman, George G., and Larry R. Mote, 1990,
“Glass–Steagall: Repeal by regulatory and judicial interpretation,” The Banking Law Journal, September/
October, pp. 388–421.
, 1989, “Securities activities of commercial banks,” in Research in Financial Services, George G.
Kaufman (ed.), Greenwich, CT: JAI Press, pp. 223–262.
Kohn, Donald L., 2007, “Industrial loan companies,” testimony before the U.S. House, Committee
on Financial Services, April 25.

Reich, John M., 2007, “Industrial loan companies,”
statement before U.S. House, Committee on Financial Services, Washington, DC: Office of Thrift Supervision, April 25.
State of Utah, Commissioner of Financial Institutions, 2006, Report to the Governor and the Legislature of the State of Utah, for the Period July 1, 2005,
to June 30, 2006, Salt Lake City, UT, available at
www.dfi.state.ut.us.
, 1987, Report to the Governor and the
Legislature of the State of Utah, for the Period July 1,
1986, to June 30, 1987, Salt Lake City, UT.
Sutton, George, 2006, “ILCs—A review of charter,
ownership, and supervision issues,” testimony before
the U.S. House of Representatives, Financial Services
Committee, Subcommittee on Financial Institutions
and Consumer Credit, Serial No. 109-106, July 12,
pp. 51–51.
, 2002, “Industrial banks,” Consumer
Finance Law Quarterly Report, Spring, pp. 178–181.
U.S. House of Representatives, Committee on
Financial Services, 2006, “ILCs—A review of charter,
ownership, and supervision issues: Hearing,” Congressional Record, 109th Congress, 2nd Session,
Serial No. 109-106, July 12.
Utah Association of Financial Services and
California Association of Industrial Banks, 2006,
“Response to the request for public comment on industrial corporations issued by the FDIC on August
29, 2006,” report, October 10, available at www.fdic.gov.
Volcker, Paul, 1987, “Statement before U.S. Senate,
Committee on Banking, Housing, and Urban Affairs,
January 21, 1987,” Federal Reserve Bulletin, March,
pp. 199–205.
Wal-Mart Stores, Inc., 2006, “Comment submitted
to FDIC,” October 20, available at www.fdic.gov.
, 2005, “Federal Deposit Insurance application to FDIC,” July 18, available at www.fdic.gov.

McWilliams, Gary, 2007, “Wal-Mart plans to sell debit
card from Visa,” Wall Street Journal, June 20, p. B2.

Weiss, N. Eric, 2007, Report to Congress: Industrial
Loan Companies/Banks and the Separation of Banking and Commerce: Legislature and Regulatory Perspectives, Washington, DC: Congressional Research
Service Report for Congress, updated January 3.

Nolan, Kevin K., 2006, “Wal-Mart’s industrial loan
company: The risk to community banks,” North
Carolina Banking Institute, Vol. 10, pp. 187–207.

Wilmarth, Arthur, Jr., 2007, “Wal-Mart and the separation of banking and commerce,” Connecticut Law
Review, May, pp. 1539–1622.

Federal Reserve Bank of Chicago

49

'">1/1kl>

CDANK

I

STRUCTURE &
QOM PETITION
Federal Heserve Bank oj Chicago

T

he Federal Reserve Bank of Chicago invites the submission of research- and

policy-oriented papers for the 44th annual Conference on Bank Structure and

Competition to be held May 14-16, 2008, at the InterContinental Hotel in Chicago.
Since its inception, the conference has fostered a dialog on current public policy is­

sues affecting the financial services industry. As in past years, the program will
highlight a conference theme (to be announced) and will also feature numerous ses­

sions on topics unrelated to the theme. We therefore welcome submissions of high-

quality research on all topics related to financial services, their regulation, and industry
structure. A nonexhaustive list of possible session topics includes:

■

Financial stability;

■

Subprime mortgage markets;

■

Credit market disruptions, market liquidity, and public policy;

■

Hedge funds, securitization, and credit market turmoil;

■

The Basel II Capital Accord;

■

Deposit insurance reform;

■

Financial industry consolidation;

■

Small business finance;

■

Nonbank financing;

■

Fair lending and the Community Reinvestment Act;

■

Hedge funds;

■

Credit derivatives;

■

Analysis of whether banks are straying too far from their core function;

■

The burden of bank regulation;

■

Retirement finance; and

■

The role and potential liability of Fannie Mae, Freddie Mac, and
pension guaranty funds.

If you would like to present a paper at the conference, please submit your completed

paper or a detailed abstract (the more complete the paper, the better), along with your
name, address, affiliation, telephone number, and email address, and those of any
co-authors, by December 21,2007. Manuscripts should be submitted via email to:

BSC_2008_submissions@frbchi.org.
Additional information will be posted to the conference website as it becomes
available:

www.chicagofed.org/BankStructureConference
or, you may contact the conference chairman directly:

Douglas Evanoff at 312-322-5814 or devanoff@frbchi.org.

Index for 2007
Title & author

Issue

Pages

BANKING, CREDIT, AND FINANCE
Transforming payment choices by doubling fees
on the Illinois Tollway
Gene Amromin, Carrie Jankowski, and Richard D. Porter

Second Quarter

22^17

Evidence on entrepreneurs in the United States:
Data from the 1989—2004 Survey of Consumer Finances
Mariacristina De Nardi, Phil Doctor, and Spencer D. Krane

Fourth Quarter

18-36

A bank by any other name ...
Christian Johnson and George G. Kaufman

Fourth Quarter

37^49

ECONOMIC CONDITIONS
Who are temporary nurses?
Andrew Goodman-Bacon and Yukako Ono

First Quarter

2-13

Bubble, bubble, toil, and trouble
Cabray L. Haines and Richard J. Rosen

First Quarter

16-35

New evidence on labor market dynamics over the business cycle
Bhashkar Mazumder

First Quarter

36-46

Asset rundown after retirement: The importance of rate
of return shocks
Eric French, Phil Doctor, and Olesya Baker

Second Quarter

48-65

Issues facing state and local government pensions
Richard H. Mattoon

Third Quarter

2-32

Understanding the evolution of trade deficits: Trade elasticities
of industrialized countries
Leland Crane, Meredith A. Crowley, and Saad Quayyum

Fourth Quarter

2-17

MONEY AND MONETARY POLICY
Against the tide—Currency use among Latin American immigrants
in Chicago
Carrie Jankowski, Richard D. Porter, and Tara Rice

Second Quarter

2-21

Government investment and the European Stability
and Growth Pact
Marco Bassetto and Vadym Lepetyuk

Third Quarter

33-13

Economic theory and asset bubbles
Gadi Barlevy

Third Quarter

44-59

To order copies of any of these issues, or to receive a list of other publications, please telephone (312) 322-5111 or write to:
Federal Reserve Bank of Chicago, Public Information Center, P.O. Box 834, Chicago, IL 60690-0834. The articles are also
available to download in PDF format from the Bank’s website at www.chicagofed.org/economic_research_and_data/
economic_perspectives.cfm.

52

4Q/2007, Economic Perspectives