(Econ, Econometrics) - Guide To Economic Indicators (4Th Ed)

Guide to
Economic
Indicators
Guide to
Economic
Indicators
4 T H E D I T I O N
N o r m a n Fru m kin
M.E.Sharpe
Armonk, New York
London, England
Copyright 2006 by Norman Frumkin
All rights reserved. No part of this book may be reproduced in any form
without written permission from the publisher, M.E. Sharpe, Inc.,
80 Business Park Drive, Armonk, New York 10504.
Library of Congress Cataloging-in-Publication Data
Frumkin, Norman.
Guide to economic indicators / Norman Frumkin.4th ed.
p. cm.
Includes bibliographical references and indexes.
ISBN 0-7656-1646-7 (hardcover : alk. paper)
1. Economic indicatorsUnited States. 2. Business cyclesUnited StatesStatistics.
3. United StatesEcomomic conditionsStatistics. I. Title.
HC103.F9 2006
330.973'0021dc22 2005013951
Printed in the United States of America
The paper used in this publication meets the minimum requirements of

American National Standard for Information Sciences
Permanence of Paper for Printed Library Materials,
ANSI Z 39.48-1984.
BM (c) 10 9 8 7 6 5 4 3 2 1
To Sarah, Jacob, Samuel, Susan, Isaac, Ann, Jonah
In memory of
Anne Frances Feldman Frumkin and Joseph Harry Frumkin

Contents
Tables, Formulas, and Figure ix

Preface xiii
Data Sharing and Confidentiality of Federal Government
Statistical Data xv
Using this Book xvii
1. Average Weekly Earnings 3

2. Average Weekly Hours 7
3. Balance of Payments 11
4. Balance of Trade 17
5. Bank Loans: Commercial and Industrial 23
6. Bankruptcies: Personal 26
7. Capacity Utilization 31
8. Consumer Attitude Indexes 35
9. Consumer Credit 42
10. Consumer Credit Delinquency 45
11. Consumer Price Index 48
12. Corporate Profits 56
13. Distribution of Income 61
14. Distribution of Wealth 68
15. Employment 73
16. Employment Cost Index 84
17. Farm Parity Ratio 88
18. GDP Price Measures 92
19. Government Economic Transactions: Expenditures,
Receipts, Surplus/Deficit 96
vii
viii GUIDE TO ECONOMIC INDICATORS
20. Gross Domestic Product 103

21. Help-Wanted Advertising Index 113
22. Home Sales: New and Existing Houses 117
23. House Prices: New and Existing Houses 122
24. Housing Affordability Index 128
25. Housing Starts 132
26. Housing Vacancy Rate 136
27. Import and Export Price Indexes 140
28. Industrial Production Index 144
29. Interest Rates 148
30. International Investment Position of the United States 155
31. InventorySales Ratios 159
32. Job Gains and Losses 163
33. Job Openings and Labor Turnover 167
34. Leading, Coincident, and Lagging Indexes 171
35. Manufacturers Orders 178
36. Mortgage Loan Applications 182
37. Mortgage Delinquency and Foreclosure 185
38. Non-Manufacturing Business Activity Index 188
39. PMI 191
40. Personal Income and Saving 195
41. Poverty 200
42. Producer Price Indexes 207
43. Productivity: Business Sector 213
44. Retail Sales 221
45. Selected Services Revenue 225
46. Stock Market Price Aggregates and Dividend Yields 229
47. Unemployment 238
48. Unit Labor Costs: Business Sector 251
49. Value of the Dollar 255
Appendix: Note on Sampling and Nonsampling Errors in

Statistical Surveys 263
References 267
Index 271
Tables, Formulas, and Figure
Tables
1.1 Average Weekly Earnings in Private Nonagricultural

Industries 5
2.1 Average Weekly Hours in Private Nonagricultural Industries 9
3.1 U.S. Balance of Payments 15
4.1 U.S. Balance of Trade 21
5.1 Bank Loans: Commercial and Industrial 25
6.1 Bankruptcy Filings: Personal 29
7.1 Capacity Utilization 34
8.1 Consumer Attitude Indexes 41
9.1 Consumer Credit Outstanding 44
10.1 Consumer Credit Delinquency Rate 47
11.1 Consumer Price Index 54
12.1 Corporate Profits 59
13.1 Distribution of Income: Household Shares 66
14.1 Net Worth and Standard Errors of Percentile Groups of
Families: 2001 71
14.2 Distribution of Dollar Net Worth 72
15.1 Employment: Alternative Definitions 82
16.1 Employment Cost Index: Private Industry 87
17.1 Farm Parity Ratio 91
18.1 GDP Price Measures 95
19.1 Government Economic Transactions 101
20.1 Gross Domestic Product and Main Components: 2004 105
20.2 Gross Domestic Product 111
21.1 Help-Wanted Advertising Index 116
ix
x TABLES, FORMULAS, AND FIGURE
22.1 Home Sales: New and Existing Houses 120

23.1 House Prices: New and Existing Houses 127
24.1 Housing Affordability Index 131
25.1 Housing Starts 134
26.1 Housing Vacancy Rates 139
27.1 Import and Export Price Indexes 143
28.1 Industrial Production Index 146
29.1 Interest Rates 153
30.1 International Investment Position of the United States 158
31.1 InventorySales Ratios 161
32.1 Job Gains and Losses 166
33.1 Job Openings and Labor Turnover: 2004 169
34.1 Leading, Coincident, and Lagging Composite Indexes
and Components 173
35.1 Manufacturers Orders 180
36.1 Mortgage Loan Applications Indexes 184
37.1 Mortgage Delinquencies and Foreclosures 187
38.1 Non-Manufacturing Business Activity Index 190
39.1 PMI 194
40.1 Personal Income and Saving 198
41.1 Poverty Measure 205
42.1 Producer Price Indexes 211
43.1 Productivity: Business Sector 219
44.1 Retail Sales 223
45.1 Selected Services Revenue 228
46.1 Stock Market Price Aggregates and Dividend Yields 236
47.1 Alternative Measures of Labor Underutilization 241
47.2 Labor Force Participation Rates 243
47.3 Unemployment Rate: Total and Insured Measures of
Unemployment 249
48.1 Unit Labor Costs: Business Sector 253
49.1 Value-of-the-Dollar Indexes: Country Coverage 257
49.2 Value-of-the-Dollar Indexes: Nominal and Real 261
Formulas
7.1 Capacity Utilization Rate 31

8.1 Consumer Confidence Index 37
8.2 Consumer Sentiment Index 39
TABLES, FORMULAS, AND FIGURE xi
24.1 Housing Affordability Index 130

26.1 Housing Vacancy Rate 138
43.1 Labor-Hour Productivity 216
43.2 Multifactor Productivity 218
47.1 Unemployment Rate 239
48.1 Unit Labor Costs 252
Figure
34.1 Leading, Coincident, and Lagging Composite Indexes 176

Preface
Guide to Economic Indicators is a reference book that gives concise

explanations of the meanings and uses of various macroeconomic indi-
cators. The selected indicators reflect the overall domestic and interna-
tional conditions of the American economy, as well as particular segments
of it. They are prepared primarily by U.S. government agencies, with a
small number prepared by private organizations.
The book explains the basic features of more than sixty statistical
measures of the U.S. economy. It is meant for economists, students,
investors, journalists, executives, and citizens interested in economic
affairs. Readers who are interested in an overall perspective on eco-
nomic indicators may find Chapter 1, Attributes of Economic Indica-
tors, in Tracking Americas Economy helpful.1
This fourth edition updates the explanations of the indicators to in-
clude their characteristics as of mid-2005 and the historical data through
2004. It also expands the coverage by adding the following new indica-
tors:
Insured Unemployment
Job Gains and Losses
Job Openings and Labor Turnover
House Prices: New and Existing Houses
Housing Vacancy Rates
Mortgage Loan Applications
Selected Services Revenue
Other new topics are data sharing among U.S. government statistical
agencies that prepare economic indicators (see below), the impact of
xiii
xiv PREFACE
labor force participation on unemployment (see Chapter 47, Unem-

ployment), and the Appendix on sampling and nonsampling errors in
statistical surveys.
The following indicators that were in the third edition are no longer pro-
vided by the source organizations and so are not included in this edition:
Business Failures
Business Starts
Business Optimism Indexes
Job Quality Index
Experimental Recession Indexes (no longer provided on a real-time
basis)
Growth Cycles (provided in occasional papers, but not on a regular
periodic schedule)
In addition, I have discontinued the Money Supply and the CRB Fu-
tures Price Index as economic indicators. In 2000, the Federal Reserve
ended its inclusion of annual money supply growth targets in its Mon-
etary Policy Report to Congress each February. The requirement by the
Full Employment and Balanced Growth Act of 1978 that money supply
targets be included in the Federal Reserves reports to Congress expired
in 2000, and the Federal Reserve concluded that the relationship be-
tween the money supply on the one hand, and economic growth and
price movements on the other, had become increasingly tenuous. I also
consider the money supply conceptually to have been a passive reflec-
tion of changes in financial assets rather than an active monetary policy
tool for influencing the economy.2 With respect to the CRB Futures Price
Index, I consider the index to primarily be a financial instrument for
traders in commodity futures index contracts; what the index measures
has little effect on overall economic growth or price movements.
The indicators appear in alphabetical order. The format of topics is
the same for all indicators: capsule explanation, where and when avail-
able (including website), content, methodology, accuracy (sampling/re-
vision error), relevance, recent trends, and the reference to the primary
data source. To facilitate cross-referencing, indicators covered in the
book are italicized when they are mentioned in chapters other than those
in which they are the primary focus.
Economic indicators are often based on survey estimates obtained
from samples of households, businesses, and governments. Such survey
PREFACE xv
estimates are subject to sampling error, as noted in the above paragraph

under accuracy. Indicators, whether based on sample data or not, are
also subject to various kinds of nonsampling error. For the effects of
sampling and nonsampling errors, see the Appendix: Note on Sampling
and Nonsampling Errors in Statistical Surveys.
I thank Lynn Taylor, economics editor of M.E. Sharpe, for initiating
this fourth edition and for facilitating the completion of the book. And I
thank Myron Sharpe for his continued confidence in my work.
Edward Steinberg, on the economics faculties of New York University
and Shippensburg University, reviewed the entire manuscript. He contrib-
uted immeasurably to the substance and clarity of the book. Others who
reviewed particular indicators are: Charles Anderson, Christopher Bach,
Bruce Baker, Steven Berman, Robert Callis, Brian Catron, Richard Curtin,
Dennis Duke, Lynn Franco, Jacob Frumkin, Samuel Frumkin, Sarah
Frumkin, Kenneth Goldstein, John Golmant, Kevin Hintzman, Thomas
Jabine, Mickey Kalavsky, Ralph Kauffman, Pamela Kelly, Kurt Kunze,
David Lassman, Patrick Lawler, Virginia Lewis, Mico Loretan, David Mead,
Charlotte Mueller, Ataman Ozyildirim, Karen Pence, David Roderer, Brooks
Robinson, Harry Rosenberg, Chris Savage, Scott Scheleur, Shelly Smith,
Tony Sznoluch, Kevin Thorpe, Katherine Wallman, and Tiffany Yanosky. I
had helpful discussions with Mary Bowler, Angie Clinton, Carmen
DeNavas-Walt, Gerhard Fries, Charles Gilbert, Daniel Ginsburg, Bruce
Grimm, Ryan Helwig, Paul Hanezaryk, James Herbert, David Hirschberg,
Patrick Jackman, Mila Kareva, Kim Kowalewski, Nellie Liang, Wayne
Lee, Peter Miller, Charles Nelson, William Nelson, Robert Parker, Al
Schwenk, Howard Silverblatt, Kenneth Stewart, Kathleen Short, and Daniel
Weinberg. Sylvia Elan assisted in proofreading the prepublication pages.
The editorial staff of M.E. Sharpe ensured the high quality of the publica-
tion: Amanda Allensworth prepared the manuscript for production; Angela
Piliouras was the production editor; Susanna Sharpe was the copyeditor;
Zeph Ernest was the typesetter; Denise Carlson prepared the index.
I thank all of the above for their help, which was essential, though
they may not agree with various aspects of the book. I am responsible
for everything in the book.
Data Sharing and Confidentiality of Federal Government

Statistical Data
A recent change affecting the production of federal government busi-

xvi PREFACE
ness statistics promises to have long-run benefits for the quality of the
data used in preparing economic indicators and for the efficiency of the
federal agencies in producing the statistics. This is the new legislation in
2002 that allows the sharing of statistical business data among the U.S.
Bureau of the Census, U.S. Bureau of Labor Statistics (BLS), and U.S.
Bureau of Economic Analysis (BEA) under prescribed procedures to
protect the confidentiality of the data provided by individual respon-
dents.3 The legislation limits the definition of the data to the descrip-
tion, estimation, and analysis of groups, while not permitting the
identification of individuals or organizations that make up the groups.
An ongoing business-data sharing activity under the new legislation
is the matching of BEAs surveys of foreign direct investment in the
United States and abroad with the Census Bureaus survey of industrial
research and development. This has led to (a) improved coverage of
companies in the Census Bureaus survey that previously were not iden-
tified as having research-and-development activities, and (b) the capa-
bility for the BEA to augment its research-and-development data with
information from the Census Bureaus survey, to identify quality issues
arising from reporting differences in the respective surveys, and to im-
prove its survey sample frames. Examples of planned data sharing in the
future are the matching of various elements of the business establish-
ment lists of the Census Bureau and BLS that will identify differences
in the lists, particularly for industry codes, which has long been consid-
ered an important problem; and the sharing of selected company and
revenue data from the BEAs international surveys with the international
price index program of the BLS, which will enable the BLS to study the
feasibility of developing a new international price index for royalties
and license fees, and subsequently, allow the BEA to compare selected
elements from its sample frame with the BLS sample frame to find
omissions in both lists. Other potential data-sharing areas include the
use of Census Bureau records to facilitate a more efficient sampling for
the BLS producer price index program, and the exchange by the BEA
and BLS of local area data within the United States to develop experi-
mental inter-area price indexes to permit comparisons of price levels
between local areas.
This business data sharing is grounded in several requirements of
the new legislation that strengthen the confidentiality of data provided
by respondents to federal government surveys. For planned data shar-
ing of mandatory surveys in which survey respondents are required by
PREFACE xvii
law to report, as distinct from voluntary surveys in which survey re-

spondents are not required by law to report, in those instances when
the respondents on mandatory surveys were not informed that the data
could be shared, the agencies must publish a notice of the proposed
data sharing in the Federal Register, specifying the data to be shared
and the statistical purposes for which the data are to be used, and al-
lowing a minimum of sixty days for public comment. Also, before
sharing any business data the agencies must enter into a written agree-
ment that specifies the data to be shared and the statistical purposes
for which the data are to be used; the officers, employees, and agents
authorized to examine the data to be shared; and the appropriate secu-
rity procedures to safeguard the confidentiality of the data provided
by individual respondents.
Using This Book
The more than sixty indicators in this book are classified in forty-
nine generic categories in alphabetic order. To help those with an
interest in particular aspects of the economy, the indicators are
grouped below under broad topics. The broad groupings are: eco-
nomic growth, household income and expenditure, business profits
and investment, labor, inflation and deflation, production, housing,
finance, government, international, cyclical indicators and forecast-
ing, economic well-being, and psychology. Obviously, several indi-
cators appear in multiple groupings.
ECONOMIC GROWTH
Gross Domestic Product

Industrial Production Index
HOUSEHOLD INCOME AND EXPENDITURE
Personal Income and Saving

Retail Sales
Consumer Credit
Consumer Credit Delinquency
Bankruptcies: Personal
Consumer Attitude Indexes
xviii PREFACE
Consumer Price Index
BUSINESS PROFITS AND INVESTMENT
Corporate Profits
LABOR
Employment
Average Weekly Hours
Average Weekly Earnings
Employment Cost Index
Job Openings and Labor Turnover
Unemployment
Help-Wanted Advertising Index
Productivity: Business Sector
Unit Labor Costs: Business Sector
INFLATION AND DEFLATION

Producer Price Indexes
Import and Export Price Indexes
GDP Price Measures
Farm Parity Ratio
Stock Market Price Indexes
PRODUCTION

Capacity Utilization
Manufacturers Orders
InventorySales Ratios
Manufacturing Business Activity Index
Non-Manufacturing Business Activity Index
PREFACE xix
HOUSING
Housing Starts
Home Sales: New and Existing Houses
Housing Vacancy Rates
Housing Affordability Index
Mortgage Delinquency and Foreclosure
FINANCE
Interest Rates
Stock Market Price Aggregates and Dividend Yields
Bank Loans: Commercial and Industrial
Consumer Credit
GOVERNMENT
Government Economic Transactions
INTERNATIONAL
Balance of Trade
Balance of Payments
International Investment Position of the United States
Value of the Dollar
Import and Export Price Indexes
CYCLICAL INDICATORS AND FORECASTING
Leading, Coincident, and Lagging Indexes

Experimental Recession Indexes
xx PREFACE
Manufacturing Business Activity Index

Non-Manufacturing Business Activity Index
ECONOMIC WELL-BEING
Distribution of Income
Distribution of Wealth
Poverty
Farm Parity Ratio
PSYCHOLOGY
Notes
1. Norman Frumkin, Tracking Americas Economy, 4th ed. (Armonk, NY: M.E.
Sharpe, 2004), ch. 1.
2. Frumkin, Tracking Americas Economy, pp. 303304.
3. Confidential Information Protection and Statistical Efficiency Act of 2002
(CIPSEA) as Title V of the E-Government Act of 2002 (Public Law 107347). For
an overall description of CIPSEA, see Executive Office of the President, Office of
Management and Budget, Statistical Programs of the United States Government:
Fiscal Year 2005. Washington, DC, 2004, pp. 4345. The Bureau of the Census and
the Bureau of Economic Analysis are in the U.S. Department of Commerce, and the
Bureau of Labor Statistics is in the U.S. Department of Labor.
Guide to
Economic
Indicators
1
Average Weekly Earnings
Average weekly earnings represent the money wages and salaries of

workers in private nonagricultural industries. Noncash fringe benefits
are excluded. Because wages and salaries reflect movements in employ-
ment between high-paying and low-paying industries and occupations,
the data are affected by changes in the composition of the industrial and
occupational job structure.
Where and When Available
Data on average weekly earnings of workers are prepared monthly by the

Bureau of Labor Statistics (BLS) in the U.S. Department of Labor. The
data are published in a news release and in two monthly BLS journals,
Monthly Labor Review and Employment and Earnings (www.bls.gov).
The data are available on the third Friday after the week containing the
twelfth of the month. Thus, the information is released on the first or
second Friday of the month following the month in question. On the day
the monthly numbers are published, the commissioner of labor statistics
reports on recent employment and unemployment trends to the Joint Eco-
nomic Committee of Congress. Preliminary data are provided for the im-
mediately preceding month; these are revised in the subsequent two months.
Annual revisions are made in February of the following year.
Content
Average weekly earnings data cover the wages and salaries of produc-
tion workers in manufacturing industries, as well as the wages, salaries,
3
4 GUIDE TO ECONOMIC INDICATORS
and commissions of nonsupervisory workers in other private nonagri-

cultural industries before the payment of income and Social Security
taxes. Earnings are excluded for office and sales workers in manufactur-
ing, for supervisors and executives in all industries, and for government
workers. The data reflect the effect of changes in the distribution of jobs
among industries and occupations. By contrast, the employment cost
index, which measures changes in labor costs to employers, maintains a
fixed composition of industries and occupations.
Earnings include wages for time at work and for paid vacations,
sick leave, holidays, and overtime (whether or not a premium is paid
for overtime). The data exclude health, retirement, and other non-
cash fringe benefits, the employer share of Social Security taxes,
bonuses, retroactive payments, tips, and in-kind payments such as
free rent and meals. The data are provided in current dollars and in
constant (1982) dollars.
The average weekly earnings data are seasonally adjusted.
Methodology
The data for weekly earnings are obtained from the establishment survey
used to measure employment. The methodology of the survey is described
under that indicator. There is no independent benchmark figure for weekly
earnings; the data are revised every February with the annual employment
benchmark to reflect revisions in the distribution of employment among
industries. Average weekly earnings are derived by multiplying average
weekly hours by average hourly earnings. Average hourly earnings are
estimated by dividing total wages and salaries by the number of hours of
production and nonsupervisory employees during the pay period. The
constant dollar data are calculated by dividing actual earnings by the con-
sumer price index for urban wage earners and clerical workers.
Accuracy
There are no sampling error estimates for the average weekly earnings data.
However, there are sampling error estimates for average hourly earnings
and average weekly hours, the components that are multiplied together to
obtain average weekly earnings (see Methodology above). The sampling
error (for one standard error) for both average hourly earning and average
weekly hours is 0.2 percent, which is indicative of what the error for average
AVERAGE WEEKLY EARNINGS 5
Table 1.1
Average Weekly Earnings in Private Nonagricultural Industries
Annual percentage change

Current dollars 1982 dollars Current dollars 1982 dollars
1995 399.53 258.43 2.3 0.6
1996 412.74 259.58 3.3 0.4
1997 431.25 265.22 4.5 2.2
1998 448.04 271.87 3.9 2.5
1999 462.49 274.64 3.2 1.0
2000 480.41 275.62 3.9 0.4
2001 493.20 275.38 2.7 0.1
2002 506.07 278.83 2.6 1.3
2003 517.30 278.72 2.2 0.0
2004 528.56 277.61 2.2 0.4
19952004 3.2 0.8
(annual average)
weekly earnings would be if it were available. For further information on

the interpretation of sampling and nonsampling errors, see the Appendix.
Relevance
Average weekly earnings data are based on job-related earnings of

workers of modest income in private nonfarm industies. This working
population of civilian family households and unrelated individuals
accounts for approximately 40 percent of the noninstitutional popula-
tion who are sixteen years of age and older (people outside of jails,
old-age homes, long-term medical care, and other sheltered housing).
The earnings data are relevant for several reasons. First, they provide a
measure of consumer purchasing power as indicated by changes in
wage earnings for an important segment of the spending public. Sec-
ond, because they gauge the economic well-being of ordinary work-
ers, their trends may suggest the direction of future wage demands.
Third, comparisons of wages and salaries in different industries indi-
cate that shifts of jobs among industries may affect wage earnings.
Recent Trends
From 1995 to 2004, average weekly earnings in current dollars in-

creased continuously, while earnings in price-adjusted (1982) dollars
declined in four of the ten years (Table 1.1). The largest increases in
price-adjusted dollars occurred in 1997 and 1998, and the second larg-
est increases were in 1999 and 2002. Price-adjusted dollars declined
in 1995, 2001, and 2004, and showed no change in 2003.
Over the entire nine-year period, the annual increases in current-
dollar earnings averaged 3.2 percent, and the annual increases and de-
creases in price-adjusted 1982 dollar earnings increased at an average
annual rate of 0.8 percent.
References from Primary Data Source
Bureau of Labor Statistics, U.S. Department of Labor. Monthly Labor Review and
Employment and Earnings. Monthly.
2
Average Weekly Hours
Average weekly hours represent the length of the workweek in private

nonagricultural industries. The data are affected by changes in both the
industrial and occupational composition of employment.
Data on average weekly hours in the workplace are prepared monthly

by the Bureau of Labor Statistics (BLS) in the U.S. Department of La-
bor. The data are published in a news release and two monthly BLS
journals, Monthly Labor Review and Employment and Earnings
(www.bls.gov).
The data are available on the third Friday after the week containing
the twelfth of the month. Thus, the information is released on the first
or second Friday of the month following the month in question. On the
day the monthly numbers are released, the commissioner of labor sta-
tistics reports on recent employment and unemployment trends to the
Joint Economic Committee of Congress. Preliminary data are provided
for the immediately preceding month; these are revised in the subse-
quent two months. Annual revisions are made in February of the fol-
lowing year.
Content
Average weekly hours measure time on the job, including straight-time

and overtime hours (whether or not a premium is paid for overtime), for
7
the average of full-time and part-time workers in private nonagricul-

tural industries. The inclusion of paid absences from work means that
hours are counted on the basis of hours paid for rather than hours
worked. The hours information covers production workers in manufac-
turing industries and nonsupervisory workers in other industries. Hours
for office and sales workers in manufacturing, for executives in all in-
dustries, and for government workers are excluded.
Separate data on overtime hours are provided for manufacturing in-
dustries. These are defined to include work time for which premium pay
is received beyond the straight-time workday or workweek. Holiday hours
are included only if premium wages are paid. Hours associated with
incentive pay for shift differentials such as night or weekend work, haz-
ardous conditions, or similar situations are excluded. Because these sepa-
rate data on overtime are limited to work for premium pay, they differ
from the overtime included in average weekly hours, which includes all
overtime even if no premium pay is involved.
The average weekly hours data are seasonally adjusted.
Methodology
The data for weekly hours are obtained from the establishment survey
used to measure employment. The methodology of the survey is described
under that indicator. There are no independent benchmark data for weekly
hours; they are revised every February with the annual employment
benchmark because of revisions in the composition of employment
among industries. Weekly hours are derived by dividing total hours paid
for by the number of employees during the pay period. These are ad-
justed for pay periods that are longer than one week so that they repre-
sent a seven-day period. Separate data are collected on the survey form
for overtime hours.
Accuracy
The sampling error (for one standard error) for the average weekly hours
data in private industries is 0.2 percent. For example, if the estimated
average weekly hours for private industries were 35 hours, in two of
three cases the true level would be somewhere between 34.93 and
35.07 hours. For further information on the interpretation of sampling
and nonsampling errors, see the Appendix.
AVERAGE WEEKLY HOURS 9
Table 2.1
Average Weekly Hours in Private Nonagricultural Industries
All private
nonagricultural Manufacturing Overtime in
industries industries manufacturing
1995 34.3 41.3 4.7
1996 34.3 41.3 4.8
1997 34.5 41.7 5.1
1998 34.5 41.4 4.8
1999 34.3 41.4 4.8
2000 34.3 41.3 4.7
2001 34.0 40.3 4.0
2002 33.9 40.5 4.2
2003 33.7 40.4 4.2
2004 33.7 40.8 4.6
Relevance
Average weekly hours is a sensitive barometer of labor demand. Em-

ployers generally prefer to increase or decrease hours worked before
hiring or laying off workers in response to movements in retail sales,
corporate profits, manufacturers orders, inventorysales ratios, or
planned production schedules. This is particularly true when the changes
in the demand for labor are small or are expected to be temporary.
Weekly hours in manufacturing is a component of the leading index
of leading, coincident, and lagging indexes. The monthly volatility of
weekly hours makes discerning a short-term trend difficult. Movements
over several months should be assessed when analyzing trends.
Recent Trends
From 1995 to 2004, average weekly hours in all private nonagricul-

tural industries peaked at 34.5 hours in 199798, and then declined to
33.7 hours in 200304 (Table 2.1). Weekly hours in manufacturing
industries were 6 to 7 hours higher than those in all nonagricultural
industries from 1995 to 2004. Manufacturing hours peaked at 41.7
hours in 1997, declined to an average of 40.4 hours during 200103,
and rose to 40.8 hours in 2004. Overtime in manufacturing industries
peaked at 5.1 hours in 1997, declined to 4.0 hours in 2001, and then
rose to 4.6 hours in 2004.
3
Balance of Payments
The balance-of-payments accounts are the most comprehensive measure

of U.S. international economic transactions with other countries. The trans-
actions include exports and imports of goods and services, income re-
ceipts and payments on foreign investments, transfer payments such as
pensions and government grants, and changes in U.S. and foreign hold-
ings of financial assets and liabilities associated with international mon-
etary reserves, direct investment transactions, banking and nonbanking
transctions, and securities transactions. The various balances focus on the
difference between exports and imports and international flows of trans-
fer payments, but exclude changes in financial assets and liabilities. When
exports exceed imports, the balance is in surplus, and when imports ex-
ceed exports, the balance is in deficit. A surplus balance is sometimes
referred to as favorable and a deficit as unfavorable.
The Bureau of Economic Analysis (BEA) in the U.S. Department of

Commerce provides quarterly measures of the balance of payments. They
are published in a news release and in the BEA monthly journal, Survey
of Current Business (www.bea.gov).
The data are available seventy to seventy-five days after the end of
the quarter to which they refer. They are initially revised in the succeed-
ing quarter and then subsequently in June of the following year as part
of the annual revisions. The annual revisions also change the data for
several of the preceding years.
11
Content
The balance of payments has two broad components. One is foreign trade
in goods, services, investment income, and unilateral transfers. The other
is the money and capital flows necessary to finance trade, transfers, and
grants. The two components are definitionally equivalent but do not match
statistically because of inadequacies in the data. The difference caused by
these data problems is noted as the statistical discrepancy.
Balance-of-payments data are provided for total U.S. transactions with
all nations and separate transactions with particular nations and regions
of the world. The United States includes the fifty states and the District
of Columbia, Puerto Rico (except transactions between the states and
Puerto Rico), and the Virgin Islands.
The balance-of-payments data are in current dollars. They are con-
verted to constant dollars for the gross domestic product.
Several elements make up the foreign trade, transfer, and grant cat-
egories of the balance of payments. Exports and imports of goods, ser-
vices, and income encompass merchandise trade in the balance of trade
plus the following services and income: transfers under the foreign mili-
tary sales program; defense purchases; travel, passenger, and freight trans-
portation between the United States and other countries provided by
American and foreign companies; other services provided by Ameri-
cans and foreigners, such as insurance, telecommunications, construc-
tion, and engineering; royalties and license fees; and dividend and interest
income paid by Americans and foreigners on foreign investments.
Unilateral transfers are transactions between U.S. residents and resi-
dents of foreign countries in which goods, services, or financial assets
are transferred and nothing of economic value is received in return. Ex-
amples include U.S. government military and nonmilitary grants for
which no payment is expected or where the payment terms are agreed to
at a future time after the transfer occurs; private and government pen-
sion payments to American workers living in foreign countries and by
other nations to foreign workers living in the United States; and gifts
sent abroad by individuals and nonprofit organizations.
Increases or decreases in U.S. assets abroad and foreign assets in the
United States measure the means of financing mentioned above foreign
trade in goods and services, unilateral transfers, and military grants.
The main elements of U.S. assets abroad are as follows. U.S. govern-
ment official reserve assets include the U.S. gold stock, special drawing
BALANCE OF PAYMENTS 13
rights and the reserve position in the International Monetary Fund (IMF),
and U.S. Treasury and Federal Reserve holdings of foreign currencies.
Other government assets include loans to foreign nations and to U.S.
private parties for investment abroad, capital contributions to interna-
tional organizations except the IMF, and U.S. holdings of foreign cur-
rencies and other short-term assets associated with foreign-aid
programs and financial operations such as guarantee programs of the
Export-Import Bank. U.S. private assets include direct investment
abroad (ownership of at least 10 percent of foreign companies) by U.S.
private parties, U.S. private holdings of foreign bonds and stocks, and
U.S. bank and nonbank loans to foreigners.
The main elements of foreign assets in the United States are as fol-
lows. Foreign official assets are investments by foreign governments in
U.S. government securities, U.S. government liabilities for foreign de-
posits in advance of delivery of foreign military sales items, and foreign
government holdings of U.S. corporate debt and equity securities and of
state and local government securities. Other foreign assets are direct
investment in the United States (ownership of at least 10 percent of
American companies) by foreign private parties; private foreign hold-
ings of U.S. Treasury securities, state and local government securities,
and corporate debt and equity securities, and loans to Americans by
foreign banks and nonbanks.
The indicator provides four separate balances of exports minus im-
ports: (1) goods trade; (2) goods and services; (3) investment income;
and (4) balance on current account (goods, services, income, and all
unilateral current transfers). Balances are not calculated for changes in
financial assets and liabilities because meaningful distinctions are diffi-
cult to make for such categories as short-term and long-term capital.
The balance-of-payments data are seasonally adjusted when seasonal
patterns are present.
Methodology
The database used in preparing the balance of payments comes from sev-
eral sources. Data for goods exports and imports are based mainly on
Census Bureau surveys (see balance of trade). The main sources for other
components are: U.S. International Trade Administration surveys of aver-
age international traveler expenditures and U.S. Department of Home-
land Security data on the number of travelers; BEA surveys of international
operations of U.S. and foreign ship operators and airlines; Census Bureau
data on the tonnage of merchandise exports and imports; reports by the
Department of Defense on foreign military sales and the Department of
Agriculture on foreign-aid shipments of food; BEA surveys of incoming
and outgoing foreign direct investment, and Treasury Department surveys
(conducted by the Federal Reserve Bank of New York) of international
assets and liabilities of U.S. banks and nonbank companies.
The quarterly measures are based on reported data for most items,
and estimates for those for which reported data are available either an-
nually or less frequently. They are revised every June when more com-
plete information is available. These revisions change some of the
components for the past three to five years.
The statistical discrepancy is defined as the accounting difference be-
tween the sums of credits and debits in the balance of payments. Credits
are exports of goods, services, and income; unilateral transfers to the United
States; capital inflows or a decrease in U.S. assets; a decrease in U.S.
official assets, and an increase in foreign official assets in the United States.
Debits are imports of goods, services, and income; unilateral transfers to
foreigners; capital outflows or an increase in U.S. assets; an increase in
U.S. official reserve assets, and a decrease in foreign official assets in the
United States. A discrepancy results from the fact that data for the various
components are developed independently and, consequently, are not fully
consistent in coverage, definition, timing, and accuracy. The discrepancy
is a net figure in which overstatement of one data element is offset by
understatement of another data element. When the discrepancy is posi-
tive, it signifies unrecorded funds entering the United States; a negative
discrepancy indicates unrecorded funds leaving the United States.
Accuracy
There are no estimates of sampling or revision error in the balance-of-

payments data. Since offsetting errors among the data elements may
reduce the statistical discrepancy, that figure provides an overall mini-
mum magnitude of the net inconsistencies in the various data sources.
Relevance
The balance of payments reflects U.S. participation in world markets

overall. It points up the relative importance of international product
BALANCE OF PAYMENTS 15
Table 3.1
U.S. Balance of Payments (billions of dollars)
Balance on goods,
services, and income Statistical discrepancy
1995 75.5 28.3
1996 81.7 12.2
1997 95.7 79.4
1998 160.7 145.0
1999 249.5 68.8
2000 357.2 69.4
2001 337.5 9.6
2002 411.2 23.7
2003 448.5 37.8
2004 587.1 85.1
Balance on goods Balance on services Balance on income

2000 452.4 74.1 21.1
2001 427.2 64.5 25.2
2002 482.3 61.1 10.0
2003 547.3 52.5 46.3
2004 665.4 47.8 30.4
markets to the American economy and indicates those markets that are
gaining or losing ground. It also highlights shifts in international invest-
ment, including the effect on interest flows and dividend flows entering
and leaving the United States. The extent to which the U.S. consumes
and produces for world markets affects the gross domestic product.
The impact of international transactions, including their financing,
affects the value of the dollar and American competitiveness. When
Americans spend and invest more money abroad than foreigners spend
and invest in the United States, the value of the dollar tends to decrease;
greater spending and investment by foreigners in the United States tends
to raise the dollar.
A large balance-of-payments deficit limits the flexibility of the Fed-
eral Reserve in conducting monetary policy (see balance of trade). Large
deficits also create a growing foreign debt that raises interest payments
to foreigners and thereby reduces the standard of living for Americans.
Recent Trends
From 1995 to 2004, the balance on goods, services, and income was
consistently negative, in which imports exceeded exports (Table 3.1).
This general pattern was accompanied by continuously increasing nega-

tive movements during the period, with the negative balance rising from
$75.5 billion in 1995 to $587.1 billion in 2004.
For the goods, services, and income components from 2000 to 2004,
goods continuously accounted for the entire increasing overall negative
balance noted above, amounting to $665.4 billion in 2004. Services had
a positive balance of $47.8 billion in 2004, though the positive balance
declined continuously from 2000. Income had a positive balance of $30.4
billion in 2004, with a fluctuating positive balance during 200004.
The statistical discrepancy fluctuated between a positive and nega-
tive position during 19952004, ranging from a positive $134.6 billion
in 1998 to a negative $95.0 billion in 2002. The statistical discrepancy
was a positive $51.9 billion in 2004.
Reference from Primary Data Source
Bureau of Economic Analysis, U.S. Department of Commerce. Survey of Current

Business. Monthly.
4
Balance of Trade
The balance of trade represents U.S. foreign trade in merchandise. Mer-

chandise is goods, as distinct from services. The balance is the differ-
ence between exports and imports. When exports exceed imports, the
balance is in surplus, and when imports exceed exports, the balance is in
deficit. A surplus is sometimes referred to as a favorable balance and
a deficit as an unfavorable balance (see balance of payments).
Two agencies provide balance-of-trade figures. The Bureau of the Cen-

sus in the U.S. Department of Commerce provides monthly data, and
the Bureau of Economic Analysis (BEA) in the U.S. Department of
Commerce provides monthly and quarterly data. The data from the two
agencies are published in a joint news release. The BEA data are pub-
lished in more detail in the BEA monthly journal, Survey of Current
Business (www.bea.gov). The Census Bureau data are published in more
detail in the FT 900 Supplement (www.census.gov).
The balance-of-trade data are available 45 to 50 days after the month
to which they refer. They are initially revised the following month and
subsequently in quarterly and annual data as part of the balance of
payments.
Content
Merchandise export and import data are provided for U.S. total foreign
trade with all nations, plus detail for trade with particular nations and
17
regions of the world, as well as for individual commodities. U.S. trade

includes that of the fifty states and the District of Columbia, Puerto
Rico, and the Virgin Islands. The trade data exclude shipments between
the United States and the commonwealth of Puerto Rico, the Virgin Is-
lands, and other U.S. possessions; however, supplementary data are pro-
vided on U.S. export and import trade with Puerto Rico, U.S. exports to
the Virgin Islands, and imports from Guam, American Samoa, and the
Northern Mariana Islands.
Constant-dollar inflation-adjusted measures of exports and imports
are also prepared.
Exports cover domestically produced goods plus imported items that
subsequently are exported without substantial physical change to the im-
ported item (referred to as a re-export). Exports are valued at the dollar
price at the U.S. port of export. This includes inland transportation, insur-
ance, and other costs to deliver the merchandise alongside the ship or
plane, but it excludes overseas transportation, insurance, and other charges
beyond the U.S. port (referred to as f.a.s., free alongside ship). The month
of exportation is the month in which the shipment leaves the United States.
Imports cover goods for immediate consumption plus those stored in
Customs bonded warehouses and in U.S. Foreign Trade Zones. They are
valued at the f.a.s. price at the foreign port of export (see exports in the
above paragraph). The month of importation is within approximately
ten days after the merchandise enters a Customs warehouse.
The distinction between the Census Bureau and the BEA data is
that BEA adjusts the Census measures to conform to the balance- of-
payments definitions. This results in three primary differences: (1) for-
eign military sales and U.S. military agencies purchases from abroad
identified in the Census Bureau documents are excluded by BEA but
included elsewhere in the balance-of-payments accounts; (2) for im-
ports from Canada, inland freight costs for transporting goods from
the point of origin in Canada to the Canadian border are excluded in
the Census data and included in the BEA data; (3) Census Bureau data
include only nonmonetary gold that is shipped across international
borders, while BEA data also include nonmonetary gold that changes
ownership through book entries without being shipped across interna-
tional borders. Nonmonetary gold represents all trade in gold in which
at least one of the parties to the transaction is a private party; it ex-
cludes gold movements between governments, central banks, and in-
ternational monetary institutions. Monetary gold represents gold
BALANCE OF TRADE 19
movements between the U.S. Treasury or the Federal Reserve Board

acting for the Treasury and foreign governments or their central banks
and the International Monetary Fund.
The balance-of-trade data are seasonally adjusted.
Methodology
The basic data on merchandise exports and imports are developed from
surveys conducted by the Census Bureau. They are adjusted by the BEA
to reflect the balance-of-payments definitions.
Bureau of the Census
The export statistics are derived mainly from mandatory information

supplied by commercial exporters to the Customs Bureau, which pro-
vides the data to the Census Bureau. The Customs Bureau checks ex-
ports requiring licenses from the State Department for military items
and from the Commerce Department for nonmilitary strategic materi-
als. These data are supplemented by data from some exporters who re-
port their shipments directly to the Census Bureau. In addition, the
Department of Defense reports military aid shipments data to the Cen-
sus Bureau. Export data for shipments over $2,500 are compiled from
the universe (100 percent sample) of all such reports. Low-valued ex-
ports of $2,500 and under accounted for 2.2 percent of all exports in
1997. Shipments of $2,500 and under are estimated from factors based
on ratios of low-valued exports to total exports in past periods. The thresh-
old dollar exemption for low-valued exports from reporting requirements
is raised from time to time, most recently in October 1989.
The import statistics are derived from mandatory information sup-
plied by importers to the Customs Bureau, which reviews the documents
for accuracy and provides the corrected data to Census. Import data for
shipments over $2,000, or over $250 for certain goods entering under
quota, are compiled from the universe (100 percent sample) of all such
reports. Low-valued imports of $2,000 and under, and under $250 for
certain imports under quotas, accounted for 0.7 percent of all imports in
1997. Shipments of $2,000 and under are estimated from factors based
on ratios of low-valued imports to total imports for past periods. The
threshold dollar exemption, exempting low-valued imports from report-
ing requirements, is raised from time to time, most recently in July 1998.
The constant-dollar measures are derived mainly by deflating the

current-dollar data by the producer price indexes.
Bureau of Economic Analysis
The statistical adjustments made to conform the Census Bureau data to

the balance-of-payments definitions are based on separate information
obtained from a variety of sources such as the Department of Defense
for military exports, the Census Bureau data for U.S. inland freight, the
BEA reconciliations with Statistics Canada for inland freight costs for
Canadian foreign trade, and the Federal Reserve Board for nonmon-
etary gold trade.
Accuracy
There are no estimates of sampling or revision error for the balance-of-trade

data. Practically all of the Census data are based on surveys of the universe
of exporters and importers.
Relevance
The balance of trade impacts the gross domestic product, employment,

and the value of the dollar. Export and import levels are influenced by
economic growth at home and abroad and by the competitive position
of American products in international markets and foreign goods in
U.S. markets.
A surplus in the trade balance or a reduction in the trade deficit
reflects increased economic growth and job expansion, while a deficit
or reduction in the surplus reflects decreased economic growth and
employment. This occurs because exports are produced in the United
States and thus generate American production and employment, while
American spending for imports stimulates production and employment
abroad.
Over the long run, imports tend to hold down inflation because
imports compete with American goods. Imports also moderate infla-
tion during temporary shortages of domestic goods by providing a
supplementary supply. Shortages may occur when drought or frost
reduces food harvests; when an unexpected surge occurs in consump-
tion; or when sudden bottlenecks appear in the production of lum-
BALANCE OF TRADE 21
Table 4.1
U.S. Balance of Trade (billions of dollars)
BEA definition Census Bureau definition

1995 174.2 158.8
1996 191.0 170.2
1997 198.1 180.5
1998 246.7 229.8
1999 346.0 328.8
2000 452.4 436.1
2001 427.2 411.9
2002 482.3 468.3
2003 547.3 532.4
2004 665.4 650.8
ber, paper, or other products for which domestic supply cannot be

expanded readily.
The balance of trade also affects the conduct of U.S. economic policy.
Thus, a large trade deficit limits the flexibility of the Federal Reserve in
conducting monetary policy for influencing the economy. The trade defi-
cit is financed by borrowing from domestic lenders or from abroad. Bor-
rowing from domestic lenders to finance the deficit could lead to higher
interest rates, unless accommodated by an increase in bank credit, which
in turn may lead to higher inflation. Borrowing from abroad can lead to a
rise in the value of the dollar: the influx of foreign funds into the United
States bids up the dollar compared with other currencies, which worsens
the deficit by making exports more expensive and imports cheaper.
The balance of trade also may affect U.S. living conditions. A con-
tinuing large deficit financed from abroad creates growing foreign debt.
Over the long run, this results in greater amounts of money paid in inter-
est as payments to foreigners. Consequently, U.S. incomes and living
conditions are reduced.
Recent Trends
From 1995 to 2004, the balance of trade under both the BEA and Cen-
sus Bureau definitions showed continuous increased negative balances,
that is, increasing deficits (Table 4.1). The trade deficit under the BEA
definition exceeded that under the Census Bureau definition by $15 to
$20 billion in all years. The BEA deficit was $665.4 billion and the
Census Bureau deficit was $650.8 billion in 2004.
References from Primary Data Sources
Bureau of the Census, U.S. Department of Commerce. FT 900 Supplement. Monthly.

Bureau of Economic Analysis and Bureau of the Census, U.S. Department of Com-
merce. U.S. International Trade in Goods and Services. Monthly.
Business. Monthly.
5
Bank Loans: Commercial and
Industrial
Commercial and industrial bank loans are loans made by commercial

banks to individuals, partnerships, and corporations for nonfarm busi-
ness use. They also include bank loans made to investors for financial
acquisitions such as company takeovers. The indicator thus focuses on
loans made to income-generating business activity, as distinct from house-
hold consumer use.
Commercial and industrial bank loan data are provided weekly and monthly
by the Federal Reserve Board (FRB). They are published in a statistical
release (H.8) and in the monthly statistical supplement to the FRB quar-
terly journal, Federal Reserve Bulletin (www.federalreserve.gov).
The weekly data are available every Friday for the week ending
Wednesday of the previous week. The monthly data are available on the
second Friday of the month after the month to which they refer. The
measures are revised on a continuing basis with the receipt of more
accurate data.
Content
Commercial and industrial bank loans represent loans outstanding. They

cover existing loans from the previous period and new loans, minus those
repaid, sold, or securitized outside the system (e.g., collateralized loan
obligations) during the period. Secured and unsecured loans are included.
In addition to traditional loans extended to borrowers, the data include
23
banks own acceptances (bills for which banks pledge their credit on
behalf of their customers).
The data exclude loans to farmers, securities and real estate firms,
other banks, and companies that mainly extend business or personal
credit; commercial paper of financial institutions bought by banks; and
loans secured by real estate. No data are available on the distribution of
short-term and long-term loans.
The bank loan data are seasonally adjusted.
Methodology
The bank loan data are obtained from weekly reports of a nonprobability
sample of Federal Reserve member and nonmember banks, both large
and small, and from quarterly reports for banks not reporting weekly.
The weekly and monthly data for all commercial banks include esti-
mates for banks not reporting weekly. Estimates for the nonweekly re-
porting banks are based on relationships developed from the quarterly
reports of all banks, those reporting weekly and those reporting quar-
terly. The bank loan data are benchmarked to the quarterly reports four
times a year.
Accuracy
Estimates of revisions to the bank loan figures are encompassed within a

larger statistical category of bank credit, which includes commercial
and industrial and other bank loans plus U.S. government and other secu-
rities owned by banks. Revisions for this much broader category are within
plus or minus 0.5 percentage point of the annual growth rate of bank credit.
The commercial and industrial bank loan component probably has a larger
revision error, although the actual range is not known.
Relevance
The bank loan data provide a clue to businesss willingness to go into

debt. For analytical purposes, the monthly movement and the monthly
level are viewed differently. A rapid increase in bank loans suggests an
optimistic outlook for business prospects, a slow rate of loan expansion
indicates a cautious business outlook, while a decrease suggests a greater
business emphasis on paying off existing loans.
BANK LOANS: COMMERCIAL AND INDUSTRIAL 25
Table 5.1
Bank Loans: Commercial and Industrial
Loans outstanding (billions of Annual percentage change

dollars) December December to December
1995 723.8 11.3
1996 784.7 8.4
1997 854.1 8.8
1998 947.4 10.9
1999 998.8 5.4
2000 1,087.0 8.8
2001 1,027.2 5.5
2002 963.1 6.2
2003 891.6 7.4
2004 911.4 1.9
By contrast, the simple existence of debt is a burden to business be-

cause of the principal and interest payments. Thus, existing debt be-
comes a depressant to further borrowing. The monthly level of existing
commercial and industrial loans (in constant dollars) is a component of
the lagging index of the leading, coincident, and lagging indexes.
Recent Trends
From 1995 to 2004, bank loans to commercial and industrial borrowers

showed diverse movements over the nine-year period (Table 5.1). Bank
loans had fluctuating increases of 5 to 11 percent from 1995 to 2000,
and after falling by a pace that accelerated from 5.5 percent in 2001 to
7.4 percent in 2003, rose by 2.2 percent in 2004. Bank loans ranged
from a low of $724 billion in 1995 to a high of $1,087 billion in 2000.
Board of Governors of the Federal Reserve System. Statistical Supplement to the

Federal Reserve Bulletin, the Statistical Supplement is monthly, and the Federal
Reserve Bulletin, is quarterly.
6
Personal bankruptcy data highlight requests that household debtors be

declared in bankruptcy and that arrangements be made for a resolution
of creditor claims for monetary resolution. The requests, which are termed
filings, may be made by the debtors themselves (voluntary filings)
or by their creditors (involuntary filings).
The personal bankruptcy data are provided quarterly by the Administra-

tive Office of the United States Courts, Statistics Division. They are
published in Statistical Tables for the Federal Judiciary (www.uscourts.
gov.bnkrptcystats/bankruptcystats.htm).
The data are available approximately seven weeks after the calendar
quarter to which they refer. They represent the total for the quarter and
the cumulated annual total of the most recent four quarters. For example,
the January 1 to December 31 data combine the four calendar quarters
from January to December, and the April 1 to March 31 data combine
the four calendar quarters from April to March. Revised data are incor-
porated as they become available.
Content
The purpose of declaring bankruptcy is to obtain a fair resolution of

overdue debts owed creditors based on the debtors financial assets and
needs, and to give the debtor a financial fresh start in life. A personal
bankruptcy occurs when a U.S. court declares an individual who has
26
BANKRUPTCIES: PERSONAL 27
overdue personal debts to be financially incapable of paying some or all

of the debts, and thus discharges those debts. The bankruptcy protects
the debtors assets from the claims of creditors in whole or in part, de-
pending on the terms of the bankruptcy.
The personal bankruptcy data (officially referred to as nonbusiness
bankruptcy filings) focus on the number of filings to the U.S. courts by
household debtors or their creditors requesting that the debtors be de-
clared in bankruptcy. The court completes its work when it acts to dis-
miss the request or refers it to the Office of the U.S Trustee of the U.S.
Department of Justice. The Office of the U.S. Trustee is responsible for
the specific resolution of how the debts will be satisfied. This indicator
is confined to filings; it does not include the later resolution of cases by
the Trustee Office.
The federal bankruptcy statute provides for different categories of
liability.1 These vary by the type and amount of a debtors assets that are
protected from being distributed to creditors. The two major variants of
the statute for households are Chapters 7 and 13 (these chapters include
business bankruptcies, but the data here are confined to personal bank-
ruptcies). Chapter 7 bankruptcies absolve household debtors from part
or all debt by discharging them from the liability of the debt that is
discharged. Chapter 7 may include the sale of the debtors properties
(liquidation) that are not exempt from the claims of creditors, with the
proceeds of the sale used to pay certain creditor claims. Chapter 13 gives
household debtors a temporary respite in paying creditors in order to
develop a financial plan such as budgeting part of the debtors future
earnings to pay part or all of the debt. Chapter 13 protection is limited to
debtors with liabilities below certain levels for secured and unsecured
debt.2 Debtors having liabilities above these thresholds must use Chap-
ter 11. Chapter 11 contains financial reorganization plans associated
with secured and unsecured debt for both personal and business bank-
ruptcies; only the personal component is included in the data here. Chap-
ter 12 gives bankruptcy protection to a family farmer with a regular
income; it is a business bankruptcy and is not included here.
Some persons who can pay part or all of their debts without suffering
undue hardship might declare bankruptcy to avoid paying their debts.
There are no comprehensive data on the extent of such abuse of the
bankruptcy system.
The personal bankruptcy data are not seasonally adjusted. However,
users of the data have estimated seasonal patterns.
Methodology
The personal bankruptcy data are prepared by the Administrative Office

of the United States Courts, Statistics Division. The data are obtained
electronically from the ninety-four U.S. bankruptcy courts around the
country. The data are obtained from the universe of all bankruptcy fil-
ings. Revised data are incorporated in each quarterly tabulation as more
complete or accurate data are received from the district courts. There is
no formal schedule for revisions.
Accuracy
There are no estimates of revision error for the personal bankruptcy

data.
Relevance
Personal bankruptcies result from many factors, such as excessive house-

hold debt in relation to income, issuance of credit cards to
noncreditworthy households, and a lessening social stigma associated
with bankruptcy.3 (Estimates of the stigma effect are based on indirect
statistical analysis, because direct measures of stigma are not available.4)
Bankruptcy typically becomes a possibility when household debt loads
(consumer and mortgage scheduled debt-service payments in relation
to disposable personal income (see personal income and saving)are
high, and persons suffer a disruption of income due to unemployment,
business losses, accidents, sickness, divorce, or other unexpected events
that significantly diminish income or increase expenses.
What is the impact of bankruptcy on the economy? First, it is impor-
tant to recognize that bankruptcy is a culmination of many problems
associated with households being overextended in debt, such as those
noted above. Thus, bankruptcy is aimed at ending the treadmill of ever-
increasing debt. Once bankruptcy has occurred, the effects on prices
(consumer price index, GDP price measures, wholesale price indexes),
interest rates, profits (corporate profits), and taxes (government eco-
nomic transactions) depend on actions taken by creditors who have lost
income due to unpaid debts.5 If creditors pass the costs along to con-
sumers, they lead to higher prices and interest rates. If creditors absorb
the costs, they lead to lower profits and in turn lower tax receipts.
BANKRUPTCIES: PERSONAL 29
Table 6.1
Bankruptcy Filings: Personal (Chapters 7, 11, 13)
Year ending December Filings Annual percentage change

1995 874,642 12.1
1996 1,125,006 28.6
1997 1,350,118 20.0
1998 1,441,701 6.8
1999 1,281,586 11.1
2000 1,217,972 5.0
2001 1,452,030 19.2
2002 1,539,555 6.0
2003 1,625,208 5.6
2004 1,563,145 3.8
19952004 (annual average) 6.7
Recent Trends
From 1995 to 2004, personal bankruptcy filings rose in each year, ex-
cept for declines in 1999, 2000, and 2004 (Table 6.1). The year-to-year
percentage changes fluctuated considerably. The average annual change
from 1995 to 2004 was an increase of 6.7 percent.
The 1,563,145 filings in 2004 included 1,117,766 under Chapter 7;
946 under Chapter 11; and 444,428 under Chapter 13 (the total filings
figure included five cases under other chapters of the federal bankruptcy
statute). There are no data distinguishing voluntary from involuntary
filings, although the bulk are thought to be voluntary.
Administrative Office of the United States Courts, Statistics Division. Statistical

Tables for the Federal Judiciary. Quarterly.
Notes
1. For a comprehensive coverage of personal bankruptcy, see Kim Kowalewski,

Personal Bankruptcy: A Literature Review, CBO Paper, Congressional Budget Of-
fice, September 2000.
The Bankruptcy Reform Act of 1978 established minimum federal standards
for the type of property and the dollar value of debt that are protected from payment
to creditors. The Bankruptcy Reform Act of 1994 raised minimum-protection lev-
els. The federal statute allows states to raise the dollar value and add to the types of
property above the federal minimums, but all bankruptcy proceedings are conducted
in federal courts. The Bankruptcy Abuse, Prevention, and Consumer Protection Act
of 2005 made it more difficult for debtors to file for bankruptcy under Chapter 7
without paying some of their debt. In addition, the Act allowed state-based asset
protection trusts (also called self-settled trusts) to be exempt from payments to credi-
tors in bankruptcy proceedings, except for those trusts that are shown by the Trustee
to have been established with the actual intent to hinder, delay, or defraud the
payment of a particular claim.
2. Secured debt represents a loan backed by a borrowers collateral, such as a
house, car, or financial securities. Credit card debt is unsecured debt.
3. Paul Paquin, and Melissa Squire Weiss, An Analysis of the Determinants of
Personal Bankruptcies, Capital One Financial Corporation, Falls Church, VA (Oc-
tober 1997); and Visa U.S.A. Inc., Consumer Bankruptcy: Causes and Implications,
July 1996.
4. Kowalewski, Personal Bankruptcy, p. xi, and Appendix B, p. 45.
5. WEFA Group, Resource Planning Service, The Financial Costs of Personal
Bankruptcy, Burlington, MA, February 1998. The study was funded by Visa and
MasterCard.
7
Capacity Utilization
Capacity utilization measures the proportion of equipment and structures

used in production by the manufacturing, mining, electric, and gas utilities
industries. It covers the same industries as the industrial production index.
When production rises faster than capacity, the capacity utilization rate (CUR)
increases, but when production rises more slowly (or declines), the CUR
decreases.
Capacity utilization is prepared monthly by the Federal Reserve Board

(FRB). It is published in a statistical release (G.17) and in the monthly
statistical supplement to the quarterly FRB journal, the Federal Reserve
Bulletin (www.federalreserve.gov).
The data are available in the middle of the month after the month to
which they refer, the same day as the industrial production index. Pre-
liminary data are provided for the preceding month; these are revised
in the subsequent three months. Annual revisions are made in the fall.
Content
The capacity utilization rate, expressed as a percentage, is the ratio of

the industrial production index to equipment and structures capacity.
The formula is:
Industrial production index

CUR = 100 (7.1)
Equipment and structures capacity
31
Because the numerator is discussed in the chapter on the industrial pro-

duction index, this section explains the denominator, or capacity.
The capacity number represents the economys ability to produce
goods and power assuming the existing equipment and structures facili-
ties are used over the normal operating period for each industrythis
ranges from a typical 35-to-40-hour workweek to continuous operations
seven days a week. The capacity measure aims at achieving a level of
sustainable maximum output within a realistic work schedule that al-
lows for normal downtime and assumes a sufficient availability of labor,
materials, and services used in production. Capacity gradually increases
over time as each year more equipment and structures investment is
added than physical facilities are scrapped. This long-term upward trend
shows no cyclical fluctuation. Industries rarely operate at a CUR of 100
percent and thus typically have unused or spare capacity available to
expand production when demand increases.
The industrial composition of the CUR is made up of manufacturing
(84 percent), mining (7 percent), and utilities (9 percent).
The CUR data are seasonally adjusted.
Methodology
Because the industrial production index in the numerator of the CUR is
already prepared, the primary task for developing the CUR is to provide
a measure of capacity. For most industries, direct measures of capacity,
such as the number of items that can be produced if the industry is oper-
ating at a CUR of 100 percent, are not available. Consequently, indirect
measures of capacity are widely used. For most manufacturing indus-
tries, these are derived from year-end surveys of capacity utilization
conducted in manufacturing industries by the Census Bureau (Survey of
Plant Capacity). Capacity is inferred from these year-end CUR survey
data by dividing them by the industrial production index for each indus-
try.1 The resultant capacity numbers are modified to reflect supplemen-
tary information on direct measures of capacity for selected manufacturing
industries and on the value of stock of existing capital facilities derived
from a perpetual inventory of investment data.
Capacity estimates in the mining and utilities industries are based on
data from the U.S. Departments of Energy and the Interior and from in-
dustry sources. Also, for those industries where direct measures of capac-
ity are not available, capacity is estimated based on long-term trends
connecting peak levels of output.
CAPACITY UTILIZATION 33
The monthly trends between the year-end levels of capacity are ob-
tained by connecting the year-end points by a straight trend line. Monthly
movements of the current year are extrapolated based on the monthly
trend of the previous year.
Accuracy
There are no estimates of revisions for the CUR data. However, the revi-
sion measures of the industrial production index (IPI) are a close ap-
proximation of revisions for the CUR data, because the volatile IPI is
the numerator and the steady growth of capacity is the denominator of
the CUR. The typical revision to the monthly IPI level between the pre-
liminary estimate and the third monthly revision is plus or minus 0.28
percent. The typical revision to the monthly movement is plus or minus
0.22 percentage point. In about 85 percent of the cases, the direction of
change in the preliminary estimate is the same as in the third revision.
Relevance
The CUR is used as an indicator of future equipment and structures in-

vestment. Generally, the higher the CUR, the greater the tendency for
equipment and structures shortages to exist, which in turns leads to addi-
tional investment. However, it is important to analyze the CUR together
with trends in the gross domestic product and business profits (see corpo-
rate profits) for clues to future investment. There is no specific CUR level
that indicates a shortage of capacity or signals additional investment. The
difference between 100 percent utilization and the CUR theoretically rep-
resents the unused capacity that is available to increase production to meet
an increased demand. In practice, however, CURs typically do not top the
90th percentile, except for utilities, in continuous processing industries
such as paper, chemicals, or petroleum refining, or during wartime when
mobilization is high and less efficient facilities are put into production.
The CUR is also used to assess future inflation in a direct relation-
ship. However, statistical linkages between the CUR and price move-
ments are not strong.
Recent Trends
From 1995 to 2004, the CUR for manufacturing, mining, and utilities
industries combined ranged from 82 to 85 percent from 1995 to 2000,
Table 7.1
Capacity Utilization (percent)
All industries Manufacturing Mining Utilities

1995 83.7 82.8 87.9 90.0
1996 82.7 81.4 90.1 90.5
1997 83.7 82.8 91.1 89.1
1998 82.9 81.8 88.9 91.2
1999 82.2 81.1 86.1 92.5
2000 82.0 80.6 90.1 92.4
2001 76.6 74.5 89.8 88.9
2002 75.3 73.5 85.5 87.6
2003 75.5 73.7 86.6 84.9
2004 78.1 76.7 86.5 85.1
declined to 75 to 76 percent during 200103, and rose to 78 percent in

2004 (Table 7.1). The movements and levels of this overall CUR were
similar to those for manufacturing industries; the most noticeable dif-
ference was that the manufacturing level was typically 1.0 to 1.5 per-
centage points lower than the total CUR for all industries.
Both the mining and utilities CURs were several percentage points
higher than those for manufacturing. The mining industries CUR typi-
cally ranged from 88 to 90 percent from 1995 to 2001, dropped to 85.5
percent in 2002, and rose to 86.5 percent in 200304. The utilities in-
dustries ranged from 90 to 92.5 percent from 1995 to 2000, and then
declined to lows of 85 percent in 200304.

Federal Reserve Bulletin. The Statistical Supplement is monthly and the Federal
Reserve Bulletin is quarterly.
Note
1. This formula works as follows:
Capacity utilization __________

_________________ Production Production 1
= = _________ _________ = Capacity
Production Capacity
____________ Capacity Production
Production
__________
1
8
Two organizations provide indicators of consumer attitudes. They focus

on households perceptions of general business and employment condi-
tions, and of their personal financial well-being, plus their attitudes to-
ward purchasing big-ticket items that last a relatively long timehomes,
cars, furniture, and major household appliances. The indicators of both
organizations are covered here: the consumer confidence index (CCI)
of The Conference Board, and the consumer sentiment index (CSI) of
the University of Michigan. Both indexes measure similar phenomena
but, because the methodologies differ and the concepts are not identical,
there are periods when their movements differ. Both are constructed as
diffusion indexes, which suggest an indication of the direction of the
movement from one period to the next, but not of the size of the move-
ment. The characteristics of The Conference Board and Michigan mea-
sures are described separately. This is followed by a summary of the
main methodological differences between them, and then by highlights
of the relevance and recent trends of both.
Consumer Confidence Index
The consumer confidence index reflects consumers attitudes toward

the economy, local job markets, and their own financial condition.
The consumer confidence index is prepared monthly by the Consumer

Research Center of The Conference Board. It is published in two Con-
35
ference Board monthly reports, Consumer Confidence Survey and Busi-

ness Cycle Indicators (www.conference-board.org).
The CCI data are available the last Tuesday of the same month to
which they refer. The data are revised for the previous month.
Content
The consumer confidence index represents the combined effects of house-

hold perceptions of local area business conditions, household percep-
tions of available jobs in local areas currently and six months ahead, and
expected household income six months ahead. A rising index means
consumers are more optimistic, and a declining one signifies greater
pessimism. While there are no absolute values that define optimism and
pessimism, comparisons of the index levels with previous periods indi-
cate whether consumers are more optimistic or pessimistic than in past
periods.
The CCI is based currently on 1985 = 100.
The CCI is seasonally adjusted.
Methodology
Data for the consumer confidence index are obtained from a monthly
household survey conducted by TNS-NFO for The Conference Board.
The survey is mailed to approximately 5,000 households in the forty-
eight mainland states and the District of Columbia, and the response
rate is about 70 percent (3,500 households). A completely new group of
households is surveyed each month.
The CCI is constructed by giving equal weight to each of five ques-
tions. There is one question each on the survey respondents local area
business conditions currently and six months ahead, one question each
on jobs in the local area currently and six months ahead, and one ques-
tion on expected household income six months ahead. In constructing
the index, the positive responses are expressed as a percentage of the
sum of the positive and negative responses. Neutral answers are not
counted. Depending on the question, positive answers are referred to as
good, better, plenty, more, higher; negative responses are re-
ferred to as bad, worse, hard to get, fewer, lower; and neutral
responses are referred to as normal, same, not so many in the
month. Mathematically, the formula is:
CONSUMER ATTITUDE INDEXES 37
Positive
CCI = 100 (8.1)
Positive + Negative
This type of index, which provides an indication of the direction but

not of the magnitude of the movement from one period to the next, is a
diffusion index (diffusion indexes are discussed further in the Content
section of the PMI chapter). In contrast, traditional indexes of economic
activity provide the actual direction and magnitude of the movement.
Accuracy
The sampling error (for one standard error) for the consumer confidence
index is 1.5 percentage points. For example, if the estimated CCI were
100.0, in two of three cases the true index would be somewhere be-
tween 98.5 and 101.5. For further information on the interpretation of
sampling and nonsampling errors, see the Appendix.
Consumer Sentiment Index
The consumer sentiment index reflects consumers attitudes toward the

economy and their own financial condition, and perceptions about buy-
ing big-ticket durable goods.
The consumer sentiment index is prepared monthly by the Survey Re-

search Center of the University of Michigan. It is published in the monthly
report, Surveys of Consumers (www.athena.sca. isr.umich.edu).
The CSI data are available within the first five to ten days of the
month after the month to which they refer. The monthly data are not
revised.
Content
The consumer sentiment index combines three main categories of

household attitudes toward the economy in one number: (1) expected
business conditions in the national economy for one and five years
ahead, (2) personal financial well-being compared with one year ear-
lier and expected one year later, and (3) whether the current period is
a good or bad time to buy furniture and major household appliances.

Upward movements of the index suggest that consumers are becom-
ing more optimistic, and downward movements suggest a growing
pessimism. While there are no absolute values that define optimism
and pessimism, comparisons of the index levels with previous peri-
ods indicate whether consumers are more optimistic or pessimistic
than in past periods.
Supplementary information for interpreting the reasons for changes
in household attitudes is included in Surveys of Consumers. The full
report includes data on attitudes toward such items as employment, prices,
interest rates, shortages, and government policies.
The CSI is based currently on 1966: 1Q = 100.
The CSI is not seasonally adjusted.
Methodology
Data for the consumer sentiment index are obtained from a telephone
survey of a sample of households conducted by the Survey Research
Center. Approximately 670 households are contacted monthly in the
forty-eight mainland states and the District of Columbia, with a response
rate of about 75 percent (500 households). The sample is designed as a
rotating panel in which one-half of the survey respondents are new each
month and one-half are carryovers from the survey panel of six months
earlier.
Five questions are used in constructing the index. There is one ques-
tion each on expected national economic conditions one year and five
years ahead; one question each on personal financial well-being con-
trasting the current period with one year earlier and with one year ahead;
and one question on whether the current period is a good time to buy
furniture and major household appliances. Equal weight is given to each
question. In valuing the answers, positive, negative, and neutral answers
are used. Depending on the question, positive answers are up, bet-
ter, good; negatives are down, worse, bad; and neutrals are
same, no change, uncertain in the month. The percentage of nega-
tive responses is subtracted from the percentage of positive responses
(the percentages of the positive and negative responses are calculated
relative to the sum of the positive, negative, and neutral responses), and
100 is added to the difference to avoid negative numbers. Mathemati-
cally, the formula is:
Positive (8.2)
CSI = 100
(Positive + Negative + Neutral)
Negative
100 + 100
(Positive + Negative + Neutral)
This type of index, which provides an indication of the direction but

not of the magnitude of the movement from one period to the next, is a
diffusion index (diffusion indexes are discussed further in the Content
section of the PMI chapter). In contrast, traditional indexes of economic
activity provide the actual direction and magnitude of the movement.
Accuracy
The sampling error (for one standard error) for the consumer sentiment
index is 1.3 percentage points. For example, if the estimated consumer
sentiment index were 100.0, in two of three cases the true index would
be somewhere between 98.7 and 101.3. For further information on the
interpretation of sampling and nonsampling errors, see the Appendix.
Main Differences Between the Conference Board and

Michigan Indexes
While the consumer confidence index and consumer sentiment index ba-
sically measure the same phenomena, there are clear differences in their
methodologies associated with the index content, wording of questions,
seasonal adjustment, household samples, data collection, and question-
naire response estimation. The main differences are summarized below.
Index content. The CCI excludes the purchase of big-ticket items,

while the CSI includes them. The CCI includes questions on current and
expected job opportunities, while the CSI excludes them.
Wording of questions. For general business conditions, the CCI fo-
cuses on the respondents local economy currently and with a short-
term six-month outlook, while the CSI focuses on the national economy
with a long-term one-to-five-year outlook. For personal financial well-
being, the CCI looks at household income six months ahead, while the
CSI asks how well off the respondent is financially compared with one
year earlier and what expectations are for one year ahead.
Seasonal adjustment. The CCI is seasonally adjusted, while the CSI

is not seasonally adjusted.
Monthly household sample. The CCI sample of respondents is 3,500
households, while the CSI sample of respondents is 500 households.
Data collection. The CCI uses a mail questionnaire, while the CSI
uses a telephone interview.
Survey response estimation. The CCI and CSI apply different weights
for positive and negative answers to survey questions.
Revisions. The CCI data are revised for the previous month, while the
CSI data are not revised.
Relevance
Perceptions by households of the strength of general business conditions

and of their personal financial conditions are closely linked to households
feelings of optimism and pessimism about the economy. In theory, when
households are optimistic, they are more willing to increase spending and
incur debt to finance the higher spending. When households are pessimis-
tic, they are likely to cut back on spending, pay off debts, and build nest-
egg savings. A households decision to buy a home or big-ticket durable
goods is typically based on advance planning and is heavily influenced by
the households perception of changing economic conditions.
When there are sustained changes in one direction over a period of time,
the consumer confidence index (CCI) and the consumer sentiment index
(CSI) are fairly good predictors of shifts in future household spending and
saving. However, the index levels are hard to distinguish one from the other.
The CCI and the CSI are classified as leading indicators of economic
activity by The Conference Board. In addition, the University of
Michigans index of consumer expectations, which is based only on
the three questions in the CSI relating to the future (the CSI has five
questions), is a component of the leading index of the leading, coinci-
dent, and lagging indexes. Because the monthly and even quarterly
movements show erratic increases and decreases, both indexes should
be viewed over longer periods to discern a change from past trends.
Recent Trends
From 1995 to 2004, the CCI and CSI had similar upward and downward
movements, except for 2002 (Table 8.1). The CCI had consistently larger
Table 8.1
Consumer Confidence Indexa Consumer Sentiment Indexb

(1985 = 100) (1966: IQ = 100)
1995 100.0 92.2
1996 104.6 93.6
1997 125.4 103.2
1998 131.7 104.6
1999 135.3 105.8
2000 139.0 107.6
2001 106.6 89.2
2002 96.6 89.6
2003 79.8 87.6
2004 96.1 95.2
aThe Conference Board. bUniversity of Michigan.
year-to-year changes than the CSI. Both indexes were at their highest
levels in 2000 and at their lowest levels in 2003.
The Conference Board. Consumer Confidence Survey and Business Cycle Indica-
tors. New York, NY. Monthly.
University of Michigan, Survey Research Center. Surveys of Consumers. Ann Ar-
bor, MI. Monthly.
9
Consumer Credit
Consumer credit represents loans by banks, finance companies, and re-

tail stores to households for financing consumer purchases of goods and
services and for refinancing existing consumer debt.
Measures of consumer credit are prepared monthly by the Federal Re-

serve Board (FRB). They are published in a statistical release (G-19)
and in the monthly statistical supplement to the quarterly FRB journal,
the Federal Reserve Bulletin (www.federalreserve.gov).
The data are available approximately five weeks after the month to which
they refer. The data are revised as revisions in the database are received.
Content
The consumer credit data include credit cards and loans for items such
as automobiles, mobile homes, education, boats, trailers, or vacations.
They include loans with a fixed repayment schedule of one or more
payments (nonrevolving credit) and loans where borrowers have the
option to repay any amount above a given minimum (revolving credit).
Securitized consumer loansloans made by finance companies, banks,
and retailers that are sold as securitiesare included. The data exclude
home mortgages, but probably include an unknown amount of consumer
loans used for business purposes. Automobile leasing is also excluded
from the consumer credit data.
The data reflect consumer credit outstanding at the end of the month.
42
CONSUMER CREDIT 43
The monthly change in consumer credit outstanding is the net effect of

credit extensions and repayments during the month. Separate data on
credit extended and repaid during the month are not available.
The consumer credit data are seasonally adjusted.
Methodology
Monthly data on consumer credit are based on the following: monthly

surveys of a sample of commercial banks conducted by the Federal Re-
serve Board; monthly surveys of consumer finance companies, includ-
ing auto finance companies, conducted by the Federal Reserve Board;
quarterly reports filed by savings institutions with the Office of Thrift
Supervision; monthly surveys of credit unions conducted by the Credit
Union National Association; monthly U.S. Treasury reports on federal
direct student loans; and monthly surveys of retail sales conducted by
the Bureau of the Census. Benchmark data are available quarterly for
commercial banks; annually for retailers (accounts receivable), and ev-
ery five years for finance companies.
Accuracy
There are no estimates of sampling or revision error for the consumer

credit data.
Relevance
Consumer credit supplements personal income and saving as a source

of consumer purchasing power. In turn, consumer purchasing power
impacts retail sales. While consumer credit outstanding typically in-
creases, the rate of increase is faster during expansions than during re-
cessions. The occasional monthly declines in consumer credit occur
mostly during recessions. Thus, consumer credit accentuates the cycli-
cal movements of consumer spending, particularly for durable goods.
Households generally borrow more during periods of rapidly growing
personal income, because prosperity leads to optimism regarding finan-
cial commitments (see consumer attitude indexes).
Consumer credit outstanding is also viewed as a burden on house-
holds because they must pay the principal and interest on the loans. The
ratio of consumer credit outstanding to personal income and household
Table 9.1
Consumer Credit Outstanding (seasonally adjusted)
Credit outstanding Annual percentage change

December (billions of dollars) (December to December)
1995 1,141.0 14.4
1996 1,242.9 8.9
1997 1,313.1 5.7
1998 1,416.8 7.9
1999 1,530.4 8.0
2000 1,705.1 11.4
2001 1,842.2 8.0
2002 1,924.2 4.5
2003 2,011.3 4.5
2004 2,109.6 4.9
debt-service payments for consumer and mortgage loans (scheduled

periodic payments of principal and interest as a percentage of dispos-
able personal income) are commonly used measures of this burden. As
this percentage rises during expansions, the growing consumer credit
can be expected to depress further consumer borrowing. The ratio of
consumer credit outstanding to personal income is a component of the
lagging index of the leading, coincident, and lagging indexes.
Recent Trends
From 1995 to 2004, the rate of increase in consumer credit outstand-

ing fluctuated considerably (Table 9.1). The annual increase declined
from 14 percent in 1995 to 6 percent in 1997, rose to 11 percent in
2000, declined to 4.5 percent in 2002 and 2003, and rose to 4.9 percent
in 2004.

Federal Reserve Bulletin. The Statistical Supplement is monthly, and the Federal
10
Consumer credit delinquency (CCD) rates are household loans with over-
due payments of thirty days or more as a percentage of all consumer loans
outstanding. The data represent loans by commercial banks and savings
and loan associations for household expenditures associated with general
personal use, vehicles, housing, and education. The loans are financed by
installment, bank card, revolving, mortgage, and home equity credit.
The consumer credit delinquency data are prepared quarterly by the

American Bankers Association. They are published in the Consumer
Credit Delinquency Bulletin (www.aba.com).
The CCD data are available approximately ten weeks after the reference
calendar quarter. For example, the data for January, February, and March
are published about the third week of June. The CCD data are not revised.
Content
The consumer credit delinquency data represent household loans extended

by commercial banks and savings and loan associations that are delinquent
as a percentage of the total number of loans outstanding at the end of each
month. Delinquency rates are provided separately for the number and the
dollar value by type of closed-end and open-end loan. Closed-end loans are
for a specified item and must be paid back within a limited period (personal,
automobile, mobile home, recreational vehicle, marine financing, property
improvement, home equity and second mortgages, and education loans).
45
Open-end loans continue so long as the necessary payments are made to

maintain them (bank card credit, revolving credit, home equity lines of credit).
The CCD data for the combined total of closed-end loans except edu-
cation is also provided. In addition, repossession rates for defaulted loans
(number of repossessions per month per 1,000 loans outstanding) are
provided for automobile, mobile home, and marine financing loans. The
data are shown for the national total and by state. The data exclude loans
by thrift institutions, consumer finance companies, charge accounts by
retail stores, and nonbank credit cards.
The CCD data are seasonally adjusted for the national total. The state
data are not seasonally adjusted.
Methodology
The consumer credit delinquency data are obtained from a quarterly sur-
vey of a sample of commercial banks and savings and loan associations
(S&Ls). Approximately 3,000 large commercial banks based on consumer
loan portfolio size are surveyed; they account for 60 to 70 percent of the
consumer loan portfolios managed by commercial banks. The surveyed
S&Ls represent approximately 25 percent of all S&Ls. The CCD data are
based solely on the reports from these banks and S&Ls.
The sample of surveyed banks and S&Ls is updated annually. The
survey sample is not a probability sample.
Accuracy
There are no estimates of sampling or revision error for the consumer

credit delinquency data.
Relevance
Analytic interest in the consumer credit delinquency data is related to

consumer credit, retail sales, personal consumption expenditures in the
gross domestic product, bankruptcies: personal, and mortgage delin-
quency and foreclosure. CCDs become more pronounced when house-
hold debt loads (consumer and mortgage scheduled debt-service
payments in relation to disposable personal income) become vulnerable
to reductions in personal income (personal income and saving) associ-
ated with unemployment, stock market losses (stock market price aggre-
CONSUMER CREDIT DELINQUENCY 47
Table 10.1
Consumer Credit Delinquency Rate (number of delinquent loans as a

percentage of all loans outstanding)
December, seasonally adjusted

1995 2.12
1996 2.34
1997 2.43
1998 2.35
1999 2.27
2000 2.40
2001 2.34
2002 2.16
2003 1.89
2004 1.68
gates and dividend yields), personal sickness, divorce, and so on. Other
factors affecting CCDs are changes in bank lending standards (the will-
ingness of banks to lend money as indicated in a monthly survey of
senior loan officers conducted by the Federal Reserve Board), and
changes in the promotion by banks and other credit card companies to
households whose financial ability to pay off large debts is considered
questionable. Reverse movements occur when CCD rates decline.
Rising CCD rates cause lenders to raise their credit standards for loans
to risky borrowers, and/or to raise interest rates to risky borrowers that
discourage them from additional borrowing, which tend to restrain spend-
ing growth. Analogously, declining CCD rates have the opposite effects
and tend to stimulate spending growth. Rising and declining CCD rates
are one precursor of movements in bankruptcies: personal.
Recent Trends
From 1995 to 2004, the consumer credit delinquency rate fluctuated

within a narrow range during 19952001, and then declined in the latter
part of the period (Table 10.1). The CCD trended downward from 2.40
in 2000 to 1.68 in 2004.
American Bankers Association. Consumer Credit Delinquency Bulletin. Washing-

ton, DC. Quarterly.
11
The consumer price index (CPI) gauges the overall rate of price change
for a fixed basket of goods and services bought by households. Because
it prices the same items (with only limited exceptions) every month, this
measure of inflation or deflation reflects the cost of maintaining the
same purchases over time. The CPI does not conform to a theoretical
cost-of-living index.
The consumer price index is prepared monthly by the Bureau of Labor

Statistics in the U.S. Department of Labor. The data are published in a
news release, the CPI Detailed Report, and the monthly BLS journal,
Monthly Labor Review (www.bls.gov).
The data are published in the middle of the month immediately fol-
lowing the month to which they refer. The monthly not-seasonally-ad-
justed data are rarely revised. The seasonally adjusted data are revised
annually for several previous years based on new seasonal factors.
Content
The consumer price index records price changes in food and beverages,
housing, apparel, transportation, medical care, recreation, education and
communication, and other goods and services. Thousands of items within
these broad groups are priced every month. It is published in two ver-
sions, the CPI-U and the CPI-W. The CPI-U represents all urban house-
holds including urban workers in all occupations, the unemployed, and
48
CONSUMER PRICE INDEX 49
retired persons; it accounts for about 87 percent of the noninstitutional

population.1 The CPI-W represents urban wage and clerical workers
employed in blue collar occupations; it accounts for about 32 percent of
the noninstitutional population. Both the CPI-U and the CPI-W exclude
rural households, military personnel, and persons in institutionalized
housing such as prisons, old-age homes, and long-term hospitals. A third
CPI measure, the C-CPI-U (chain CPI-U) introduced in 2000, is dis-
cussed below under Methodology.
CPIs are calculated for the nation as a whole, for broad geographic
regions, and for large metropolitan areas. They therefore provide differ-
ential national and geographic measures of price movements over time.
However, the CPI does not reflect the actual dollar level of living costs
in the nation or in one area compared with anotherit reveals only the
inter-area differences in price movements.2
The weights of the CPI are based on the proportions of household
budgets that consumers actually spend for particular goods and ser-
vices. The spending patterns are updated every two years. For ex-
ample, the CPI for 2004 and 2005 reflects spending patterns during
200102 and the CPI for 2006 and 2007 reflects spending patterns
during 200304.
The CPI is based on actual transaction prices, which take into ac-
count such variations as premiums or discounts from the list price, sales
and excise taxes, import duties, and trade in allowances when the used-
car trade-in is part of the new-car price. The CPI reflects price move-
ments for the same or similar item exclusive of enhancement or reduction
in the quality or quantity of the item (see Methodology below).
The CPI is not a cost-of-living (COL) index. A COL index is a theo-
retical construct associated with the minimum expenditures necessary
to maintain a constant standard of living over time. The CPI is not a
cost-of-living (COL) index because the CPI measures price movements
for purchasing only the same items between two periods, regardless of
the availability of lower-priced substitutes, except for substitution among
certain product categories, as noted below under Methodology. The
CPI also does not include other attributes of a COL index, such as ac-
counting for household preferences between work and leisure, how
changes in income tax rates affect the households after-tax income and
consequently the financial ability to buy the same goods and services
over time, and social problems of crime and pollution that cause house-
holds to move to other areas or buy protective items. Also, the CPI is not
a standard-of-living concept that prices spending patterns aimed at achiev-

ing certain standards of nutrition, housing, health, and so on that society
considers appropriate.
The CPI is currently based on 1982 84 = 100.
The CPI data are seasonally adjusted.
Methodology
The monthly consumer price index (CPI) data are obtained primarily
from surveys of retail and service establishments, utilities, and house-
holds. Surveyors visit or telephone the same retail and service estab-
lishments and price the same items if still available (or close
substitutes) every month or bimonthly, depending on the city and
item in the survey sample. For a small number of items such as used
cars, airfares, and postal rates, the Bureau of Labor Statistics receives
monthly reports on prices from trade sources and the Postal Service.
Because housing rents are not volatile, and in order to reduce survey
costs, rent information is obtained by less frequent visits (every six
months) to one of six survey panels of apartments and single-family
homes. The monthly rent change represents the change between rent-
als in the current month and six months earlier for the same panel of
housing units. For example, the same panels are surveyed at six-month
intervals for January and July, February and August, and so on, with
each months change represented by the sixth root of the six-month
change. This procedure ensures that each panel is representative of
the entire sample.
The current CPI weights for 2006 and 2007, which represent the pro-
portion of household budgets spent on the various components, reflect
consumer purchasing patterns during 200304. Weights for the main
product categories are based on surveys of households to determine their
actual purchasing patterns. As noted previously under Content, the
CPI spending patterns are revised every two years.
In order to more closely approximate the substitution effects of price
shifts within generic product categories that represent closely related
products, the Bureau of Labor Statistics uses geometric-mean estimat-
ing for these items. The geometric mean maintains the item share of
expenditures, as distinct from quantities, constant from the base period.
For example, in the category of laundry equipment, consumer prefer-
ences between washers and dryers are updated every year. This substitu-
tion within generic categories is done where it is considered most ap-

propriate, which occurs in categories accounting for about 61 percent of
the CPI. Food and beverages, apparel, and other goods and services are
based on the geometric mean exclusively; in addition, selected shelter
and government services charges are based on the geometric mean.
The remaining 39 percent of the CPI is based on arithmetic-mean
indexing of fixed quantity weights between major spending pattern re-
visions. Use of the arithmetic means for selected categories reflects the
fact that there is little ongoing substitution for these items. For example,
rental housing is often based on a contract period; alternative electric,
natural gas, and water utilities under deregulation typically are not avail-
able to households; and medical services are often determined by insur-
ance plans.
It is important to keep in mind that this substitution is confined to
generic product categories; it is not done between dissimilar categories
such as between laundry equipment and kitchen appliances. Geometric
averaging, in contrast to arithmetic averaging, treats price increases and
decreases symmetrically, without the distortion of shifting bases.3
A new type of CPI, the C-CPI-U (chain CPI-U), was introduced in
2000. The C-CPI-U has a more comprehensive use of geometric averag-
ing than the CPI-U and CPI-W. In addition to geometric averaging within
generic product categories, the C-CPI-U includes geometric averaging
at the overall index level from the base period to the current period. This
is referred to as a superlative index.
If the quality or quantity of an item in the monthly survey has
changed, an adjustment is made to reflect the improvement or de-
cline. The goal is to price products having the same functional char-
acteristics over time. For example, when an apartment building is
renovated to include an elevator, if the rent increases by the cost of
installing the elevator, the market-rent increase is represented as zero
price change in the CPI. By contrast, if a loaf of bread becomes
smaller but the price remains the same, the price of bread per unit
has in fact increased and is represented as a price increase in the
CPI. And when a new car with better safety features than those in the
previous model increases in price less than the cost attributable to
the improved safety features, the difference between the market price
increase and the cost of making the safety improvements is repre-
sented as a price decrease in the CPI. The rent index assumes a small
loss of quality as a housing unit ages. Thus, if the money rent of an
apartment remains constant, the CPI would assume that the rent on
an apartment of unchanging quality has increased.
Use of hedonic price measurement substantially lessens the problem of
accounting for quality change. The word hedonic reflects its root mean-
ing of pleasure, as the characteristics of a product in a hedonic index are
assessed for the pleasure, or utility, they give the buyer. Hedonic price
measurement, which requires considerable amounts of data, substantially
lessens the problem of accounting for complex issues of quality change
for which the production cost method is inadequate. A hedonic price in-
dex traces the effects of a group of attributes of a product that influences
the price of an item, through both (a) the utility of the attributes to the
buyer (demand), and (b) the cost of providing the attributes to the pro-
ducer (supply). Currently hedonic indexes in the CPI are used for apparel,
television sets, audio equipment, and college textbooks.
Because the data needed to make the necessary adjustments are not
always available, the CPI contains an unknown amount of price change
caused by quality and quantity changes. More generally, adjustments
for quality change in the CPI are often complex and require special at-
tention by the analyst preparing the estimates. Thus, the basis for some
quality adjustments is ambiguous or relies on assumptions that are diffi-
cult to verify, which limits the accuracy of the adjustments. For example,
scientific breakthroughs in various human diseases have resulted in longer
life spans and an improved quality of life. But the task of putting a dol-
lar value on the extended life spans and better quality of life, versus the
increased costs of medical procedures and medicines in achieving these
benefits, has not yet been solved for inclusion in the CPI. Another ex-
ample is the greater variety of products available in stores from more
countries resulting from the increased globalization of foreign trade.4
But estimates of the impact of a greater variety of products that increase
the quality of such goods, and consequently lower their CPI-measured
price, are based on broad assumptions that need more refined statistical
measurement for inclusion in the CPI.
Accuracy
The sampling error (for one standard error) of the monthly percentage
change in the consumer price index (CPI) is plus or minus 0.06 of a
percentage point. For example, if the estimated increase in the CPI from
one month to the next were 0.30 percent, in two of three cases the true
increase would be somewhere between 0.24 and 0.36 percent. For fur-
ther information on the interpretation of sampling and nonsampling er-
rors, see the Appendix.
Relevance
The consumer price index (CPI-U) is the most widely quoted number
on price movements. In the formulation of macroeconomic fiscal and
monetary policies (see Relevance under gross domestic product), trends
in the CPI are a major guide in determining whether economic growth
should be stimulated or restrained. The CPI is also contrasted with un-
employment to analyze the tradeoff between inflation and unemploy-
ment, which is referred to as the Phillips Curve. Low inflation and low
unemployment are primary goals of economic policies, as formulated
in the Full Employment and Balanced Growth Act of 1978 (Humphrey-
Hawkins Act).
The CPI is used in a variety of ways to adjust for cost escalation in
commercial activities and in government programs: price change ad-
justments to wages, pensions, and income maintenance payments for
cost-of-living allowances; price change adjustments in business con-
tracts; and indexing of federal individual income tax brackets to limit
inflation-induced bracket creep. Many labormanagement union con-
tracts are based on the CPI-W. In addition, the CPI is used to deflate
various economic indicators to adjust for price change such as the con-
sumer expenditure component of the gross domestic product.
Supplementary CPI measures are published that exclude price move-
ments of food and energy products. Because these products sometimes
have volatile price movements that are unrelated to cost pressures in the
overall economy, their exclusion provides the core inflation rate, which
is also referred to as the underlying rate of inflation.
Recent Trends
From 1995 to 2004, the CPI-U and the CPI-W both increased within a
range of approximately 1.5 to 3.5 percent annually (Table 11.1). The
annual percentage changes in both indexes were the same in two years,
differed by one percentage point in six years, and differed by 0.2 and
0.3 percentage point each in one year. The CPI-U increased more than
the CPI-W in 1996, 1998, 2001, 2002, 2003, and 2004, and the CPI-W
Table 11.1
Consumer Price Index (annual percentage change)
CPI-U C-CPI-U CPI-W

1995 2.8 NA 2.9
1996 3.0 NA 2.9
1997 2.3 NA 2.3
1998 1.6 NA 1.3
1999 2.2 NA 2.2
2000 3.4 NA 3.5
2001 2.8 2.3 2.7
2002 1.6 1.2 1.4
2003 2.3 2.1 2.2
2004 2.7 2.2 2.6
19952004 20012004 19952004
(annual average) (annual average) (annual average)
2.4 1.95 2.3
NA = not available
increased more than the CPI-U in 1995 and 2000. The C-CPI-U in-
creased less than the CPI-U and the CPI-W in all years since the C-
CPI-U was published. Over the entire nine-year period, the CPI-U
increased at an average annual rate of 2.4 percent, and the CPI-W in-
creased at an average annual rate of 2.3 percent. From 2001 to 2004,
the C-CPI-U increased at an average annual rate of 1.9 percent (see
Table 11.1).
Bureau of Labor Statistics, U.S. Department of Labor. CPI Detailed Report and
Monthly Labor Review. Monthly.
Notes
1. The noninstitutional population represents people who are not in hospitals,

nursing homes, jails, etc.
2. For possible future work on developing experimental inter-area price indexes
to permit comparisons of price levels between local areas, see the Preface under
Data Sharing and Confidentiality of Federal Government Statistical Data. Also,
for research on preparing inter-area price levels for the consumer price index, see
Bettina H. Aten, Report on Interarea Price Levels, Bureau of Economic Analysis,
U.S. Department of Commerce, April 25, 2005. Available from the Bureau of Eco-
nomic Analysis, www.bea.gov.
3. For example, a price increase from $4 to $5 in the ratio of 1.25 is 25 percent,

while a price decrease from $5 to $4 in the ratio of .80 is 20 percent. The arithmetic
mean shows the price increase as more important than the price decrease ([1.25 +
.80]/2 = 1.025), while the geometric mean gives the same weight to both the price
increase and the price decrease (1.25 x .80 = 1). Technically, geometric averaging is
the nth root of the product of n numbers, and arithmetic averaging is the sum of n
numbers divided by n.
4. Christian Broda, and David Weinstein, Are We Underestimating the Gains
from Globalization for the United States? Current Issues in Economics and Fi-
nance, Federal Reserve Bank of New York, April 2005; idem, Globalization and
the Gains from Variety, NBER Working Paper 10314, National Bureau of Eco-
nomic Research, February 2004.
12
Corporate Profits
Corporate profits are the returns to corporate enterprise from current

operations. Profits occur when operating income (receipts) exceeds op-
erating expenses (costs), and losses (negative profits) occur when ex-
penses exceed income. Because profits are the difference between income
and expenses, both elements affect profits. From one year to the next,
for example, profits decline when income rises less than expenses or
when income declines more than expenses, while profits increase when
income rises more than expenses or when income declines less than
expenses.
Corporate profits measures are prepared quarterly for all industries by

the Bureau of Economic Analysis in the U.S. Department of Commerce
as part of the gross domestic product. The data are published in a news
release and in the BEA monthly journal, Survey of Current Business
(www.bea.gov). The profits data are available approximately fifty-five
days after the quarter to which they refer; they are revised one month
later and annually every summer.
Content
The corporate profits data are based on those used in calculating the
gross domestic product (GDP). They are closer to the definitions of cor-
porate profits in federal income tax returns than to those in company
financial reports to stockholders. Profits are measured both before and
56
CORPORATE PROFITS 57
after the payment of federal, state, and local income taxes. Undistrib-
uted corporate profits are the profits retained in the business after corpo-
rate income taxes are paid and dividends are distributed to stockholders.
(Profits also accrue to unincorporated sole proprietorships and partner-
ships; corporations differ from unincorporated businesses in the method
of designating ownership in the company, liability of owners, and in-
come taxes.)
The following items highlight the main characteristics of corporate
profits in the GDP that diverge from the definitions used in corporate
federal income tax returns:
Corporations encompass the following: (1) institutions required to

file federal corporate income tax returnsfor-profit corporations,
savings and loan associations, mutual savings banks, and coopera-
tives, and (2) certain institutions not required to file federal corpo-
rate income tax returnsFederal Reserve Banks, federally
sponsored credit unions, private noninsured pension funds, and
nonprofit organizations that primarily serve business.
The national measure of corporate (gross national product) profits
shown in Table 12.1 includes profits from subsidiaries of U.S. com-
panies abroad and excludes profits from subsidiaries of foreign
companies in the United States. This measure is modified in the
GDP through the rest-of-the-world sector, which conforms profits
to the domestic measure by excluding profits from subsidiaries of
U.S. companies abroad and including profits from subsidiaries of
foreign companies in the United States.
Inventory profits and losses are modified in the inventory valua-
tion adjustment to eliminate profits or losses on inventory hold-
ings due to changes in inventory prices.
Depreciation allowances on equipment and structures are modi-
fied in the capital consumption adjustment to reflect the eco-
nomic lifetime of capital facilities actually used in business
practice in place of the service lives specified for these facilities
in the income tax laws, and to reflect the current cost of replac-
ing the existing equipment and structures in place of the origi-
nal acquisition cost.
Capital gains and losses from the sale of property are excluded from
profits, except for the bid/ask price spreads on sales of securities by
security dealers and brokers.
Depletion allowances for using up nonreplaceable mineral reserves

of mining corporations are added to profits.
Charges for bad debts are treated as changes in asset valuation rather
than as a deduction from profits.
The corporate profits data are seasonally adjusted.
Methodology
Corporate profits are estimated every quarter from the Quarterly Finan-
cial Report (QFR) prepared for manufacturing, mining, and trade cor-
porations by the U.S. Bureau of the Census. The QFR profits data are
based on definitions used in company reports to stockholders. These are
supplemented by company stockholder reports published in the press
for certain industries not covered in the QFR, and indirect data on the
economic activity for other industries such as construction. The above
quarterly data are revised in subsequent years based on the annual Sta-
tistics of Income information for all industries, which is derived from
federal corporate income tax returns and prepared by the U.S. Internal
Revenue Service. These include the effects of IRS audits of corporate
income tax returns. Special adjustments to the reported data for the GDP
estimates of profits are made for institutional coverage, inventories, de-
preciation, capital gains and losses, depletion, and bad debts, as noted
above under Content.
Accuracy
The average revision error of corporate profits in the gross domestic

product as a percentage of the quarterly change in profits is plus or mi-
nus 11.6 percentage points from the estimates eighty-five days after the
quarter to the annual estimates three years later.
Relevance
Profits are the returns to investment and risk taking and are the prime
motivating factor of the private-enterprise economy. Past profits and an-
ticipated future profits directly affect business actions on employment,
inventory-sales ratios, and equipment and structures investment. When
business conditions are buoyant, entrepreneurs and executives are opti-
mistic about the future and likely to expand their work force, inventories,
CORPORATE PROFITS 59
Table 12.1
Corporate Profits (billions of dollars)
Profits before taxes,

including IVA and CCAdj Annual percentage change
1995 696.7 16.1
1996 786.2 12.8
1997 868.5 10.5
1998 801.6 7.7
1999 851.3 6.2
2000 817.9 3.9
2001 767.3 6.2
2002 886.3 15.5
2003 1,031.8 16.4
2004 1,161.5 12.6
19952004
(annual average) 5.8
Note: Based on gross national product measure. See text under Content for differ-
ence with gross domestic product measure.
IVA = inventory valuation adjustment; CCAdj = capital consumption adjustment
and investment in capital facilities. By contrast, when business conditions

are depressed, entrepreneurs and executives are pessimistic about the fu-
ture and are likely to retrench employment, inventories, and investment.
Profits are one of the more volatile elements of the economy in that
they tend to rise faster than the overall economy during business expan-
sions and to decline more sharply than the rest of the economy during
business recessions. While some corporations lose money, it is rare for
all corporations in total to lose money. The last time total corporate
profits of all companies were negative was in the depth of the Great
Depression in 1931 and 1932.
Undistributed profits, also referred to as retained earnings, are profits
after the payment of dividends to stockholders. Undistributed profits are
internally generated funds available to business for use in operations, in
investment, or in the balance sheet as an addition to surplus (external
funds are obtained from bank loans: commercial and industrial and sell-
ing new equity stock). Undistributed profits fluctuate more than divi-
dends since companies do not change dividend payments frequently.
Company actions to change dividends are an indicator of business opti-
mism, as a dividend increase indicates an optimistic outlook, while a
dividend decrease indicates a pessimistic outlook.
Recent Trends
From 1995 to 2004, corporate profits fluctuated with annual percentage

increases in double digits during 199597 and 200204, and annual de-
creases of 4 to 8 percent in 1998, 2000, and 2001 (Table 12.1). Over the
entire nine-year period, corporate profits increased at an average annual
rate of 6 percent.

Business. Monthly.
13
Distribution of Income
The distribution of income data represent relative measures of in-

come inequality. They show the proportions of total money income
received by households in ascending steps on the income ladder. The
data are often referred to with the qualitative designation of low-,
middle-, and high-income groups. They are typically shown in sta-
tistical quintiles, which array the number of households from the
lowest to the highest fifths based on income. For example, in 2003,
the 20 percent of households with the lowest incomes received 3.4
percent of all money income, while the 20 percent of households
with the highest incomes received 49.8 percent of all money income.
Quintile income groups provide a relative measure of income distri-
bution by comparing the position of one income group with that of
others. In contrast, a single number, referred to as the Gini index of
income inequality, signifies the overall extent of inequality in the
income distribution.
The Bureau of the Census in the U.S. Department of Commerce pre-

pares annual measures of the distribution of income. The data are pub-
lished annually in Income, Poverty, and Health Insurance in the United
States (www.census.gov).
The income measures are available in August/September after the
year to which they refer. Revisions for previous years are made in the
annual publication.
61
Content
The official measures of the distribution of income represent money

income of households both before and after the payment of such items
as federal and state income taxes, Social Security taxes, federal em-
ployee retirement taxes, property taxes, Medicare deductions, and union
dues. Money income is defined as regularly received cash income,
such as wages and salaries, profits from self-employment, Social Se-
curity, retirement, unemployment insurance, other income maintenance
benefits, interest, dividends, rents, royalties, estates and trusts, educa-
tional assistance, alimony, child support, and financial assistance from
outside the household (excluding gifts and sporadic assistance). Non-
cash benefits, such as food stamps, Medicare, Medicaid, and rent
supplements, as well as income from nonrecurring sources such as
capital gains and life insurance settlements, are excluded from money
income. In addition to this official money-income measure, the Cen-
sus Bureau prepares seventeen alternative income-distribution mea-
sures based on varying definitions of items included and excluded as
cash and noncash income and on the inclusion and exclusion of cer-
tain taxes; these also are the basis for alternative estimates of the pov-
erty population (see poverty).
Statistical quintiles, which are featured in this description, are one of
various measures of income distribution (e.g., deciles, percentiles).
Quintiles array the number of households from the lowest to the highest
fifths based on income and are a commonly used method for highlight-
ing income shares in the distribution of income data. In contrast, the
Gini index of income inequality is a single overall measure of the in-
come distribution. The Gini index is a number that indicates the extent
of inequality throughout the entire distribution. The Gini index ranges
from zero to one, with zero indicating perfect equality (all recipients
receive equal amounts of income), and 1.0 indicating perfect inequality
(one recipient receives all of the income).
Household income data are also published in dollar amounts be-
fore the payment of taxes in selected increasing intervals up to
$99,999 (under $5,000, $5,000$9,999, $10,000$14,999, $15,000
$24,999, $25,000$34,999, $35,000$49,999, $50,000$74,999,
$75,000$99,999, $100,000 and over). Unpublished income inter-
vals from $100,000 to $249,999, in one total for $250,000 and over,
and in uniform intervals of $2,500 for the entire distribution up to
DISTRIBUTION OF INCOME 63
$249,999 are available on the Census Bureaus Web site. Income in-
tervals above $250,000 are not available, in order to maintain the
confidentiality of the data.
The household income data represent the income received by the
unit defined as a household. A household consists of all persons who
occupy a housing unit. A household may include one or more families
and/or one or more unrelated individuals. A family refers to two or
more persons related by birth, marriage, or adoption and living to-
gether in a house, apartment, a group of rooms, or a single room in-
tended for separate living quarters. An unrelated individual is a person
fifteen years old and older who does not live with any relatives. A
housing unit has direct access from the outside or through a common
hall. The occupants of a housing unit do not live or eat with any other
people in the structure. The definition of households excludes people
living in group quarters (e.g., hotels, dormitories), or institutions (e.g.,
hospitals, jails, shelters, halfway houses), or having no residence
(people living on the street).
The members of a household may or may not share their incomes for
personal consumption.1 Consequently, the household is not a perfect
unit for measuring income distribution. But the household is a practical
device for measuring the incomes of people living in housing units.
Methodology
The before-tax household income data are based on the Current Popula-
tion Survey (CPS) conducted by the Census Bureau. The information is
collected every March in an income supplement for the previous calen-
dar year. The survey sample is approximately 60,000 households. Typi-
cally, 55,500 are interviewed and 4,500 are not available for interviews.
For additional detail on the CPS, see employment.
Estimates of household income after the payment of taxes are based
on tax simulations incorporating the CPS income figures with several
other data sources. Federal and state income taxes are simulated based
on data from the Internal Revenue Services Statistics of Income and the
Commerce Clearinghouses State Tax Handbook; Social Security and
federal employee retirement taxes are estimated using the legal percent-
age rates for these taxes; and property taxes are estimated from infor-
mation in the U.S. Department of Housing and Urban Developments
American Housing Survey.
The quintiles and the Gini index for the household income data are
calculated using two methodologiesactual sorted data and grouped
data. The actual sorted data are used for the before-tax estimates. The
grouped data are in income intervals of $2,500 and are used for the
after-tax estimates (see Content, above).
Accuracy
The sampling error (for one standard error) for the quintile shares of
income before taxes in 2004 rose in each quintile from 0.02 of a per-
centage point for the lowest quintile to 0.34 percentage point for the
highest quintile of households. In the lowest quintile of households, the
share of income before taxes in 2004 was 3.4 percent. For example, in
two of three cases the true share of income was somewhere between
3.398 and 3.402 percent. For further information on the interpretation of
A comparison of the household income estimates of the Census Bu-
reau with the personal income estimates of the U.S. Bureau of Eco-
nomic Analysis (see personal income and saving) for the year 2001,
after placing both income estimates on a comparable definitional ba-
sis, indicated that total household income was 11 percent below total
personal income.2 The difference is attributed to underreporting by
survey respondents on the CPS, which is the source of household in-
come data, as contrasted with administrative records of income tax,
unemployment insurance, Social Security, and other income programs
that are the source of the personal income data. This overall
underreporting is not taken into account in developing the income dis-
tribution data because determining the variations in underreporting
among income groups is difficult.
Relevance
The income distribution focuses on differences in economic well-

being among groups in the population. The data show the inequality of
income shares in the population and how it changes over time. The
designation of low-, middle-, and high-income group is qualitative. It
varies with the income thresholds used to arrive at these classifica-
tions, and with the perception individuals have of who is rich, poor, or
in the middle.3 The user of income distribution data should be aware

of the income thresholds associated with references to low-, middle-,
and high-income groups.
A large disparity in the income distribution suggests a society that is
divided into haves and have nots, which raises both economic and
social concerns. Economic growth is hindered when purchasing power
and profit-motivated incentives are not broadly based. Socially, a large
disparity results in increasing discord and despair among the popula-
tion. Economic growth and social harmony are regarded as essential to a
democratic and stable society, even while political and economic phi-
losophies for achieving these goals differ.
The distribution of income is related to poverty and the distribution of
wealth, as well as to productivity, average weekly earnings, and price move-
ments in the consumer price index, producer price indexes, and GDP price
measures. Inequality of income stems from many sources, such as the match
of workers having particular skills with the job market for those skills, dis-
crimination, inheritance, innate abilities, entrepreneurial spirit, and luck.
Recent Trends
From 1980 to 2004 (based on five-year intervals to 2000 and then 2003
and 2004), the share of money income received by each of the first four-
fifths of the households declined (Table 13.1). Conversely, the top fifth
of households showed an increasing share of the income from 1980 to
2004. The only exception occurred in the fourth-fifth category, which
increased slightly from 2000 to 2004.
The same general pattern occurred both before and after the payment
of federal and state income taxes, Social Security and federal employee
retirement taxes, and property taxes, although the inequality increased
slightly less for income after taxes. Before taxes, the share of total income
going to the top fifth of households rose from 43.7 percent in 1980 to 49.8
percent in 2003 (6.1 percentage points). After taxes, the top fifths share
rose from 40.6 percent in 1980 to 46.2 percent in 2003 (5.6 percentage
points). Income share estimates after taxes for 2004 were not available at
the time of this writing.
The Gini index for income before taxes rose from 0.401 in 1980 to
0.466 in 2004. Thus, on an overall basis, income inequality increased
over the twenty-three-year period.
66
Table 13.1
Distribution of Income: Household Shares (in 2003 dollars)
Income (percentage)
Households (quintiles) 2004 2003 2000 1995 1990 1985 1980
Income before taxes
Lowest fifth 3.4 3.4 3.6 3.7 3.9 4.0 4.3
Second fifth 8.7 8.7 8.9 9.1 9.6 9.7 10.3
Third fifth 14.7 14.8 14.8 15.2 15.9 16.3 16.9
Fourth fifth 23.2 23.4 23.0 23.3 24.0 24.6 24.9
Highest fifth 50.1 49.8 49.8 48.7 46.6 45.3 43.7
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Gini index of income inequality 0.466 0.464 0.462 0.450 0.428 0.419 0.403
GUIDE TO ECONOMIC INDICATORS
Mean money income in quintiles in 2004: lowest fifth: $10,264; second fifth: $26,241; third fifth: $44,455; fourth fifth: $70,085; highest
fifth: $151,593
Income after taxes NA

Lowest fifth NA 4.0 4.5 4.5 4.5 4.6 4.9
Second fifth NA 10.0 10.5 10.5 10.8 11.0 11.6
Third fifth NA 15.9 16.5 16.4 16.9 17.2 17.9
Fourth fifth NA 24.0 24.1 23.9 24.3 24.7 25.1
Highest fifth NA 46.2 44.4 44.6 43.5 42.6 40.6
Total 100.0 100.0 100.0 100.0 100.0 100.0
Note: Components may not sum to totals, due to rounding. NA = Not available at time of writing.
Bureau of the Census, U.S. Department of Commerce. Income, Poverty, and Health
Insurance Coverage in the United States. Annual.
Notes
1. Paul Ryscavage, Income Inequality in America (Armonk, NY: M.E. Sharpe,

1999), p. 29.
2. John Ruser, Adrienne Pilot, and Charles Nelson, Alternative Measures of
Household Income: BEA Personal Income, CPS Money Income, and Beyond, May
2004. Available from the Census Bureau, www.census.gov.
3. Ryscavage, Income Inequality in America, pp. 9 and 35.
14
Distribution of Wealth
The distribution of wealth represents the material assets, liabilities, and

net worth (assets minus liabilities) of households. The distribution fo-
cuses on the percentage shares of wealth accounted for by households.
Assets cover financial and nonfinancial categories: cash, securities, and
other financial holdings, and personal property, real property, and other
nonfinancial holdings. Liabilities comprise debt associated with con-
sumer credit, mortgage loans, and other borrowing.
The Federal Reserve Board (FRB) prepares the distribution of wealth

data for a single year every three years. A summary of the data is pub-
lished in the quarterly FRB journal, Federal Reserve Bulletin
(www.federalreserve.gov).
The wealth data are available approximately two years after the year
to which they refer. The wealth measures are not scheduled to be re-
vised; they are revised on an ad hoc basis when a specific issue arises.
Content
Distribution-of-wealth data are provided as a balance sheet of the as-

sets, liabilities, and net worth for families. They are available on a statis-
tically consistent basis for 1989, 1992, 1995, 1998, and 2001. Earlier
wealth data for the 1960s, 1970s, and 1983 are not statistically consis-
tent with the 1989 and later data. Wealth data are planned for every three
years for 2004, 2007, and so on.
68
DISTRIBUTION OF WEALTH 69
In the distribution of wealth data, a family is defined as a couple

whether married or living together as partners, as well as individuals
living in the household who are financially interdependent with the couple
or with other individuals in the household. Thus, this definition of a
family is closer to the definition of a household in the U.S. Census
Bureaus data on income distribution, in which an individual is not clas-
sified as a family, than it is to the definition of family in the Census
Bureaus data (see distribution of income).
Assets
Financial and nonfinancial asset data represent the cumulation of previ-

ous incomes that have been saved or invested.
Financial Assets
Transaction accounts: checking, savings, and money market ac-

counts, other
Certificates of deposit
Savings bonds
Other government and commercial bonds
Publicly traded corporate stocks
Mutual funds
Retirement accounts
Cash value life insurance
Other financial assets
Nonfinancial Assets
Vehicles: automobiles, sport utility vehicles, trucks, motorcycles,

motor homes, recreational vehicles, airplanes, boats
Primary residence
Other residential property
Equity in nonresidential property
Business equity: sole proprietorships, partnerships, corporations, other
Other nonfinancial assets: artwork, jewelry, precious metals, antiques,
other tangible assets
The values of these assets incorporates the effects of capital gains and
losses. Capital gains and losses are the difference between the current and
sales price of an asset. Capital gains or losses that result from changes in
the value of such assets as stocks, real estate, and businesses are realized
as income only when the asset is sold. Unrealized capital gains and losses
are those in which the asset has not yet been sold and are included in the
wealth measures. Thus, as the prices of stocks, businesses, or real estate
assets rise or fall, the wealth estimates rise or fall accordingly.
Liabilities
Liabilities data refer to debts incurred from borrowings. Some of the

liabilities are financial obligations associated with the financial and non-
financial assets mentioned above.
Home secured debt: home mortgages and home equity borrowing

Other residential debt
Installment borrowing
Other lines of credit
Credit card balances
Other debt: loans on insurance policies, loans against pension ac-
counts, borrowings on margin accounts, other loans
Methodology
The distribution-of-wealth data are obtained from the Survey of Consumer

Finances sponsored by the Federal Reserve Board. The NORC at the Uni-
versity of Chicago (formerly the National Opinion Research Center at the
University of Chicago) has conducted the survey for the FRB since 1992.
The 2001 survey is based on information obtained from a probability
sample of 4,499 households. The sample includes an oversampling of
wealthy households that hold a disproportionately large share of such
assets as noncorporate businesses and tax-exempt bonds to ensure that
sufficient survey returns are available for developing these estimates.
The data were gathered during MayDecember 2001, primarily from
interviews conducted in person, plus telephone interviews when more
convenient for the respondent.
Accuracy
Standard errors (for one standard error) due to sampling and imputations
for the distribution-of-wealth data based on the net worth held by various
DISTRIBUTION OF WEALTH 71
Table 14.1
Net Worth and Standard Errors of Percentile Groups of Families: 2001

(percentiles of families)
049.9 5089.9 9094.9 9598.9 99100 Total

Share of 2.8 27.4 12.1 25.0 32.7 100.0
dollar net
worth
(percentage)
Standard 0.1 0.7 0.7 1.1 1.4
errors
Source: Arthur B. Kennickell, A Rolling Tide: Changes in the Distribution of Wealth

in the U.S., 1989-2001, Federal Reserve Board, Washington, DC. September 2003,
Table 5, p. 9.
percentile groupings of families for 2001 are shown in Table 14.1. In the
049.0 percentile of families, the standard error was 0.1 of a percentage
point and the share of dollar net worth was 2.8 percent. For example, in
two of three cases the true share of the dollar net worth was somewhere
between 2.7 and 2.9 percent. For further information on the interpretation
of sampling and nonsampling errors, see the Appendix.
Relevance
The distribution of wealth shows the disparity of economic well-being

among households. In highlighting differences in the material well-be-
ing of the population, the wealth measures summarize the relative eco-
nomic well-being and economic power derived from income and saving
flows over time and the intergenerational transfer of wealth. The wealth
data are the resultant inventory of haves and have nots emanating
from income and saving flows and intergenerational transfers.
The distribution of wealth is related to the distribution of income and
poverty. Inequality of income stems from many sources, such as the match
of workers having particular skills with the job market for those skills,
discrimination, inheritance, innate abilities, entrepreneurial spirit, and luck.
Recent Trends
From 1989 to 2001 (based on three-year points), movements in the dis-

tribution of net worth (assets minus liabilities) among families grouped
Table 14.2
Distribution of Dollar Net Worth (percentage)
Percentiles of families
049.9 5089.9 9094.9 9598.9 99100 Total
1989 2.7 29.9 13.0 24.1 30.3 100.0
1992 3.3 29.7 12.6 24.4 30.2 100.0
1995 3.6 28.6 11.9 21.3 34.6 100.0
1998 3.0 28.4 11.4 23.3 33.9 100.0
2001 2.8 27.4 12.1 25.0 32.7 100.0
Note: Components may not sum to totals, due to rounding.

Source: See Table 14.1.
by percentile showed a general increasing inequality of wealth, though

the inequality patterns fluctuated among the five survey years (Table
14.2). In 2001, the lowest 50 percent of families in terms of net worth
accounted for 2.8 percent of the net worth held by all families, while the
highest one percent of families in terms of net worth held 32.7 percent
of the net worth held by all families.

Ana M. Aizcorbe, Arthur B. Kennickell, and Kevin B. Moore, Recent Changes in
U.S. Family Finances: Evidence from the 1998 and 2001 Survey of Consumer
Finances, Federal Reserve Bulletin, January 2003.
Arthur B. Kennickell, A Rolling Tide: Changes in the Distribution of Wealth in the
U.S., 19892001, Federal Reserve Board, September 2003.
EMPLOYMENT 73
15
Employment
Employment represents workers engaged in gainful work. There are two

official measures of employment prepared by the U.S. Bureau of Labor
Statistics. One is a count of jobs and is based on a survey of employer
establishments, and the other is a count of employed persons and is
based on a survey of households. Both measures reflect similar phe-
nomena, but because the definitions and methodologies differ, there are
periods when their movements differ substantially.
The two surveys are described separately, first the establishment sur-
vey and then the household survey. This is followed by a summary of
the main technical differences between them, and then by highlights of
the relevance and recent trends of both.
Establishment Survey
Employment data are provided monthly by the Bureau of Labor Statis-

tics (BLS) in the U.S. Department of Labor. The data are published in a
news release and in two BLS monthly journals, Monthly Labor Review
and Employment and Earnings (www.bls.gov).
the twelfth of the month. Thus, the information is released on the first
or second Friday of the month following the month in question. The
exception to the third Friday rule is that if January 1, 2, or 3 or a fed-
eral holiday falls on the third Friday, the release date is moved one day
earlier. On the day the data are released, the commissioner of labor
73
statistics reports on recent employment and unemployment trends to

the Joint Economic Committee of Congress. Preliminary data are pro-
vided for the immediately preceding month; these are revised in the
subsequent two months. Annual revisions are made in February of the
following year.
Content
Employment data cover the number of paid nonfarm civilian jobs on

employer payrolls in U.S. enterprises and governments. To be counted,
a job must be on the payroll of a business, a nonprofit organization, or
the federal or a state or local government. All paid jobs are counted
equally, from the lowest pay scales to company executives and officers,
as are full-time and part-time jobs. Since some individuals hold two or
more jobs, the number of jobs exceeds the number of working persons.
Included in the employment data are residents of Canada and Mexico
who commute to jobs on employer payrolls in the United States, and
institutionalized persons (e.g., those confined to penal or mental facili-
ties) and residents of old-age homes on payroll jobs. The employment
data exclude farm workers, self-employment, jobs in private households,
railroad employment, religious organization employment, elected offi-
cials, unpaid family work, military personnel on active duty in the armed
forces (but uniformed military personnel who hold civilian jobs are in-
cluded in the job count), and employees of the Central Intelligence
Agency, Defense Intelligence Agency, National Geospatial-Intelligence
Agency, and National Security Agency.
Persons on paid leave for illness or vacation are counted as employed
because the job continues as a payroll cost. Those temporarily not work-
ing because of illness, vacation, strike, or lockout and who are not paid
are not counted as employed.
The establishment employment data are seasonally adjusted.
Methodology
The employment data are based on employer payroll records that rep-
resent employees on payrolls during pay periods that include the twelfth
day of the month. The data are obtained from a survey of a sample of
approximately 160,000 businesses and government agencies includ-
ing 400,000 individual worksite establishments in 2004. The sample
EMPLOYMENT 75
includes employers with only one work location as well as those with
several establishments (an establishment is defined as the physical lo-
cation of an employers operations, with companies and other organi-
zations operating in more than one location having several
establishments). The survey sample covered approximately one-third
of all nonfarm employment in 2004.
The survey sample became a full probability sample in 2003. The
monthly estimates are based on changes in employment by the same
establishments reporting in the preceding month. In order to reduce the
reporting burden, all but the largest establishments are rotated out of the
sample after participating in the survey for an extended period of time.
Employers having single or multiple establishments in their unemploy-
ment insurance (UI) tax accounts that have more than 1,000 employees
are in the establishment sample continuously, that is, they are not ro-
tated out of the sample. Employers pay their UI taxes to the Bureau of
Public Debt in the U.S. Department of the Treasury.
The surveys of industry, nonprofits, and state and local governments
are conducted for BLS by the state governments employment agencies.
The monthly data are collected primarily by electronic data interchange,
touch data entry, and computer-assisted telephone interviewing, and sec-
ondarily by Fax and mail. Data covering all federal civilian workers,
including the Department of Defense, are provided by the U.S. National
Finance Center. Armed forces military personnel and employees of the
Central Intelligence Agency, Defense Intelligence Agency, National
Geospatial-Intelligence Agency, and National Security Agency are ex-
cluded from the job count.
The monthly employment data are benchmarked and revised every
year. The benchmark is based primarily on the universe of the UI em-
ployment data, which account for approximately 97 percent of the total
employment of the establishment survey. The remaining 3 percent of
total employment is obtained mainly from records of the Railroad Re-
tirement Board and from the U.S. Census Bureaus County Business
Patterns (CBP); the CBP data reflect employment statistics from ad-
ministrative records of the U.S. Internal Revenue Services Employers
Quarterly Federal Tax Return (Form 941) for small employers, and from
the Census Bureaus annual Company Organization Survey for large
employers. The benchmark data are prepared for March of the previous
year, and the relative revisions for March are carried back through the
previous eleven months and extrapolated forward to the current period.
The benchmark data become available in February of the current period

with the release of the January data. For example, the employment data
beginning with January 2006 that are released in February 2006 are based
on the March 2005 benchmark data.
Because of the difficulty of obtaining timely information on the startup
of new firms, the establishment survey collection is late in capturing the
employment in these firms. To compensate for this understatement of
jobs in new establishments, the survey data are augmented by an esti-
mate derived from a statistical model. The model also implicitly adjusts
for the late identification of establishments going out of operations.
Accuracy
The sampling error (for one standard error) for the establishment em-
ployment data is 0.2 percent for the monthly level and 67,700 jobs for
the monthly change. For example, if the estimated monthly level of es-
tablishment employment were 130 million workers, in two of three cases
the true level would be somewhere between 129,740,000 and
130,260,000 workers. And if the estimated increase in establishment
employment from one month to the next were 200,000 workers, in two
of three cases the true increase would be somewhere between 132,300
and 267,700 workers. For further information on the interpretation of
Household Survey
Employment data are provided monthly by the Bureau of Labor Statis-

tics in the U.S. Department of Labor. The data are published in a news
release and in two BLS monthly journals, Monthly Labor Review and
Employment and Earnings (www.bls.gov).
the twelfth of the month, which falls on the first or second Friday of the
month following the month to which they refer. Thus, the information is
released on the first or second Friday of the month following the month
in question. The exception to the third Friday rule is that if January 1, 2,
or 3 or a federal holiday falls on the third Friday, the release date is
moved one day earlier. On the day the data are released, the commis-
EMPLOYMENT 77
sioner of labor statistics reports on recent employment and unemploy-

ment trends to the Joint Economic Committee of Congress. The monthly
data are revised every January for the previous five years based on up-
dated seasonal factors.
Content
Employed persons represent noninstitutionalized individuals sixteen

years old and older living in the United States who worked at least one
hour as wage- and salary-paid employees, self-employed persons work-
ing in their own businesses, unpaid workers in a family business who
worked at least fifteen hours a week, and civilian government workers
in nonfarm and farm activities.1 They include U.S. residents who com-
mute to jobs in Canada and Mexico. Thus, the employed civilian popu-
lation consists of wage and salary employees and those who work for
profit, the latter being the self-employed and unpaid workers in family
businesses who are assumed to share in the profits. All persons are
counted equally if they are paid for an hour or more per week. If a per-
son has two or more jobs, the job with the most hours worked in the
week is the only one counted, and the hours worked for all jobs are
assigned to that job. Supplemental data on the number of multiple job-
holders are provided. Also, detail is provided by age, gender, race, His-
panic origin, educational attainment, marital status, and broad
occupational and industry groupings.
Persons are defined as employed who are temporarily absent because
of vacation, illness, bad weather, child-care problems, maternity or pa-
ternity leave, strike or lockout, job training, or other family or personal
reasons, regardless of whether they are paid for the time off or are seek-
ing other jobs. Included as employed are citizens of foreign countries
who are temporarily in the United States but not living on embassy pre-
mises. Excluded from the employment data are all military personnel
on active duty in the armed forces (including uniformed military per-
sonnel holding civilian jobs); residents of Canada and Mexico who com-
mute to jobs in the United States; persons who work solely around their
house (painting, repairing, other home housework); those who do vol-
unteer work for religious, charitable, or other organizations; and institu-
tionalized persons (e.g., those confined to penal or mental facilities) and
residents of homes for the aged on payroll jobs.
The household employment data are seasonally adjusted.
Methodology
The household survey employment data are obtained from a monthly

survey of households called the Current Population Survey (CPS). The
CPS is a sample of about 60,000 households, which the U.S. Bureau of
the Census conducts for the Bureau of Labor Statistics.2 Responses are
actually obtained from about 55,500 households; no responses are ob-
tained from the remaining 4,500 households due to absence, impassable
roads, refusals, or other reasons. The sample is representative of the
distribution of households in small and large metropolitan areas and in
rural areas. It undergoes a major revision every ten years to be consis-
tent with the most recent decennial population census. The sample cur-
rently is based on the decennial 2000 census of population. The sample
is also updated during intercensal years on a limited basis to reflect
current changes in residential locations due to new construction based
on housing starts data prepared by the Census Bureau and on estimates
of international migration.
In order to reduce the reporting burden on any group of households,
the sample is divided into eight subsamples (panels) that are rotated
over a sixteen-month period. Each subsample is surveyed for four con-
secutive months, then dropped from the survey for eight months, and
subsequently resurveyed for the following four months. At the end of
the sixteen months, the subsample is eliminated from the sample and is
replaced with a new panel of households. The result of this procedure is
that every month 25 percent of the households in the sample are either
new to the survey or are returning after an eight-month hiatus. Corre-
spondingly, 25 percent of the sample households drop out of the survey
every month.
The survey refers to the individuals employment status during the
calendar week (Sunday through Saturday) that includes the twelfth of
the month. The survey is conducted mainly by telephone interviews,
supplemented by personal interviews as necessary.
The CPS survey data are also used to estimate unemployment.
Accuracy
The sampling error (for one standard error) for the household employ-
ment data is 326,000 workers for the monthly level and 212,000 work-
ers for the monthly change. For example, if the estimated monthly level
EMPLOYMENT 79
of household employment were 150 million workers, in two of three

cases the true level would be somewhere between 149,574,000 and
150,326,000 workers. And if the estimated increase in household em-
ployment from one month to the next were 200,000 workers, in two of
three cases the true change would be somewhere between 12,000
and 412,000 workers. For further information on the interpretation of
Comparison of the Establishment and Household

Employment Measures
This section summarizes the main differences between the employment

data obtained from the establishment and household surveys.
Coverage
Employment data based on the establishment survey are limited to em-

ployees in nonagricultural industries (including government civilian
workers) who are paid for their work or for their absence from the job.
By contrast, employment data based on the household survey cover a
broader range of employment, including farm workers, the self-employed,
private household workers, unpaid workers in family businesses, plus
those temporarily absent from work due to illness, vacation, strike, or
lockout, even if they are not paid during their absence. The establish-
ment data partially compensate for their smaller coverage by including
workers of all ages, multiple jobs of workers, residents of Canada and
Mexico who commute to the United States for work, and institutional-
ized persons and residents of homes for the aged on payroll jobs. By
contrast, the household survey is limited to workers sixteen years and
older, counts each worker only once regardless of how many jobs he or
she may hold (supplementary data on multiple jobs of workers are avail-
able), excludes Canadian and Mexican residents commuting to jobs in
the United States, but includes U.S. residents commuting to Canada and
Mexico, and excludes institutionalized persons and residents of homes
for the aged on payroll jobs.
Also, there are differences in the treatment of persons in the armed
forces on active duty. The establishment survey distinguishes between
uniformed personnel holding military or civilian jobs. Those holding
military jobs are excluded and those holding civilian jobs are included
in the employment count. By contrast, the household survey excludes

both categories from the employment count.
In addition, the survey reference period differs between the two sur-
veys. The reference period for the establishment survey is the pay pe-
riod that includes the twelfth of the month, while the reference period
for the household survey is the calendar week (Sunday through Satur-
day) that includes the twelfth of the month. Because pay periods vary in
length, typically from one to two weeks but in some cases also longer,
such as with monthly pay, the establishment survey counts workers over
a longer period than the household survey, which can cause differences
in the two measures if the employment movement is upward or down-
ward during the pay period.
The net effect of these differences is that the household survey shows
more employment than the establishment survey. Reconciliation of both
measures, when they are put on as similar a basis as the available data
will allow, reduces the difference substantially.3 For example, in 2003,
household survey employment exceeded establishment survey employ-
ment by 7.74 million before the reconciliation and only 258,000 after
the reconciliation; and in 2004, household survey employment exceeded
establishment survey employment by 7.77 million before the reconcili-
ation and only 345,000 after the reconciliation.
Component Detail
The establishment survey details the industry composition and geo-

graphic location of jobs, which is useful in analyses of the industrial
structure and location of employment. The household survey details
the age, gender, race, Hispanic origin, educational attainment, marital
status, and broad occupational and industry groupings of workers,
which is useful for analyzing the worker characteristics of employ-
ment trends. Therefore, each measure has its particular uses because
of the different detail provided.
Accuracy
Statistically, both surveys have strengths and weaknesses. The estab-

lishment survey has a better information source because the data are
obtained from employer payroll records, which are used for tax and
accounting records. But its use of a statistical model to compensate for
EMPLOYMENT 81
the late introduction of new employer establishments into the survey

sample may cause the monthly movements to be smoother than actually
occurs in the workplace.
In contrast, the household survey obtains undocumented answers to
survey questions from household members that suffer from such prob-
lems as the inability to obtain information on all persons in the sample
household,4 differences in the interpretation of the questions by differ-
ent respondents in the same household from month to month, inability
or unwillingness of respondents to provide correct information, and in-
ability to recall information. The change in respondents from month to
month by the rotation of the survey samples may cause more volatility
in the changing monthly direction of employment than actually occurs
in the workplace.
Overall, while each survey has shortcomings, I agree with the con-
sensus view of the economics community that the establishment survey
provides more reliable measures of monthly and annual employment
movements than those of the household survey. In addition to the greater
reliability of the employer payroll records of employment used in the
establishment survey compared with the personal interviews in the house-
hold survey, I believe the much larger sample and the much smaller
sampling error in the establishment survey make its employment esti-
mates superior to those of the household survey.
Relevance
Employment is the main source of household incomes, which in turn,

are spent on consumer goods and services. Because household spending
(retail sales represent the goods component) accounted in the first half
of the 2000s for 70 percent of the gross domestic product, employment
is a key factor affecting economic growth. In addition, employment data
based on the establishment survey are used as inputs to other indicators
such as the gross domestic product, industrial production index, and
leading, coincident, and lagging indexes.
The distribution of employment between high- and low-paying jobs
also affects personal income and saving. The types of jobs held influ-
ence economic growth as well as living conditions (distribution of in-
come and distribution of wealth), because the bulk of the population
depends on employment as its main source of income.
The distinction in detail provided by both surveysindustry and ge-
Table 15.1
Employment: Alternative Definitions
Establishment survey Household survey

(nonfarm civilian jobs) (All civilian workers)
Level (millions)
1995 117.3 124.9
1996 119.7 126.7
1997 122.8 129.6
1998 125.9 131.5
1999 129.0 133.5
2000 131.8 136.9
2001 131.8 136.9
2002 130.3 136.5
2003 130.0 137.7
2004 131.5 139.3
Change (millions)
19952000 14.5 12.0
20002004 0.3 2.4
19952004 14.2 14.4
200001 0.0 0.0
200102 1.5 0.4
200203 0.3 1.2
200304 0.5 1.6
Annual percentage change
19952000 2.36 1.85
20002004 0.06 0.44
19952004 1.28 1.22
ography for establishment survey and demographic for the household

surveymeans that at the disaggregated level, each survey has its unique
uses, as noted above.
Recent Trends
From 1995 to 2004, establishment employment increased 14.2 million,

and household employment increased 14.4 million (Table 15.1). Both
surveys showed similar directional movements in employment in all
years except 2003, when employment in the establishment survey de-
clined while that in the household survey rose.
But there were sharp differences in the movements of both surveys
between the intermediate periods of 19952000 and 200004 (both sur-
veys showed zero change between 2000 and 2001). Over the 19952000
period, establishment employment increased 14.5 million and house-
EMPLOYMENT 83
hold employment increased 12.0 million, while over 200004, estab-

lishment employment decreased 0.3 million and household employment
increased 2.4 million. Also, the 200204 establishment employment lev-
els were below those in 2000 and 2001, while the 200304 employment
levels in the household survey were above those in 2000 and 2001.
These differences also appeared in the relative measures of annual
percentage change between 19952000 and 200004. These showed
negative growth in establishment employment and marked lesser growth
in household employment in 200004 compared with 19952000.
Notes
1. Self-employed persons whose businesses are unincorporated are classified as

self-employed, but those whose businesses are incorporated are classified as wage and
salary employees because, technically, they are paid employees of a corporation.
2. A household consists of all personsrelated family members and unrelated
individualswho occupy a housing unit and have no other usual address. A housing
unit is intended as separate living quarters, and encompasses single-family houses,
townhouses, condominiums, apartments, mobile homes, single rooms, and group
quarters where residents share common facilities or receive formal or authorized
care or custody. There were 112.0 million households in the United States in 2003.
3. There are a variety of irreconcilable statistical differences between the es-
tablishment and household employment data, including sampling errors, benchmarks,
estimation methodologies, pay period versus calendar week data collections, and
some independent contractors classified as wage and salary workers rather than as
self-employed in the household survey.
4. Particularly minority men and undocumented aliens.
16
Employment Cost Index
The employment cost index (ECI) measures changes in labor costs to

employers for money wages and salaries and noncash fringe benefits in
nonfarm private industry and state and local governments for workers at
all levels of responsibility. The ECI is not affected by shifts in the com-
position of employment between high-wage and low-wage industries or
between high- and low-wage occupations within industries. Thus, the
ECI represents labor costs for the same jobs over time. This contrasts
with the shifting types of jobs in average weekly earnings,
The employment cost index is prepared quarterly by the Bureau of La-

bor Statistics (BLS) in the U.S. Department of Labor. The data are pub-
lished in a news release and in the BLS monthly journal, Monthly Labor
Review (www.bls.gov).
The data are available during the last week of the month immediately
following the quarter to which they refer (April for the first quarter, July
for the second quarter, October for the third quarter, and January for the
fourth quarter). No revisions are made to the ECI data, except for the re-
estimation of seasonal factors for the most recent five years.
Content
The employment cost index data cover labor costs to employers. The
costs include money wages and salaries; commissions; bonuses; fringe
benefits such as paid leave for vacations, illness, holidays, and so on,
84
EMPLOYMENT COST INDEX 85
and noncash health, retirement, and other fringe benefits. The data are
provided for private nonfarm industries, including union and nonunion
workers separately, and for state and local governments. Costs are in-
cluded for all workersproduction, nonsupervisory, supervisory, and
executive. The wage and salary component of labor costs reflects straight-
time pay only before payroll deductions, excluding premium rates for
overtime, holidays, night work, and hazardous conditions. Production
bonuses, incentive earnings, commission payments, and cost-of-living
adjustments are included in straight-time wage and salary rates.
The benefit cost component reflects the cost of benefits. Benefit costs
include paid vacation, sick, and holiday leave; life, health, and disability
insurance; higher pay for overtime, weekends, holidays, shift differen-
tials, and nonproduction bonuses such as referral bonuses and lump-
sum payments in lieu of wage increases; legally required benefits for
Social Security, Medicare, federal and state unemployment insurance,
and workers compensation, and other benefits for severance pay and
supplemental unemployment plans.
Wages and salaries plus benefit costs are called compensation. Data
are provided separately for compensation, wages and salaries, and ben-
efit costs.
The ECI represents a fixed composition of industries and of occupa-
tions within industries. Therefore, movements in the index over time are
not affected by shifts between higher- and lower-paying industries and
occupations. By contrast, average weekly earnings are affected by such
changes. In addition to industry and occupational detail, the ECI distin-
guishes compensation costs between union and nonunion workers.
The ECI is currently based on an index base of June 1989 = 100.
The ECI data are seasonally adjusted.
Methodology
The employment cost index data are based on an ECI survey of em-
ployer payrolls in the third month of the quarter (March, June, Septem-
ber, and December) for the pay period including the twelfth day of the
month. The survey is a probability sample of approximately 9,700 pri-
vate industry employers and 800 state and local governments, public
schools, and public hospitals obtained for 10 occupational categories.
The index weights represent the wage and salary and fringe benefit
costs of each occupation within an industry. This is average compensa-
tion per worker multiplied by the number of workers in each occupa-

tion/industry group. The employment data are obtained from the census
of population and the compensation data are from the ECI survey. The
composition of industry and occupational employment currently prima-
rily reflects the distribution of the triennial occupational employment
survey of the Bureau of Labor Statistics. The occupational/industry
weights derived from that survey will be updated to 2002 in 2006. In
contrast to the overall ECI, the component indexes for union and non-
union workers are based on current period distributions rather than fixed
weights because of the changing union status of workers within a com-
pany; these are updated every quarter based on the current ECI survey
of employer payrolls.
Accuracy
The sampling error (for one standard error) of the twelve-month per-
centage change in the employment cost index for compensation both for
private industry workers and for state and local government workers is
0.2 of a percentage point. For example, if the estimated increase in the
ECI over a twelve-month period were 4 percent, in two of three cases
the true increase would be somewhere between 3.8 and 4.2 percent.
For further information on the interpretation of sampling and non-
sampling errors, see the Appendix.
Relevance
The employment cost index is the most comprehensive measure of un-

derlying trends in employee compensation as a cost of production. It is
used for analyzing changes in wages and salaries and fringe benefits in
labor markets for various categories of occupations and industries, col-
lective bargaining negotiations, and cost escalators in union and other
business contracts, and for adjusting pay of federal government em-
ployees including members of Congress (if Congress elects to use it),
federal judges, and senior government executives. In distinguishing be-
tween union and nonunion workers, it also provides data for contrasting
compensation trends between union and nonunion companies.
The ECI is not strongly related to overall price movements in the
consumer price index, producer price indexes, and GDP price measures.
This probably reflects the fact that it does not adjust for productivity
EMPLOYMENT COST INDEX 87
Table 16.1
Employment Cost Index: Private Industry (annual percentage change)
12 months ending Wages and Benefit

December Compensation salaries costs
1995 2.6 2.8 2.2
1996 3.1 3.4 2.0
1997 3.4 3.9 2.3
1998 3.5 3.9 2.4
1999 3.4 3.5 3.4
2000 4.4 3.9 5.6
2001 4.2 3.8 5.1
2002 3.2 2.7 4.7
2003 4.0 3.0 6.4
2004 3.8 2.4 6.9
changes.1 By contrast, unit labor costs data adjust for productivity

changes and tend to parallel price movements.
Recent Trends
From 1995 to 2004, the employment cost index showed greater increases
in annual compensation costs during 200004 compared with those dur-
ing 199599 (Table 16.1). From 1995 to 1999, annual increases in com-
pensation were in the 3.5-percent-and-under range, and from 2000 to
2004, the annual compensation increases were typically in the 4-percent
range, though they declined from the peak 4.4 percent in 2000 to 3.8
percent in 2004. The greater compensation costs during 200004 re-
flected the much larger increases in benefit costs compared with those
in wage and salary costs during the period. This reversed the earlier
pattern of 199599, when wages and salaries showed greater increases
than benefit costs.
Bureau of Labor Statistics, U.S. Department of Labor. Monthly Labor Review.

Monthly.
Note
1. C. Alan Garner, A Closer Look at the Employment Cost Index, Economic

Review, Federal Reserve Bank of Kansas City, third quarter 1998.
17
Farm Parity Ratio
The farm parity ratio provides a general indication of farmers economic

well-being. The ratio shows the relationship between (a) the prices farmers
receive for sales of crop and livestock products, and (b) the prices farmers
pay for production and living expenses. This relationship is a limited mea-
sure of the change in income because it does not include the effects of im-
provements in production technology or of the changing quantities and quality
of farm products sold. The 191014 period is used as the base for compari-
son because prices of farm and nonfarm items were considered to have been
generally in balance in that period by the legislators who passed the Agri-
cultural Adjustment Act of 1939, establishing the parity ratio.
The farm parity ratio is prepared monthly by the National Agricultural

Statistics Service of the U.S. Department of Agriculture. It is published
in the report Agricultural Prices (www.usda.gov/nass).
The data are available at the end of the month to which they refer.
Revisions are made to each month in the following monthly estimates,
and annually in January of the following year.
Content
The farm parity ratio is composed of the index of prices received for
sales of crop and livestock products in the numerator, and the index of
prices paid for farm production and living expenses in the denominator.
The percentage change in both indexes reflects the movement from 1910
88
FARM PARITY RATIO 89
14 to the current period. Currently, the ratio using 199092 = 100 is also
provided to facilitate comparisons with other price indexes. Basing the
ratio on 199092 does not affect percentage changes from one period to
another, but the levels are different. Thus, when using the 199092 base,
a comparison with 191014 is not readily observable unless back data
are shown for the earlier period. When either ratio is above 100, farm-
ers purchasing power is higher than in the base period, and when either
ratio is below 100, their purchasing power is less than in the base period.
The current weights in both the prices received and the prices paid
indexes reflect the relative dollar importance of sales and expenses of the
components of each index during 199092. In the prices-received index,
crops account for 48 percent of the weight and livestock products for 52
percent. Crop products include food and feed grains, cotton, tobacco, oil-
bearing crops (e.g., soybeans and peanuts), and fruits and vegetables. Live-
stock products include meat animals, dairy products, poultry, and eggs.
The index represents about 90 percent of the cash receipts from all farm
products. Of the excluded commodities, livestock products such as wool,
horses, goats, and ducks account for 2 percent and crop products such as
forest, nursery, greenhouse, and specialty crops account for 8 percent.
In the prices-paid index, farm production expenses are weighted 82
percent and living expenses are weighted 18 percent. Farm production
costs include such items as feed, feeder livestock, seed, fertilizer, fuels,
chemicals, equipment, cash rent, wages, interest, and real estate taxes.
The living expense component is based on the consumer price index
(CPI-U).
The farm parity ratio is currently based both on 191014 = 100 and
199092 = 100.
The farm parity ratio is not seasonally adjusted.
Methodology
Weights for the prices-received and prices-paid indexes are based on five-
year moving averages. The moving-average weights capture changes over
time in the composition of crop and livestock items that farmers produce
and sell, and of the goods and services items used in their production, thus
maintaining the weights on a generally up-to-date basis. Prices-received
weights are based on farm cash receipts prepared by the Department of
Agricultures Economic Research Service, and prices-paid weights are based
on the departments annual farm finance survey on costs and returns.
Current-period prices-received data are derived from department sur-

veys of marketings and prices for various crop and livestock products.
The data incorporate changes in the relative proportions of the various
grades or qualities of the products sold. The data are not adjusted for
changes in the quality of farm products.
Current-period prices-paid data are based on surveys of firms that
sell to farmers and firms that purchase the item directly from farmers,
such as feeder pigs, on the producer price indexes and the consumer
price index, and on a quarterly department survey of farm labor wage
rates (interim months are based on U.S. Bureau of Labor Statistics wage
data). The prices-paid data for farm expenditures are not adjusted for
changes in the quality of goods and services bought. The living-expense
component of prices paid is adjusted for changes in quality in the con-
sumer price index.
Accuracy
There are no estimates of sampling or revision error for the farm parity
ratio.
Relevance
The farm parity ratio is a limited indicator of the economic well-being

of farmers. While the ratio contrasts prices of farm products sold with
production and living costs, it does not reflect improvements in farm
production technology or changes in the quantity and quality of farm
products sold. Thus, it is not an indicator of farm income. However, it
portrays whether price movements are more or less favorable to farm-
ers. Used with projections of the production of crop and livestock prod-
ucts, the parity ratio foreshadows the likely direction of changes in
income. In addition, by focusing attention on the price component of
farm income, the farm parity ratio is used as one guide for initiating or
terminating farm marketing quotas for restricting shipments of particu-
lar products to market in order to bolster farm prices.
Recent Trends
From 1995 to 2004, the farm parity ratio generally declined, though it
increased in 2003 and 2004 (Table 17.1). From peaks of 93 and 98 in
FARM PARITY RATIO 91
Table 17.1
Farm Parity Ratio (199092 = 100)
Prices-received
Farm parity ratio index Prices-paid index
1995 93 102 109
1996 98 112 115
1997 90 107 118
1998 89 102 115
1999 83 96 115
2000 80 96 120
2001 83 102 123
2002 79 98 124
2003 84 107 128
2004 90 119 133
Note: The parity ratio does not always equal the division of prices received by prices
paid, due to rounding.
199596, the parity ratio declined to 79 in 2002, and then rose to 90 in

2004. The prices-received index fluctuated from year to year, while the
prices-paid index increased continuously, except for 1998 and 1999.
National Agricultural Statistics Service, U.S. Department of Agriculture. Agricul-

tural Prices. Monthly.
18
GDP Price Measures
There are two measures of price change associated with the gross domestic
product (GDP): the chain-type price index and the implicit price deflator.
The GDP price measures are the most comprehensive indicators of price
change in the U.S. economy. They include the goods and services elements
of consumer, investment, government, and international economic transac-
tions in the GDP. Because of different methodologies used in their calcula-
tion, the two measures sometimes result in slightly different price movements.
The GDP price measures are prepared quarterly by the Bureau of Eco-
nomic Analysis (BEA) in the U.S. Department of Commerce. The data
are published in BEA News, a monthly news release and in the BEA
monthly journal, Survey of Current Business (www.bea.gov).
The data are available during the fourth week of every month. Pre-
liminary data for the immediately preceding quarter are provided in the
month following the quarter (April for the first quarter, July for the sec-
ond quarter, and so on). These are initially revised in the subsequent two
months. More detailed revisions are made annually every August, and
comprehensive benchmark revisions based largely on the quinquennial
economic censuses are published about every five years.
Content
Two gross domestic product price measuresthe chain-type index and

the implicit price deflatorare developed for the entire GDP and for
92
GDP PRICE MEASURES 93
the consumer, investment, government, and international components.

The two measures are based on the same composition of the goods and
services and are virtually identical. The Bureau of Economic Analysis
considers the chain-type index as the featured measure; the implicit price
deflator is also provided because of its widespread use in the past.
The GDP price measures are currently based on 2000 = 100.
The GDP price measures are seasonally adjusted.
Methodology
The GDP chain-type price index weights change annually according to

the composition of expenditures for goods and services items in each
year. Two expenditure weights in adjacent years are used in calculating
price movements. Price movements from one period to the next are based
on the Fisher ideal index number formula.
The chain-type price measure is developed by applying price move-
ments of the various goods and services items contained mainly in the
consumer price index, producer price indexes, and the import and ex-
port price indexes to the expenditure weights noted above. These are
supplemented with other price and cost indexes, including those for con-
struction and defense prices and costs.
The chain-type price index movements are calculated using the geo-
metric mean of two price movements based on alternative goods and
services quantity weights (Fisher ideal index). The quantity weights for
the most recent years and quarters are based on the annual GDP revi-
sions each August. For years and for historical quarters, quantity weights
for two different years are used. Price changes are calculated separately
using the two weighting patterns, and the actual movement is the geo-
metric mean of the two movements. Geometric averaging, in contrast to
arithmetic averaging, treats price increases and decreases symmetrically,
without the distortion of shifting bases.1
The implicit price deflator is calculated by dividing the index of
current-dollar GDP by the chain-type quantity index of GDP, and
multiplying by 100. Since it is a by-product of the estimation of real
GDP, it does not involve the conventional index number construction
of multiplying price movements by the weights and summing the
products, as in the chain-type price index. Nevertheless, if it were
not for certain technicalities, the chain-type index and the implicit
price deflator would be identical. An example of these technicalities
is that the quarterly price and quantity chain-type indexes are ad-
justed so that they average to their corresponding annual values. As a
result, the formulas for quarterly chain-type prices and quantities
are not exactly symmetrical.
Accuracy
There are no estimates of revision error for the GDP price measures.
Relevance
A unique feature of the GDP price measures is their inclusion of all

components of the economy. This enables comprehensive analyses of
the sources of price movements in the consumer, investment, govern-
ment, and international components integrated in a statistically consis-
tent framework.
In analyzing consumer price movements, it is useful to compare
the movements of the chain GDP price for consumer expenditures
with the consumer price index (CPI) and with the chain CPI that was
introduced in 2000. Differences between the chain GDP price index
and the CPI reflect variations in the coverage of certain goods and
services items. The chain GDP price index represents spending of all
consumers while the CPI refers only to spending by urban civilian
households, the expenditure weights of the chain GDP price index
are updated annually while the CPI weights are updated every two
years, and the geometric averaging of the chain GDP price index
contrasts with the arithmetic averaging of the CPI. While the chain
CPI is based on similar geometric averaging as the chain GDP price
index, differences between the chain GDP price index and the chain
CPI still reflect variations in their coverage of certain goods and ser-
vices items, spending of all consumers as distinct from spending by
urban civilian households, and annual versus biennial updating of
the expenditure weights.
Recent Trends
From 1995 to 2004, the GDP price measures ranged from annual in-
creases of 1.1 percent in 1998 to 2.6 percent in 2004 (Table 18.1). The
rate of price increase fluctuated throughout the period. The chain-type
GDP PRICE MEASURES 95
Table 18.1
GDP Price Measures (annual percentage change)
Chain-type price index Implicit price deflator

1995 2.0 2.0
1996 1.9 1.9
1997 1.7 1.7
1998 1.1 1.1
1999 1.4 1.4
2000 2.2 2.2
2001 2.4 2.4
2002 1.7 1.7
2003 2.0 2.0
2004 2.6 2.6
19952004
(annual average) 1.9 1.9
index and the implicit price deflator had identical movements in all years.
Over the entire nine-year period, both the chain-type index and the im-
plicit price deflator increased at an average annual rate of 1.9 percent.

Business. Monthly.
Note
1. For example, a price increase from $4 to $5 in the ratio of 1.25 is 25 percent,

while a price decrease from $5 to $4 in the ratio of .80 is 20 percent. The arithmetic
mean shows the price increase as more important than the price decrease ([25%
20%]/2 = 2.5%), while the geometric mean gives the same weight to both the price
increase and the price decrease (1.25 .80 = 1). Technically, geometric averaging is
the nth root of the product of n numbers, and arithmetic averaging is the sum of n
numbers divided by n.
19
Government Economic
Transactions: Expenditures,
Receipts, Surplus/Deficit
Government economic transactions represent expenditures, receipts, and

the resultant surplus/deficit for all levels of governmentfederal, state,
and local. Government expenditures and receipts are underwritten by
legislation appropriating funds to be spent and by tax laws specifying
what items are to be taxed and the taxation rates. Actual expenditures
and receipts are also influenced by the state of the economy. Expendi-
tures and receipts are in balance when expenditures equal receipts, in
surplus when receipts exceed expenditures, and in deficit when expen-
ditures exceed receipts.
This discussion centers on the government expenditure and receipts
data that are part of the national income and product accounts (see gross
domestic product). This integrates government economic transactions
with the macroeconomic measures of the overall economy. The govern-
ment transactions data differ in content and timing from government
budgets data of the federal, state, and local governments, though the
differences are relatively small.

Commerce prepares quarterly measures of government expenditures,
receipts, and the surplus/deficit of the federal government and of the
total of all state and local governments in the BEA monthly journal,
Survey of Current Business (www.bea.gov).
The BEA data on government economic transactions are available
in the month after the quarter to which the data refer (April for the
96
GOVERNMENT ECONOMIC TRANSACTIONS 97
first quarter, July for the second quarter, and so on). The exception is
data on corporate income taxes and thus the budget surplus/deficit
position, which are prepared and revised in the subsequent months.
These are revised annually every summer, and in the subsequent
benchmark revisions of the national income and product accounts
(see gross domestic product).
Content
Government expenditures consist of consumption expenditures, social

benefits payments, interest payments on the public debt, and subsidies.
Consumption expenditures are composed of compensation of govern-
ment employees, purchases of goods and services from business, depre-
ciation of government structures and equipment, and other smaller items.
Social benefits payments include Social Security, Medicare, Medicaid,
unemployment insurance, veterans benefits, and other income-mainte-
nance payments to individuals. Grants-in-aid from the federal govern-
ment to state and local governments are a federal expenditure, and
grants-in-aid from state governments to local governments are a state
and local expenditure. Government expenditures exclude transactions
in financial assets and land.
Government receipts encompass government revenue from income,
sales, and property taxes; contributions for social insurance; customs
duties; grants-in-aid received by state and local governments, and fees,
licenses, and other miscellaneous sources of revenue.
The main components of expenditures and receipts differ in content
and magnitude between the federal government and state and local gov-
ernments. For example, only the federal government spends for Social
Security and for farm subsidies, and collects customs duties, while the
federal government expenditures for defense far outweigh the state gov-
ernment outlays for the National Guard. By contrast, only the state and
local governments spend for local schools and collect sales and prop-
erty taxes, while state and local government expenditures for police far
outweigh federal outlays for the Park Police, Federal Protective Ser-
vice, Capital Police, and the Secret Service.
The surplus/deficit measure is referred to in the government eco-
nomic transactions data as net government saving. The surplus/defi-
cit designation is used here because of the common understanding of
these terms.
Government debt, which is not included in government economic

transactions, results from borrowing. The debt primarily represents the
cumulative excess of annual deficits over annual surpluses in previous
years, with an addition for any deficits in the current year. Also, some
borrowing occurs in anticipation of spending in future years, particu-
larly for capital construction projects of state and local governments.
Federal government economic transactions are generally on a cash
basis, but with several exceptions. The primary exceptions are: For spend-
ing, payment of interest on the public debt is on an accrual basis, pay-
ment for large defense items such as airplanes and missiles are recorded
on a delivery basis, and expenditures for ships and construction are re-
corded on a work-put-in-place basis, which is similar to an accrual ba-
sis. For receipts, business taxes are recorded on an accrual basis.
State and local government economic transactions are also generally
on a cash basis. The exceptions are for interest payments and business
taxes, which are on an accrual basis.
The government economic transactions data are seasonally adjusted.
Methodology
Federal government economic transactions for expenditures are based

on reports of government agencies to the Office of Management and
Budget. The Treasury Department provides monthly data on federal tax
revenues, which are revised on an annual basis. The Bureau of Eco-
nomic Analysis modifies these for the accrual, delivery-based, and work-
put-in-place elements.
State and local government economic transactions for quarterly data
are based on survey data of the U.S. Bureau of Labor Statistics and the
U.S. Bureau of the Census for payrolls, construction, and taxes, and are
supplemented by the trend of less recent actual data for the other com-
ponents. The Bureau of Labor Statistics conducts a monthly survey of
state and local government worker payrolls. The Census Bureau con-
ducts a monthly survey of state and local governmentowned new con-
struction and a quarterly survey of state and local tax revenues. Annual
state and local government economic transactions data are based on
Census Bureau surveys of all state governments and a sample of local
governments that provide spending and revenue information on a yearly
basis. The Bureau of Economic Analysis estimates the accrual elements
for the government economic transactions data.
Accuracy
There are no estimates of revisions of government economic transac-

tions. Estimates of government current expenditures and gross invest-
ment that are part of the gross domestic product show a revision in the
quarterly real (constant dollars) growth rate from the third month after
the quarter to which they refer to succeeding annual revisions each Au-
gust of an average of plus or minus 1.3 percentage points.
Relevance
Government economic transactions are integrated with the gross do-

mestic product, which facilitates their use in analyzing overall economic
trends. A surplus or deficit in government economic transactions data
affects spending in the entire economy both directly and indirectly. High
economic growth tends to raise tax receipts and lower outlays for unem-
ployment insurance. This results in the direct effect of an increase in the
surplus or a reduction in the deficit, with the effect of removing money
from the income stream, and thus restraining spending, though not nec-
essarily proportionately. In fact, the relationship is not always linear.
For example, during the high economic growth for most of the 1980s,
the deficit also rose for a few years, as the tax reductions of that decade
significantly lessened the tax receipts that were generated by economic
growth. To the extent that low economic growth leads to an increasing
deficit or a reduction in the surplus, the direct effect is to add money to
the income stream and thus stimulate spending, though not necessarily
proportionately.
Government economic transactions also affect spending indirectly
through interest rates. These effects are in opposition to the more di-
rect effects. Because a surplus (or reduction in the deficit) reduces the
debt (or slows the increase in the debt), it tends to lower interest rates
and consequently stimulate private spending. Analogously, a deficit
(or reduction in the surplus) increases debt (or slows the reduction in
debt), resulting in higher interest rates, which tend to restrain spend-
ing. Again, the interaction of the surplus/deficit with interest rates is
not always linear. For example, during periods of uncertainty when
investors park their funds in safe U.S. government securities, the re-
sultant smaller pool of funds for business and household borrowers
leads to higher interest rates for the borrowers because of lower sup-
plies of loanable funds. Thus, borrowers have to pay an interest pre-

mium in order to get lenders to provide loans. The net result of these
contrasting direct and indirect effects determines the overall impact of
government budgets on the economy.
In addition to government economic transactions impacting the
economy, the economy impacts government transactions through feed-
back effects. Because economic growth (measured by the gross domes-
tic product) affects tax revenues and spending for unemployment
insurance, periods of high economic growth generally tend toward a
budget surplus (or reduction in the deficit) and low growth periods tend
toward a deficit (or reduction in the surplus).
The size and components of expenditures and receipts also influence
the economy. The amount of total expenditures, as well its composition
of civilian and defense programs, impacts employment, average weekly
earnings, and industrial production (industrial production index). The
amount of total receipts, as well as the component shares of income,
sales, and property taxes, affects business and employment incentives
and the distribution of income.
Recent Trends
From 1995 to 2004, government economic transactions fluctuated no-

ticeably in relationship to the gross domestic product (Table 19.1). The
federal government had both greater levels and greater fluctuations in
expenditures and receipts than those of state and local governments.
Consequently, the pattern of total government, the sum of the federal
and the state and local governments, reflected the movements of the
federal government more than those of the state and local governments.
As a percentage of the GDP, federal government expenditures de-
clined from 21.7 percent in 1995 to 19.0 percent in 2000, rose to 20.4
percent in 2003, and declined to 20.0 percent in 2004. As a percentage
of the GDP, state and local government expenditures declined from 13.2
percent in 1995 to 12.7 percent in 199798, rose to 13.7 percent in 2002,
and declined to 13.4 percent in 2004.
As a percentage of the GDP, federal government receipts rose from
19.0 percent in 1995 to 20.9 percent in 2000, and then declined to 16.8
percent in 2004. As a percentage of the GDP, state and local government
receipts rose from an average of 13.3 percent during 19952000 to 13.5
13.6 percent during 200104.
Table 19.1
Government Economic Transactions (percentage of the gross domestic product)
Expenditures Receipts Surplus/Deficit()

Calendar State Total State Total State Total
year Federal and local government Federal and local government Federal and local government
1995 21.7 13.2 32.4 19.0 13.4 29.9 2.7 0.2 2.5
1996 21.3 13.0 31.9 19.5 13.3 30.4 1.8 0.3 1.5
1997 20.6 12.7 30.9 19.9 13.2 30.7 0.7 0.5 0.2
1998 19.8 12.7 30.1 20.3 13.3 31.1 0.4 0.6 1.0
1999 19.3 12.8 29.6 20.4 13.3 31.2 1.1 0.5 1.7
2000 19.0 12.9 29.4 20.9 13.4 31.8 1.9 0.5 2.4
2001 19.4 13.5 30.2 19.9 13.6 30.7 0.5 0.0 0.5
2002 20.0 13.7 30.8 17.6 13.5 28.2 2.4 0.2 2.7
2003 20.4 13.6 30.9 17.1 13.6 27.6 3.3 0.0 3.3
2004 20.0 13.4 30.3 16.8 13.5 27.3 3.2 0.1 3.1
For the surplus/deficit measures, as a percentage of the GDP, the

federal government deficit declined from 2.7 percent in 1995 to 0.7
percent in 1997, shifted to a surplus that peaked at 1.9 percent in
2000 that dropped to 0.5 percent in 2001, and reverted to deficits
during 200204 that were 3.3 and 3.2 percent in 200304. As a per-
centage of the GDP, the state and local government surplus was typi-
cally 0.5 percent during 19952000, dropped to zero surplus/deficit
in 2001 and 2003, with a deficit of 0.2 percent in 2002, and was a
surplus of 0.1 percent in 2004.
Reference from Primary Data Sources

Business. Monthly.
GROSS DOMESTIC PRODUCT 103
20
The gross domestic product (GDP) is the broadest indicator of economic

output and growth. It covers the goods and services produced and consumed
in the private, public, domestic, and international sectors of the economy.
Two measures of the GDP are provided, one from the viewpoint of demand
that shows the market for goods and services, and the other from the view-
point of supply, showing the resource costs in producing the goods and
services. The GDP is a summary measure of the national income and prod-
uct accounts, which are also referred to as the national accounts.
In addition, the GDP is presented in two ways with respect to price
levels. One is in current dollars that represent actual prices in every
period, and the other is in chained dollars that abstract from changing
prices over time. The current-dollar GDP is the market value of goods
and services produced, which is the product of quantities and prices.
The chained-dollar GDP, which is referred to as real GDP, represents
the quantity of economic output and is the measure used to define the
rate of economic growth. There are several alternative measures of real
GDP growth based on variations in the component items. Indicators of
price movements are provided for the total GDP and its major compo-
nents (see GDP price measures).
The gross domestic product is prepared quarterly by the Bureau of Eco-

nomic Analysis (BEA) in the U.S. Department of Commerce. The data
are published in a monthly news release and in the BEA monthly jour-
nal, Survey of Current Business (www.bea.gov).
The data are available during the fourth week of every month. Initial
103
(referred to as advance) data are provided in the month after the quar-
ter to which they refer (April for the first quarter, July for the second
quarter, and so on). These are revised in the subsequent two months
(referred to as preliminary and final, respectively), with more de-
tailed revisions made annually every summer, and still more compre-
hensive benchmark revisions made about every five years.
Content
The composition of the two gross domestic product measures is shown

in Table 20.1. The product side reflects demand or markets for goods
and services, and the income side reflects the supply or costs of pro-
ducing the goods and services. The two measures are conceptually equal,
but they differ statistically because they are estimated independently
and because of inadequacies in the data; this difference is called the
statistical discrepancy. As noted below under Accuracy, each mea-
sure is equally valid for calculating economic growth rates.
The GDP is measured on a value-added basis. Only the value that is
added in each stage of production, from raw materials to semifinished
goods to final products, is counted. This prevents endless double-count-
ing that would occur if goods and services purchased from other busi-
ness for use in production were included.
The GDP data are seasonally adjusted.
Product Side: Demand Components
The component markets for the nations output represent the demand
aspects of the economy and are referred to as the product-side of GDP.
The product-side total is the official GDP measure. It has the following
main components:
Personal consumption expenditures represent spending by households

for durable goods, nondurable goods, services, and the operating ex-
penses of nonprofit organizations.
Gross private domestic investment represents business spending for equip-
ment, nonresidential structures, and software by for-profit and nonprofit
organizations; residential construction; and the change in business invento-
ries, excluding profit or loss due to cost changes between the time of pur-
chase and sale of inventoried goods (inventory valuation adjustment).
Table 20.1
Gross Domestic Product and Main Components: 2004
Product side Income side

$ Billions Percentage $ Billions Percentage
Gross domestic Gross
product 11,735.0 100.0 domestic
product 11,735.0 100.0
Personal Compensation
consumption of employees 6,632.0 56.5
expenditures 8,229.9 70.1 Wages and
Durable goods 993.9 8.5 salaries 5,355.7 45.6
Nondurable Supplements
goods 2,377.0 20.3 to wages
Services 4,859.0 41.4 and salaries 1,276.3 10.9
Gross private Proprietors

domestic incomed 902.8 7.7
investment 1,927.3 16.4
Nonresidentiala 1,220.5 10.4 Corporate
Residentialb 663.4 5.7 profitse 1,181.6 10.1
Inventory
changec 43.4 0.4 Rental income
of personsf 165.1 1.4
Net exports of Net interest 549.5 4.7

goods and
services 606.2 5.2 Taxes on
Exports 1,175.5 10.0 production
Imports 1,781.6 15.2 and importsg 841.1 7.2
Government Consumption
consumption of fixed capitalh 1,407.3 12.0
expenditures
and gross Business
investment 2,183.9 18.6 transfer 82.1 0.7
Federal 809.9 6.9 payments
State and
local 1,373.9 11.7
(continued)
Table 20.1 (continued)

Product side Income side
$ Billions Percentage $ Billions Percentage
Less:
Government
subsidies
less surplus of
government
enterprises 33.5 0.3
Rest of the worldi 43.9 0.4
Statistical
discrepancyj 50.9 0.4
Source: Bureau of Economic Analysis, U.S. Department of Commerce, Survey of

Current Business, March 2005.
Notes: Components do not sum to totals, due to rounding. a. Business purchases of
structures, equipment, and software. b. New housing construction and improvements. c.
With inventory valuation adjustment. d. Profits of unincorporated businesses with inven-
tory valuation and capital consumption adjustments. e. With inventory valuation and capi-
tal consumption adjustments. f. With capital consumption adjustment. g. Mainly sales and
property taxes. h. Mainly depreciation allowances with capital consumption adjustment
(CCA sign reversed with those of d, e, f above). i. Adjustment to place sum of incomeside
components on a domestic income basis, rather than a national income basis. j. Product
side total less sum of incomeside components shown above. Balancing item between
productside and incomeside totals.total less sum of incomeside components shown
above. Balancing item between productside and incomeside totals.
Government consumption expenditures and gross investment repre-

sent the federal, state, and local wages of government workers; pur-
chases of civilian and defense goods, services, and structures, and the
value of current services derived from government-owned equipment
and structures (depreciation). Transfer payments for Social Security and
other income-maintenance payments, federal grants to state governments
and state grants to local governments, interest on the public debt, and
subsidy payments to business are excluded. These latter items are part
of total government spending (see government economic transactions).
Net exports of goods and services represent the international balance
of exports minus imports in goods and services.
Income Side: Supply Components
The labor, capital, and tax costs in producing the nations output are
reflected in the supply aspects of the economy and are referred to as the
income side of GDP. This side has the following main components:
Compensation of employees represents the money wages and salaries

and noncash fringe benefits of workers that accrue at the time they are
earned (supplements to wages and salaries).
Proprietors income and corporate profits represent business profits
of unincorporated businesses and corporations, excluding the profit or
loss due to cost changes between the time of acquisition and time of
sales of inventories (inventory valuation adjustment) or to cost changes
in replacing existing capital facilities since their acquisition (capital con-
sumption adjustment).
Rental income of persons represents the net income (income less ex-
penses) to owners of residential and nonresidential real property who
are not primarily engaged in the real estate business. Rental income
includes nonmarket imputations of income for owner-occupied housing
as if such housing were rented at the market price, plus royalties paid by
businesses to individuals.
Interest is interest paid by the business sector to households and gov-
ernments, minus interest received by businesses from these nonbusiness
sectors. Interest paid and received between businesses cancels out. In-
terest includes nonmarket imputations for banking services provided to
household and government customers without charge.
Consumption of fixed capital represents the charge for the using up
(depreciation) of privately owned and government-owned equipment and
structures, including owner-occupied housing. It includes adjustments
for the changing costs of replacing existing capital facilities from their
original purchase cost (capital consumption adjustment).
Taxes on production and imports represent sales and property taxes,
customs duties, user fees, fines, rents and royalties paid to governments,
and other miscellaneous payments to governments.
Real GDP
The GDP in chained dollars, referred to as real GDP, represents the

quantity of goods and services. Real GDP is the preeminent measure of
the nations economic growth. Real GDP eliminates the effect of price
increases or decreases from one period to the next. By contrast, the GDP
in current dollars, referred to as nominal GDP, represents the value of
goods and services produced in the prices of each period, and thus in-
cludes price and quantity.
The quantity index for real GDP growth is based on a chain-type
index. The chain-type index utilizes priceweights, which change an-

nually according to the prices of goods and services items in each year.
Quantity movements from one period to the next are based on the Fisher
ideal index number formula described below under Methodology. This
is the obverse of the chain-type price index, which employs quantity
weights for its calculation (see GDP price indexes).
Other Summary GDP Measures
In addition to GDP, several variants of total GDP are provided to assist

in economic analysis. These are based on adjustments related to inven-
tories, international transactions, and statistical problems. The most
widely cited one is final sales, which excludes the effect of inventory
increases or decreases. This results in highlighting underlying demand
as represented by purchases of goods and services by households, busi-
ness, government, and foreigners, independent of whether the highly
volatile business inventories in stores, warehouses, and factories are ac-
cumulating or depleting.
These summary measures of the nations output that are closely re-
lated to the GDP are gross domestic income, final sales, final sales to
domestic purchasers, gross domestic purchases, gross national product
(GNP), and command-basis GNP. Typically, the largest difference in the
quarterly and annual movements between GDP and the other summary
measures occurs with final sales.
Methodology
The gross domestic product is calculated using secondary data that

are initially compiled for other purposes, which limits control of the
quality of the data for GDP requirements. Because of this depen-
dence on secondary data, the BEA focuses sharply on unusual move-
ments in this database and raises questions with the organizations
providing the data to determine whether errors or special circum-
stances affect the figures. This close attention to the data base is
done when the GDP components are first estimated and is repeated a
second time when the total product and income sides are compared,
particularly if there is a large difference (statistical discrepancy) be-
tween the two GDP aggregates.
This process sometimes uncovers data problems that in turn lead to
modifying the initial estimates. In addition, special formulas are used to

adjust inventories and depreciation to exclude profit or loss due to chang-
ing costs of inventories or of capital facilities between the time of pur-
chase and their sale or replacement. These are referred to as the inventory
valuation adjustment and capital consumption adjustment, which are
included in the product side under private investment and government
investment, and in the income side under proprietors income, corporate
profits, rental income of persons, and consumption of fixed capital.
The data used in constructing the GDP come from many government
agencies and private organizations that provide statistics obtained from
surveys, income tax returns, and regulatory reports. The items in this da-
tabase vary considerably in definition, collection technique, and timeli-
ness, and thus are of uneven quality. Because the two measures of GDP on
the product and income sides are developed independently from different
data sources, the statistical discrepancy between the two GDP totals
indicates the extent of the inconsistency in the two databases. Formally,
the statistical discrepancy is the product side minus the income side.
Real GDP is prepared by dividing the current-dollar data for the vari-
ous goods and services items primarily by the price levels of the con-
sumer price index, producer price indexes, and the import and export
price indexes. These are supplemented with other indicators of price
change such as those for construction and defense prices and costs.
The movements in the chain-type quantity index for real GDP (which
is an example of a Fisher ideal index) are calculated using the geomet-
ric mean of two more basic measures of these movements. For annual
estimates and for recent quarters, these more basic measures are a
Laspeyres index, which measures quantity movements in terms of the
weights of previous-period prices, and a Paasche index, which mea-
sures quantity movements in terms of the weights of current-period
prices. In each of these instances, quantities in the current and preced-
ing periods are multiplied by prices for a single period (the current or
preceding one).
Quarterly quantity indexes for historical periods use a slightly differ-
ent formula, in that the two sets of price weights are for years instead of
quarters. The first two quarters of the year use prices for the current and
preceding year, while the last two quarters use prices for the current and
following year. In the most recent quarters, prices in the following year
are not yet known, so price weights are for the current and preceding
quarter.
Accuracy
There are no estimates of sampling error for the gross domestic product
data. The statistical discrepancy indicates the extent to which unknown
errors in the databases, in which some are above and others below the
correct values, do not cancel each other. However, because some of
these errors are offsetting, the statistical discrepancy is a net figure of
the consistency of the databases rather than a gross measure of all errors
regardless of whether the high and low figures are offsetting. As noted
earlier, the product side GDP is regarded as the official measure, al-
though the availability of the product and income sides allows the calcu-
lation of alternative growth rates that provide a lower and upper range
for use in analysis (by comparing movements of the product side against
the product side minus the statistical discrepancy). Thus, one way of
viewing the accuracy of the GDP data is to treat it as being within the
growth rate range indicated by the product and income sides movements.
Another perspective of GDP accuracy is provided by considering the
size of the revisions to the provisional GDP data.1 These are shown in terms
of the confidence that the percentage growth rates in GDP are likely to be
revised upward or downward within a specified range based on past experi-
ence. For example, the growth rate of the estimate of real GDP that is pub-
lished in the third month after the quarter to which it refers is revised in the
succeeding annual revisions each August as follows: in two of three cases in
a range of 1.2 to 2.0 percentage points and in nine of ten cases in a range of
2.6 to 3.1 percentage points. Thus, if the growth rate reported in the third
month after the quarter is 3.0 percent, there is a two-thirds probability that
after the annual revisions, the figure will be in the range of 1.8 to 5.0 per-
cent, and a 90 percent chance that it will be in the range of 0.4 to 6.1 percent.
Relevance
The gross domestic product provides the overall framework for analyz-
ing and forecasting economic trends. It has the unique attribute of inte-
grating the markets for goods and services (demand or spending) with
the production of the goods and services (supply or costs) in one format.
Because the costs of production also generate wage and profit incomes,
the GDP measures are the basis for analyzing the feedback effects be-
tween spending and incomes from one period to the next.
The analyses used to assist the president and Congress in formulating
Table 20.2
Gross Domestic Product (billions of dollars)
Real GDP (billions of Annual percentage

Current dollars chained 2000 dollars) change in real GDP
1995 7,397.7 8,031.7 2.5
1996 7,816.9 8,328.9 3.7
1997 8,304.3 8,703.5 4.5
1998 8,747.0 9,066.9 4.2
1999 9,268.4 9,470.3 4.5
2000 9,817.0 9,817.0 3.7
2001 10,128.0 9,890.7 0.8
2002 10,469.6 10,048.8 1.6
2003 11,971.2 10,320.6 2.7
2004 11,734.3 10,755.7 4.2
19952000 4.1
(annual average)
20002004 2.3
(annual average)
19952004 3.3
(annual average)
fiscal policies and the Federal Reserve in formulating monetary policies

focus on real GDP growth. These economic policies are aimed at maximiz-
ing employment growth and minimizing unemployment and inflation (con-
sumer price index, producer price indexes, and GDP price measures). Fiscal
policies refer to federal spending and taxes (government economic transac-
tions), and monetary policies refer to interest rates. Analyses of the cyclical
expansion and recession movements and of the longer-term periods that
span several business cycles are the backdrop for deriving implications of
adopting particular fiscal and monetary policies for moderating cyclical fluc-
tuations and stimulating noninflationary long-term economic growth. The
GDP is the overall framework of such analyses, although it is supplemented
importantly with assessments from other indicators.
Recent Trends
From 1995 to 2004, the real gross domestic products annual growth
rate accelerated from 2.5 percent in 1995 to a range of 3.7 to 4.5 percent
during 19962000 (Table 20.2). Growth rates during the recession of
2001 and the recovery years of 2002 and 2003 were 1 to 3 percentage
points below those of the 19952000 period, although the growth rate
accelerated from the 2001 recession low of 0.8 percent, rising to 4.2
percent in 2004. Over the entire nine-year period, real GDP increased at
an average annual rate of 3.3 percent. Within this period, real GDP in-
creased at an average annual rate of 4.1 percent during 19952000, and
2.3 percent during 200004.

Business. Monthly.
Note
1. For a discussion of the magnitude of the revisions, see Dennis J. Fixler and
Bruce T. Grimm, Reliability of the NIPA Estimates of U.S. Economic Activity,
Survey of Current Business, February 2005.
HELP-WANTED ADVERTISING INDEX 113
21
Help-Wanted Advertising Index
The help-wanted advertising index tracks employers advertisements for

job openings in the classified section of newspapers in fifty-one labor
market areas. The index represents job vacancies resulting from turn-
over in existing positions due to such reasons as workers changing jobs
or retiring, and from the creation of new jobs. It excludes unadvertised
job vacancies and jobs advertised in nonclassified sections of newspa-
pers such as display ads in business or news sections.
Measures of the help-wanted advertising index are provided monthly by

The Conference Board. The data are published in a news release and in
The Conference Boards monthly report, Business Cycle Indicators
(www.conference-board.org).
The data are available twenty-five to thirty days after the month to
which they refer. The data are revised in the following month.
Content
The help-wanted advertising index covers jobs in many fieldsprofes-

sional, technical, crafts, office, sales, farm, custodial, and so on. They
include a higher proportion of all junior and middle-level vacancies than
of managerial, executive, or unskilled levels. In addition to the national
help-wanted index, local indexes for fifty-one labor markers are pro-
vided. The help-wanted index does not distinguish job vacancies by oc-
cupational skills.
113
The help-wanted advertising index is currently based on 1987 = 100.

The help-wanted data are seasonally adjusted.
Methodology
The help-wanted advertising data are obtained from classified adver-

tisements in one daily (including Sunday) newspaper in each of the fifty-
one labor markets (fifty-one cities including their suburbs). Newspapers
are selected according to how well their ads represent total jobs in the
local labor market area. The fifty-one labor markets accounted for ap-
proximately 45 percent of nonagricultural employment in 2002.
The index reflects the number of job advertisements. Each advertise-
ment is weighted equally regardless of whether it is an ad for one job or
for multiple positions or whether for full-time or part-time work. Adver-
tisements of both employers and employment agencies and advertise-
ments for the same job on successive days are included in the count.
Index weights for the fifty-one labor markets are based on the pro-
portion of nonagricultural employment accounted for by each of the
labor markets. The index is based on 1987 weights. Within each market
area, help-wanted advertisements in the Sunday newspaper are weighted
according to the ratio of the average Sunday advertising volume to aver-
age daily advertising volume.
The help-wanted advertising index does not include employer online
advertising for jobs, although future research may be done on the feasi-
bility of incorporating online advertising into the index. Beginning in
2005, The Conference Board instituted a monthly data series on new
online job advertisements, The Conference Board Help-Wanted Online
Data Series. This new online series is considered developmental.
Accuracy
There are no estimates of sampling or revision error for the help-wanted

advertising index.
Relevance
The help-wanted advertising index indicates the direction of employers

hiring plans. In theory, it provides an advance signal of future changes in
employment and cyclical turning points. In practice, the help-wanted in-
HELP-WANTED ADVERTISING INDEX 115
dex leads the downturn from the expansion peak to a recession, but it lags
the turning point in moving from recession to expansion, based on analy-
ses conducted as part of the leading, coincident, and lagging indexes. The
lag in the recovery from a recession results from the tendency of employ-
ers to increase average weekly hours of existing workers when business
improves or to call back workers on layoff before advertising for new
workers. In the past two recoveries of 199192 and 2002, the lag also
reflects the much smaller increase in employment than in previous recov-
eries.1 Also, during the past two recoveries of 199192 and 200203, job
losses were far more structural than cyclical as compared to previous re-
coveries, resulting in proportionately more permanent job losses that make
it more difficult to find new employment than cyclical job losses, in which
at least some rehiring occurs along with a cyclical upturn.2
The help-wanted index tends to be inversely related to unemployment.
When help-wanted advertisements increase, unemployment usually de-
clines, while when help-wanted advertisements decrease, unemployment
usually rises. But the help-wanted movements typically do not parallel
unemployment movements because of changing advertising practices. For
example, during periods of low unemployment, employers may rely more
heavily on help-wanted advertisements than on alternative means of find-
ing workers. During periods of high unemployment, employers may find
workers more easily through alternative means such as through workers
initiating the contact on their own or on the advice of a friend.
Some advertised jobs may not be filled because employers are not
satisfied with the applicants, there is an overall shortage of applicants,
or employers decide not to fill the jobs.
The help-wanted index does not measure overall job vacancies, nor
does it provide data on occupational skills. Therefore, the index cannot
be compared with unemployment data to assess job shortages and sur-
pluses in labor markets.
The Job Openings and Labor Turnover survey provides a more com-
prehensive measure of employer recruitment for specific job openings
in the overall economy than that of the help-wanted advertising index.
Recent Trends
From 1995 to 2004, the help-wanted advertising index fluctuated in a

narrow range of 83 to 89 during 19952000, and then dropped sharply
to 38 in 2003 and 2004 (Table 21.1).
Table 21.1
Help-Wanted Advertising Index (1987 = 100)
Year Index count

1995 85
1996 83
1997 87
1998 89
1999 87
2000 83
2001 58
2002 44
2003 38
2004 38
The Conference Board. Business Cycle Indicators. New York, NY. Monthly.
Notes
1. Stacey L. Schreft and Aarti Singh, A Closer Look at Jobless Recoveries,

Economic Review, Federal Reserve Bank of Kansas City, second quarter 2003.
2. Erica L. Groshen and Simon Potter, Has Structural Change Contributed to a
Jobless Recovery? Current Issues in Economics and Finance, Federal Reserve Bank
of New York, August 2003.
HOME SALES: NEW AND EXISTING HOUSES 117
22
Home Sales: New and
Existing Houses
Sales of new and existing privately owned homes represent the number
of single-family unattached houses and townhouses for new home sales;
and the number of single-family unattached houses and townhouses,
plus the number of condominium and cooperative apartments in multi-
family buildings for existing home sales. Each house or apartment is
counted as one unit regardless of the sale price. New homes are newly
constructed houses that are sold by the developer to the first owner. Ex-
isting homes are houses that are at least one year old.
(Data on the inventories of new and existing homes for sale are pro-
vided by the same organizations that prepare the sales data. I note these
data on unsold homes to alert the reader to their availability, though they
are not covered here.)
New Homes
Measures of new home sales are prepared monthly by the Bureau of the
Census in the U.S. Department of Commerce and the U.S. Department
of Housing and Urban Development. They are published in a news re-
lease, New Residential Sales (www.census.gov).
The data are available approximately one month after the month to
which they refer. They are revised in the three succeeding months.
117
Content
A new home sale is recorded when a sales contract is signed or a buyers

deposit is accepted. Although some contracts or deposits are conditional
and subsequently canceled, the data are not revised to show this change
The new-home-sales data are seasonally adjusted.
Methodology
Data on new home sales are obtained from a monthly survey of a sample
of builder or owner developers conducted by the Bureau of the Census.
Imputations are made to the new-home-sales data to allocate late re-
ports of home sales data to the month when the sale occurred, and to
account for those sales taking place prior to the issuance of a building
permit, where one is required.
Accuracy
The sampling error (for one standard error) for new home sales data is 6
percent. For example, if the monthly new-home-sales were estimated at
one million units, in two of three cases the true sales would be some-
where between 940,000 and 1,060,000 units. For further information on
the interpretation of sampling and nonsampling errors, see the Appendix.
Existing Homes
Existing home sales are prepared monthly by the National Association

of Realtors (NAR). They are published in a news release and in the
monthly NAR newsletter, Real Estate Outlook: Market Trends and In-
sights (www.realtor.org).
The data are available twenty-five days after the month to which they
refer. They are revised one month later and annually every February.
Content
Existing homes are houses that are at least one year old. An exception is
a newly constructed house that is first sold more than one year after
being put on the market, in which case it is defined as a new house.

Sales of existing homes include transactions conducted through a real
estate broker and those made directly by the owner.
The existing home sales data are seasonally adjusted.
Methodology
Data on existing home sales are obtained from monthly surveys of local
areas based on a sample of over 600 boards/associations of realtors and
multiple-listing systems conducted by the National Association of Real-
tors (NAR). The NAR sample and its multiple-listing systems account for
approximately 35 percent of all existing home sales. Through the meth-
odology, the data capture estimates of sales made directly by the owner.
The local area data are primarily for metropolitan areas, but they are
considered to be representative of nonmetropolitan counties surrounding
the metropolitan areas, based on biennial information in the American
Housing Survey of the U.S. Department of Housing and Urban Develop-
ment. The local area data are summed to the four broad Census Bureau
regions of the United States, the Northeast, South, Midwest, and West.
The regional data are in turn augmented by an inflation factor to account
for total sales of each region, including those not captured in the sample
of realtor and multiple-listing reporting systems. The inflation factor, which
is revised approximately every five years, is based on sales data in the
American Housing Survey and the decennial Census of Housing. The sales
data for the four regions are then summed to the national total.
Accuracy
There are no estimates of sampling or revision error for the existing

home sales data.
New and Existing Homes
Relevance
Close to 70 percent of all households own single-family homes. A home

is typically the largest single item bought by householders (outlays of
comparable magnitude are associated with financing college educations
and major medical bills). Economic output increases far more by the
Table 22.1
Home Sales: New and Existing Houses (thousands of units)
Ratio, existing homes

New homesa Existing homesb to new homes
1995 667 3,852 5.8
1996 757 4,167 5.5
1997 804 4,371 5.4
1998 886 4,966 5.6
1999 880 5,190 5.9
2000 877 5,171 5.9
2001 908 5,332 5.9
2002 973 5,631 5.8
2003 1,086 6,183 5.7
2004 1,200 6,784 5.7
a
Single-family unattached houses and townhouses; bSingle-family unattached houses
and townhouses, and condominium and cooperative apartments in multifamily buildings.
purchase of a new house than of an existing house because of the mate-

rials and construction work required in building a new house, although
renovation work is sometimes done when an existing house is purchased.
While existing home sales have a much smaller direct impact on the
economy than new home sales, existing and new home sales are in fact
closely linked because existing home owners often can afford to buy a
new home only by selling their existing home. Also, both new and exist-
ing home sales generate purchases of furniture, appliances, and other
house furnishings, which is a secondary stimulus to the economy.
Home sales are sensitive to changes in economic conditions related
to employment, personal income and saving, interest rates, housing starts,
housing affordability index, and mortgage delinquency and foreclosure.
Although housing is a necessity of living, home sales are highly cyclical
because households are most likely to purchase a home during prosper-
ous times when they can best afford it, but they tend to defer a home
purchase during depressed times when they can least afford it.
Recent Trends
From 1995 to 2004, there were far more sales of existing homes than of
new homes (Table 22.1). Sales of existing homes were five to six times
as high as sales of new homes.
New home sales rose from 667,000 in 1995 to 1.2 million in 2004.
These sales increased continuously over the nine-year period, except for
small declines in 1999 and 2000.
Sales of existing homes rose from 3.9 million in 1995 to 6.8 million
in 2004. These sales increased continuously over the nine-year period,
except for a small decline in 2000.
Bureau of the Census, U.S. Department of Commerce, and U.S. Department of

Housing and Urban Development. New Residential Sales. Monthly.
National Association of Realtors. Real Estate Outlook: Market Trends and Insights.
Washington, DC. Monthly.
23
House Prices: New and
Existing Houses
There are two measures of house prices for privately owned single-fam-
ily houses. One represents sales prices for newly constructed houses
and adjusts for the price effect of changes in the size and various charac-
teristics of new houses. However, this measure does not adjust for the
price effect of building new houses in less developed locations that low-
ers the price of the purchased lot. The other price measure represents the
resale price and the house revaluations required to obtain mortgage refi-
nancing of existing houses. This measure adjusts for the price effect of
resales and mortgage refinancing related to changes in the characteris-
tics of the same neighborhoods. However, the measure does not adjust
for the price effect of structural and landscaping improvements made to
the same houses.
The price measures for new and existing houses are described sepa-
rately. This is followed by a comparison of the relevance and recent
trends of both price measures.
New Houses
The new house price index is prepared quarterly by the Bureau of the
Census in the U.S. Department of Commerce. It is published in the
news release, Price Index of New One-Family Houses Sold (www.
census.gov).
The data are available four weeks after the quarter to which they re-
fer. They are revised in the succeeding quarter.
122
HOUSE PRICES: NEW AND EXISTING HOUSES 123
Content
The new house price index is based on the actual sales price of single-
family houses (the price index excludes houses built for the exclusive use
of the land owner). The sales price covers the value of the house structure,
the developed lot, selling expenses, and the sellers profit. The index cov-
ers detached single-family houses and attached townhouses, including
condominiums and cooperatives, whether financed with conventional,
Federal Housing Administration, or Veterans Administration mortgages.
The sale is recorded when a sales contract is signed or a deposit is ac-
cepted, regardless of the stage of construction. If that sale falls through and
the house is sold to a different buyer, the index still records the sales data of
the initial transaction and thus does not reflect the price or timing of the sale
to a different buyer. Analogously, subsequent price changes due to changes
in amenities after the initial sale is recorded are not included in the index.
The new house price index represents a constant set of structural char-
acteristics and amenities within different regions of the United States.
Examples of the characteristics are floor space, number of bedrooms,
garage or carport, kitchen appliances, heating system, exterior wall ma-
terial, and number of fireplaces. This maintenance of a constant quality
house means that index changes over time reflect price changes only.
The physical characteristics that are used to develop the constant qual-
ity price index are updated approximately every ten to fifteen years.
However, no direct adjustment is made in the index for the price effect
of new houses being constructed in less developed geographic areas. Less
developed areas have lower land costs than do more developed areas (the
Henry George effect), which lowers the price of the purchased lot. An
indirect adjustment for the lower cost of lots in the new house price index
may be reflected in the inclusion of four separate regional indexes (North-
east, Midwest, South, West) and in a metropolitan area variable in the
statistical regression used to calculate the index (see Methodology be-
low). These indirect measures distinguish between regions, such as the
much greater construction of new housing in the South than in the North-
east, highlighting the greater availability of undeveloped land and lower
land costs in the South. The indirect measure also distinguishes between
metropolitan areas and the entire region, with the region having lower
land costs because it includes the rural component. But these indirect
measures do not adjust for differences in land costs within regions and
within the metropolitan area of a region.
The new house price index is prepared for the United States and four
broad regions of the country.
The new house price index is currently based on 1996 = 100.
The new house price index is not seasonally adjusted.
Methodology
The new house price index data on the physical characteristics of the
house (floor space, number of bedrooms, etc.) are obtained from monthly
interviews with homebuilders or owners of a sample of houses that were
sold. Approximately 13,000 interviews are conducted each year. The
new house price index is referred to as a hedonic price index (for a brief
description of hedonic price indexes, see the Methodology section of
the consumer price index).
A statistical regression model is used to relate the characteristics of
each house with its sale price. The model is developed for detached
houses in separate strata for four regions of the country (Northeast, Mid-
west, South, West) and in a fifth stratum for all attached townhouses in
the country. The U.S. index is derived by combining the four regional
indexes and the attached townhouse index based on the following per-
centage weights that reflect the relative importance of new house con-
struction among the five categories:
Northeast 6.2
Midwest 15.9
South 40.3
West 27.1
Attached townhouses 10.5
Total 100.0
The data on house sales in the new house price index are obtained from
the Census Bureaus Survey of Construction. The sales data are based on a
monthly survey of a sample of homebuilders. The quarterly sales data used
in the index are the three-month total in each calendar quarter (e.g., first-
quarter sales are the sum of January, February, and March sales).
Accuracy
The statistical regression model error in the new house price index in
two of three cases is plus or minus 0.5 percent.
Existing Houses
The price index for existing houses is prepared quarterly by the Office
of Federal Housing Enterprise Oversight, an independent agency under
the U.S. Department of Housing and Urban Development. The data on
the existing house price index are published in a news release, House
Price Index (www.ofheo.gov).
The data are available two months after the quarter to which they
refer. All data are revised each quarter as revised or more complete in-
formation is received.
Content
The existing house price index is based on repeat sales or house revalu-
ations required to obtain mortgage refinancing of existing mortgages.
The index covers the same houses that have been purchased or securitized
through secondary mortgage transactions by the Federal National Mort-
gage Corporation (Fannie Mae) or the Federal Home Loan Mortgage
Corporation (Freddie Mac). The index is confined to house sales of single-
family detached houses and to financing by conventional mortgages.
Thus, the index excludes townhouses and houses financed with Federal
Housing Administration or Veterans Administration mortgages.
The existing house price index adjusts for the price effect of chang-
ing demographic, socioeconomic, or land use characteristics in the same
neighborhoods over time. However, the index does not adjust for im-
provements to the house or landscaping over time, for depreciation of
the house, or for maintenance of the house.
The existing house price index is prepared for the United States, re-
gions, states, and metropolitan areas.
The existing house price index is currently based on 1980 (first quar-
ter) = 100.
The existing house price index is seasonally adjusted.
Methodology
The existing house price index incorporates repeat mortgage refinancings

of repeat sales or refinancing of the same house that may occur years
apart, in the quarter that the transaction occurs. The data are based on a
sample of secondary mortgage transactions that are purchased or
securitized by Fannie Mae or Freddie Mac. The repeat transactions data
for the same house are identified by matching street addresses consis-
tent with U.S. Postal Service standards.
Greater weight is given in the index to sales price changes of repeat
sales or refinancing of houses that occur in short rather than long time
periods between transactions because (a) differential depreciation rates
for similar properties are more likely over longer periods of time, and
(b) local area real estate values are affected by changing demographic
and socioeconomic characteristics in individual neighborhoods over time.
The existing house price index is developed using a statistical regres-
sion model.
The existing house price index is limited to mortgage loans that do
not exceed a conforming loan limit. Purchases of mortgages by Fannie
Mae or Freddie Mac may not exceed the conforming loan limit. The
conforming loan limit is based on the amount of the mortgage loan, not
on the value of the house. In 2004, the conforming loan limit was
$333,700, which thus included houses of higher value.
Accuracy
The statistical regression model error in the existing house price index
in two of three cases is less than plus or minus one percent.
New and Existing Houses
Relevance
House prices are one factor affecting the demand for housing. Other
factors include mortgage interest rates (interest rates), unemploy-
ment, disposable personal income (personal income and saving),
household debt (consumer credit), and housing affordability (hous-
ing affordability index).
Recent Trends
From 1995 to 2004, the new house price index increased at a lesser rate
than that of the existing house price index, except for 1996, 1997, and
Table 23.1
House Prices: New and Existing Houses (year-to-year percentage change)
New houses Existing houses

(annual average) (fourth quarter to fourth quarter
1995 2.7 4.5
1996 4.9 2.6
1997 5.9 4.6
1998 3.2 5.0
1999 7.5 5.2
2000 5.8 7.6
2001 3.0 7.6
2002 7.3 7.6
2003 7.7 8.2
2004 10.6 11.2
1999, when the new house price index increased at a greater rate than
the existing house price index (Table 23.1). The directional year-to-year
movements of the two indexes also occasionally varied. In 1998 and
2000, the rate of increase in the new house price index declined, while
that of the existing house price index rose, and in 1996, the rate of in-
crease in the new house price index rose, while that in the existing house
price index declined. Another variation occurred in 2001 and 2002, when
the rate of increase in the new house price index first declined and then
rose, while the increase in the existing house price index was stable in
both years.
Bureau of the Census, U.S. Department of Commerce. Price Index of New One-
Family Houses Sold. Quarterly.
Office of Federal Housing Enterprise Oversight, U.S. Department of Housing and
Urban Development. House Price Index. Quarterly.
24
Housing Affordability Index
The housing affordability index (HAI) measures the extent to which fami-
lies with the median income can afford an existing single-family, median-
price house. The index gauges whether (a) the required monthly mortgage
payments are below the threshold income/price relationship (house is af-
fordable), or (b) the required mortgage payments are above the income/
price relationship (house is not affordable). The index includes the effects
of family incomes, house prices, thirty-year fixed-rate mortgages, adjust-
able-rate mortgages, a down payment of 20 percent, and a qualifying in-
come standard for prospective buyers to obtain a mortgage.
The housing affordability index is prepared monthly by the National

Association of Realtors. The data are published in the Real Estate Out-
look: Market Trends and Insights (www.realtor.org).
The index is published the last week of the month following the month
to which the data refer, and it is revised every month. Annual revisions
are made every March.
Content
The housing affordability index indicates a threshold for determining the

extent to which monthly mortgage payments for an existing single-family
median-price house can be met by the median-income family. The thresh-
old number is 100. When the index is above 100, the median-price house
is affordable for a median-income family; and when the index is below
128
HOUSING AFFORDABILITY INDEX 129
100, the house is not affordable. For example, an index of 100 means the
family has the exact amount of income required to finance the house; an
index of 115 means the family income is 15 percent higher than necessary
to finance the house; and an index of 90 means the income is 10 percent
lower than necessary to finance the house. In addition to the composite
index, supplementary indexes are provided separately for fixed-rate mort-
gages, adjustable-rate mortgages, and first-time buyers.
The HAI is derived from the interrelationships of the median-price
house, a down payment of 20 percent, fixed and adjustable mortgage
interest rates, median family income, and a qualifying income standard
for prospective buyers to obtain a mortgage.
The median-price house means that the prices of 50 percent of exist-
ing single-family houses sold are above the median threshold price and
50 percent of the homes sold are below the threshold price. The median-
price house is determined by the sales prices of houses of all sizes (e.g.,
houses with varying amounts of square footage) and of all amenities
(e.g., houses with and without central air conditioning).
Family income is defined as gross income before the payment of in-
come taxes. The median family income means that 50 percent of the
families have incomes above the threshold and 50 percent have incomes
below the threshold. The median-income family is based on families of
all sizes. A family refers to two or more individuals related by birth,
marriage, or adoption and living together in a house, apartment, or rooms
intended for separate living quarters.
Mortgage interest rates cover conventional loans for thirty-year fixed-
rate mortgages and conventional loans for adjustable-rate mortgages.
The qualifying income standard for a prospective house buyer to ob-
tain a mortgage is based on the percentage of gross monthly family in-
come that monthly mortgage principal and interest expenses (assuming
a down payment of 20 percent) would be. Under the standard, in order
for a family to obtain a mortgage, the monthly housing expense may not
exceed 25 percent of the monthly gross income. Housing expense ex-
cludes real estate taxes, homeowner insurance, and housing maintenance.
The HAI is not seasonally adjusted.
Methodology
The housing affordability index is calculated as the percentage that me-

dian gross monthly income is of the qualifying income standard for ob-
taining a mortgage. Mathematically, the formula is:
Housing
Median family monthly gross income
affordability = _________________________________ 100 (24.1)
Qualifying monthly income standard
index
The median family income data are based on annual income information
obtained in the Current Population Survey (CPS) of the U.S. Bureau of
the Census (see distribution of income for the CPS methodology). The
annual income data from the previous one to two years are updated to the
current month by using statistical regressions to project historical trends.
The median house price data are based on a monthly survey of realtors
and multiple listing systems of housing sales by the National Associa-
tion of Realtors. See home sales: new and existing houses for the meth-
odology of the NAR survey.
The mortgage interest data are based on the effective interest rate for
loans closed from a monthly survey of mortgage lenders by the Federal
Housing Finance Board. The effective interest rate includes the contract
rate plus fees and charges.
The housing expense data are derived from other existing data. The
mortgage principal is the house price noted above less 20 percent for the
down payment. The interest rate is the effective interest rate noted above.
The composite HAI (fixed-rate plus adjustable-rate mortgages) is a
weighted average of the fixed-rate and adjustable-rate HAIs. The weights
are based on the proportion of mortgages closed on fixed-rate and ad-
justable-rate loans obtained from the Federal Housing Finance Board.
Accuracy
There are no estimates of sampling or revision error for the housing

affordability index.
Relevance
The housing affordability index suggests prospects for owners and buy-
ers to match a house sale price with the income required to finance the
purchase. The higher the index is above 100, the greater the pool of
prospective buyers who can afford to buy a house, and so the greater the
number of likely home sales. By contrast, the lower the index is below
HOUSING AFFORDABILITY INDEX 131
Table 24.1
Housing Affordability Index (actual income as a percentage

of required income)
Fixed-rate
Composite mortgages Adjustable-rate mortgages
1995 132.4 126.6 143.3
1996 133.3 129.6 142.9
1997 134.0 130.8 145.3
1998 141.1 139.7 151.0
1999 139.1 136.3 150.4
2000 129.2 127.6 141.3
2001 135.7 135.7 145.5
2002 133.9 131.6 147.1
2003 138.4 125.7 140.5
2004 132.6 121.1 135.4
100, the smaller the pool of prospective buyers and the smaller the num-
ber of likely house sales. Because sales of existing houses provide own-
ers with the income to buy new or other existing houses, sales of existing
houses generate income and employment through the construction of
new houses plus the purchase of furniture, appliances, and other house
furnishings typically associated with moving into a newly purchased
house (see home sales: new and existing houses).
Recent Trends
From 1995 to 2004, the annual averages of the housing affordability

indexes for the composite, fixed-rate mortgages, and adjustable-rate mort-
gages were well above 100 (Table 24.1). The composite HAI fluctuated
within a range of 129.2 (2000) and 141.1 (1998), with peak levels in
1998, 1999, and 2003, and low levels in 1995, 2000, and 2004. The
fixed-rate HAI ranged between 121.1 (2004) and 139.7 (1998). The ad-
justable-rate HAI ranged between 135.4 (2004) and 151.0 (1998).
Washington, DC. Monthly.
25
Housing Starts
The housing starts data represent the beginning of construction of new

privately owned single-family homes, townhouses, and multifamily apart-
ment buildings. Each single-family house and each separate apartment
within apartment buildings (including cooperative and condominium
buildings) is counted as one housing start. Housing starts exclude pub-
licly owned housing, mobile homes, group quarters such as hotels and
dormitories, additions and alterations to existing housing, and conver-
sions from nonresidential structures to residential use.
Housing starts data are prepared monthly by the Bureau of the Census
in the U.S. Department of Commerce. The data are published in a news
release and in the report Housing Starts (www.census.gov).
The data are available during the third week of the month after the
month to which they refer. Each monthly report contains revised data for
the two previous months. The seasonally adjusted data are revised every
year for the preceding two years based on revised seasonal factors.
Content
A housing start is the beginning of construction of a privately owned

residential structure containing one or more housing units. A housing
unit is a dwelling of one or more rooms intended for occupancy as sepa-
rate living quarters by a family or group of unrelated individuals living
together. Each housing start represents a housing unit. A housing start is
132
HOUSING STARTS 133
counted as occurring in the month that excavation work begins for the
foundation, or of a residential structure rebuilt on an existing founda-
tion. While each single-family house and apartment unit is counted as
one housing start, the effect of differences in size and amenities of each
start with respect to the volume of construction work is captured only in
data on the dollar value of new housing construction put in place.
In addition to national totals, housing starts data are published for the
Northeast, Midwest, South, and West regions of the country.
The housing starts data are seasonally adjusted.
Methodology
Housing starts data are estimated separately for housing in local ar-
eas that require building permits for construction and for housing in
non-permit-issuing localities. For the permit-issuing areas, two
monthly sample surveys are used: (1) a mail survey of 8,700 of 20,000
permit-issuing localities to determine the total number of permits
issued, and (2) a survey of 900 areas by on-site interviewers to deter-
mine in which month construction started on housing units that were
authorized in previous months and the current month. The informa-
tion obtained on-site about the rates of construction started for per-
mits issued for each month to date is applied to the current permit
figures to develop the total number of housing starts every month.
The data are adjusted upward to reflect housing starts for which late
reports are received and for housing starts begun before a building
permit was issued. These upward adjustments are based on factors
derived from annual reviews plus more up-to-date monthly modifi-
cations of the extent to which these events occur.
Private housing starts in nonpermit areas are estimated from monthly on-
site surveys of ongoing construction work in a sample of eighty localities.
Accuracy
The sampling error (for one standard error) of the housing starts data is
3 percent. For example, if the monthly housing starts were estimated at
two million units, in two of three cases the true housing starts would
be somewhere between 1,940,000 and 2,060,000 units. For further in-
formation on the interpretation of sampling and nonsampling errors, see
the Appendix.
Table 25.1
Housing Starts (thousands)
Single-family Multifamily
Total percentage of total percentage of total
1995 1,354 79 21
1996 1,477 79 21
1997 1,474 77 23
1998 1,617 79 21
1999 1,641 79 21
2000 1,569 78 22
2001 1,603 79 21
2002 1,705 80 20
2003 1,848 81 19
2004 1,957 82 18
Relevance
New housing construction is important to the overall economy. Con-

struction results in the hiring of workers, the production of construc-
tion materials and equipment, and the sale of large household appliances
such as ranges and refrigerators. In addition, when owners or tenants
occupy the housing, they often buy new furniture, carpeting, and other
furnishings.
The rate of new housing construction is heavily influenced by
growth in the number of households in the long run, and by the growth
of inflation-adjusted household incomes and by the level and move-
ment of mortgage interest rates over shorter periods. Because hous-
ing lasts for many years and there is little need to replace it frequently,
the purchase of new housing usually is deferred until incomes and
interest rates make it affordable (see housing affordability index).
Housing starts are stimulated when incomes of workers at all earn-
ings levels rise steadily or interest rates are low or declining, and
they are restrained when incomes of workers at all earnings levels do
not rise steadily or interest rates are high or rising (see distribution
of income and interest rates).
Housing starts are related to home sales: new and existing houses.
Building permits for the construction of new housing (see Methodol-
ogy), which are a precursor of housing starts, are a component of the
leading index of the leading, coincident, and lagging indexes.
HOUSING STARTS 135
Recent Trends
From 1995 to 2004, housing starts rose from 1.35 million units in 1995
to 1.96 million units in 2004 (Table 25.1). Increases occurred in all years
except 1997 and 2000. The single-family share of all housing units, typi-
cally 79 percent during 19952001, rose to 82 percent in 2004.
Bureau of the Census, U.S. Department of Commerce. Housing Starts. Monthly.

26
Housing Vacancy Rate
The housing vacancy rate represents housing units that are not occupied
and that are physically suitable for occupancy. The housing vacancy
data cover year-round and seasonal housing in multifamily rental apart-
ments, single-family houses, townhouses, condominiums, cooperatives,
and mobile homes.
The housing vacancy rate is prepared quarterly by the Bureau of the Cen-
sus in the U.S. Department of Commerce. The data are published in a
news release and in the report, Housing Vacancy Survey (www.census.gov).
The data on housing vacancies are available about the fourth week of
the month following the quarter to which they refer. They are revised
approximately at ten-year intervals with the incorporation of new bench-
mark data from the decennial censuses.
Content
Housing vacancies comprise year-round and seasonal housing units that are
not occupied and are physically suitable for occupancy. Housing vacancy
data are provided for the United States, four Census geographic regions,
metropolitan areas, and outside metropolitan areas, and for selected charac-
teristics of the house, such as the number of rooms and contract rent.
A housing unit is a single-family house, an apartment in a multifam-
ily building (including townhouses, condominiums, and cooperatives),
or a mobile home that is owned or rented and is occupied or intended for
136
HOUSING VACANCY RATE 137
occupancy as separate living quarters by a family or a group of indi-

viduals living together. A housing unit excludes people living in group
quarters (e.g., dormitories, barracks, transient hotels or motels, except
in cases where people consider the hotel their usual residence,) and in
institutions (e.g., hospitals, jails, shelters, halfway houses).
A housing unit is defined as vacant if no one is living in it when the
housing enumerator conducts the survey, except when the household oc-
cupants are only temporarily absent when the housing survey is taken. A
vacant unit may be one that is entirely occupied by persons who have a
usual residence elsewhere. A household consists of all personsrelated
family members and unrelated individualswho occupy a housing unit
and have no other usual address.
When a seasonal housing unit, which is intended for use only during
certain seasons of the year, is occupied by a household that has a usual
residence elsewhere, it is counted as vacant, while the housing unit that
is the usual residence of the household is counted as occupied. An unoc-
cupied housing unit is not considered vacant if it is exposed to the ele-
ments so that the roof, walls, windows, or doors do not protect the interior
from the elements, or if there is positive evidence (such as a sign on the
house or block) that the unit is condemned or is to be demolished.
The housing vacancy data are not seasonally adjusted.
Methodology
The housing vacancy data are obtained from a monthly sample survey
of households, called the Current Population Survey (CPS), which is
conducted by the U.S. Bureau of the Census. The CPS is the same sur-
vey that is used to obtain the monthly household employment and un-
employment data (see employment and unemployment). The housing
occupancy and vacancy data are obtained by the enumerators at the same
time that they obtain the employment and unemployment data. The CPS
is a probability sample of households that is currently drawn from the
2000 census of population and is updated in subsequent years using
changes in residential locations associated with new housing construc-
tion data prepared by the Census Bureau.
Survey responses for the employment data in the CPS are obtained
from approximately 55,000 households each month. About 6,400 addi-
tional vacant units are surveyed for the housing vacancy data. The count
of occupied housing units is the same as the count of households.
Information on vacancies is obtained from neighbors, real estate agents,

property managers of rental property, or the owner of the property.
The housing vacancy rate is calculated as a percentage separately for
rental housing and for homeowner housing:
Rental
Rental Vacant for-rent housing units
vacancy = 100
vacancy (26.1)
raterate Renter occupied housing units +
vacant for-rent housing units +
rented units awaiting occupancy
Homeowner
Homeowner Vacant for-sale housing units
vacancy = 100
vacancy
raterate Owner occupied housing units +
vacant for-sale housing units +
sold units awaiting occupancy
Accuracy
The sampling error (for one standard error) for the rental housing va-
cancy rate is 0.2 of a percentage point. The sampling error (for one stan-
dard error) for the home owner housing vacancy is less than 0.05 of a
percentage point. For example, if the estimated rental housing vacancy
rate were 10 percent, in two of three cases the true vacancy rate would
be somewhere between 9.8 and 10.2 percent. For further information on
the interpretation of sampling and nonsampling errors, see the Appen-
dix. These error ranges may vary as the vacancy rate varies.
Relevance
The housing vacancy rate is a broad indicator of the relative shortage or

surplus of existing housing. The lower the vacancy rate, the greater the
potential for additional housing, while the higher the vacancy rate, the
smaller the potential for additional housing. This generalization may be
weakened if there is a shift in preference between rental and homeowner
housing. For example, a rise in the rental vacancy rate may reflect an
increasing preference for homeowner housing, in which case new single-
family construction could increase despite the rise in rental vacancies.
In assessing the need for additional housing, vacancy rates should be
HOUSING VACANCY RATE 139
Table 26.1
Housing Vacancy Rates (percent)
Rental housing Homeowner housing

1995 7.6 1.5
1996 7.8 1.6
1997 7.7 1.6
1998 7.9 1.7
1999 8.1 1.7
2000 8.0 1.6
2001 8.4 1.8
2002 8.9 1.7
2003 9.8 1.8
2004 10.2 1.7
supplemented with information on household growth, incomes, and rent-

als and sales by locality. Vacancy rates may also suggest a strengthening
or lessening of rental levels and housing prices.
Recent Trends
From 1995 to 2004, the rental housing vacancy rate generally rose from
7.6 percent in 1995 to 10.2 percent in 2004 (Table 26.1). The increase
was particularly pronounced from 2001 to 2004. The homeowner hous-
ing vacancy rate fluctuated around 1.61.8 percent over the 19952004
period. In general, the rental housing vacancy rate was about five times
the level of the homeowner vacancy rate.
Bureau of the Census, U.S. Department of Commerce. Housing Vacancy Survey.

Quarterly.
27
Import and Export Price
Indexes
The import and export price indexes measure price changes in agricul-
tural, raw material, and manufactured products for goods bought from
and sold to foreigners, plus a limited number of transportation services.
They represent increases and decreases in prices of internationally traded
goods due to changes in the value of the dollar and changes in the mar-
kets for the items.
Import and export price indexes are provided monthly by the Bureau of
Labor Statistics (BLS) in the U.S. Department of Labor. The data are
published in a news release and in the BLS monthly journal, Monthly
Labor Review (www.bls.gov).
The data are published two to three weeks after the month to which
they refer. They are revised in the following three months. Major bench-
mark revisions, which include updating the weighting structure, are made
every year.
Content
The import and export price indexes cover most internationally traded
goods. The broad product categories of the indexes are food, feeds, and
beverages; industrial supplies and materials; capital goods; automotive
vehicles, parts, and engines; and consumer goods, excluding automo-
tive. Goods exclusively for military use (based on U.S. Customs Bureau
definitions), works of art, commercial aircraft, and ships are excluded.
140
IMPORT AND EXPORT PRICE INDEXES 141
Supplementary international price indexes for transportation services

are provided for air freight rates, air passenger fares, crude oil tanker
freight rates (imports only), and ocean liner freight rates (imports only).
These services are not included in the above indexes for goods.
The indexes reflect movements for the same items exclusive of en-
hancement or reduction in the quality or quantity of the item. Prices repre-
sent the actual transaction value including premiums and discounts from
list prices and changes in credit terms and packaging. Prices usually are
based on the time the item is delivered, not the time the order is placed.
The price definition for imports is the value at the foreign port of
export loaded on the carrier (free on board, f.o.b.) or the value at the
U.S. border including overseas transportation and insurance costs (cost,
insurance, freight, c.i.f.). Import duties are excluded from the price.
The price definition for exports is the value at the U.S. port of export
before loading on the carrier (free alongside ship, f.a.s.) or loaded on the
carrier (free on board, f.o.b). Regardless of variations in the pricing basis,
it is more important that the same products maintain consistent defini-
tions over time. Thus, the same definition of prices is maintained for indi-
vidual products to ensure consistency in the monthly price movements.
The import and export price indexes are currently based on 2000 = 100.
The import and export price indexes are not seasonally adjusted.
Methodology
The price data are obtained by a Bureau of Labor Statistics mail survey
of a sample of importers and exporters, including a limited number of
foreign trade brokers. In addition, prices of crude petroleum imports are
based on U.S. Department of Energy data, and those for grain exports
(excluding rice) are based on U.S. Department of Agriculture data. The
overall response rate to the BLS survey rises from 7580 percent for the
initial estimate for the month to 8590 percent as late responses to the
survey are incorporated in the third revised estimate of the month.
Price quotations are sought for the first transaction of the month, which
typically occurs within the first week of the month. The weights of the
indexes are obtained from the Census Bureaus foreign trade data and
are reweighted annually with a two-year lag.
If the reported import or export price includes a change in the quality
(performance) or quantity (size) of the item, the Bureau of Labor Statis-
tics attempts to adjust the price to compensate for the improvement or
decline in quality or quantity. The adjustment is made to ensure that price

movements reflect items having the same functional characteristics over
time. The quality and quantity adjustments are based on performance and
size data in relation to production costs supplied by the importers and
exporters. Because the data to make the necessary adjustments are not
always available, the import and export price indexes contain an unknown
amount of price change caused by quality or quantity changes.
Accuracy
There are no estimates of sampling or revision error for the import and
export price indexes.
Relevance
The import and export price indexes are used in analyzing the effect of
changes in the competitive position of U.S. imports and exports. This
includes their linkage to the volume of imports and exports and the bal-
ance of trade, and their relationship to domestic price change in the
The international price indexes are also used in evaluating the effect of
changes in the value of the dollar on import and export prices. The ex-
tent of pass-throughs of price changes that partly or fully offset changes
in the value of the dollar can be calculated through the preparation of
supplementary trade-weighted exchange rates.
In addition, the international price indexes are used to adjust other
economic indicators for price change such as the merchandise trade data
in the balance of trade, the balance of payments, and the exports and
imports components of the gross domestic product.
Recent Trends
From 1995 to 2004, all imports prices fluctuated within an annual

range of 6.1 to 6.5 percent, compared with nonpetroleum imports prices
that fluctuated within a range of 3.6 to 3.9 percent (Table 27.1). The
narrower range of nonpetroleum imports reflects the volatility of petro-
leum prices. All imports prices changed more than nonpetroleum im-
ports prices in all years except 1999. In most years, all exports prices
and nonagricultural exports prices changed less than all imports and
IMPORT AND EXPORT PRICE INDEXES 143
Table 27.1
Import and Export Price Indexes (annual percentage change)
Import prices Export prices

Nonpetroleum Nonagricultural
All imports imports All exports exports
1995 4.6 3.9 5.1 4.6
1996 1.0 0.7 0.6 0.9
1997 2.5 2.2 1.3 0.3
1998 6.1 3.6 3.3 2.5
1999 0.9 1.4 1.3 0.6
2000 6.5 1.0 1.6 1.8
2001 3.5 1.5 0.8 1.0
2002 2.5 2.3 1.0 1.3
2003 3.0 1.1 1.5 1.1
2004 5.6 2.6 3.9 3.3
19952004
(annual average) 0.2 0.8 0.03 0.04
nonpetroleum imports prices. Over the entire nine-year period, the an-
nual increases and decreases in all imports prices averaged 0.2 per-
cent and for nonpetroleum imports prices averaged 0.8 percent, which
indicated a relatively small change in the price levels from 1995 to 2004.
The nine-year change for all exports averaged 0.03 percent and for
nonagricultural exports averaged 0.04 percent, which indicated virtu-
ally zero change in the price levels from 1995 to 2004.

Monthly.
28
The industrial production index (IPI) measures the change in output in

U.S. manufacturing, mining, and electric and gas utilities industries.
Output refers to the physical quantity of items produced, as distinct from
sales value, which combines quantity and price. The index covers the
production of goods and power for domestic sales in the United States
and exports. It excludes production in the agriculture, construction, trans-
portation, communication, trade, finance, and service industries; gov-
ernment output; and imports. While the excluded industries and imports
are not directly in the index, they are indirectly incorporated to the ex-
tent that the manufacturing, mining, and utilities industries use them as
intermediate items, in which case they are a component of the product
or power produced.
The industrial production index is prepared monthly by the Federal Re-

serve Board (FRB). It is published in a statistical release (G. 17) and in
the monthly statistical supplement to the quarterly FRB journal, the Fed-
eral Reserve Bulletin (www.federalreserve.gov).
The data are available in the middle of the month after the month
to which they refer, the same day as the capacity utilization measure.
Preliminary data are provided for the preceding month; these are re-
vised in the subsequent three months. Annual revisions are made in
the fall.
144
INDUSTRIAL PRODUCTION INDEX 145
Content
The industrial production index is provided from two perspectives: (a)

output originating in the producing industries (supply), and (b) selected
consumer and business markets of the items produced (demand). The
component groups of the supply perspective are the producing indus-
tries: manufacturing, mining, and electric and gas utilities. For the mar-
ket perspective, these components are categorized as products according
to their typical usage. These are: consumer goods, business equipment,
defense and space equipment, intermediate products including construc-
tion and business supplies, and materials including energy and nonenergy
materials, parts, and containers.
The IPI is currently based on 1997 = 100.
The IPI data are seasonally adjusted.
Methodology
The industrial production index weights are updated annually based on an

index number formulation in which the geometric mean of the value-added
weights in the previous year and the current-year is used (Fisher ideal
index). This is similar to the weighting used in the GDP price indexes and
in the chain index of the consumer price index (C-CPI-U). Because the
value-added data from the Census Bureau annual surveys are available
with a lag of approximately two years, weights in the current year are
extrapolated by the movements of producer price indexes in each indus-
try for the most recent period. This procedure utilizes price movements in
order to account for significant shifts in the value-added weights on an
up-to-date basis, such as sharp declines in computer and semiconductor
prices. Value added generally refers to the wages, profits, and deprecia-
tion of capital facilities in the producing industry, that is, the value an
industry adds to goods and services it buys from other industries, although
there are technical differences in the definition of value added as it is used
in the Census Bureau surveys from that in the gross domestic product.
The monthly movements of the index are based on the following source
information in the third monthly revision: production of actual items (48
percent); production worker hours in producing industries (30 percent);
electric kilowatt-hour consumption by producing industries (18 percent),
and various estimating techniques (4 percent). In the annual revisions,
monthly movements of the IPI components that were estimated from in-
Table 28.1
Industrial Production Index (1997 = 100)
All industries Manufacturing Mining Utilities

1995 89.4 88.1 96.7 97.2
1996 93.2 92.2 98.3 100.0
1997 100.0 100.0 100.0 100.0
1998 105.8 106.6 98.5 102.6
1999 110.6 112.2 93.6 105.5
2000 115.4 117.3 95.8 108.6
2001 111.3 112.3 96.7 108.1
2002 111.0 111.9 92.6 111.4
2003 110.9 111.9 92.2 111.9
2004 115.5 117.2 91.4 114.8
19952004 2.9 3.2 0.6 1.9
(annual average
percentage change)
direct sources (kilowatt hours and worker hours) are corrected to reflect
more extensive direct data on production, which are obtained mainly from
the Census Bureaus Annual Survey of Manufactures.
Accuracy
The typical revision to the monthly industrial production index level

between the preliminary estimate and the third monthly revision is plus
or minus 0.28 percent. The typical revision to the monthly movement is
plus or minus 0.22 of a percentage point. In about 85 percent of the
cases, the direction of change in the preliminary estimate is the same as
in the third revision.
Relevance
The coverage of the industrial production index makes it a sensitive

gauge of the most cyclical aspects of the economy. Although the indus-
tries covered amounted to only 16 percent of the gross domestic product
in 2003, they account for the bulk of the more volatile movements in
expansions and recessions. Consequently, the IPI tends to rise more in
expansions and fall more in recessions than the overall economy. The
IPI is a component of the coincident index of the leading, coincident,
and lagging indexes.
INDUSTRIAL PRODUCTION INDEX 147
Recent Trends
From 1995 to 2004, the industrial production index for all industries
combined increased from 89.4 in 1995 to 115.4 in 2000, declined to
110.9 in 2003, and rose to 115.5 in 2004 (Table 28.1). Manufacturing
had similar movements as the all-industries total. Mining peaked at 100
in 1997, and then declined to 91.4 in 2004. Utilities rose in all years,
from 97.2 in 1995 to 114.8 in 2004, except for zero change in 1997 and
a decline in 2001. The average annual percentage change in the IPI over
the nine-year period was: all industries (2.9 percent), manufacturing (3.2
percent), mining (0.6 percent), and utilities (1.9 percent).

29
Interest Rates
Interest is the cost of borrowing money, and interest rates are the price
of money. An interest rate, also referred to as a yield, is the annualized
percentage that interest is of the principal of the loan. Loans are ex-
tended and paid back through the use of debt instruments. The value
(i.e., the price) of the debt instrument, such as a bond, fluctuates due to
actual and anticipated inflation movements until its maturity date, when
the principal is repaid; during this period, the price of the debt instru-
ment moves inversely to the interest rate, with a price rise accompanied
by an interest rate decline, and vice versa. Interest rates differ for vari-
ous loans due to the length and risk of the loan. Generally, short-term
loans have lower interest rates than long-term loans, and loans subject
to little risk of not being repaid have lower interest rates than those with
higher risk.1
Measures of interest rates for different types of debt instruments are

reported on a daily, weekly, or monthly basis by the Federal Reserve
Board, U.S. Department of the Treasury, Federal Reserve Bank of New
York, Federal Housing Finance Board, Moodys Investors Service, and
Standard & Poors Corporation. Interest rates are published in news re-
leases of the source organizations. Several interest rates are published in
the Federal Reserves weekly statistical release H.15 (www.
federalreserve.gov). Other general Web sites for interest rates include:
Bloomberg Financial (www.bloomberg.com), Wall Street Journal Inter-
active Edition (www.wsj.com), Barrons (www.barrons.com),
148
INTEREST RATES 149
MarketWatch (www.marketwatch.com), Bankrate (www.bankrate.com),

and Interest (www.interest.com).
The data are available within a day to a week after the period to which
they refer.
Content
The nine annualized interest rate measures covered here show the dif-
ferent costs of borrowing for short-term, medium-term, and long-term
loans of high quality that are authorized by debt instruments such as
notes and bonds. Loans of high quality have the least risk of nonpay-
ment. While all nine types are high-quality loans, some are more secure
than othersfor example, U.S. Treasury securities are default-free, and
thus are the highest quality. Loan periods may be broadly defined as up
to one year (short-term), one to three years (medium-term), and more
than three years (long-term). Some interest rates are for new loans, while
others are for outstanding loans that are traded in securities markets.
The loans are made in transactions involving households, nonbank in-
dustries, commercial banks, and federal, state, and local governments.
Depending on the type of debt instrument, interest rates are measured
according to one of three methods: (a) paying a certain amount at regu-
larly specified intervals with a bond coupon or through negotiated terms
of the loan, (b) the extent to which the par value (redemption price when
the security expires) of a noncoupon security is above the discounted
market price of the security, or (c) a hybrid of regular interest and a
premium or discounted market price from the par value of the security.
Interest rates are not seasonally adjusted.
Selected Debt Instruments and Interest Rates
The nine interest rates are summarized below (U.S. Treasury notes and
bonds for three- and ten-year maturities make up two of the nine interest
rates). Sources of the interest rate data are in parentheses.
U.S. Treasury Three-Month Bills (U.S. Department of the Treasury):
Short-term default-free borrowing of new issues sold at a discount from
the par value.
U.S. Treasury Notes and Bonds with Average Constant Maturities of
Three and Ten Years (Federal Reserve): Yields on actively traded medium-
term and long-term default-free outstanding issues sold with a coupon
interest rate and at a premium or discount from the par value. These are
averages of securities that encompass a range of remaining maturities
from under one year to twenty years and do not represent a particular
issue.
Federal Funds (Federal Reserve Bank of New York): Loans between
commercial banks to enable the borrowing bank to meet its reserve re-
quirements with the Federal Reserve. They are primarily overnight loans
but also include term loans ranging from a few days to over one year.
The daily effective rate is a composite of the varying interest rates on
the different loan maturities.
Discount Rate, primary credit (Federal Reserve): Short-term borrow-
ing by commercial banks from regional Federal Reserve. This rate is
available to banks only with adequate capitalization and supervisory
ratings for soundness. The borrowing, which ranges from overnight to a
few weeks, is used variously to maintain certain reserve levels over a
two-week period, to meet huge outflows at the end of a day, to keep
bank reserves from falling close to or below legal minimum require-
ments, or for any other purpose including financing the sale of federal
funds. The primary credit discount rate is set above the federal funds
rate (immediately above), so it is a backup to the lower federal funds
rate in the event of particular short-term needs. Federal Reserve second-
ary credit is available at interest rates above the primary credit discount
rate to banks lacking satisfactory capitalization or supervisory ratings.2
High-Grade Municipal Bonds (Standard & Poors): Long-term out-
standing general obligation and revenue issues of municipalities sold
with a coupon interest rate and at a premium or discount from the par
value. The interest from the coupon rate is exempt from federal taxes.3
Corporate AAA Bonds (Moodys Investors Service): Long-term public
utilities and other nonfinancial outstanding issues judged to be the best
quality with the smallest degree of investment risk. They are sold with a
coupon interest rate and at a premium or discount from the par value.
Prime Rate Charged by Commercial Banks (Federal Reserve): Refer-
ence rate for small business loans, home equity loans, and credit card
loans. Bank loans to large businesses are more often priced with refer-
ence to the London Interbank Offered Rate (LIBOR) denominated in
dollars and other currencies as well as with reference to other rates.
New-Home Mortgage Yields (Federal Housing Finance Board): The
effective rate at closings of conventional first mortgage loans for fixed-
and variable-rate mortgages for newly built, single-family, nonfarm
INTEREST RATES 151
houses amortized, on average, over ten years. This rate includes the con-
tract interest rate and all fees, commissions, discounts, and points paid
by the borrower and/or seller to the lender in order to obtain the loan.
The ten-year amortization period is an approximation of the average life
of a conventional mortgage.
(The Federal Home Loan Mortgage Corporation provides a real-time
home mortgage rate based on mortgage commitments [prospective in-
terest rates] for new and existing single-family homes. This differs from
the above FHFB rate, which represents closings [actual interest rates]
for new homes on a monthly basis. The FHLMC rate is noted here to
acquaint the reader with an indicator of future mortgage interest rates.)
Methodology
Procedures used in calculating interest rates for the nine debt instru-
ments are summarized below.
U.S. Treasury Three-Month Bills: The discounted price of the auction
conducted every Monday as a percentage of the par value of the securi-
ties is the weekly yield. These weekly yields are averaged to obtain the
monthly figure.
U.S. Treasury Notes and Bonds with Average Constant Maturities of
Three and Ten Years: Yield curves are constructed by plotting interest
rates on the vertical axis of a graph and years to maturity on the horizon-
tal axis. A line is drawn through the middle of the plotted points and the
interest rate is read from the line that corresponds to three and ten years.
The daily closing market bid yields are averaged to obtain the weekly
number, and the weekly numbers are averaged for the monthly yield.
Federal Funds: Daily rates for federal funds of varying maturities are
obtained from federal funds brokers in New York City. The daily effec-
tive rate is the average of these rates weighted by the volume of loans
transacted through brokers at each rate.
Federal Reserve Primary Credit Discount Rate: These administered
rates of the twelve regional Federal Reserve Banks must be approved by
the Federal Reserve Board. The rates of all twelve regional banks are
the same, except for periods of a day or two when changes are not made
simultaneously. Officially, the rate for the Federal Reserve Bank of New
York is used, but in practice the New York rate is the same as those of the
other regional banks.
High-Grade Municipal Bonds: The yields of general obligation and
revenue issues for fifteen municipalities with remaining maturities of

approximately twenty years are averaged using equal weights for each
issue. The Wednesday closing is used as the weekly yield.
Corporate AAA Bonds: These are yields for public utilities and other
nonfinancial issues with remaining maturities of at least twenty years.
The number of bonds included in the indicator changes as the universe
of bonds meeting the AAA standard changes. They are averaged using
equal weights for each issue and calculated daily.
Prime Rate Charged by Commercial Banks: The daily rate charged
by the majority of the twenty-five largest domestically chartered com-
mercial banks based on assets. The rate is identical in at least thirteen of
the banks, and is referred to as the predominant rate.
New Home Mortgage Yields: A sample of savings and loan associa-
tions, commercial banks, mutual savings banks, and mortgage bankers
is surveyed for yields on mortgage loans for newly built, single-family
nonfarm houses. The survey covers mortgage closings during the last
five working days of the month. Interest rates of the various lenders are
weighted by the share of the mortgages originated by each type of lender.
The weights are updated quarterly.
Accuracy
There are no estimates of sampling or revision error for the interest rate
data.
Relevance
Interest rates have a significant impact on borrowing and spending. Des-

ignations of low or high interest rates are based on borrowers as-
sessments of past levels and prospective movements. If interest rates are
expected to rise, there is an incentive to borrow immediately, but if in-
terest rates are expected to fall, there is an incentive to delay borrowing.
Interest rates react to and influence movements of the gross domestic
product and the value of the dollar. The Federal Reserve focuses on
interest rates as the ultimate tool in conducting monetary policy to fos-
ter economic growth and moderate inflation.
Yield curves represent differential interest rates for short-term, me-
dium-term, and long-term debt instruments. The yield curve reflects
expectations of future interest rates, inflation, and business cycle move-
Table 29.1
Interest Rates (percent)
U.S. U.S. U.S. Primary credit

Treasury Treasury Treasury discount Prime rate
3-month 3-year 10-year rate: Federal High-grade Corporate charged by New home
bills: new constant constant Federal Reserve Bank municipal AAA commercial mortgage
issues maturities maturities funds of New York bonds bonds banks yields
1995 5.51 6.25 6.57 5.83 5.21* 5.95 7.59 8.83 7.87
1996 5.02 5.99 6.44 5.30 5.02* 5.75 7.37 8.27 7.80
1997 5.07 6.10 6.35 5.46 5.00* 5.55 7.26 8.44 7.71
1998 4.81 5.14 5.26 5.35 4.92* 5.12 6.53 8.35 7.07
1999 4.66 5.49 5.65 4.97 4.62* 5.43 7.04 8.00 7.04
2000 5.85 6.22 6.03 6.24 5.73* 5.77 7.62 9.23 7.52
2001 3.45 4.09 5.02 3.88 3.40* 5.19 7.08 6.91 7.00
2002 1.62 3.10 4.61 1.67 1.17* 5.05 6.49 4.67 6.43
2003 1.02 2.10 4.01 1.13 2.12 4.73 5.67 4.12 5.80
2004 1.38 2.78 4.27 1.35 2.34 4.63 5.63 4.34 5.77
*The primary credit discount rate was instituted in 2003. This rate differs from the discount rate that was used before 2003, which included
secondary credit.
INTEREST RATES
153
ments. The interest rate spread of the ten-year Treasury bond less the
federal funds rate is a component of the leading index of the leading,
coincident, and lagging indexes. The prime interest rate is a component
of the lagging index of the leading, coincident, and lagging indexes.
The federal funds rate and the primary credit discount rate are the
only indicators in this book that are acted on by a government authority
solely to affect their performance and consequently that of the economy.
The federal budget is acted on directly, but the federal budget is not
associated solely with influencing the economy (see government eco-
nomic transactions).
Recent Trends
From 1995 to 2004, interest rates drifted lower, except for increases in
some issues in 1999, 2000, and 2004. The declines were greater in short-
term and medium-term issues than in long-term issues (Table 29.1). In-
terest rates in 2004 ranged from 1.35 percent for federal funds to 5.77
percent for new home mortgage yields.
General Reference for Several Interest Rates
Board of Governors of the Federal Reserve System. Selected Interest Rates, Statis-
tical Release H.15. Weekly.
Notes
1. In the early 1960s, an attempt was made to lessen the differential between
short-term and long-term interest rates, which was referred to as Operation Twist.
See Franco Modigliani and Richard Sutch, Innovations in Interest Rate Policy,
American Economic Review, March 1966.
2. The primary credit discount rate was instituted in 2003. Before 2003, the
discount rate included banks in todays definition with both primary and secondary
credit ratings.
3. The yield includes both the coupon rate and the purchase price. When the
purchase price is the same as the principal of the bond (par value), the yield is the
coupon rate. When the purchase price is above par (premium) or below par (dis-
count), the yield differs from the coupon rate. When the purchase price is at a pre-
mium, the yield is below the coupon rate; when the purchase price is at a discount,
the yield is above the coupon rate. Capital gains and losses, which are associated
with the difference between the par value and the purchase price of the bond, are
subject to the liability for federal taxes.
INTERNATIONAL INVESTMENT POSITION OF THE UNITED STATES 155
30
International Investment
Position of the United States
The international investment position of the U.S. represents the difference

between the value of foreign assets held by U.S. parties abroad and the
value of U.S. assets held by foreign parties in the United States. It is com-
monly referred to as the overall creditor or debtor status of the nation,
although technically it only partially represents debt because it also in-
cludes equity ownership of foreign companies. Nevertheless, in this dis-
cussion, the creditor and debtor designations will be used because they
reflect the general perception of the meaning of the data. If American
assets abroad exceed foreign assets in the United States, the United States
is a creditor nation, and if foreign assets in the United States exceed Ameri-
can assets abroad, the United States is a debtor nation.

Commerce provides annual measures of the U.S. international invest-
ment position. They are published in a news release and in the BEA
The data are available each July for the preceding year. Annual revi-
sions to the historical data for several of the preceding years are also
provided each July.
Content
International investments include capital equipment and structures in

manufacturing and other industries, stocks, bonds, loans, the official U.S.
155
gold stock, and the U.S. reserve position and special drawing rights in the
International Monetary Fund. The net investment position of the U.S. is
defined as foreign assets held by U.S. parties abroad (assets) minus U.S.
assets held by foreigners (liabilities). The net investment position reflects
the total U.S. and foreign assets outstanding at the end of each year.
The investment position data include the actual capital flows at cur-
rent costs in the balance of payments plus valuation changes to the ex-
isting holdings of assets from previous acquisitions. Investment position
data consist of transactions throughout the year and valuation adjust-
ments to positions outstanding at the beginning of the year.
The annual change in the investment position is attributable to four
factors: (1) capital flows of private and government assets, (2) valuation
adjustments due to price changes, (3) valuation adjustments due to ex-
change rate movements, and (4) methodological changes of coverage,
statistical discrepancies, and valuation of assets. Methodological changes
result in a break in the comparability of the data series.
Methodology
The main data sources used in developing the investment position indi-
cator are: U.S. Treasury Department surveys conducted by the Federal
Reserve Bank of New York on international assets and liabilities; BEA
surveys of foreign direct investment abroad and in the United States;
bilateral financial data provided by other countries; and the value of the
dollar based on Federal Reserve Board and Treasury Department mea-
sures. Breakdowns of the factors contributing to the changes in position
from year to year are also derived from these data.
Direct investment and portfolio investment are two major components
of international investment. Direct investment is associated with a long-
term interest in and control of corporate and noncorporate business enter-
prises. It is defined as when a foreign investor owns 10 percent or more of
the voting securities or equivalent equity of an enterprise. Portfolio in-
vestment is associated with short-term activity in financial markets that
emphasizes the ability to move funds between countries and investments.
It is defined as when a foreign investor owns less than 10 percent of the
voting securities or equivalent equity of a business enterprise. Portfolio
investment also includes the total amount of an investors holdings of
foreign private and government bonds and other debt instruments.
The valuation of the investment position is made in current-cost and
INTERNATIONAL INVESTMENT POSITION OF THE UNITED STATES 157
market-value prices. Current cost is the dollar outlay necessary to re-

place the tangible equipment, structures, and land assets of direct in-
vestments. Market value is the dollar worth of the direct investment
tangible assets as measured by stock market prices, which also implic-
itly includes the value of intangible assets such as patents, trademarks,
management, and name recognition.
Portfolio investments are measured only in market-value prices. Thus,
the alternative current-cost and market-value price measures of the in-
vestment position reflect these distinctions only for direct investment.
Separate measures of historical cost of direct investment represent the
original purchase price of tangible assets; however, historical-cost esti-
mates are not included in the investment position measures.
Accuracy
There are no estimates of sampling or revision error in the international

investment position data. Increases and decreases in the position from
year to year attributable to methodological changes in the data sources
indicate the net effect of inconsistencies in the various data sources. This
is an overall minimum assessment of the inconsistencies because offset-
ting errors among the data elements reduce the net effect. In addition, the
statistical discrepancy in the balance of payments shifts between a net
inflow of unrecorded funds into the United States and a net outflow of
unrecorded funds from the United States. If the unrecorded flows are capital
funds, foreign assets in the United States would be understated (inflows)
or overstated (outflows). For all of these reasons, the international invest-
ment position is an order of magnitude rather than a precise number.
Relevance
The international investment position reflects the international indus-

trial and financial base of the United States. A creditor status signifies
that Americans own more capital abroad than foreigners own in the United
States, while a debtor status indicates that Americans own less capital
abroad than foreigners own in the United States. Because a creditor na-
tion is less dependent on outside sources of financing than a debtor na-
tion, its capital funding requirements are less vulnerable to changes in
international financial markets. Therefore, a creditor nation is more in-
dependent than a debtor nation in conducting monetary policies that
Table 30.1
International Investment Position of the United States (billions of dollar)
Net investment Net investment U.S. assets Foreign assets

position position abroad in the U.S.
(current cost) (market value) (current cost) (current cost)
1995 458.5 305.8 3,486.3 3,944.7
1996 495.1 360.0 4,032.3 4,527.4
1997 820.7 822.7 4,567.9 5,388.6
1998 895.4 1,070.8 5,095.5 5,990.9
1999 766.2 1,037.4 5,974.4 6,740.6
2000 1,381.2 1,581.0 6,238.8 7,620.0
2001 1,919.4 2,339.4 6,308.7 8,228.1
2002 2,107.3 2,445.1 6,645.7 8,752.9
2003 2,156.7 2,372.4 7,641.0 9,797.7
2004 2,484.2 2,542.2 9,052.8 11,537.0
influence its economy with respect to interest rates and the value of the
dollar.
A creditor nation has a net inflow and a debtor nation has a net out-
flow of interest and dividend incomes paid on international loans and
investments. If a nation is a creditor, the income flows increase exports,
which results in a surplus (or reduction in the deficit) in its balance of
payments. If the nation is a debtor, the income flows increase imports,
which results in a deficit (or reduction in the surplus) in its balance of
payments. The income flows also tend to raise the wealth and standard
of living in creditor nations relative to debtor nations. However, creditor
nations risk adverse actions by debtor nationsdefault on foreign debt
and expropriation of foreign-owned properties.
Recent Trends
From 1995 to 2004, the United States became an increasingly net debtor
nation (Table 30.1). The net debt in 2004 was $2.48 trillion (current
cost) and $2.54 trillion (market value). The growing net debt reflected a
continuous trend of slower growth of U.S. assets abroad than of foreign
assets in the United States.

Business. Monthly.
INVENTORYSALES RATIOS 159
31
InventorySales Ratios
The inventorysales ratio represents the monthly turnover rate of inven-

tories at current sales rates. For example, a ratio of 1.6 means that exist-
ing inventories will be used up in 1.6 months if sales continue at the
current rate (assuming that inventories are not replenished during the
period). The ratio is calculated with inventories in the numerator and
sales in the denominator.

pares monthly inventorysales ratios. They are published in the re-
port, Manufacturing and Trade Inventories and Sales (www.
census.gov).
The data are available forty-five days after the month to which they
refer; they are revised in the next month. Annual revisions are made in
the spring of the following year.
Content
The Census Bureaus inventorysales ratios reflect inventories and sales

of manufacturers, merchant wholesalers, and retailers. They include
materials and supplies, work in process, and finished goods of manufac-
turers, including service installation costs; they also include merchan-
dise of wholesalers and retailers who resell them without further
processing, including service installation costs of retailers.
The inventorysales ratios are provided in current dollars.
159
The inventorysales ratios are seasonally adjusted.

(An alternative measure of inventorysales ratios, referred to as pri-
vate inventories to final sales ratios, is provided as part of the gross
domestic product measures. These measures are not included in this
book, but are briefly noted here to alert the reader to their existence.
Final sales exclude sales of crude materials, supplies, and semifinished
goods that become intermediate products in the production process.
They also exclude sales of finished goods from manufacturers to whole-
salers and retailers and from wholesalers to retailers who resell them
in the same state, and thus provide unduplicated sales values of prod-
ucts from manufacturers to wholesalers to retailers. This contrasts with
the inventorysales ratios in this book that include the duplication of
such sales. The inventoryfinal sales ratios are provided quarterly in
current and in price-adjusted dollars. The inventoryfinal sales ratios
are relevant for analysts who prefer the use of unduplicated sales of
products as described above.)
Methodology
The inventory and sales data are obtained from the Census Bureaus
monthly surveys of manufacturers, merchant wholesalers, and retail-
ers. The inventory data are defined at current cost valuation, which is
the book-value acquisition cost before the companies convert invento-
ries to a LIFO (last-in, first-out) valuation. The sales data for manu-
facturers represent shipments and the sales data for wholesalers and
retailers represent sales.
Accuracy
There are no estimates of sampling or revision error for the inventory

sales ratios for the combined total of manufacturing and trade indus-
tries. Estimates of sampling error are available for the wholesale and
retail trade components separately.
Relevance
Inventorysales ratios impact future production levels. Inventories are a

business cost, which businesses finance by borrowing funds or tying up
their own money. Thus, high inventorysales ratios suggest that busi-
INVENTORYSALES RATIOS 161
Table 31.1
Inventory-Sales Ratio (manufacturing and trade industries)
1995 1.48
1996 1.46
1997 1.42
1998 1.43
1999 1.40
2000 1.41
2001 1.44
2002 1.40
2003 1.38
2004 1.31
nesses will tend to cut back on orders to suppliers because it is expen-

sive to hold goods that are not selling rapidly. The resulting lower manu-
facturers orders lead to declines in the industrial production index and
employment. Analogously, low inventorysales ratios suggest increased
orders to replenish inventories, because the ability to furnish items readily
off the shelf and maintain a wider selection of goods for customers will
promote sales. In this case, the growth in orders results in higher pro-
duction and employment.
These general tendencies hold true, more or less, depending on the
extent to which inventory movements result from deliberate action by
businesses to build up or deplete inventories through sales incentives,
cost cutting, changes in production and orders (planned inventory
change), or from unanticipated inventory accumulation or depletion
due to unexpected surges or weaknesses in customer demand (un-
planned inventory change). Unplanned inventory changes may affect
future production more because of the surprise effect they have on
business expectations. Unanticipated changes are not quantifiable be-
cause the inventory data do not distinguish the planned and unplanned
components. The Census Bureaus inventorysales ratio for manufac-
turing and trade combined is converted to price-adjusted dollars for
inclusion as a component of the lagging index of the leading, coinci-
dent, and lagging indexes.
Recent Trends
From 1995 to 2004, the inventorysales ratio for manufacturing and

trade inventories combined generally declined from 1.48 in 1995 to 1.31
in 2004 (Table 31.1). The nine-year decline was interrupted by increases

in 1998 and 2001.
Bureau of the Census, U.S. Department of Commerce. Manufacturing and Trade

Inventories and Sales. Monthly.
JOB GAINS AND LOSSES 163
32
Job gains and losses represent gross job increases and decreases, not
just the employment net change of the gains minus the losses. Employ-
ment gains cover job increases in existing business establishments and
job increases in establishments newly starting in business. Employment
losses cover job decreases in existing businesses and job decreases in
businesses going out of business. The formal title of this program is
Business Employment Dynamics. Job gains and losses is a new eco-
nomic indicator that was first published in 2004.
The job gains and losses data are prepared quarterly by the Bureau of
Labor Statistics in the U.S. Department of Labor. The data are published
in Business Employment Dynamics, a news release (www.bls.gov).
The data are available in the eighth month after the quarter to which
they refer. Although the published data are described as preliminary,
revisions are presently not published because they are minor and do not
significantly affect the published data.
Content
Job gains and losses data cover private industry jobs and are derived
from the federalstate unemployment insurance system. The main cat-
egories of workers excluded from the employment gains and losses data
are: government workers, self-employed workers, private household
workers, railroad industry workers, most agricultural workers on small
163
farms, religious organization workers, and some not-for-profit organi-

zation workers. Also excluded are businesses with zero employment in
two consecutive quarters.
The job gains and losses data reflect changes in employment for the
same group of business establishments from quarter to quarter (see
Methodology for the definition of a business establishment). These
cover job increases in new and existing businesses, including businesses
with positive job levels in one quarter, up from previous zero job levels.
Similarly, job decreases are covered in existing businesses, business es-
tablishments that go out of business, and those with zero job levels in
one quarter, down from previous positive levels.
The job gains and losses data are provided for the United States and
for broad industry divisions. Separate data for geographic regions are
planned in the future.
The job gains and losses data are provided in absolute levels of the
number of jobs and the number of establishments, and in relative rates
as a percentage of total employment. The absolute and relative data are
classified separately for establishments with gross job gains and gross
job losses, with further divisions within the gains and losses categories
for increases and decreases other than zero (expansions and contrac-
tions), and increases and decreases from zero to positive job levels (job
openings) and positive to zero job levels (job closings).
The job gains and losses data represent the last month of each calen-
dar quarter (March, June, September, December). The provision of an-
nual data is planned for a later time.
The job gains and losses data are seasonally adjusted.
Methodology
The job gains and losses data are based on jobs in business establish-
ments. A business establishment is an employers place of business in
a particular geographic location that produces the same or comple-
mentary commodities and/or services. For a company that has more
than one establishment, each establishment is counted separately. This
contrasts with company data, in which all establishments of the com-
pany are consolidated, regardless of the variety of products made in
each establishment.
The data are obtained from the universe of employers that report
their employment and wages to the State Unemployment Insurance
JOB GAINS AND LOSSES 165
agencies every quarter. This is a mandatory program that is part of

funding the Unemployment Insurance system. It is referred to as the
ES-202 program.
The jobs data cover all corporation officials, executives, supervisory
personnel, clerical workers, wage earners, pieceworkers, and part-time
workers who were on the payroll for the pay period of the establishment
that includes the twelfth day of the month. Persons on the payroll for
more than one establishment are counted for each job, because this is a
count of jobs as distinct from a count of persons (see employment for
this distinction). The employment data exclude workers who earned zero
wages during the pay period (e.g., due to strikes or lockouts, temporary
layoffs, illness, unpaid vacations, or wages earned during the month but
not during the pay period).
Accuracy
There are no sampling errors with the job gains and losses data because
they are obtained from the universe of employers.
Relevance
The gross increases and decreases of jobs in the gains and losses data
portray a far more dynamic reality of job markets than only the net changes
in employment (increases minus decreases) in the traditional employment
data. The vast amount of job creation and destruction highlights the con-
tinuously shifting strengths and weaknesses in job markets.
Assessments of the job gains and losses data together with the job
openings and labor turnover data would enhance the understanding of
job markets. These assessments should be reflected in shaping fiscal
and monetary policies and jobs programs.
Recent Trends
From 1995 to 2004, job gains peaked at 9.1 million in 1999, fell to 7.6
million in 2003, and rose to 8.1 million in 2004. Job losses rose from 7.4
million in 1996 to 8.8 million in 2001, and fell to 7.2 million in 2004.
The net change of job gains less job losses peaked at 1.1 million in
1999, fell to a negative 871,000 in 2001, declined to a smaller negative
of 175,000 in 2002, and then rose to a positive 344,000 in 2003 and
Table 32.1
Job Gains and Losses (OctoberDecember quarter)
Net changea Job gains Job losses

Thousands
of jobs
1995 407 7,877 7,470
1996 861 8,278 7,417
1997 702 8,731 8,029
1998 759 8,576 7,817
1999 1,105 9,144 8,039
2000 336 8,691 8,354
2001 871 7,893 8,764
2002 175 7,702 7,877
2003 344 7,646 7,302
2004 869 8,081 7,212
Percentage of
total employmentb
1995 0.4 8.1 7.7
1996 0.9 8.3 7.4
1997 0.6 8.4 7.8
1998 0.7 8.1 7.4
1999 1.1 8.5 7.4
2000 0.3 7.9 7.7
2001 0.8 7.3 8.1
2002 0.2 7.1 7.3
2003 0.4 7.2 6.8
2004 0.7 7.4 6.7
a
Job gains minus job losses; bJob gains and job losses as a percentage of the average
of the previous and current employment levels.
869,000 in 2004. The percentages of total employment for job gains and
losses and the net change in jobs showed a similar pattern as the above
absolute movements (see Table 32.1).
Job gains exceeded job losses in the 1995 to 2000 period and in 200304.
Job losses exceeded job gains in 2001 and 2002.
Bureau of Labor Statistics, U.S. Department of Labor. Business Employment

Dynamics. Quarterly.
JOB OPENINGS AND LABOR TURNOVER 167
33
Job Openings and Labor
Turnover
The job openings and labor turnover survey (JOLTS) data represent the
extent to which business establishments and government agencies ac-
tively recruit for specific job openings, the extent of job hires, and the
extent of job separations.
The JOLTS data cover private nonfarm industries, federal govern-
ment civilian agencies (excluding uniformed armed forces), state gov-
ernment agencies, and local government agencies. JOLTS is a new
economic indicator that was first published in 2002.
The JOLTS data are prepared monthly by the Bureau of Labor Statistics
in the U.S. Department of Labor. The data are published in a news re-
lease, Job Openings and Labor Turnover (www.bls.gov).
The JOLTS data are available five to six weeks after the month to
which they refer. The initial data are revised in the subsequent month.
Annual revisions are made in the spring of the following year.
Content
The JOLTS data cover employees on employer payrolls for the pay pe-
riod that includes the twelfth day of the month. All employees are cov-
ered, except proprietors and/or partners of unincorporated businesses
and unpaid family workers. Employees of temporary help agencies,
employee leasing companies, outside contractors, and consultants are
counted as employed by the employer that records them on its payroll,
167
which may not be the employer where they are working. The JOLTS
data are published for the United States, four regions of the country,
sixteen broad industry categories, the federal government, and state and
local governments.
Job Openings
A job opening is defined as: (a) a specific position for which work is
available, where (b) work could start within thirty days, and (c) the em-
ployer is actively recruiting from outside the establishment to fill the
position. The job openings data exclude jobs to be filled from within the
establishment such as transfers, promotions, or recall from layoff. The
job openings rate is the percentage that job openings are of total em-
ployment plus job openings.
Hires
Hires are employee additions to employer payrolls of both new and

rehired workers. The hires data do not include transfers or promotions
from within the business establishment or the government agency. The
hires rate is the percentage that hires are of total employment.
Separations
Separations are job terminations due to quits, layoffs and discharges, re-
tirement, transfers to other locations, disability, and death. The separation
rate is the percentage that separations are of total employment.
The JOLTS data are seasonally adjusted for job openings, hires, quits,
total separations, and for selected industries.
Methodology
The JOLTS data are obtained from a probability sample of 16,000 non-
farm business establishments and government agencies in all states
and the District of Columbia. Large business establishments and gov-
ernment agencies are always included in the sample; that is, they are
not subject to random selection. The JOLTS sample is selected from
the same sampling frame as the employer establishment survey cov-
ered in employment.
JOB OPENINGS AND LABOR TURNOVER 169
Table 33.1
Job Openings and Labor Turnover: 2004 (seasonally adjusted)
Level (thousands)
Job openings Hires Separations
January 2,864 4,310 3,994
February 2,961 4,159 4,196
March 3,105 4,838 4,289
April 3,111 4,509 4,334
May 3,181 4,339 4,254
June 3,140 4,492 4,235
July 3,231 4,297 4,190
August 3,206 4,504 4,271
September 3,265 4,406 4,214
October 3,300 4,552 4,215
November 3,277 4,990 4,266
December 3,507 4,639 4,435
Rate (percent)
January 2.1 3.3 3.1
February 2.2 3.2 3.2
March 2.3 3.7 3.3
April 2.3 3.4 3.3
May 2.4 3.3 3.2
June 2.3 3.4 3.2
July 2.4 3.3 3.2
August 2.4 3.4 3.2
September 2.4 3.3 3.2
October 2.4 3.4 3.2
November 2.4 3.8 3.2
December 2.6 3.5 3.3
The job openings data are based on the last business day of the month.
The hires and separations data are based on the entire month.
Accuracy
The sampling error (for one standard error) for the JOLTS monthly rate
data is: job openings0.051 of a percentage point; hires0.085 of a
percentage point; separations0.083 of a percentage point. For example,
if the monthly hires rate were 2 percent, in two of three cases the true
rate would be somewhere between 1.949 and 2.051 percent. For further
information on the interpretation of sampling and nonsampling errors,
see the Appendix.
Relevance
The JOLTS data are an approach at the macroeconomic level to com-

pare available job vacancies with unemployment. This is referred to as
the Beveridge Curve, which shows an inverse relationship between job
vacancies and unemployment. Shifts in the Beveridge Curve indicate
whether the efficiency of matching jobs with workers is improving or
worsening. Shifts in the efficiency are affected by such factors as occu-
pational skills, geographical location of job openings and unemployed
workers, quality of labor, and hiring discrimination.
Ideally, the JOLTS data would match job openings with job skills of
unemployed workers. But occupational data are not provided in the
JOLTS data nor are the industry or geographic detail sufficient for
such usage.
Recent Trends
From January to December 2004, job openings generally rose to a

high in December, except for declines in August and November. Hires
exceeded separations in all months except February, with the largest
differential in November of 700,000. Job opening levels were typi-
cally 1.2 to 1.5 million below hires, and around 1.0 million below
separations.
As a percentage of total employment, job openings were typically 2.3
to 2.4 percent, with a high of 2.6 percent in December; hires were mostly
3.3 to 3.4 percent, though with noticeably higher rates in March and
November; and separations were mostly 3.2 to 3.3 percent. (See Table
33.1.)
Bureau of Labor Statistics, U.S. Department of Labor. Job Openings and Labor
Turnover. News Release. Monthly.
LEADING, COINCIDENT, AND LAGGING INDEXES 171
34
Leading, Coincident, and
Lagging Indexes
The leading, coincident, and lagging (LCLg) indexes are an analytic

system for assessing current and future economic trends, particularly
cyclical expansions and recessions. The system is based on grouping
some key indicators according to their tendency to change direction be-
fore, during, or after the general economy turns from a recession to an
expansion or from an expansion to a recession. Indicators in the leading
index change direction before a cyclical turning point in the general
economy; those in the coincident index change direction at the same
time as the general economy; and those in the lagging index change
direction after the change in the general economy. Substantively, the
leading index reflects business commitments and expectations, the co-
incident index reflects the current pace of economic growth, and the
lagging index reflects business production costs.
The Conference Board prepares monthly measures of the leading, coin-

cident, and lagging indexes. They are published in the monthly report of
The Conference Board, Business Cycle Indicators (www.conference-
board.org).
The data are available one month after the month to which they refer;
they are revised in the subsequent five months as new source data be-
come available. Annual revisions of the source data are made every
December. Comprehensive revisions in the components and the meth-
odology do not follow a regular schedule. The last revision of this type
was made in December 1996.
171
Content
The leading, coincident, and lagging system reflects the business cycle
concept that each phase of the business cycle contains the seeds of the
subsequent phase.1 The phases of the business cycle are widely referred
to as recovery, expansion, and recession, and less often with an addi-
tional phase of contraction. Recovery is the upturn in economic growth
from the low point of the previous recession. Expansion is the upward
continuation of economic growth from the recovery above the high point
attained in the previous expansion before the economy turned down into
recession. Recession is the downturn in economic growth from the high
point of the previous expansion. Contraction is the downward continua-
tion of economic growth in the recession below the low point of the
previous recession (this is the counterpart of the expansion definition),
although a recession typically does not become a contraction.
A complete business cycle includes all upward and downward phases.
Average economic growth over each cycle is typically measured from
the high point of the previous expansion to the high point of the current
expansion, and less often from the low point of the previous recession to
the low point of the current recession.
The LCLg system is based on the idea that profits are the prime mover
of a private enterprise economy and that the recurring business cycles of
expansion and recession are caused by changes in the expectation for
profits.2 When anticipated profits are positive, business expands pro-
duction and investment, but when they are negative, business retrenches.
The outlook for profits is reflected in the LCLg system in the leading
index and in the ratio of the coincident index to the lagging index.
The leading index represents business commitments and expectations
regarding labor, product, and financial markets, and thus points to future
business actions. The coincident index represents current movements of
production and sales, and so should coincide with business cycle turning
points from recovery to expansion to recession to contraction to recovery.
The lagging index represents costs of business production. The ratio of
the coincident index to the lagging index suggests whether profits will
rise or fall in the future due to the differential movements of sales and
costs. If the coincident index increases more than the lagging index, or if
the coincident index increases and the lagging index falls, profits are likely
to rise. Analogously, if the coincident index increases less or declines more
than the lagging index, profits are likely to fall. In this way, the coinci-
Table 34.1
Leading, Coincident, and Lagging Composite Indexes and Components
Leading Composite Index

1. Weekly hours of manufacturing production workers (average weekly hours)
2. Weekly initial claims for unemployment insurance, state programsinverted
scale (unemployment)
3. Manufacturers new orders for consumer goods and materials, in price-
adjusted dollars (manufacturers orders)
4. Vendor performance (percentage of companies receiving slower deliveries)
5. Manufacturers new orders for nondefense capital goods industries, in price-
adjusted dollars (manufacturers orders)
6. New private-housing units authorized by local building permits (housing starts)
7. Prices of 500 common stocks, index (stock market price aggregates and
dividend yields)
8. Money supply (M2), in price-adjusted dollars
9. Interest rate spread, ten-year Treasury bonds less federal funds (interest rates)
10. Consumer expectations, index (consumer attitude indexes)
Coincident Composite Index

1. Employees on nonagricultural payrolls (employment)
2. Personal income less transfer payments, in price-adjusted dollars (personal
income and saving)
3. Industrial production, index (industrial production index)
4. Manufacturing and trade sales, in price-adjusted dollars
Lagging Composite Index

1. Duration of unemployment, weeksinverted scale (unemployment)
2. Inventories divided by sales for manufacturing and trade, ratio, in price-
adjusted dollars (inventory-sales ratios)
3. Labor cost per unit of output in manufacturing, monthly percentage change
(unit labor costs)
4. Prime rate charged by banks, percent (interest rates)
5. Commercial and industrial loans outstanding, in price-adjusted dollars (bank
loans: commercial and industrial)
6. Consumer installment credit outstanding divided by personal income, ratio
(consumer credit and personal income and saving)
7. Consumer price index for services, monthly percentage change (consumer
price index)
dentlagging ratio, which is internally generated from the LCLg system

and is independent of the leading index, is an alternative leading index.
The LCLg indexes are called composite indexes because they group
several component indicators. Table 34.1 lists the component indicators
in each of three composite indexes. The leading index components re-
flect: the degree of tightness in labor markets due to employer hiring
and layoffs; the effect of new orders for manufactured products on fu-
ture production; financial conditions associated with short-term and long-
term interest rate differentials that indicate the stance of Federal Re-
serve monetary policy; optimism or pessimism reflected by price move-
ments in the stock market, and consumer psychology that affects
household spending plans. The coincident index components reflect:
employment; real personal income generated from production; output
in the cyclically sensitive manufacturing, mining, and electric power
industries; and real manufacturing and trade sales that encompass the
flow of goods between manufacturers, from manufacturers to wholesal-
ers, from wholesalers to retailers, and from retailers to households and
businesses. The lagging index components reflect: the effect of the du-
ration of unemployment on business wage pressures; the cost of main-
taining inventories; labor costs of production in manufacturing; interest
payments as a cost of production; the burden of existing business debt in
taking on new loans; and prices of consumer services as an indication of
production-cost pressures in labor-intensive industries.
The LCLg indexes are currently based on 1996 = 100.
Most of the component indicators of the LCLg indexes are season-
ally adjusted. The three composite indexes are not seasonally adjusted
at the overall level.
Methodology
The component indicators of the leading, coincident, and lagging in-

dexes are selected based on tests conducted for the following criteria:
theoretical rationale for the leading, coincident, and lagging properties;
differences in the timing of their change in direction in relation to the
cyclical turning points of the economy; consistency with the general
upward and downward direction of the business cycle; clear upward or
downward trends as distinct from erratic monthly movements from which
it is difficult to discern a trend; the quality of the methodology used in
collecting the data; promptness of the availability of monthly data; and
the extent of revisions to preliminary data.
Data for the components of the LCLg indexes are based on many of
the indicators discussed elsewhere in this book. These indicators are
referenced in Table 34.1, although they are not always definitionally
identical to the LCLg components.
The components are combined in the three indexes and weighted
equally. A statistical procedure is used in calculating the indexes to pre-
vent the components that have sharp monthly movements from domi-
nating the indexes. The long-run trends of the leading and lagging in-
dexes are equated to the long-run trend of the coincident index.
Accuracy
There are no estimates of sampling or revision error for the leading,

coincident, and lagging indexes.
Relevance
The leading, coincident, and lagging indexes help in assessing the current
momentum and future direction of the economy. Because the coincident
index reflects actual economic growth, it indicates whether the economy is
currently expanding or in recession (the actual designation of a recession
period is determined by the Business Cycle Dating Committee of the non-
profit National Bureau of Economic Research). Movements in the leading
index and the coincidentlagging ratio suggest whether the existing trend
measured by the coincident index will continue. A change in direction in the
leading index and the coincidentlagging ratio tends to foreshadow move-
ment in the coincident index. But it is important to note that while the LCLg
system is a forecasting tool, it does not provide actual forecasts.
Figure 34.1 shows that the lead time between the change in direction of
the leading measures and the coincident index is noticeably longer at the
cyclical downturn than at the upturn. It also indicates that the lead times
vary considerably from cycle to cycle. And in the 199091 and 2001 re-
cessions, the lagging index ran counter to the theory when it showed a
lead before the onset of each recession; in fact, the lead was twice as long
as the lead of the leading index. Changes in the lagging index follow the
coincident index and are thus used to confirm that a directional change
has occurred, except as just noted. The above-noted variations in timing
from cycle to cycle are examples of the continually changing economic
landscape that gives each business cycle unique characteristics.
The LCLg indexes are a useful tool for gauging strengths and weak-
nesses in the economy. However, they are limited for forecasting future
economic trends. First, the preliminary contemporaneous data that are
available during the months before a downturn into a recession or an
upturn into a recovery do not always provide advance signs of a cyclical
turning point. The leading index systematically leads at cyclical turning
points only in the recalculated historical measures; these incorporate
Figure 34.1 Leading, Coincident, and Lagging Composite Indexes

(1996 = 100)
T PT PT PT PT PT PT PT
Source: The Conference Board, Business Cycle Indicators, June 2005.

Note: Vertical bars are recession periods. Numbers on bars are monthly leads (-) and
lags (+) from cyclical turning points. Series numbers are component indicators of the
composite index.
P = Peak: High point of expansion.
T = Trough: Low Point of recession
data revisions in the components and changes in the methodology that

are made several years later. At the same time, the real time anticipation
of changes in economic growth rates has improved as a result of new
procedures for incorporating more timely data for selected component
indicators of the composite indexes.3 Second, the LCLg indexes do not
forecast quantitative economic growth rates or the timing of future cy-
clical turning points. Third, the indexes occasionally give false signals
of a pending change in the direction of the economy, such as prospective
recessions in 1966, 1984, and 1995 that did not materialize.
These complexities make it difficult to develop LCLg system guide-
lines for forecasting. The Conference Board suggests a nuanced approach
for identifying a pending recession.4 This provides for a decline in the
leading index of 1 to 2 percent over a six-month period together with
declines in at least half of the components of the leading index.
Recent Trends
From the 1960s to 2004 (Figure 34.1), the leading index led at all cycli-
cal turning points, although by varying numbers of months. The coinci-
dentlagging ratio led at all cyclical turning points except in the recovery
in 1960, the recovery in 1975, and the onset of the recession in 1990.
The coincident index was exactly at the cyclical turning points, except
for divergences of one month in three cases, two months in one case,
and three months in one case. The lagging index lagged at all cyclical
turning points except at the onset of the 199091 and 2001 recessions.
. Business Cycles Indicators Handbook. 2001.
Notes
1. For an insightful explanation of business cycles, including why they are likely
to continue into the twenty-first century, see Victor Zarnowitz, Theory and History
Behind Business Cycles: Are the 1990s the Onset of a Golden Age? Journal of
Economic Perspectives (Spring 1999).
2. Ibid., p. 241.
3. Robert H. McGuckin, Ataman Ozyildirum, and Victor Zarnowitz, A More
Timely and Useful Index of Leading Indicators (New York: The Conference Board),
October 2004.
4. The Conference Board, Business Cycle Indicators (monthly). Inside front cover.
35
Manufacturers Orders
Manufacturers orders measure commitments by customers to pay for

the subsequent delivery of durable and nondurable goods produced by
manufacturers. They include new orders received each month and the
backlog of unfilled orders from previous months.
Manufacturers new orders and unfilled orders are prepared monthly by

the Bureau of the Census in the U.S. Department of Commerce. They
are published in a news release and in the monthly report that covers
both durable and nondurable goods industries, Manufacturers Shipments,
Inventories, and Orders (www.census.gov).
The data are available one month after the month to which they refer.
An advance report for durable goods is available three weeks after the
month to which the data refer. The monthly data are initially revised in
the following month; they are subsequently revised based on annual
benchmark information in the following year.
Content
Manufacturers orders data provide the dollar value of all durable and
nondurable industries, industry detail, and market categories of various
consumer goods, capital goods, defense products, and materials, plus
the delivery, installation, repair, and other services that are associated
with the goods in the orders contract.
Orders are defined to be legally binding documents such as signed
178
MANUFACTURERS ORDERS 179
contracts, letters of award, or letters of intent, although this definition

may not apply in some industries. In the case of letters of intent, the
sales value is included if the parties are in substantial agreement on the
amount; otherwise, only the funds specifically authorized to be expended
are included.
New orders represent the dollar value of orders received each month
for delivery during that month or later. They are the net of contract
changes, including cancellations, that raise or lower the value of unfilled
orders received in previous months.
Unfilled orders represent the dollar value of the backlog of orders
that have accumulated from previous months for goods that have not yet
been delivered. They are a running total from one month to the next of
the backlog at the beginning of the month, plus new orders received
during the month, minus shipments of goods to customers and
cancellations of existing orders during the month.
The manufacturers orders data are seasonally adjusted.
Methodology
Manufacturers orders data are based on monthly surveys of

manufacturers. These surveys also obtain information on shipments and
inventories (see inventorysales ratios). While respondents are asked
about both new and unfilled orders, the survey on new orders is
incomplete. Due to a lack of readily accessible records, some survey
respondents do not report new orders and others do not report new orders
for goods that were shipped from existing inventories in the same month.
Consequently, new orders are estimated indirectly from unfilled orders,
shipments, and cancellations as follows:
unfilled orders (end of current month)

plus: shipments (during month)
minus: unfilled orders (end of previous month)
equals: new orders (during month)
The survey sample includes most manufacturing companies with annual

shipments of $500 million or more and a limited number of smaller
companies. The survey has response problems that prevent it from being
a probability sample.1 The survey sample accounts for about 60 percent
of all manufacturing shipments.
Table 35.1
Manufacturers Orders (billions of dollars)
New orders Unfilled orders

(monthly average) (end of year)
1995 285.5 447.3
1996 297.3 488.8
1997 315.0 513.2
1998 317.3 496.5
1999 329.8 505.9
2000 346.8 550.0
2001 322.9 517.6
2002 316.7 485.8
2003 329.2 506.3
2004 365.8 552.2
The monthly data are revised every year to reflect more complete
information in the Annual Survey of Manufactures.
Accuracy
There are no estimates of sampling error for the manufacturers orders

data. Based on annual revisions, monthly new orders are typically revised
by plus or minus 0.3 percent, and monthly unfilled orders are typically
revised by plus or minus 0.1 percent.
Relevance
Manufacturers orders indicate current demand, which translates into

future production (see industrial production index) and employment for
manufacturing industries. There is a lead time between orders on the
one hand and production and employment on the other, although
determining when orders result in production is not an exact science.
Rising orders are associated with higher current demand and subsequent
increased production and employment, while falling orders indicate lower
current demand followed by decreased production and employment.
This relationship occurs for both new orders and unfilled orders, with
the distinction that new orders reflect current demand and unfilled orders
reflect the cumulation of demand from previous periods. The greater the
ratio of unfilled orders is to new orders, the more is future production
sustained by the backlog of unfilled orders, while the smaller the ratio
MANUFACTURERS ORDERS 181
of unfilled orders is to new orders, the more is future production linked

to current new orders. New orders for consumer goods and materials
industries in price-adjusted dollars are a component of the leading index
of the leading, coincident, and lagging indexes.
Recent Trends
From 1995 to 2004, manufacturers new orders and unfilled orders

increased in all years, except for 2001 and 2002 (Table 35.1). After peaking
in 2000, new and unfilled orders first exceeded the 2000 levels in 2004.
Unfilled orders were consistently 1.5 to 1.6 times as large as new orders.
Bureau of the Census, U.S. Department of Commerce. Manufacturers Shipments,

Inventories, and Orders. Monthly.
Note
1. The response problem with this survey is long-standing. While this is a vol-
untary survey, the response in many other voluntary surveys is adequate to provide a
probability sample. If this survey were made mandatory, which would require new
legislation, the response would likely increase. For the attributes of probability
samples, see the Appendix.
36
Mortgage loan applications indexes represent the number of mortgage ap-

plications for single-family houses both for the purchase of a house and for
refinancing an existing house mortgage. The data cover conventional, Fed-
eral Housing Administration (FHA), and Veterans Administration (VA)
mortgages combined, and fixed-rate, adjustable-rate, and balloon mortgages.
The mortgage loan application indexes are prepared weekly by the Mort-
gage Bankers Association of America. They are published in the report,
Weekly Mortgage Applications Survey (www.mortgagebankers.org).
The data are available for the calendar week Monday to Friday on the
following Wednesday at 7:00 A.M. The data are not revised.
Content
The mortgage loan applications indexes combine the number of ap-

plications for the purchase of a single-family house and for refinanc-
ing single-family existing house mortgages, and separate indexes for
house purchases and refinancing. Several component indexes are pre-
pared: conventional, FHA, and VA mortgages are combined into
Government indexes, fixed-rate, adjustable-rate, and indexes for
balloon mortgages, and more detailed cross classifications of the
above types.
The mortgage applications data comprise mortgages originating with
commercial banks, thrift institutions, and mortgage banking companies.
182
MORTGAGE LOAN APPLICATIONS 183
They are estimated to account for approximately 50 percent of all first

mortgages of single family houses.
The mortgage applications data give equal weight to each mortgage
application. No distinction is made for the different dollar value of the
various applications.
The mortgage applications indexes are currently based on March 16,
1990 = 100.
The mortgage applications indexes are seasonally adjusted.
Methodology
The mortgage loan applications data are obtained from a sample of com-
mercial banks, thrift institutions, and mortgage banking companies. The
data are not based on a probability sample. In terms of the base-period
weights of the indexes, the house purchase index accounts for 85 per-
cent and the refinancing index accounts for 15 percent of total the mort-
gage applications index.
Accuracy
There are no estimates of sampling or revision error for the mortgage

loan applications indexes.
Relevance
Mortgage loan applications indexes provide a partial early indication of

home sales (house purchase index) and of retail sales (mortgage refi-
nance index). Mortgage applications movements should be compared
with other determinants of home sales and retail sales to assess the
strength of those indicators.
Recent Trends
From 1995 to 2004, there were sharp differences in the movements of

the mortgage loan applications indexes between the house purchase in-
dex and the refinancing index (Table 36.1). The house purchase index
showed a much steadier annual rate of growth, while the refinancing
index was volatile with large increases and decreases. The house pur-
chase index did not decline in any year, and except for a nominal in-
crease in 2001, it increased noticeably every other year. In contrast, the
Table 36.1
Mortgage Loan Applications Indexes (March 16, 1990 = 100)
Total applications House purchase Refinancing

1995 190.3 165.2 336.0
1996 222.1 183.3 447.6
1997 252.6 205.6 525.5
1998 482.9 266.0 1,745.2
1999 352.0 275.8 795.3
2000 322.7 302.7 438.8
2001 625.7 304.8 2,492.3
2002 799.7 354.7 3,388.0
2003 1,067.9 395.1 4,981.8
2004 736.1 454.1 2,376.4
Annual average percentage change
19952000 11.1 12.9 5.5
20002004 22.9 10.7 52.6
19952004 16.2 11.9 24.3
refinancing index rose sharply in 1998, declined sharply in 1999 and

2000, rose greatly during 200103, and declined sharply in 2004.
On an average annual basis, the house purchase index rose more during
19952000 than during 200004, while the refinancing index rose more
during 200004 than during 19952000. During 19952000, the house
purchase index rose more than the refinancing index, 12.9 percent com-
pared with 5.5 percent. During 200004, the house purchase index rose at
an average annual rate of 10.7 percent, but the extraordinarily large aver-
age annual increase in the refinancing index of 52.6 percent during 2000
04 resulted in a much greater differential than the differential by which
the house purchase index exceeded the refinancing index during 1995
2000. Thus, over the entire 19952004 period, the house purchase index
rose at an average annual rate of 11.9 percent, while the refinancing index
rose at an average annual rate of 24.3 percent.
The total applications index showed movements closer to those of
the house purchase index than to those of the refinancing index because
the house purchase index has a much greater weight than the refinanc-
ing index in the construction of the index (see Methodology).
Mortgage Bankers Association of America. Weekly Mortgage Applications Survey.

Washington, DC. Weekly.
MORTGAGE DELINQUENCY AND FORECLOSURE 185
37
Mortgage Delinquency and
Foreclosure
Mortgage debt delinquency and foreclosure data represent payment diffi-

culties of mortgages on residential properties of one to four housing units
for conventional, FHA, and VA fixed-rate and adjustable-rate mortgages.
Mortgage delinquency rates are the number of payments on mortgage
loans overdue thirty days or more as a percentage of all mortgage loans
outstanding. Mortgage foreclosure rates are the number of mortgages in
foreclosure as a percentage of all mortgage loans outstanding (foreclo-
sure terminates a mortgagors right to redeem a mortgaged property).
The mortgage delinquency and foreclosure data are provided quarterly

by the Mortgage Bankers Association of America. They are published
in the report, National Delinquency Survey (www.mbaa.org).
The data are available approximately ten weeks after the calendar
quarters ending March 31, June 30, September 30, and December 31.
For example, data for the JanuaryMarch quarter are published about
the second week of June. The delinquency and foreclosure data are oc-
casionally revised.
Content
The mortgage delinquency (MD) data show delinquency rates as of the

end of March 31, June 30, September 30, and December 31 for the na-
tion, regions, and states. MD rates are provided separately for payments
overdue 30 days, 60 days, and 90 days or more. They show the totals of
185
all loans, prime loans, and subprime loans, and within these categories
fixed-rate loans and adjustable-rate loans. Additional MD data are pro-
vided for FHA and VA loans.
The national, regional, and state mortgage foreclosure (MF) data show
foreclosure rates for foreclosures started during the quarter, and the in-
ventory of all mortgages in the process of foreclosure as of March 31,
June 30, September 30, and December 31. Foreclosures started during
the quarter include mortgages placed in the process of foreclosure, the
voluntary relinquishment of the deed in lieu of proceeding with the fore-
closure process, and mortgages assigned to FHA, VA, and other insurers
or investors during the quarter. The inventory of all foreclosures at the
end of the quarter covers all mortgages in the process of foreclosure.
Foreclosures are excluded from the MD data.
The national and regional MD data are seasonally adjusted. The state
MD data are not seasonally adjusted.
The national and regional MF data for foreclosures started during the
quarter are seasonally adjusted. MF data for the inventory of foreclo-
sures and state MF data are not seasonally adjusted.
Methodology
The mortgage delinquency and mortgage foreclosure data are obtained

from a quarterly survey of approximately 38 million mortgage loans
from a sample of more than 140 financial institutions that provide mort-
gage servicesmortgage bankers, commercial banks, savings and loan
associations, mutual savings banks, life insurance companies, and credit
unions. Of these, prime and subprime conventional loans account for
approximately 86 percent, FHA loans for 10 percent, and VA loans for 4
percent. The mortgage loans provided by this sample of financial insti-
tutions account for approximately 80 percent of all mortgage loans ex-
tended. The survey sample is not a probability sample.
The MD and MF data are developed from the sample of outstanding
loans. Thus, the sample is updated every quarter for the addition of new
loans and the deletion of repaid loans.
Accuracy
There are no estimates of sampling or revision error for the mortgage

delinquency and mortgage foreclosure data.
MORTGAGE DELINQUENCY AND FORECLOSURE 187
Table 37.1
Mortgage Delinquencies and Foreclosures
Mortgage delinquencies
(percent of loans with Mortgage foreclosures
installments past due) (percentage of loans in foreclosure)
1995 4.24 0.88
1996 4.33 0.99
1997 4.30 1.09
1998 4.45 1.16
1999 4.26 1.17
2000 4.38 1.11
2001 5.11 1.33
2002 5.11 1.48
2003 4.74 1.33
2004 4.35 1.17
Relevance
Analytic interest in the mortgage delinquency and mortgage foreclosure data

is associated with their effect on home sales, housing starts, the housing
affordability index, the gross domestic product (residential investment and
consumer expenditure components), and interest rates (new home mort-
gage yields). These effects stem from the behavior of mortgage lenders.
When MD and MF rates rise, lenders tend to raise interest rates and eligibil-
ity standards for mortgage applicants, if the demand for mortgage loans
sustains the higher rates. When MD and MF rates decline, the reverse occurs
with lower interest rates and less stringent eligibility standards. The thrust is
to balance the generation of interest income revenues with the risk of losses
on unpaid loans for their net effect on the profits of mortgage lenders.
Recent Trends
From 1995 to 2004, the mortgage delinquency rate rose from 4.24 per-
cent in 1995 to highs of 5.11 percent in 2001 and 2002, and then fell to
4.35 percent in 2004. The mortgage foreclosure rate rose from 0.88 per-
cent in 1995 to a high of 1.48 percent in 2002, and then fell to 1.17 in
2004. (See Table 37.1.)
Mortgage Bankers Association of America. National Delinquency Survey. Wash-

ington, DC. Quarterly.
38
Non-Manufacturing Business
Activity Index
The non-manufacturing business activity index (NMBAI) represents

various aspects of the health of goods and services industries except
manufacturing. The business activity index is constructed as a diffusion
index, which suggests an indication of the direction of the movement
from one period to the next, but not of the size of the movement. The
PMI is a counterpart diffusion index for manufacturing industries.
The non-manufacturing business activity index is prepared monthly by

the Institute for Supply Management. It is published in a news release
and in the ISM Non-Manufacturing Report on Business (www.ism.ws).
The data are available the third business day of the month after the
month to which they refer. The index is not revised.
Content
The non-manufacturing business activity index represents sales, receipts,

or revenues of selected industries within agriculture, mining, construc-
tion, trade, transportation, communications, utilities, finance, and services.
It is one of several diffusion indexes prepared for non-manufacturing in-
dustries. The others are new orders, backlog of orders, inventory change,
inventory sentiment, new export orders, imports, prices, supplier deliver-
ies, and employment. Because these indexes were first introduced in 1997,
several more years of data are required in order to prepare a composite of
some or all of the indexes. The non-manufacturing business activity index
188
NON-MANUFACTURING BUSINESS ACTIVITY INDEX 189
is featured here because it is more representative of overall economic ac-

tivity than the other non-manufacturing indexes. By contrast, the PMI for
manufacturing is based on several years of data from which a composite
index has been developed.
The non-manufacturing business activity index is seasonally adjusted.
Methodology
The non-manufacturing business activity index is prepared from a

monthly sample of more than 370 companies in 17 non-manufacturing
industries that are members of the Institute for Supply Management.
The number of companies reporting for each industry reflects the pro-
portion of the non-manufacturing gross domestic product accounted for
by the industry.
The survey obtains data on directional movements of each item in
comparison with its level in the previous month. Depending on the item
surveyed, positive responses are designated higher, better, or faster
than last month; negative responses are designated lower, worse, or
slower than last month; and no-change responses are designated the
same as last month. Because the responses are supplied by the third
week of the month, the survey is based on only partial information for
which the entire month is estimated.
The index numbers for the composite and for each component are the
proportion of the surveyed companies that report a positive change in
activity from the previous month; this includes one-half of the compa-
nies reporting no change. For example, if 56 percent report a positive
change, 10 percent report no change, and 34 percent report a negative
change, the index is 61 (i.e., 56 plus one-half of 10).
This type of index, which suggests an indication of the direction
but not of the size of the movement from one period to the next, is a
diffusion index (diffusion indexes are discussed further in the Con-
tent section of the chapter on PMI). In contrast, traditional indexes
of economic activity provide the actual direction and magnitude of
the movement.
Accuracy
There are no estimates of sampling or revision error for the non-manu-

facturing business activity index.
Table 38.1
Non-Manufacturing Business Activity Index (seasonally adjusted)
2003 2004
January 53.4 January 64.2
February 54.3 February 60.8
March 47.9 March 64.0
April 52.6 April 66.9
May 54.7 May 63.3
June 60.3 June 61.1
July 63.1 July 63.4
August 64.9 August 59.3
September 63.4 September 58.7
October 65.0 October 61.5
November 60.2 November 61.9
December 58.9 December 63.9
Relevance
Non-manufacturing business activity accounts for the major part of the

economy. But there is insufficient experience with the index to develop
relationships between it and the overall economy.
Recent Trends
From 2003 to 2004, the non-manufacturing business activity index rose

from lower levels during JanuaryMay 2003 to similar higher levels
from June 2003 to December 2004 (Table 38.1). There were several
monthly changes in direction both from the low of 47.9 in March 2003
to the high of 66.9 in April 2004.
Institute for Supply Management. ISM Non-Manufacturing Report on Business.

Tempe, AZ. Monthly.
PMI 191
39
PMI
The PMI (formerly the Purchasing Managers Index) reflects several

aspects of the direction of economic activity in manufacturing indus-
tries. It is constructed as a diffusion index, which suggests an indication
of the direction of the movement from one period to the next, but not of
the size of the movement. The non-manufacturing business activity in-
dex is a counterpart diffusion index for non-manufacturing industries.
The PMI is prepared by the Institute for Supply Management. It is pub-

lished in a news release and in the ISM Manufacturing Report on Busi-
ness (www.ism.ws).
The data are available on the first business day of the month after the
month to which they refer. The seasonal factors of the index are updated
by the U.S. Department of Commerce; they are revised for the previous
four years every January. No other revisions are made to the index.
Content
The PMI is a composite of five indexes of manufacturing activity: new

orders (manufacturers orders), production (industrial production index),
employment (employment), the promptness of manufacturers deliver-
ies (the difference between the time a purchased item is scheduled to
arrive and the time when it actually arrives), and total purchased inven-
tories. The index numbers for the composite and for each component
are the proportion of the surveyed companies that report a positive change
191
in activity from the previous month. This includes one-half of the com-
panies reporting no change. For example, if 56 percent report a posi-
tive change, 10 percent report no change, and 34 percent report a negative
change, the index is 61 (i.e., 56 plus one-half of 10).
By showing the proportion of survey respondents that report rising
economic activity, a diffusion index suggests that if more respondents
follow the dominant upward or downward pattern in successive months,
it will lead to similar movements in actual sales, employment, prices,
and so on. Thus, as increasing proportions of firms report a rise or de-
cline in economic activity, similar patterns of change would be expected
in the percentage rates of growth or decline for the total of all firms.
A diffusion index suggests an indication of the direction, but not of
the size of the movement from one period to the next. This contrasts
with traditional indexes of economic activity that provide the actual di-
rection and magnitude of the movement.
The index weights all surveyed firms equally regardless of size (or
for diffusion indexes of surveyed individuals, all are weighted equally
regardless of income). It indicates the pervasiveness of increases and
decreases among a surveyed population. Use of a diffusion index as-
sumes that the cyclical movements of small and large firms are similar
in terms of their percentage rates of growth and decline, although the
timing of the cyclical movements varies among the firms. This implies a
direct relationship between the proportion of firms reporting activity
moving in a particular direction and changes in the magnitude of the
rate of growth or decline.
A diffusion index of 50 occurs when equal numbers of firms have
increases and decreases, and therefore the rate of growth from the previ-
ous period is approximately zero. When the index is above 50 the rate of
growth tends to be positive, and when the index is below 50 the rate of
growth tends to be negative. In addition, the farther the index rises above
50, the greater is the magnitude of growth, while the farther the index
falls below 50, the greater is the magnitude of decline.
The PMI is seasonally adjusted based on separate seasonal adjust-
ments for each component.
Methodology
The PMI is a composite formed from the five component indexes that
reflect various aspects of operations of manufacturing companies. The
PMI 193
component weights are assigned judgmentally based on qualitative as-

sessments of their importance. The components and their weights are:
New orders 0.30

Industrial production 0.25
Employment 0.20
Delivery schedules 0.15
Inventories 0.10

Total 1.00
The data for each component are obtained monthly from a survey of
a sample of more than 350 manufacturing companies in 20 broad manu-
facturing industry groups that are members of the Institute for Supply
Management. The number of companies included in the sample for each
manufacturing industry is based on the proportion of the gross domestic
product accounted for by each industry. The industries represent food,
chemicals, machinery, and other broad manufacturing groups, includ-
ing finer industry classifications within each broad group such as dairy,
meat, and grains within the food group. Each company is weighted
equally regardless of the size of the firm.
The survey obtains data on directional movements of each item in
comparison to its level in the previous month. Depending on the item
surveyed, positive responses are designated higher, better, or faster
than last month; negative responses are designated lower, worse, or
slower than last month; and no-change responses are designated the
same as last month. Because the responses are supplied by the third
week of the month, the survey is based on only partial information for
which the entire month is estimated.
Accuracy
There are no estimates of sampling or revision error for the PMI.
Relevance
For the overall economy as measured by the real gross domestic product
(GDP), the threshold of the PMI is approximately 43. Thus, a PMI above
43 suggests an expanding GDP, while a PMI below 43 suggests a de-
Table 39.1
PMI (seasonally adjusted)
2003 2004
January 52.8 January 62.8
February 49.9 February 62.1
March 46.4 March 62.3
April 46.1 April 62.3
May 49.8 May 62.6
June 50.4 June 61.2
July 52.5 July 61.6
August 55.6 August 59.6
September 55.1 September 59.1
October 57.7 October 57.5
November 61.3 November 57.6
December 62.1 December 57.3
clining GDP. The figure of 43, rather than 50, reflects the tendency for
continued growth in the non-manufacturing part of the economy.
Recent Trends
From 2003 to 2004, the PMI rose from lower levels during January
July 2003 to higher levels during August 2003July 2004, and then de-
clined to slightly lower levels during AugustDecember 2004 (Table
39.1). There was only one monthly change in direction from the low of
46.1 in April 2003 to the high of 62.8 in January 2004, while a few more
occurred from January to December 2004.
Institute for Supply Management. ISM Manufacturing Report on Business. Tempe,

AZ. Monthly.
PERSONAL INCOME AND SAVING 195
40
Personal Income and Saving
Personal income mainly measures the income received by house-

holds from employment, self-employment, investments, and trans-
fer payments. It also includes small amounts for expenses of nonprofit
organizations and income of certain fiduciary activities. Disposable
personal income (DPI) refers to personal income after the payment
of income, estate, and certain other taxes and payments to govern-
ments. Personal saving is the residual of DPI minus consumer out-
lays, and the saving rate is saving as a percentage of DPI. The personal
income and saving measures are definitionally consistent with those
for the gross domestic product.
The personal income and saving measures are prepared monthly by the
Bureau of Economic Analysis (BEA) in the U.S. Department of Com-
merce. The data are published in a monthly news release and in the BEA
The personal income, disposable personal income, and personal
saving data are available during the fourth week of the month after
the month to which they refer. The data are revised initially in the
subsequent two months. Subsequent revisions are made annually
every summer, and in the subsequent benchmark revisions of the
national income and product accounts (see the gross domestic prod-
uct). Data on the saving rate are available one month after the above-
noted dollar measures.
195
Content
Personal income (PI) mainly measures income of households. Household

income is derived from wages and salaries; fringe benefits; profits from
self-employment; rental profits from real property (for persons not prima-
rily engaged in real estate), plus patent, copyright, and natural resource
royalties; noncash rent imputed from owner-occupied homes; interest;
dividends; Social Security and unemployment insurance benefits, food
stamps, and other income maintenance programs. PI also includes operat-
ing expenses (excluding depreciation) of nonprofit organizations that pri-
marily serve individuals, and investment income of life insurance
companies, private noninsured welfare funds, nonprofit organizations, and
private trust funds. PI excludes income changes from capital gains and
losses associated with the difference between the purchase and sales price
of financial assets, real property, and personal property.
PI reflects income flows before the payment of income, estate, gift,
and personal property taxes plus fees, fines, and penalties paid to fed-
eral, state, and local governments. Social Security taxes paid by em-
ployees and employers are excluded from PI.
Disposable personal income (DPI) is income excluding the tax and
nontax payments to governments included in PI. DPI is provided in cur-
rent and constant dollars.
Personal saving is the income remaining from DPI after deductions
for consumer spending for goods and services, interest payments on
consumer loans (excluding home mortgage interest), and money sent as
gifts abroad (net transfer payments to foreigners). This is part of the
national income and product accounts (see gross domestic product). The
total of consumer spending, interest payments, and foreign gifts is re-
ferred to as personal outlays. The saving rate is saving as a percentage
of DPI. Saving is not affected by gifts between households (such as
when parents give a house to their children) and by sales of homes, cars,
and other assets between households (except for payments to intermedi-
aries such as brokers commissions and used car dealer markups).
The personal income and saving data are seasonally adjusted.
(An alternative measure of saving, referred to as household saving, as
distinct from personal saving described above, is provided by the Fed-
eral Reserve Board as part of its flow-of-funds accounts. I note the house-
hold saving measure to alert the reader to its existence, though it is not
included in the book. The household saving measure is based on the
change in net worth of financial and tangible assets of householdsthat

is, assets less liabilities. This differs from the personal saving measure
that is part of the national income and product accounts, which is ob-
tained as the difference between income and outlays in each period.
Other differences between the two saving measures include, but are not
limited to, capital gains and losses and changes in the net worth of unin-
corporated businesses, which are included in the household saving but
not in the personal saving measure. Differences in the uses of the two
measures include: personal saving seems more relevant for assessing
the contribution of personal saving to national saving, while household
saving seems more relevant for assessing whether households in the
aggregate are preparing adequately for financial needs of retirement.1)
Methodology
Data for the components of personal income are obtained from several
government and nongovernment sources with varying degrees of cur-
rency. For example, wages are based on the monthly employment pay-
roll survey, and Social Security and unemployment insurance benefit
payments are based on monthly reports from the Social Security Ad-
ministration and the U.S. Department of Labor. Stock dividend income
is derived from the U.S. Census Bureaus Quarterly Financial Report
and from corporate quarterly reports to stockholders (these quarterly
data are interpolated through the three months of the quarter in estimat-
ing the monthly figures). Data for other PI components, such as income
from fringe benefits, self-employment, rent, interest, and life insurance
benefits, typically are available only annually, and the monthly histori-
cal and current data are mainly estimated indirectly.
Disposable personal income is calculated by subtracting income, es-
tate, gift, and personal property taxes plus miscellaneous fines, fees,
and penalties from PI. Data for these deductions are obtained from two
sources. The U.S. Department of the Treasury provides monthly data
for the federal component, and the Census Bureau provides a quarterly
survey of state and local governments. However, the state and local sur-
vey data are used only in the historical quarterly data because they are
available too late for the current data every quarter; indirect estimates
are made for the current data. DPI in chained dollars is calculated by
dividing DPI in current dollars by the chain-type price index for per-
sonal consumption expenditures (see GDP price measures).
Table 40.1
Personal Income and Saving (billions of dollars)
Personal Disposable Saving rate

income personal income (percentage)
1995 6,152.3 5,408.2 4.6
1996 6,520.6 5,688.5 4.0
1997 6,915.1 5,988.8 3.6
1998 7,423.0 6,395.9 4.3
1999 7,802.4 6,695.0 2.4
2000 8,429.7 7,194.0 2.3
2001 8,724.1 7,486.8 1.8
2002 8,881.9 7,830.1 2.4
2003 9,169.1 8,169.2 2.1
2004 9,713.3 8,664.2 1.8
Personal saving represents the difference between DPI and the sum
of consumer spending for goods and services, interest payments on con-
sumer loans, and net transfers to foreigners (referred to as personal out-
lays). The saving rate for each month is calculated as a three-month
moving average of saving as a percentage of DPI in order to dampen
erratic month-to-month movements.
Accuracy
Revisions to the monthly estimates of personal income indicate that the

first three estimates for each month provide the correct indication of the
direction of change in 90 percent of the cases.
Relevance
Personal income is the main component of consumer purchasing

power, and thus has a prime influence on consumer spending. PI is
supplemented by consumer credit as a source of financing for con-
sumer spending. Consumer spending accounts for approximately 70
percent of the gross domestic product (GDP) and figures prominently
in retail sales. Consequently, PI has a major effect on overall eco-
nomic activity and employment. Disposable personal income in
chained dollars (adjusted for inflation) provides a better analytic
measure of consumer purchasing power and its effect on real GDP
than current-dollar personal income. Because personal saving indi-
cates consumers willingness to spend, the saving rate is an element

in predicting future spending trends.
Recent Trends
From 1995 to 2004, personal income and disposable personal income

increased in all years (Table 40.1). The personal saving rate declined
from 4.6 percent in 1995 to 1.8 percent in 2004, with the general level
declining from the 4-percent range during 199598 to the 2 percent range
from 1999 to 2004.

Business. Monthly.
Note
1. Maria G. Perozek and Marshall B. Reinsdorf, Alternative Measures of Per-

sonal Saving, Survey of Current Business, April 2002, pp. 1314.
41
Poverty
Measures of poverty count the number of all people (people in families

plus unrelated individuals), families, and unrelated individuals with in-
comes below a specified minimum level. This income threshold defines
subsistence living conditions according to societal standards as they
existed in the 1960s, adjusted for changes in the cost of living since
then. Incomes below the threshold are regarded as subjecting the recipi-
ents to living conditions below currently accepted standards of decency.
At any point in time, poverty is defined at a specific threshold of income
inadequacy (how much is too little). The threshold is typically fixed
at this level for a period of time. However, what is regarded as poverty
also has an evolving dimension over time. Evolving standards typically
are raised over time to reflect improved living conditions and higher
aspirations afforded by advancements in technology. When the standard
is raised, the number of persons defined as living in poverty increases.
There is one official measure of poverty. It counts cash income only, as
distinct from noncash benefits such as Medicare; cash income reflects the
income before the payment of certain taxes and other items. Sixteen alter-
native measures now published are based mainly on various treatments of
cash and noncash income maintenance payments, taxes, and capital gains.

pares annual measures of poverty. The data are published annually in
Income, Poverty, and Health Insurance Coverage in the United States
(www.census.gov).
200
POVERTY 201
The data are available in the fall after the year to which they refer.
Revisions for previous years are made in the annual publications.
Content
The poverty data indicate the number of persons and families with money
incomes below the poverty threshold based on the number of persons in
the family and their ages. The threshold has two components. One is a
standard of inadequacy for food, housing, and other living conditions.
The other is the income associated with the standard.
The official poverty threshold follows the U.S. Office of Manage-
ment and Budgets Statistical Policy Directive No. 14. It is based on the
poverty standard developed in 1963. Refinements made to the threshold
since 1963 have not significantly changed the official count of the pov-
erty population.1 The threshold measures households money income
before the payment of such items as federal and state income taxes,
Social Security taxes, federal employee retirement taxes, property taxes,
Medicare deductions, and union dues. Money income is defined as regu-
larly received cash income, such as wages and salaries, profits from
self-employment, Social Security, retirement, unemployment insurance,
other income maintenance benefits, interest, dividends, rents, royalties,
estates and trusts, educational assistance, alimony, child support, and
financial assistance from outside the household (excluding gifts and
sporadic assistance). Noncash benefits, such as food stamps, Medicare,
Medicaid, and rent supplements, as well as income from nonrecurring
sources such as capital gains and life insurance settlements, are excluded
from money income.
A household consists of all persons who occupy a housing unit. A
household may include one or more families and/or one or more unre-
lated individuals. A family refers to two or more persons related by birth,
marriage, or adoption and living together in a house, apartment, a group
of rooms, or a single room intended as separate living quarters. An unre-
lated individual is a person fifteen years and older who does not live
with any relatives. A housing unit has direct access from the outside or
through a common hall. The occupants of a housing unit do not live or
eat with any other people in the structure. The definition of households
excludes people living in group quarters (e.g., hotels, dormitories), or
institutions (e.g., hospitals, jails, shelters), or having no residence (people
living on the street).
The members of a household may or may not share their incomes for
personal consumption. Consequently, the household is not a perfect unit
for measuring poverty. But the household is a practical device for mea-
suring the incomes of people living in housing units.
The poverty standard used in the first decade of the twenty-first cen-
tury reflects the same minimum living conditions specified when the
standard was developed in 1963. In order to maintain the 1963 income
threshold of living conditions, it is routinely updated only for inflation.
For example, the annual threshold income before taxes for a four-person
family rose from $3,128 in 1963 to $18,810 in 2003. (There was no
official poverty standard before the 1960s; as noted below under Rel-
evance, there was an implied standard in the 1930s.)
Sixteen alternative nonofficial poverty measures that use different
definitions of household income are also provided by the Census Bu-
reau. The definitions vary in terms of the treatment of government cash
transfer payments, noncash benefits, federal and state income taxes,
Social Security payroll taxes, capital gains, and an imputed return on
the equity of home ownership. In 2003, under the official measure, 12.5
percent of the population was defined as being in poverty, with the alter-
native measures ranging from 9.0 to 20.6 percent of the population.
Methodology
The poverty standard is based on estimates made by the U.S. Depart-

ment of Agriculture in 1961 of the least expensive of four food plans to
meet nutritional adequacy standards. The food plan assumed that all
meals are prepared at home from foods purchased at retail. This number
was multiplied by three to determine total income necessary to meet all
living expenses, including housing, health, transportation, and other
nonfood items, except medical expenditures. This multiplier is based
on a 1955 study indicating that food accounts for one-third of the aver-
age household budget for families of three or more persons at all in-
come levels, not just low-income families. Other procedures were used
to calculate thresholds for one- and two-person units. Thus, the nonfood
components are estimated indirectly as a statistical aggregate rather than
by estimating each living expense component separately with specific
minimum standards for each (e.g., housing, clothing, transportation).
The money income data are based on the Current Population Survey
(CPS) conducted by the Census Bureau. The information is collected
POVERTY 203
every March in an income supplement for the previous year. The survey
sample is approximately 60,000 households, of whom about 55,500 are
interviewed and 4,500 are not available for interviews. For additional
detail on the CPS, see employment.
Accuracy
The sampling error (for 1.6 standard errors) of the people in poverty as
a percentage of the total population is 0.2 of a percentage point. For
example, if the estimated poverty population were 13 percent, in nine of
ten cases the true rate would be somewhere between 12.8 and 13.2
percent. For further information on the interpretation of sampling and
nonsampling errors, see the Appendix.
A comparison of the household income estimates of the Census Bu-
reau with the personal income estimates of the U.S. Bureau of Eco-
nomic Analysis (see personal income and saving) for the year 2001,
after placing both income estimates on a comparable definitional basis,
indicated that total household income was 11 percent below total per-
sonal income.2 The difference is attributed to underreporting by survey
respondents on the CPS, which is the source of household income data,
as contrasted with administrative records of income tax, unemployment
insurance, Social Security, and other income programs that are the source
of the personal income data. This overall underreporting is not taken
into account in developing the poverty data because determining the
variations in underreporting among income groups is difficult.
Relevance
The poverty measure reflects societal concerns about how well the na-
tion is providing for the minimal subsistence needs of the people at the
bottom of the income ladder. While it does not reflect the social stan-
dards of the 2000s, it focuses attention on the most needy in the popula-
tion and on the progress made in alleviating their condition. As an absolute
number, it quantifies the magnitude of the poverty problem for purposes
of political debate regarding appropriate ways to deal with it.
There is extensive American historical evidence that poverty income
thresholds are raised over time as absolute poverty lines show a pattern
of rising income adjusted for inflation (real income) as the real income
of the general population rises. The publics estimate of the amount of
real income required for a family to get along also rises as the real
income of the general population rises. On the basis of that empirical
evidence, there is general agreement that poverty is an evolving concept
that changes over time to reflect the evolving living conditions and aspi-
rations of society. Thus, the poverty standard adopted in 1964 by Presi-
dent Lyndon Johnsons Council of Economic Advisers was approximately
75 percent higher in real terms than the standard implicit in President
Franklin Roosevelts 1937 statement that one-third of the nation was ill-
housed, ill-fed, and ill-clothed. One economist estimated that if the
Johnson standard of the 1960s had been used in the 1930s, Roosevelts
one-third of a nation would have been close to two-thirds.3 This illus-
trates how changing perceptions of what constitutes a minimally ac-
ceptable standard of living affect the count of persons in poverty.
These perceptions have also changed between the 1960s and the
first decade of the twenty-first century, although the current official
poverty measure has remained constant between the two periods. If
the poverty measure were raised on the basis of a political consensus
of higher minimal subsistence needs in the twenty-first century, the
number of persons defined to be in poverty would increase. The fed-
eral budget (government budgets and debt) is perceived as being a
major constraint to reviewing the poverty standard. Many fear that
even with some noncash benefits included in the income definition, a
new poverty standard would raise the poverty count, and thus increase
federal spending for social programs.
Over the years, the president ultimately has decided on the definition
of the poverty standard, as the adoption of a poverty standard has been
an executive branch function. But this does not preclude that in the fu-
ture another procedure may be used if a new poverty measure is adopted.
A 1995 report by the National Research Council (NRC, associated
with the National Academy of Sciences) recommended a methodology
for establishing a new poverty threshold for the United States.4 While
the proposed methodology has several variants, the general approach
would result in a poverty threshold that would come close to one-half of
the median income after taxes of a four-person family. The methodol-
ogy incorporates nonmoney income from government programs such as
food stamps, housing subsidies, and energy assistance; calculates the
income needs threshold using a designated percentage of median house-
hold expenditures for food, clothing, shelter, and utilities, augmented
by a factor for outlays on additional items such as household supplies
POVERTY 205
Table 41.1
Poverty Measure (official)
All people in poverty Families in poverty

Number Percentage of Number Percentage of
(millions) population (millions) families
1995 36.4 13.8 7.5 10.8
1996 36.5 13.7 7.7 11.0
1997 35.6 13.3 7.3 10.3
1998 34.5 12.7 7.2 10.0
1999 32.8 11.9 6.8 9.3
2000 31.6 11.3 6.4 8.7
2001 32.9 11.7 6.8 9.2
2002 34.6 12.1 7.2 9.6
2003 35.9 12.5 7.6 10.0
2004 37.0 12.7 7.9 10.2
and personal care, and adjusted for family size and geographic location;
supplements the income needs threshold by the value of the Earned In-
come Tax Credit (EITC); and deducts child care, work-related transpor-
tation, and medical out-of-pocket expenses. It also includes a procedure
for updating the poverty standard over time based on yearly movements
in the income needs threshold, the EITC, and the expense items. A study
by the Census Bureau estimates that using the recommended methodol-
ogy in the NRC report would raise the poverty rate in 2003 from the
official level of 12.5 percent to a range of 12.6 to 14.5 percent.5 Poverty
estimates based on the recommended methodology in the NRC report
for 2004 were not available at the time of this writing.
Poverty is related to the distribution of income and the distribution
of wealth. Inequality of income stems from many sources, such as the
match of workers having particular skills with the job market for those
skills, discrimination, inheritance, innate abilities, entrepreneurial
spirit, and luck.
Recent Trends
From 1995 to 2004, the count of all people and families living in pov-
erty showed different movements between the entire two intermediate
periods within the nine-year span (Table 41.1). The number of people in
poverty declined from 1995 to 2000, and rose from 2000 to 2004. This
resulted in the poverty population increasing from 36.4 million in 1995
to 37.0 million in 2004. The same pattern occurred for the count of
families in poverty.
In 2004, the poverty population rate was 12.7 percent and the families
in poverty rate was 10.2 percent. Over the 19952004 period, the poverty
rates showed the same movements as those for the number in poverty.
Bureau of the Census, U.S. Department of Commerce. Income, Poverty, and Health
Insurance Coverage in the United States. Annual.
Notes
1. Mollie Orshansky did the seminal work on the 1960s poverty standard that
still serves as the official standard. See Mollie Orshansky, Children of the Poor,
Social Security Bulletin (July 1963); Orshansky, Counting the Poor: Another Look
at the Poverty Profile, Social Security Bulletin (January 1965) (reprinted in the
Bulletin in October 1988). Gordon Fisher has written the definitive history of the
evolution of poverty standards: From Hunter to Orshansky: An Overview of (Unof-
ficial) Poverty Lines in the United States from 1904 to 1965, mimeo, revised 1997.
(For a condensed version, see Fisher, The Development and History of the Poverty
Thresholds, Social Security Bulletin [Winter 1992]); Fisher, The Development of
the Orshansky Poverty Thresholds and Their Subsequent History as the Official
U.S. Poverty Measure, mimeo, revised 1997; Fisher, Is There Such a Thing as an
Absolute Poverty Line Over Time? Evidence from the United States, Britain, Canada,
and Australia on the Income Elasticity of the Poverty Line, mimeo, 1995.) These
papers are not official government documents. They can be found on the Internet at
www.census.gov/hhes/poverty/povmeas/papers.html, or obtained from Fisher at the
Office of the Assistant Secretary for Planning Evaluation, U.S. Department of Health
& Human Services. Links to these and other papers on the history of poverty lines
can be found at http://aspe.hhs.gov/poverty/contacts.shtml.
2. John Ruser, Adrienne Pilot, and Charles Nelson, Alternative Measures of
Household Income: BEA Personal Income, CPS Money Income, and Beyond, May
2004. Available from the Census Bureau, www.census.gov.
3. Victor Fuchs, Toward a Theory of Poverty, in The Concept of Poverty (Wash-
ington, DC: Chamber of Commerce of the United States, 1965), p. 73.
4. Constance F. Citro and Robert T. Michael, eds., National Research Council,
Measuring Poverty: A New Approach (Washington, DC: National Academy Press, 1995).
5. Bureau of the Census, U.S. Department of Commerce, Alternative Poverty
Estimates in the United States: 2003, June 2005.
PRODUCER PRICE INDEXES 207
42
The producer price indexes (PPIs) measure the rate of price change of
domestically produced goods in the manufacturing, mining, agriculture,
fishing, forestry, and selected services industries. The PPIs exclude prices
of construction and imports, although imported goods are indirectly in-
cluded when they are components of domestically produced items. The
PPIs most often used for economic analysis are stage-of-processing in-
dexes for commodities, which are covered in this chapter. Other PPIs
are industry net-output indexes and non-stage-of-processing commod-
ity indexes.
The producer price indexes are prepared monthly by the Bureau of La-
bor Statistics (BLS) in the U.S. Department of Labor. The data are pub-
lished in a news release, the PPI Detailed Report, and in the BLS monthly
journal, Monthly Labor Review (www.bls.gov).
The data are prepared around the middle of the month following the
month to which they refer. The monthly data are revised for the preced-
ing fourth monthfor example, revisions to January data are included
with the release of the April data in May.
Content
The producer price indexes (PPIs) are price measures of U.S.-produced

commodities that represent three stages of processing in the production
of different commodities. The stage-of-processing indexes are: (a) fin-
207
ished goods; (b) intermediate materials, supplies, and components, and

(c) crude materials for further processing. In addition, the all-commodi-
ties index combines the three stage-of-processing indexes into one over-
all index. The PPIs cover commodities sold in the United States and in
foreign markets (exports).
The PPIs reflect the first sale of the goods resulting in net revenue to
the producer, and thus exclude price changes associated with resales
and markups for the same item through wholesalers, retailers, or other
producers, as well as excise taxes. They represent the actual producer
transaction price of goods meant for immediate delivery, including pre-
miums and discounts from list prices and changes in the terms of sale
such as distinctions for household and business customers and the size
of the order. Prices of items with long production lead times are based
on the time the item is delivered, not when the order is placed. The price
quote is from the site of the producer (f.o.b., free on board), unless the
price quote includes transportation charges when the producer provides
such services directly and not through an outside transportation com-
pany or contractor. Exceptions to this include the use of list prices when
transaction prices are not available, and prices quoted at central mar-
kets, particularly for farm products.
The PPIs reflect price movements for the same or similar items that
are adjusted for enhancement or reduction in the quality or quantity of
the item. Prices on futures markets are excluded.
The PPIs are currently based on 1982 = 100.
The producer price indexes are seasonally adjusted.
Stage-of-Processing Components
There are three stage-of-processing components of the commodities PPIs:

finished goods; intermediate materials, supplies, and components; and
crude materials for further processing.
Finished goods. These are items used by a household, business, gov-
ernment, or foreign buyer in the form in which they were sold, without
further fabrication. They include household goods ranging from fresh
foods to cars and capital goods such as tractors, trucks, and machine
tools.
Intermediate materials, supplies, and components. These are items
that have been fabricated but are not ready for independent use. They
become part of other products, require further fabrication, or are other-
wise used as inputs, such as cotton yarns, chemicals, containers, office

supplies, electric power, and internal combustion engines.
Crude materials for further processing. These items are not sold di-
rectly to households. They are either sold for the first time in their initial
state of production, such as livestock or crude petroleum, or are reused,
such as scrap metal.
The three indexes reflect a theoretical typology of goods based on
production, moving sequentially from a products initial state to its end
result. Classifying products by end user (household, business, govern-
ment, or foreigner) and by degree of fabrication is referred to as stage-
of-processing classification. Although the stage-of-processing concept
theoretically represents a step-by-step flow from crude to intermediate
to finished products, in practice this flow does not always occur. For
example, there are reverse flows of intermediate containers to crude
materials and of finished equipment to both intermediate and crude
material groups; there are also products that skip the intermediate stage,
such as when crude live poultry becomes finished processed poultry.
Methodology
Monthly price data for the producer price indexes are obtained from a
mail survey conducted by the Bureau of Labor Statistics. The survey
samples prices for more than 3,200 commodities. These data are supple-
mented by price information provided by the U.S. Department of Agri-
culture for farm products. Nearly all price quotes are reported by the
sellers rather than the buyers. Prices of most commodities are obtained
between the ninth and fifteenth days of the month for which the PPIs are
calculated. Examples of exceptions are compact disks and audiotapes,
whose prices are obtained one month later.
The weights for the stage-of-processing and the all-commodities PPIs
are based on the value of sales of the component commodities and in-
dustries. These reflect data in the five-year economic censuses, unless
the industry is not covered in the censuses. For example, sales weights
for electric power are based on U.S. Department of Energy data. The
weights are currently based on 1997 sales volumes. The structuring of
commodities in a stage-of-processing chain is based on the commodity
ordering in the inputoutput tables prepared by the Bureau of Economic
Analysis in the U.S. Department of Commerce. The weighting structure
is updated approximately every five years.
If the reported monthly price includes a change in the quality of the

item, an adjustment is made to reflect the improvement or decline. Thus,
the PPIs aim at measuring price movements of items having the same
functional characteristics over time. For example, if better brakes are
included in a car, the price increase attributable to the improved brakes
does not appear as a price increase in the PPI, but if an auto bumper is
weakened because of relaxed standards and there is no change in market
price, the weaker auto bumper is considered a price increase in the PPI.
Because product cost data required to make the necessary adjustments
are not always available, the PPI contains an unknown amount of price
change caused by quality and quantity changes.
One approach for adjusting for quality change in price indexes when
product cost data are inadequate or not available is through hedonic
price indexes. Hedonic indexes are based on statistical regressions that
reflect historical relationships between quality change and price change.
Hedonic indexes require vast amounts of data for their calculation. They
are currently used for computers and in peripherals in the PPIs. For a
brief further description of hedonic price indexes, see the Methodol-
ogy section of the consumer price index.
Accuracy
There are no estimates of sampling or revision error for the producer

price indexes.
Relevance
Producer price indexes provide a basis for analyzing whether infla-

tion is caused by burgeoning demand or supply bottlenecks. One
method is by identifying the stages of processing in which sharp price
increases occur.
The PPIs theoretically help predict potential price changes in the se-
quential development of crude materials to intermediate materials to
finished goods. However, the stage-of-processing concept does not es-
tablish a complete unidirectional flow of materials in the production
process from crude to intermediate to finished goods. While, theoreti-
cally, crude prices should predict intermediate prices, and intermediate
prices should predict finished prices in the commodities indexes, these
leadlag relationships are not exact because of the failure to maintain a
Table 42.1

(percentage change)
Intermediate Crude Finished

materials, materials for goods less
Finished supplies, and further foods and
goods components processing energy
1995 1.9 5.4 0.9 2.1
1996 2.7 0.6 10.8 1.4
1997 0.4 0.1 2.4 0.3
1998 0.8 2.1 12.9 0.9
1999 1.8 0.2 1.4 1.7
2000 3.8 4.9 22.8 1.3
2001 2.0 0.4 0.3 1.4
2002 1.3 1.5 10.7 0.1
2003 3.2 4.6 25.2 0.2
2004 3.6 6.6 17.5 1.5
19952004 1.7 1.5 5.0 1.2
(annual
average)
unidirectional flow in the stage-of-processing groups. More research on

the properties of these classifications is needed to realize their analyti-
cal potential in the transmission of price change.1
The PPIs are also used to determine cost escalation in business con-
tracts and to deflate the gross domestic product to constant dollars.
Supplementary PPI measures are published that exclude price move-
ments of food and energy products. Because these products sometimes
have volatile price movements that are unrelated to underlying infla-
tionary or deflationary forces in the economy, their exclusion provides
the core inflation rate, which is also referred to as the underlying rate
of inflation.
Recent Trends
From 1995 to 2004, producer price indexes for finished goods and for
finished goods less foods and energy showed the least year-to-year vola-
tility of all the PPIs (Table 42.1). And prices of finished goods less foods
and energy were less volatile than those of finished goods.
Prices of crude materials were much more volatile and increased far
more than those of finished goods and of intermediate materials, sup-
plies, and components. Some of these differences result from fluctua-
tions in oil prices, which have their initial and heaviest impact on crude
materials. The effect of price changes of crude oil is successively damp-
ened as an increasing number of nonpetroleum products are incorpo-
rated in intermediate products and in finished goods.
Bureau of Labor Statistics, U.S. Department of Labor. PPI Detailed Report and
Monthly Labor Review. Monthly.
Note
1. Tae-Hwy Lee and Stuart Scott, Investigating Inflation Transmission by Stages

of Processing, in Cointegration, Causality, and Forecasting: A Festschrift in Honor
of Clive W.J. Granger, ed. Ralph Engle and Halbert White (Oxford, UK: Oxford
University Press, 1999).
PRODUCTIVITY: BUSINESS SECTOR 213
43
Productivity: Business Sector
Productivity represents the nations efficiency in producing goods and

services. There are two economywide productivity measures, labor-hour
productivity and multifactor productivity. Labor-hour productivity, which
is the traditional measure, encompasses the aggregate effects of em-
ployee schooling, experience, and worker skills; equipment, structures,
and other capital services used in production; worker effort; technol-
ogy; and all other factors. Multifactor productivity abstracts from em-
ployee schooling, experience, and worker skills and the use of capital
services, thus focusing on the aggregate effects of worker effort, tech-
nology, and all other factors.
Productivity is calculated as the ratio of output (numerator) to input
(denominator), with the quotient being output per unit of input. When
output increases more or decreases less than input, productivity rises;
and when output increases less or decreases more than input, produc-
tivity declines. The two measures covered here, labor-hour productiv-
ity and multifactor productivity, have the same output measure, with
the small exception of the exclusion of government enterprises for
multifactor productivity. Therefore, the different input measures ac-
count for practically all of the differences in productivity between the
two measures.
Separate measures of labor-hour productivity and multifactor pro-
ductivity are prepared for the business sector and the nonfarm busi-
ness sector. The business sectors are derived from the gross domestic
product (GDP) definitions. Business sectors exclude not-for-profit
organizations, household output, rental value of owner-occupied hous-
213
ing, and general government. The nonfarm business sector also ex-
cludes farming.
The productivity output definitions for the business sectors are based
on the value-added definition of the GDP. This method counts output
as the sum of the incomes associated with employee compensation,
business profits, interest payments, depreciation allowances, and taxes
on production and imports. It excludes purchased materials and ser-
vices used in production, and thus prevents the duplication that would
occur by double counting the value of the purchased materials and
services in all stages of production. The productivity measures are
expressed as an index.
This chapter covers labor-hour productivity and multifactor produc-
tivity separately, except that the categories of Relevance and Recent
Trends are discussed jointly at the end of the chapter for both productiv-
ity measures.
Labor-Hour Productivity
Labor-hour productivity data are prepared quarterly by the Bureau of

published in a news release and in the BLS monthly journal, Monthly
Labor Review (www.bls.gov).
Preliminary data are provided in the second month following the quar-
ter to which the data refer (May for the first quarter, August for the
second quarter, and so on). The data are available at the same time as the
unit labor costs data, and follow soon after publication of the gross do-
mestic product measures. The data are initially revised in the subse-
quent month, and then two months later along with the release of the
preliminary data for the following quarter. Annual revisions are made
every year as more data become available.
Content
Labor-hour productivity is defined as output per hour of labor ex-

pended. Output per hour encompasses the combined influences of
all factors affecting the use of labor, such as the services derived
from capital equipment and structures, the substitution of capital
services for labor, worker skills and effort, executive direction and
managerial skills, technology, level of output, capacity utilization,
energy consumption, materials quality, public sector infrastructure,
and the interactions among them.
Labor-hour productivity includes the quantity of labor hours
worked as inputs, and thus eliminates the effect on output of these
worker inputs. Labor-hour productivity does not separate the spe-
cific contributions to productivity of worker schooling, experience,
skills, and effort, capital quantity and quality inputs, and all other
inputs contributing to output. Consequently, it reflects the joint ef-
fects of all inputs, except the quantity of labor hours worked, includ-
ing the interactions among them.
The labor-hour productivity indexes are currently based on 1992 = 100.
The labor-hour productivity data are seasonally adjusted.
Methodology
This section concentrates on the inputs of the productivity ratio, be-

cause the outputs follow the definitions of the gross domestic product
business sector, as noted at the beginning of the chapter.
Labor-hour productivity inputs are represented by labor hours
worked. Labor hours worked are the product of employment multi-
plied by average weekly hours, converted to average annual hours.
The employment data are based on the sum of paid jobs counted in the
establishment survey and the number of self-employed and unpaid fam-
ily workers counted in the household survey (see employment).
Labor hours obtained from the establishment survey are based on
hours at work (called hours worked), which is limited to time at the
job site, including paid time to travel between job sites, coffee breaks,
and machine downtime. It excludes time associated with paid vacation
and sick leave. By contrast, the sum of hours worked and of time associ-
ated with paid vacation and sick leave is called hours paid. The data
on hours for the self-employed and family workers are less clear, and so
their labor input can be described as either hours worked or hours paid.
Beginning in 2005, labor hours data for the nonfarm self-employed,
nonfarm unpaid family workers, and farm workers were refined to bet-
ter account for the hours worked by multiple jobholders in their primary
as distinct from their secondary jobs.
Mathematically, labor-hour productivity is expressed as follows:
Labor-hour Real
Real (43.1)
Labor-hour Output Real business GDP* business
productivity =
productivity
Input
= =
Labor hours
business
GDP per
per
GDP
labor hour
labor hour
Accuracy
In nineteen of twenty cases, labor-hour productivity in the second quar-

terly revision (four months after the preliminary data) differs from the
preliminary index by 1.4 to 1.4 index points.
Multifactor Productivity
Multifactor productivity data are prepared annually by the Bureau of

Labor Statistics in the U.S. Department of Labor. The data are pub-
lished in a news release, Multifactor Productivity Trends
(www.bls.gov).
The data are available about six months after the end of the year to
which they refer.1 Revisions are made with each annual release, based
on new data or improved methodology for all earlier years and thus
provide a consistent long-term time series.
Content
Multifactor productivity is defined as output per unit of labor and

capital combined. It excludes the inputs of labor hours worked; and
labor-hour inputs adjusted for worker schooling, experience, and
skills, and for the capital services inputs of equipment, structures,
land, and inventories. Therefore, the resultant productivity measure
centers on worker skills and effort, executive direction and manage-
rial skills, technology, level of output, capacity utilization, energy
*Gross domestic product, excluding households, not-for-profit organizations,

rental value of owner-occupied housing, and general government (adjusted for price
change).
Hours worked by paid employees, the self-employed, and unpaid family workers.
consumption, materials quality, public sector infrastructure, all other

factors, and the interactions among them. By excluding the labor-
hour and capital services inputs, multifactor productivity also ex-
cludes the effect on output of the substitution of capital services for
labor.
The multifactor productivity indexes are currently based on 1996 =
100.
Methodology
This section on multifactor productivity concentrates on the inputs of

the productivity ratio, because the outputs follow the definitions of the
gross domestic product business sector, as noted at the beginning of the
chapter. The one divergence of the output measure of multifactor pro-
ductivity from that in labor-hour productivity is that multifactor produc-
tivity excludes government enterprises (government enterprises are
financed with user fees, in contrast to general government, which is
financed with taxes). Thus, multifactor productivity is referred to as the
private business sector, as distinct from labor hour productivity, which
is referred to as the business sector.
Multifactor productivity inputs are the weighted average of the
dollar value of labor hours and of capital services. The labor-hour
inputs begin with those for labor-hour productivity described above.
These are modified by giving greater weight to workers with more
schooling, experience, and skills to reflect differences in the capa-
bilities between workers. The assumption is that as the workforce is
composed of an increasing proportion of workers with a greater
amount of schooling, experience, and skills, the workforce itself be-
comes more productive.
The capital inputs represent the services that flow from the stocks
of capital. They include the rental value of the services of the capital
facilities used in the production of goods and services, adjusted for
price changeequipment, structures, land, and inventories. Modifi-
cations for the composition of equipment and structures facilities give
more weight to short-lived equipment items than to long-lived ones,
because per dollar short-lived assets provide more services per year
than long-lived assets.
Mathematically, multifactor productivity is expressed as follows:
Multifactor Output Real business GDP* (43.2)

Multifactor
productivity =
productivity
Input
=
Weighted labor hours
and capital services
= Real
Real business
business GDP
GDP per
per unit
unit of
of labor
labor and
and
capital services
capital services
Accuracy
There are no estimates of sampling or revision error for multifactor

productivity.
Relevance
Productivity is important because greater efficiency increases the quan-

tity of goods and services available for civilian and defense needs, and
over time is a key ingredient for raising the material living conditions
and the security of the population. The relationship of productivity to
price change (consumer price index, producer price indexes, GDP price
measures), average weekly earnings, and employment is also important.
When productivity increases by relatively large amounts, production costs
fall, and more goods and services tend to be available at lower or smaller
increases in prices than in the absence of the large productivity increases.
Analogously, rising productivity permits higher wages by limiting in-
creases in production costs, without lowering profit margins.
But productivity increases can cause employment dislocation, because
the introduction of new technology changes or eliminates some jobs. Dis-
placed workers with outmoded skills may not be able to find new jobs or
may find work only at lower rates of pay. While some individuals may
therefore be adversely affected by productivity growth, rising productivity
does not lead to unemployment or lower wages at the economywide level.
Because quarterly movements in productivity are heavily influenced
*Same as output in labor-hour productivity above, except that government enter-

prises are excluded.
Hours worked by paid employees, the self-employed, and unpaid family work-
ers, modified for schooling, experience, and skills of different groups of workers.
Rental value of equipment, structures, land, and inventories adjusted for price
change. Depreciated value of equipment and structures modified for short-lived and
long-lived items.
Table 43.1
Productivity: Business Sector*

(annual percentage change)
Labor-hour productivity Multifactor productivity

1995 0.2 0.2
1996 3.0 1.7
1997 1.9 0.9
1998 2.8 1.1
1999 3.0 1.3
2000 2.8 1.4
2001 2.5 0.1
2002 4.0 1.9
2003 3.9 3.1P
2004 3.4 3.3P
19952004
(annual average) 3.0 1.6P
*Labor-hour productivity is referred to as the business sector, and multifactor pro-

ductivity is referred to as the private business sector.
P = Preliminary.
by cyclical changes in output, short-term changes in labor-hour produc-

tivity mainly reflect cyclical changes in economic growth rather than
basic changes in efficiency. Such basic changes are discerned by exam-
ining trends over at least several quarters that have relatively steady rates
of economic growth as reflected in the gross domestic product. Over the
long run, basic changes in productivity are seen more directly in the
annual movements of multifactor productivity noted below.
Multifactor productivity suggests the extent to which labor, capital,
materials, and other aspects of production are improving, in terms of
both technology and efficient usage. Changes in multifactor productiv-
ity indicate the extent of fundamental changes in the aggregate that im-
pact productivity. However, because multifactor productivity
encompasses all of the causal factors, further analysis is necessary to
identify which elements are changing significantly.
The productivity measures are prepared from many different data
sources that are inconsistent and also have known data problems. The
statistical estimation is also based on indirect procedures, which are
used in the absence of direct measures. The most serious problems ap-
pear in the service industries because of the difficulty of calculating the
output of some services. Because of these problems, there is an unknown
amount of error in the productivity numbers.
Recent Trends
From 1995 to 2004, productivity in the business sector showed diver-

gent movements between labor-hour productivity and multifactor pro-
ductivity (Table 43.1). Labor-hour productivity increased significantly
more than multifactor productivity in all years. After increasing by only
0.2 percent in 1995, labor-hour productivity increased in the 2 to 3 per-
cent range from 1996 to 2001, rose to increases of 4 percent in 200203,
and dropped to an increase of 3.4 percent in 2004. Multifactor produc-
tivity growth declined from 1.7 percent in 1996 to 0.1 percent in 2001,
though with interruptions in this downward trend in the intervening years,
and increased to 2 percent in 2002 and 3 percent in 200304. The only
decline in multifactor productivity occurred in 1995. Over the entire
nine-year period, labor-hour productivity increased at an average an-
nual rate of 3 percent, and multifactor productivity increased at an aver-
age annual rate of 1.6 percent.

Monthly.
. Multifactor Productivity Trends. Annual.
Peter B. Meyer, and Michael J. Harper, Preliminary estimates of multifactor pro-
ductivity growth, Monthly Labor Review, June 2005.
Note
1. The six-month schedule, which provides preliminary estimates, began in 2005

and is a speedup from the previous lag of one to two years. See Meyer and Harper
under References from Primary Data Source.
RETAIL SALES 221
44
Retail Sales
Retail sales represent the dollar receipts of retail establishments prima-

rily from selling merchandise, and secondarily from services incidental
to the sale of goods. Retailers sell to the public, which, in addition to
households, includes businesses and governments. Retail establishments
cover stores, mail-order houses, vending machines, and house-to-house
canvass. Retailing includes a wide range of businesses, such as food,
drug, liquor, department, variety, apparel, building material, hardware,
furniture, sporting goods, book, jewelry, camera, and optical goods stores;
automotive dealers, gasoline stations, restaurants, bars, and florists.
Retail sales data are prepared monthly by the Bureau of the Census in the
U.S. Department of Commerce. They are published in news releases, in
the Monthly Retail Trade and Food Services Survey, and in the Annual
Benchmark Report for Retail Trade and Food Services (www.census.gov).
Advance data are available about nine working days after the month
to which they refer. Preliminary monthly data, showing additional kinds
of business detail, are available six weeks after the reference month;
these are accompanied by revised data for the previous two months.
Annual revisions are made in March for the previous year.
Content
Retailers mainly resell merchandise purchased from manufacturers and

wholesalers, with markups from the purchase price. Their receipts are
221
also derived from delivery, installation, repair, and other services asso-
ciated with the merchandise. In addition, retailers make goods on their
own premises, such as a retail bakery, but such baking is subordinate to
selling to the public.1 Merchandise is composed of nondurable goods
such as food and clothing, which last less than three years, and durable
goods such as cars and furniture, which last more than three years.
Retail sales are the dollar value of receipts of retail establishments
after deductions for refunds, allowances for merchandise returned by
customers, and rebates by the retailer. Sales reflect the full price of the
item whether sold for cash or on credit, but they exclude receipts from
interest and other credit charges to the customer.
Receipts exclude sales and excise taxes collected directly from the
customer, but include gasoline, liquor, tobacco, and other excise taxes
collected by the manufacturer or wholesaler and passed along to the
customer. Merchandise sold at retail by manufacturers, wholesalers, and
service establishments is not included in the retail sales data.
The retail sales data are available both seasonally adjusted and not
seasonally adjusted.
Methodology
The retail sales data are obtained from a monthly sample survey of re-
tailers. The survey sample is updated quarterly to account for new firms
that start in business and existing firms that go out of business. New
firms starting in business are added to the survey sample nine months or
more after starting in operation, due to lags in obtaining notification of
the startup and data on the new firm.
The survey sample includes all firms defined as large based on sales
volume. Other firms are selected for the sample randomly based on their
kind of business and sales volume. The approximate response rate is 80
percent for the monthly retail sales survey and 89 percent for the annual
retail sales survey.
Accuracy
The sampling error (for one standard error) in the monthly percentage
range in the retail sales data is 0.5 of a percentage point. For example, if
the estimated increase in retail sales from one month to the next were 1
percent, in two of three cases the true increase would be somewhere
RETAIL SALES 223
Table 44.1
Retail Sales
Billions of dollars Annual

(monthly average) percentage change
1995 189.0 5.2
1996 201.1 6.4
1997 210.0 4.5
1998 220.4 4.9
1999 239.9 8.9
2000 255.8 6.6
2001 263.1 2.8
2002 269.2 2.3
2003 283.3 5.2
2004 305.4 7.8
between 0.5 and 1.5 percent. For further information on the interpreta-
tion of sampling and nonsampling errors, see the Appendix. Revisions
between the monthly advance and revised monthly data in two of three
cases range from 0.3 to 0.6 percent, and between the monthly prelimi-
nary and the monthly revised data from 0.2 to 0.2 percent.
Relevance
Retail sales are a key indicator of the strength of consumer spending. Con-
sumer spending typically accounts for 70 percent of the gross domestic
product, and consumer spending for durable and nondurable goods (the
items covered in the retail sales data) accounts for approximately 43 percent
of total consumer spending (the remainder of 57 percent represents con-
sumer spending for services). Because the ultimate purpose of economic
production is to provide for the well-being of people, consumer spending is
a bedrock of the economy. Through its impact on economic growth, con-
sumer spending is also an underlying factor affecting capital investment in
equipment and structures. In addition to being an economic indicator in
their own right, retail sales are a major data source used in preparing the
consumer expenditures component of the gross domestic product.
Recent Trends
From 1995 to 2004, retail sales rose continuously to a monthly average

of $305 billion in 2004 (Table 44.1). The annual increases peaked at 8.9
percent in 1999 and bottomed at 2.3 percent in 2002. The most typical
increases were in the 5 to 7 percent range.
Bureau of the Census, U.S. Department of Commerce. Monthly Retail Trade and
Food Services Survey, and Annual Benchmark Report for Retail Trade and Food
Services.
Note
1. For example, a bakery that makes bread and sells it to stores for resale to the
public is classified as a manufacturer, while a bakery that makes bread on the same
premises as the store that sells it to the public is classified as a retailer.
SELECTED SERVICES REVENUE 225
45
The data on selected services revenue represent private industry operat-

ing revenues for three broad service industry categories: (a) information
services, (b) professional, scientific, and technical services, (c) admin-
istrative and support, waste management, and remediation services, and
(d) hospitals and nursing and residential care facilities. These industries
sell their services to households, businesses, and governments. The se-
lected services revenue data are a new economic indicator that was first
published in 2004.
The selected services revenue data are prepared quarterly by the Bureau of
the Census in the U.S. Department of Commerce. The data are published in
a news release, Quarterly Revenue for Selected Services (www.census.gov).
The services revenue data are available seventy-five days after the
quarter to which they refer. The initial data are revised in the subsequent
quarter. Annual revisions are made in the following year.
Content
The selected services private industry operating revenue data cover many
component industries within the broad categories of information ser-
vices; professional, scientific, and technical services; and administra-
tive and support, waste management, and remediation services. Revenues
are the charges or billings for the services rendered by a firms opera-
tions, though payment may be received at a later date. Revenues ex-
225
clude income from nonoperating activities such as interest, investments,

sales of securities and real estate, loans, grants, sale of merchandise
from retail establishments, and sales and other taxes collected from cus-
tomers and submitted to government tax agencies.
Industry revenue data are also shown separately for sales to three
types of customers: households, businesses, and governments. As this
new indicator is developed, expanded industry coverage is planned for
transportation and warehousing; finance and insurance; real estate and
rental leasing; arts, entertainment, and recreation services; and others.
The services data are provided in current dollars (not adjusted for
price change) and as a percentage change from quarter to quarter.
The selected services data are not yet seasonally adjusted. However,
there are plans to provide seasonally adjusted data after 16 to 20 quar-
ters of data have been obtained.
Methodology
The selected services data are obtained from a probability sample of

approximately 6,000 private industry firms and in some cases from ag-
gregations of several establishments within a firm. An establishment is
a single physical location of an employers operations where payroll
and employment records are kept, with firms operating in more than
one location having several establishments. The sample covers estab-
lishments of all sizes and is updated quarterly for new firms starting in
business, for existing firms going out of business, and for other changes
to the universe of firms, such as from mergers and acquisitions.
Imputations are made for the firms that do not respond to the survey
based on reported revenues for firms of a similar size and in the same
industry. Imputations account for the following approximate shares of
total revenues: information services (9 percent); professional, scientific,
and technical services (26 percent); administrative and support, waste
management, and remediation services (30 percent), and hospitals and
nursing and residential care facilities (21 percent).
The selected services data are revised annually based on data from
the more comprehensive Service Annual Survey.
Accuracy
The reliability of the selected services revenue data for the dollar levels is
based on the coefficient of variation. The coefficient of variation is the
SELECTED SERVICES REVENUE 227
percentage that the sampling error is of the total dollar level. The coeffi-
cient of variation (for 1.65 standard errors) is: information services (0.5
percent); professional, scientific, and technical services (1.4 percent); ad-
ministrative and support, waste management, and remediation services
(2.6 percent), and hospitals and nursing and residential care facilities (1.1
percent). For example, if the estimated quarterly revenue for information
services were $230,000 million, in nine of ten cases the true level would
be somewhere between $228,850 million and $231,150 million.
The sampling error (for one standard error) for the quarter-to-quarter
change in the services revenue data is: information services (0.3 of a per-
centage point); professional, scientific, and technical services (0.9 of a
percentage point); administrative and support, waste management, and
remediation services (0.9 of a percentage point, and hospitals and nursing
and residential care facilities (0.4 of a percentage point). For example, if
the estimated increase in information services from one quarter to the
next were 5 percent, in two of three cases the true increase would be
somewhere between 4.7 and 5.3 percent. For further information on the
interpretation of sampling and nonsampling errors, see the Appendix.
Relevance
Services are the largest industrial component of the U.S. economy as

well as having the greatest growth over time. They are composed of
both labor-intensive and high-technology production operations, and run
the gamut from low-wage to high-wage employment. The initial cover-
age of the new quarterly selected services revenue data accounts for
about 15 percent of the gross domestic product. This coverage will in-
crease considerably as the selected services data are expanded to in-
clude many more industries, as noted above under Content.
Recent Trends
From the fourth quarter of 2004 to the first quarter of 2005, total se-
lected services revenues decreased 1.5 percent; information services
decreased 0.5 percent; professional, scientific, and technical services
decreased 0.5 percent; administrative and support, waste management,
and remediation services decreased 4.3 percent, and hospitals and nurs-
ing and residential care facilities increased 2.4 percent. Total selected
services revenue was $792 billion in the first quarter of 2005. Among
Table 45.1
Selected Services Revenue, Billions of Dollars

(not seasonally adjusted)
Percentage
change, 2004: 4Q
2004: 4Q 2005: 1Q to 2005: 1Q
Information services 237.0 227.4 4.1
Professional, 251.5 250.2 0.5
scientific, and
technical services
Administrative and 129.6 124.0 4.3
support, waste
management, and
remediation services
Hospitals and
nursing and
residential care 185.6 190.1 2.4
facilities
Total 803.7 791.7 1.5
the components, professional, scientific, and technical services was the

largest, followed in order by information services, hospitals and nursing
and residential care facilities, and administrative and support, waste
management, and remediation services (Table 45.1).
Bureau of the Census, U.S. Department of Commerce. Quarterly Revenue for Se-
lected Services. Quarterly.
STOCK MARKET PRICE AGGREGATES AND DIVIDEND YIELDS 229
46
Stock Market Price Aggregates
and Dividend Yields
Stock market aggregate price measures represent the overall price move-
ments of common stocks and certain other securities such as investment
funds and real estate investment trusts of corporations traded on U.S.
stock markets. The price performance and dividend yields of four dif-
ferent composite price measures are covered here.
Corporate stockholders are owners who have shares of stock in the
company. Stockholders share in company profits through dividends, and
in the case of common stocks, also have the right to vote on company
policies. Rising and falling stock prices affect capital gains and losses to
investors.
The stock market price aggregates and the dividend yields are cov-
ered separately, first the price aggregates and then the dividend yields.
Stock Market Price Aggregates
The selected four stock market price aggregates are prepared daily by the
New York Stock Exchange Inc., Nasdaq (National Association of Securi-
ties Dealers Automated Quotations), Standard & Poors Corporation, and
Dow Jones & Co. Inc. They are published in daily newspapers and elec-
tronically. General Web sites include: Bloomberg Financial
(www.bloomberg.com), Wall Street Journal Interactive Edition
(www.wsj.com), Barrons (www.barrons.com), MarketWatch
(www.marketwatch.com), MoneyCentral (www.moneycentral.msn.com),
and Finance Yahoo (www.finance.yahoo.com).
229
The stock market price aggregates are disseminated continuously

during the trading day.
Content
The four stock market price aggregates covered here are: the New
York Stock Exchange composite index, Nasdaq composite index,
Standard & Poors 500 composite index, and the Dow Jones indus-
trial average. The first three aggregates, which are referred to as in-
dexes, are referenced to a base period. By contrast, the Dow Jones
industrial average is not referenced to a base period. The four stock
price aggregates represent different groups of companies in which
price movements of each company are combined into a single num-
ber. Of the four aggregates, the New York Stock Exchange index and
the Nasdaq index are derived from the buying and selling transac-
tions of company shares that are listed on their exchanges. The Stan-
dard & Poors index and the Dow Jones industrial average are based
on stock prices of particular companies that are especially selected
for inclusion in each price measure.
The stock market price aggregates exclude the effect of changes in
the capitalized financial structure of companies such as corporate re-
structuring, stock splits, mergers, and spinoffs, so that price movements
are not distorted by changes in the underlying value of a share of stock
following a new capitalization of the company. Therefore, the price ag-
gregates cannot be compared with an average of actual current market
prices for the same companies because the latter would reflect the effect
of new capitalizations on the price per share without adjusting for changes
in the underlying value of each share.
The four stock price aggregates differ in the coverage of companies
that are priced and in the methodology used in calculating the aggregates.
The stock market price aggregates are not seasonally adjusted.
New York Stock Exchange Composite Index
The New York Stock Exchange composite index covers prices of ap-
proximately 2,100 company issues of common stocks, open-end funds,
real estate investment trusts, American depositary receipts, and tracking
stocks listed on the New York Stock Exchange. The composite index
excludes preferred stocks, closed-end funds, and certain other securi-
ties. Component price indexes of the composite are provided for com-
panies in a technology, media, telecom aggregate, and energy, finance,
and healthcare groupings.
Nasdaq Composite Index
The Nasdaq composite index covers the prices of over 3,000 company
stock issues traded on the Nasdaq: common stocks, real estate invest-
ment trusts, American depositary receipts, tracking stocks, and certain
other securities. The composite index excludes preferred stocks, closed-
end funds, and certain other securities. Component indexes of the com-
posite are provided for companies in industrial, insurance, bank, other
finance, transportation, computer, biotechnology, telecommunications,
and social groupings.
Standard & Poors 500 Composite Price Index
The Standard & Poors 500 composite price index covers prices of
500 companies on the New York Stock Exchange, Nasdaq, and AMEX
(American Stock Exchange). As a proportion of the market value of
the total index, companies on the New York Stock Exchange accounted
for 85.1 percent, the Nasdaq for 14.8 percent, and the AMEX for 0.1
percent as of January 2005. Component indexes of the 500 composite
are provided for companies in industrial, transportation, utilities, and
financial groupings.
Dow Jones Industrial Average
The Dow Jones industrial average covers prices of thirty U.S. com-
panies. They are widely held large companies in manufacturing, min-
ing, communications, finance, services, and retail industries. Dow
Jones indexes are also provided for companies in transportation and
utilities groupings.
Methodology
As noted above, the four stock price aggregates are adjusted to eliminate
the price effect of changes in the capitalized financial structure of compa-
nies, such as corporate restructuring, stock splits, mergers, and spinoffs.
This maintains the capitalized structure of each company as it was when

the company was first included in the price measure. The New York Stock
Exchange, Nasdaq, and Standard & Poors 500 price indexes are weighted
using the market value of the companies as the weights. The market value
of a company is the number of its common stock shares outstanding mul-
tiplied by its market price per share. By contrast, the Dow Jones industrial
average gives each company equal weight.
New York Stock Exchange Composite Price Index
The New York Stock Exchange composite price index reflects the mar-
ket value of the securities issues in the index. The stocks are averaged in
proportion to the market value of each company, which gives price move-
ments of firms with large market values more weight than those with
small market values. The market values are adjusted to reflect the num-
ber of shares actually available to investors. New companies are added
and old companies are deleted from the index as the companies are listed
and delisted on the New York Stock Exchange.
The New York Stock Exchange composite index currently is based
on December 31, 2002 = 5,000.
Nasdaq Composite Price Index
The Nasdaq composite price index reflects the market value of the secu-
rities issues in the index. The stocks are averaged in proportion to the
market values of each company, which gives price movements of firms
with large market values more weight than those with small market val-
ues. New companies are added and old companies are deleted from the
index as the companies are listed and deleted on the Nasdaq.
The Nasdaq composite index currently is based on February 5,
1971 = 100.
Standard & Poors 500 Composite Price Index
The Standard & Poors 500 composite index reflects the market value
of stocks in the index. Examples of the criteria used for including com-
panies in the index are: they are U.S. operating companies, have ad-
equate liquidity, a reasonable per-share price, market capitalization of
$4 billion or more, public float of at least 50 percent of the stock, and
are not a closed-end fund. The stocks are averaged in proportion to the
market values of each company, which gives price movements of firms
with large market values more weight than those with small market
values. New companies are substituted for old companies because of
mergers, bankruptcies, and capital restructuring, and to update the rep-
resentation of stocks to more closely represent important industries in
the U.S. economy.
The Standard & Poors 500 composite index currently is based on
194143 = 100.
Dow Jones Industrial Average
The Dow Jones industrial average reflects the average price per share
of thirty stocks. The index gives each stock an equal base weight of
3.33 percent. This results in more weight to price movements of com-
panies with high prices per share than those with low prices per share.
New companies are substituted for old companies because of merg-
ers and to update the index to better represent large, widely held
companies.
Accuracy
There are no estimates of sampling or revision error for the stock mar-
ket price aggregates.
Dividend Yields
Dividend yields represent cash dividend payments to stockholders as a

percentage of the market price of each companys stock; stock divi-
dends are excluded from the dividend yield measures. The dividend yield
includes all companies in the stock market price aggregatesthose com-
panies that pay dividends and those that do not. The annual dividend
yield is calculated as the dividends paid for the entire twelve months of
the calendar year divided by the stock market price aggregates at the
year-end December 31 level.
Company payouts of regular cash dividends and special cash divi-
dends (dividends paid from time to time with no set schedule) are re-
flected in the three price aggregates based on whatever the dividend
effects are on the market price of the companys stock. This differs from
total return stock price measures, which add the dividend payments to
the change in the stock prices for a given time period.
Dividend yields are provided for the Nasdaq composite price index
(the Nasdaq estimate first became available in 2002), Standard & Poors
500 composite price index, and the Dow Jones industrial average. Divi-
dend yields are not provided for the New York Stock Exchange compos-
ite price index.
The dividend yield for the Nasdaq composite price index and the Dow
Jones industrial average includes both regular and special cash divi-
dends. The dividend yield for the Standard & Poors 500 composite in-
dex includes regular cash dividends only.
Company buybacks of corporate stock are an implicit dividend yield
that supplements the above cash dividend yield. The buyback yield is
calculated as the percentage that the value of the repurchased stock is of
the companys balance sheet equity value. The buyback yield is a net
cash company outflow of the repurchases minus the exercise of em-
ployee stock options. A buyback yield is available only for a sample of
the largest companies in the Standard & Poors 500 composite index.1
Relevance
Stock market prices influence the course of future economic growth

through their effect on perceptions of the health of the economy and the
wealth of stockholders. High or rising stock prices encourage consumer
and investment spending because they reflect and/or promote optimism
about the economy and additions to the wealth of stockholders either as
paper capital gains or actual (realized) capital gains when the stocks are
sold. Low or falling stock prices discourage such spending because of
the pessimistic outlook they foster and a diminished wealth of stock-
holders either as a paper capital loss or actual (realized) capital loss
when the stocks are sold. The Standard & Poors 500 composite price
index is a component of the leading index of the leading, coincident,
and lagging indexes.
Stock price movements affect consumer spending through their
wealth effect on households (see distribution of wealth). Households
feel richer and more willing to spend when the value of their paper stock
holdings is high or when they sell the stock to obtain the actual capital
gains. The opposite occurs when the value of their stock holdings is low.
Stock prices also influence investment spending because high stock prices
make it easier for businesses to finance new investment by selling new
equity stock or by obtaining loans through new bond sales or other debt
financing. The choice of equity (stock) or debt (bond) financing is de-
termined by differences in the cost of raising funds (equity versus debt)
and by the effect of selling new stock on the choice of the capital struc-
ture of the company. Generally, it is easier to sell new stock when stock
prices are high or rising than when they are weak.
Dividend yields affect investor perceptions of future trends in stock
prices. Low yields suggest expectations of large or long-term price in-
creases, and high yields indicate anticipated small price increases or long-
term price declines. Dividend yields also affect patterns of investment.
Thus, low yields may lead investors out of stocks and into bonds, real
estate, or other investments that have an expected higher return, leading to
a fall in stock prices. By contrast, high yields may entice investors into
stocks, both because of the high return and the anticipation that the high
yields will stimulate higher stock prices (the latter is counter to the above
notion that high yields indicate weak future stock prices).
Recent Trends
From 1995 to 2004, price movements of the four stock market price
aggregates were similar over the entire period, but varied during the
19952000 and 200004 intermediate periods (Table 46.1). The aver-
age annual increase over the 19952004 period ranged from 8.5 to 9.7
percent in the four aggregates. The New York Stock Exchange compos-
ite index and the Dow Jones industrial average showed the closest an-
nual movements during the 19952000 and 200004 intermediate
periods. The Nasdaq composite index showed the most volatile move-
ments and the Standard & Poors 500 composite index showed the next
most volatile movements during 19952000 and 200004.
From 1995 to 2004, dividend yields typically were around the 1.5 to
2.0 percent range. The Standard & Poors 500 yield dropped from 2.56
percent in 1995 to a low of 1.15 percent in 2000, and then rose to a high
of 1.77 percent in 2003. The Dow Jones industrial average yield dropped
from 2.27 percent in 1995 to a low of 1.47 percent in 1999 and then rose
to 2.27 percent in 2002 (the same high as in 1995). The Nasdaq yield
was 0.48 percent in 2002 and 2003 and rose to 1.5 percent in 2004.
Table 46.1
New York Stock Standard &

Exchange Nasdaq Poors Dow
Index Market 500 Index Jones
(12/31/02 Index (2/5/71 (194143 Industrial
= 5,000) = 100) = 10) Average
1995 3,079 925 542 4,494
1996 3,787 1,165 671 5,743
1997 4,827 1,469 873 7,441
1998 5,818 1,795 1,086 8,626
1999 6,547 2,728 1,327 10,465
2000 6,806 3,784 1,427 10,735
2001 6,398 2,035 1,194 10,189
2002 5,579 1,540 994 9,226
2003 5,447 1,647 965 8,994
2004 6,613 1,987 1,131 10,317
Annual
percentage
change
19952000 17.2 32.5 21.4 19.0
20002004 0.7 14.9 5.6 1.0
19952004 8.9 8.9 8.5 9.7
Buyback
Cash dividend yield (percent) dividend
Cash Cash Cash Cash yield
dividend dividend dividend dividend (percent)
1995 NA NA 2.56 2.27 2.41
1996 NA NA 2.19 2.03 2.06
1997 NA NA 1.77 1.72 1.69
1998 NA NA 1.49 1.65 1.48
1999 NA NA 1.25 1.47 1.19
2000 NA NA 1.15 1.60 1.07
2001 NA NA 1.32 1.81 1.26
2002 NA 0.48 1.61 2.27 NA
2003 NA 0.48 1.77 2.00 NA
2004 NA 1.50 1.72 2.22 NA
NA = Not available.
The buyback dividend yield, when added to the above cash dividend
yield, gives an implicit total yield. A buyback yield, which is available
only for a sample of the largest companies in the Standard & Poors 500,
declined from 2.41 percent in 1995 to 1.07 percent in 2000, and then rose
to 1.26 percent in 2001. Data for 2002 to 2004 are not available.
Note
1. Estimates of the buyback yield in Table 46.1 are based on an analysis of

company reports to the Securities and Exchange Commission (10-K reports) of a
sample of the largest companies in the Standard & Poors 500 composite index. See
J. Nellie Liang and Steven A. Sharpe, Share Repurchases and Employee Stock
Options and Their Implications for S & P 500 Share Retirements and Expected
Returns, Finance and Economics Discussion Series no. 59, Federal Reserve Board,
1999. The buyback yields in Table 46.1 reflect revised and updated data since the
above paper was published.
47
Unemployment
Unemployment represents the number of persons without jobs who are

available for and actively seeking work. There are two measures of un-
employment. One is confined to unemployed workers who collect un-
employment insurance benefit payments through the federal-state
unemployment insurance system. The other is a more comprehensive
definition of unemployment and thus covers many more persons other
than those who collect unemployment insurance benefit payments
This chapter covers both unemployment measures. The comprehen-
sive measure is the total rate of unemployment and is referred to here as
total unemployment. It is the familiar one that is used in economic
analysis and in developing fiscal and monetary policies for the macro
economy. The insured unemployment measure is referred to here as in-
sured unemployment. It focuses on insurance benefit payments for the
unemployed.
Recent trends are covered at the end of the chapter for both total
unemployment and insured unemployment.
Total Unemployment
Total unemployment covers all persons sixteen years and older who are
available for and actively seeking work who lost or quit previous jobs,
and others with no work experience or who have re-entered the work-
place. It includes but is not limited to workers who qualify for unem-
ployment insurance benefit payments. The unemployment rate is equal
to the number of unemployed persons as a percentage of the sum of the
employed and the unemployed. As a measure of additional workers avail-
238
UNEMPLOYMENT 239
able for employment, the unemployment rate reflects the slack or tight-
ness in labor markets.
Total unemployment measures are prepared monthly by the Bureau of

published in a news release and in two BLS monthly journals, Monthly
Labor Review and Employment and Earnings (www.bls.gov).
the twelfth of the month. Thus, the information is released on the first or
second Friday of the month following the month in question. On the day
the data are released, the commissioner of labor statistics reports on
recent unemployment and employment trends to the Joint Economic
Committee of Congress. The monthly data are revised every January for
the previous five years, based on updated seasonal factors.
Content
The labor force is defined as the sum of employed and unemployed per-
sons living in the United States. Both citizens and foreigners are included.
Unemployment data measure the number of persons sixteen years
and older who do not have jobs and are available for and actively seek-
ing work. The unemployment rate is the percentage of persons in the
labor force who are unemployed and is calculated by the formula below.
This calculation is done separately for several demographic groups and
weighted together to arrive at the overall rate. The demographic groups
represent distinctions by age, gender, race, and ethnicity.
(47.1)
Unemployed persons
Unemployment rate = 100
Employed +
unemployed persons (labor force)
Employed persons are defined as nonfarm and farm workers aged

sixteen years and older who are not institutionalized and who are not
residents of homes for the aged.1 The definition includes full-time and
part-time wage and salary employees at paid jobs who work at least one
hour a week, self-employed persons working in their own business, and
unpaid workers in family businesses who work at least fifteen hours a

week (family workers are assumed to share in the profits of the busi-
ness).2 Because the employment measures count persons rather than jobs,
individuals holding two or more jobs are counted only oncein the job
they work the most hours during a week. Persons temporarily absent
from their jobs because of illness, vacation, strike, or lockout are in-
cluded as employed whether or not they are paid while they are absent
from work. (This definition of employment conforms to the house-
hold measure of employment discussed in the employment chapter.)
Unemployed persons are defined as those who actively sought a job
at least once in the previous four weeks through such actions as having
a job interview; contacting an employer for a job interview; answering a
job advertisement; sending out resumes; contacting an employment
agency, a friend, or a relative; placing an advertisement in a newspaper;
checking with a union or professional register; obtaining assistance from
a community organization, or waiting at a designated labor pickup point.
By contrast, looking at job advertisements or attending training pro-
grams or courses is defined as a passive search for work and does not
meet the criterion of being unemployed. The unemployed include indi-
viduals who collect unemployment insurance as well as those who are
not eligible for unemployment insurance (for example, formerly em-
ployed workers who have exhausted their unemployment insurance or
unemployed persons who did not work long enough to qualify for un-
employment benefits). Students are counted as unemployed if they sought
work and are available at least for part-time jobs.
Discouraged workers are not in the labor force and so are not counted
as unemployed. These are workers who say they want a job, but are not
seeking work because they think there are no jobs available in the local
labor market or believe they do not qualify for the existing job vacancies
due to lack of schooling or training, employers thinking they are too
young or too old, or other types of discrimination. Persons are classified
as discouraged only if they looked for a job at least once during the past
twelve months, or since the end of their last job if they held one within
the past twelve months.
Discouraged workers, a subset of the category of marginally attached
workers, are outside the labor force for the economic reasons noted above.
Like discouraged workers, marginally attached workers other than dis-
couraged workers have looked for a job at least once during the past twelve
months, or since the end of their last job if they held one within the past
UNEMPLOYMENT 241
Table 47.1
Alternative Measures of Labor Underutilization

(seasonally adjusted)
February 2005
(percent)
U-1. Persons unemployed 15 weeks or longer, as a 1.9

percentage of the civilian labor force
U-2. Job losersa and persons who completed temporary jobs, 2.7
as a percentage of the civilian labor force
U-3. Total unemployed, as a percentage of the civilian 5.4
labor force (official unemployment rate)
U-4. U-3 plus discouraged workers, as a percentage of the 5.7
civilian labor force plus discouraged workers
U-5. U-4 plus all other marginally attached workers, as a 6.4
percentage of the civilian labor force plus all marginally
attached workersb
U-6. U-5 plus total employed part-time because full-time jobs 9.3
are not available
a
Job losers are unemployed because they were laid off or fired; bFor the distinction
between discouraged and other marginally attached workers, see text.
twelve months. The only difference is that marginally attached workers

other than discouraged workers are not currently looking for work be-
cause of noneconomic reasons, such as illness or medical limitations, child-
care problems or other family or personal obligations, school, or training.
Unemployment rates are also calculated by length of time out of work,
by demographic components of the workforce such as age, race, gender,
and marital status of adults in the household, and for large states and
metropolitan areas.
The unemployment data are seasonally adjusted.
Table 47.1 shows the total unemployment rate, plus five alternative
measures of labor underutilization. The official number is referred to as
U-3. The five alternative measures calculate rates based on the duration
of unemployment; number of workers who have lost their jobs; number
of temporary jobs that ended; and number of discouraged workers, other
marginally attached workers, and part-time workers. These measures
provide a range of labor underutilization rates significantly below and
above the official rate (discussed further under Relevance).
The alternative measures of labor underutilization under Total Un-
employment show a range of 7.4 percentage points in February 2005
(Table 47.1). The lowest rate of U-1, which consists of persons unem-
ployed for fifteen weeks or longer, was 1.9 percent. The highest rate of
U-6, which consists of all unemployed workers, plus all marginally at-
tached workers, plus part-time workers who want a full-time job, was
9.3 percent. The official rate, U-3, was 5.4 percent.
Impact of Labor Force Participation on Unemployment
As noted above, the labor force is the sum of employed and unem-
ployed persons. The labor force as a percentage of the civilian
noninstitutionalized population (see note 1 for noninstitutionalized per-
sons) is referred to as the labor force participation rate (LFPR). Thus,
the LFPR is not used in calculating the unemployment rate. However,
trends in the LFPR are important in understanding trends in the unem-
ployment rate.
In 2004, the unemployment rate was 5.5 percent and the LFPR was
66.0 percent. This compares with the unemployment rate of 4.0 percent
and the LFPR of 67.1 percent in 2000. Thus, from 2000 to 2004, the
unemployment rate rose and the LFPR fellthe unemployment rate
rose from 2001 to 2003 and declined in 2004, while the LFPR declined
each year during 200104. Ordinarily, it would be expected that a fall-
ing LFPR over several years would be associated with lower unemploy-
ment. This reflects the fact that a falling LFPR would lessen the increase
in the number of noninstitutionalized persons not in the labor force (the
noninstitutionalized population increased in all years since the end of
World War II), and consequently lessen the number of unemployed per-
sons looking for work.
Moreover, during 200104, trends in the LFPR reversed or sharply
accentuated movements in previous years (Table 47.2). The overall LFPR
rose continuously in the decades from the 1960s up to the year 2000,
and then declined continuously during 200104. The overall increase in
the LFPR up to the year 2000 was the net effect of divergent trends in
the demographic components of the labor force, with the LFPR for men
twenty years and older declining, the LFPR for women twenty years
and older rising, and the LFPR for teens rising in the 1960s and 1970s
and declining in the 1980s and 1990s up to 2000. Then, from 200104,
the long-run rate of decline in the LFPR for men increased, the long-run
increase in the LFPR for women reversed to a decline, and the decline in
the LFPR for teens that began in the 1980s intensified.
UNEMPLOYMENT 243
Table 47.2
Labor Force Participation Rates (percent)
Men Women Teens Total

1979 79.8 50.6 57.9 63.7
1980 79.4 51.3 56.7 63.8
1989 78.1 57.7 55.9 66.5
1990 78.2 58.0 53.7 66.5
2000 76.7 60.6 52.0 67.1
2001 76.5 60.6 49.6 66.8
2004 75.8 60.3 43.9 66.0
Notes: Cyclical expansion peaks: 1979, 1989, 2000; Cyclical recession lows: 1980,
1990, 2001; Cyclical expansion: 2004.
I have re-estimated the LFPR in 2004 assuming that trends from

1989 to 2000 (years of cyclical expansion peaks) continued at the same
rate during 200104, both for the demographic components of men,
women, and teens, and in the aggregate. These calculations result in
total unemployment rates in 2004 from 7.5 to 7.7 percent. Using the
same procedure with the LFPR trends from 1990 to 2001 (years of
cyclical recession lows), the resultant unemployment rates in 2004 are
7.4 to 7.9 percent. Thus, at a minimum, if the LFPRs had continued at
their long-run trend from 2001 to 2004, the unemployment rate in 2004
would have been at least two percentage points higher than the actual
unemployment rate of 5.5 percent.
The focus on the LFPR highlights the discouraged worker effect in
the long run. As noted above, the term discouraged workers, as mea-
sured in the alternative measures of labor underutilization, refers to
workers who have not looked for work in the past twelve months, the
short run, due to economic reasons (Table 47.1). The use of the LFPR in
my analysis over several years relates to what the behavior of prospec-
tive workers would have been if the job situation had been stronger dur-
ing 200104.
While there are anecdotal reports of mothers with young children
who have decided to be at home with their children and thus are not in
the labor force, which would tend to lower the LFPR for women, these
do not appear to have been great enough to have changed the LFPR for
women as sharply as shown in Table 47.2. At the other end of the spec-
trum, there are anecdotal reports of retired men and women who con-
tinue to work at other jobs after retirement, which would tend to raise
the LFPR for men and women. but these too do not appear to have caused
the sharp changes in Table 47.2.
Methodology
Unemployment data are obtained from a monthly survey of households

called the Current Population Survey (CPS). The CPS is a sample of
about 60,000 households, which the U.S. Bureau of the Census con-
ducts for the Bureau of Labor Statistics.3 Responses are actually ob-
tained from about 55,500 households; no responses are obtained from
the remaining 4,500 households due to absence, impassable roads, re-
fusals, or for other reasons. The sample is representative of the distribu-
tion of households in small and large metropolitan areas and in rural
areas. It undergoes a major revision every ten years to be consistent with
the most recent decennial population census. The sample currently is
based on the 2000 census of population. The sample is also updated
annually on a limited basis to reflect current changes in residential loca-
tions due to new construction based on housing starts data prepared by
the Census Bureau.
In order to reduce the reporting burden on any group of households, the
CPS is divided into eight subsamples (panels) that are rotated over a six-
teen-month period. Each subsample is surveyed for four consecutive
months, then dropped from the survey for eight months, and subsequently
resurveyed for the following four months. At the end of the sixteen months,
the subsample is eliminated from the sample and is replaced with a new
panel of households. The result of this procedure is that every month 25
percent of the households in the sample are either new to the survey or are
returning after an eight-month hiatus. Correspondingly, 25 percent of the
sample households drop out of the survey every month.
The survey refers to the individuals employment status during the
calendar week that includes the twelfth of the month. The survey is con-
ducted mainly by telephone interviews, supplemented by personal in-
terviews as necessary.
Accuracy
Assuming an unemployment rate of 5.4 percent, in nine of ten cases

(1.6 standard errors), a monthly change in the unemployment rate of
at least plus or minus 0.23 of a percentage point is regarded as statisti-
UNEMPLOYMENT 245
cally significant. Although a change of zero or plus or minus 0.10 of a

percentage point is not statistically significant for one month, cumula-
tive changes of 0.12 of a percentage point in the same direction for
two or more consecutive months are statistically significant. The only
difference in these estimates if the unemployment rate is 6.0 percent is
that a monthly change in the unemployment rate of at least plus or
minus 0.24 is regarded as statistically significant (compared with 0.23
of a percentage point at a 5.4 percent unemployment rate). For further
information on the interpretation of sampling and nonsampling errors,
see the Appendix.
Relevance
The unemployment rate is a major indicator of the degree to which the

economy provides jobs for those seeking work. It is a key consideration
when the president, Congress, and Federal Reserve determine whether
economic growth should be stimulated or restrained through fiscal and
monetary policies (see government budgets and debt for fiscal policy
and interest rates for monetary policy).
In general, there is an inverse relationship between unemployment and
the gross domestic product (GDP), which is referred to as Okuns Law. In
the early 2000s, this relationship functioned roughly as follows: the yearly
unemployment rate remains the same if the annual real GDP increases by
2.0 percent, with every 1 percentage point growth in real GDP above 2.0
percent lowering the unemployment rate by 0.5 percentage point, and ev-
ery 1 percentage point growth in real GDP below 2.0 percent raising un-
employment by 0.5 percentage point. These estimates reflect the
relationship that averages out over the years, but they do not hold in every
year. The formulation of Okuns Law takes price movements into account
not only by the utilization of real GDP, but also by limiting the concept to
a nonaccelerating inflation rate of unemployment (referred to as NAIRU).
The unemployment rate is also used to analyze the tradeoff between un-
employment and inflation (e.g., the consumer price index), which is re-
ferred to as the Phillips Curve. In theory, there is an inverse relationship
between unemployment and inflation: when unemployment declines, infla-
tion rises, and when unemployment rises, inflation declines. But this is not
a one-to-one relationship, nor is it equally applicable when the economy is
functioning at high and low unemployment. In general, the Phillips Curve is
most consistent with the theory at low rates of unemployment and inflation.
Perhaps more significant than the year-to-year movements are long-

term shifts in the improvement or worsening of the tradeoff shown in
the Phillips Curve. An improvement in the tradeoff means that over time,
the same rate of unemployment generates a lower rate of inflation than
it had in the past, while a worsening of the tradeoff means that the same
rate of unemployment generates more inflation than it had in the past.
Current formulations of the Phillips Curve include the effects of infla-
tion expectations on the movement of the curve. This kind of analysis is
used to assess goals for minimum unemployment and inflation rates
such as are included in the Full Employment and Balanced Growth Act
of 1978 (Humphrey-Hawkins Act).
Alternative measures of labor underutilization indicate a range of esti-
mated slackness in the economy, from the lowest rate at the U-1 end of the
spectrum to the highest rate at the U-6 end. Depending on the social and
political perspectives of those characterizing the extent of unemployment,
persons highlighting the economys strength would point to the U-1 end,
while persons highlighting the economys weakness would point to the
U-6 end. Total unemployment (defined as the official rate) is U-3.
In addition, the unemployment rate determines when federally fi-
nanced supplementary unemployment benefits go into effect for par-
ticular localities when unemployment is persistently high. These benefits
supplement regular state-provided unemployment benefits that have been
exhausted, and are triggered by a formula that includes both the national
unemployment rate and state and metropolitan area unemployment rates
(see Insured Unemployment below).
Insured Unemployment
Insured unemployment represents unemployed workers who are avail-

able for and are actively seeking work, provided they are receiving un-
employment insurance benefit payments under the federal-state
unemployment insurance system. Unemployment insurance benefit pay-
ments provide some income subsistence for unemployed workers and
also function as automatic stabilizers during cyclical expansions and
recessions in the macro economy.
Insured unemployment data are prepared weekly by the Employment

UNEMPLOYMENT 247
and Training Administration in the U.S. Department of Labor. The data

are published in a news release, Unemployment Insurance Weekly Claims
Report (www.ows.doleta.gov).
The data are available each Thursday following the week to which
they refer. The data are revised in the subsequent three weeks.
Content
Insured unemployment applies solely to unemployed workers who are

eligible for and apply for unemployment benefit payments through the
federal-state unemployment insurance system. The federal-state system
allows states to pay benefits up to a maximum of twenty-six weeks of
unemployment, though individual states vary in determining the eligi-
bility of unemployed persons to obtain benefit payments. The unem-
ployment insurance program also triggers additional benefit payments
when state unemployment is persistently high during recessions. These
extended benefit payments go beyond the first twenty-six weeks to an
additional thirteen to twenty weeks among individual states.
The benefit payments are financed by a federally mandated tax that
employers in each state must pay into a state trust fund. In addition,
three states (Alaska, New Jersey, and Pennsylvania) require employees
to pay into the state trust fund. State governments administer the unem-
ployment insurance program under general federal guidelines that give
the states complete discretion in determining who is eligible for benefit
payments, the dollar amount of the benefits, and how long the benefits
are paid while the recipient is unemployed. There is considerable varia-
tion among the states in applying these aspects of the program. Eligibil-
ity for unemployment insurance, benefit amounts, and the length of time
benefits are available are determined by each states law under which
unemployment insurance claims are established.
In order to qualify for unemployment insurance benefit payments, the
person must have worked a specified period of time for a private em-
ployer (in a place of business or as a household worker) or for a state or
local government. The unemployment insurance data exclude other un-
employment insurance programs that cover federal workers or railroad
workers. There is no federal-state unemployment insurance coverage for
employees of religious organizations or for self-employed workers.
As an incentive to unemployed workers to create their own jobs by
starting their own businesses, Self-Employment Assistance programs
allow states to pay a self-employed allowance to help these workers

while they are establishing businesses and becoming self-employed.
Participants receive weekly allowances while they are getting their busi-
nesses off the ground. This is a voluntary program, and in 2004 fewer
than ten states had Self-Employment Assistance programs.
Insured unemployment data are prepared for the United States,
the fifty states and the District of Columbia, Puerto Rico, and the
Virgin Islands. The insured unemployment rate, which is calculated
weekly, is the number of persons that receive unemployment insur-
ance benefit payments as a percentage of all persons covered under
the unemployment insurance program. The unemployment rate for
the week that includes the twelfth of the month represents the monthly
unemployment rate. In addition, weekly data are provided on the
number of persons filing initial claims for unemployment insurance
benefit payments.
The nationwide insured unemployment data are seasonally adjusted. In-
sured unemployment data for individual states are not seasonally adjusted.
Methodology
The data on insured unemployment are obtained by the Employment

and Training Administration in the U.S. Department of Labor from the
universe of records of the state employment agencies that administer
the Unemployment Insurance program.
Accuracy
There are no sampling errors in the insured unemployment data because

the data are collected from the universe of insured workers.
Relevance
In addition to providing some subsistence income for unemployed work-

ers, unemployment insurance benefit payments function as automatic
stabilizers for the macro economy. During economic recessions, benefit
payments maintain some income for unemployed workers, which in turn
contributes to household expenditures and economic growth. And when
unemployment declines during economic expansions, benefit payments
decline and thus reduce the stimulus to household spending. In eco-
UNEMPLOYMENT 249
Table 47.3
Unemployment Rate: Total and Insured Measures of

Unemployment (percent)
Total unemployment Insured unemployment

1995 5.6 2.3
1996 5.4 2.2
1997 4.9 1.9
1998 4.5 1.8
1999 4.2 1.7
2000 4.0 1.6
2001 4.7 2.3
2002 5.8 2.8
2003 6.0 2.8
2004 5.5 2.3
Notes: Cyclical expansion peaks: 1979, 1989, 2000; Cyclical recession lows: 1980,
1990, 2001; Cyclical expansion: 2004.
nomic analysis, the number of persons filing initial claims for unem-
ployment insurance benefit payments when they first become unem-
ployed is a component of the leading index (see leading, coincident,
and lagging indexes).
Recent Trends
From 1995 to 2004, the total measure of unemployment was typically

2.5 times the level of insured unemployment during 19952000 and in
2004, and 2.0 times that of insured unemployment during 200103 (Table
47.3). The two measures moved in the same upward and downward yearly
direction over the 19952002 period and in 2004. In 2003, the total
measure rose, while the insured measure did not change.
Employment and Training Administration, U.S. Department of Labor. Unemploy-
ment Insurance Weekly Claims Report. News Release. Weekly.
Notes
1. Institutionalized persons include those confined to penal or mental facilities.

2. Self-employed persons whose businesses are unincorporated are classified as

self-employed, but the self-employed whose businesses are incorporated are classi-
fied as wage and salary workers, because technically they are paid employees of a
corporation.
3. A household consists of all personsrelated family members and unrelated
individualswho occupy a housing unit and have no other usual address. A housing
unit is intended as separate living quarters, and encompasses single-family houses,
townhouses, condominiums, apartments, mobile homes, single rooms, and group
quarters where residents share common facilities or receive formal or authorized
care or custody. There were 112 million households in the United States in 2003.
UNIT LABOR COSTS: BUSINESS SECTOR 251
48
Unit Labor Costs:
Business Sector
Unit labor costs (ULC) represent the relationship of labor costs per hour
to productivity. When compensation per hour increases more (or de-
clines less) than productivity, ULC increase. Similarly, when compen-
sation increases less (or declines more) than productivity, ULC decline.
ULC may also be considered as compensation per unit of output, or the
share that compensation is of output. The level and movements of ULC
reflect one aspect of cost pressures on prices.
The measure of unit labor costs is prepared on a quarterly basis by the

Bureau of Labor Statistics (BLS) in the U.S. Department of Labor. The
data are published in a monthly news release and in the BLS monthly
journal, Monthly Labor Review (www.bls.gov).
Preliminary data are prepared in the second month following the quar-
ter to which the data refer (May for the first quarter, August for the second
quarter, November for the third quarter, and February for the fourth quar-
ter). The data are available at the same time as the productivity data and
follow soon after publication of the gross domestic product measures. The
data are initially revised in the subsequent month, and then two months
later along with the release of the preliminary data for the following quar-
ter. Annual revisions are made every year as more data become available.
Content
Unit labor costs are defined as the ratio of compensation per hour to
productivity. This is equivalent to the ratio of compensation to output
251
(real business gross domestic product), because the labor hours terms in
the numerator and denominator cancel out algebraically, as shown in
the following formula:
(48.1)
Compensation*

Labor hours Compensation* Compensation per
ULC = = =
Productivity Real GDP# unit of real GDP
#
Real GDP

Labor Hours
Compensation for work covers the wage, salary, and fringe benefit
income of employees, plus the income of self-employed persons, part of
which is attributable to wages as distinct from profits.
The business sectors are derived from the gross domestic product (GDP)
definitions. The GDP value-added concept counts output as the sum of the
incomes associated with employee compensation, business profits, interest
payments, depreciation allowances, and indirect business taxes. It excludes
purchased materials and services used in production. Business sectors ex-
clude not-for-profit organizations, household output, rental value of owner-
occupied housing, and general government. The nonfarm business sector
also excludes farming. The ULC measures are expressed as an index.
The ULC indexes are currently based on 1992 = 100.
The ULC data are seasonally adjusted.
Methodology
Compensation data for unit labor costs are obtained from the income side of
the gross domestic product, which is based on employment from the estab-
lishment and household surveys, average weekly hours, and average hourly
earnings. An additional estimate is made for the wage component of self-
employment income based on data on hours worked by proprietors from
the household survey of employment; the estimate assumes that proprietors
*Wages, salaries, and fringe benefits and the wage component of income from
self employment.
Hours worked by paid employees, the self-employed, and unpaid family workers.
#
Gross domestic product, excluding households, not-for-profit organizations, rental
value of owner-occupied housing, and general government (adjusted for price change).
UNIT LABOR COSTS: BUSINESS SECTOR 253
Table 48.1
Unit Labor Costs: Business Sector (annual percentage change)
1995 1.9
1996 0.5
1997 1.3
1998 3.2
1999 1.8
2000 4.0
2001 1.6
2002 1.1
2003 0.3
2004 1.0
19952004 (annual average) 1.3
work for the same hourly earnings as employees in the industry. Productiv-
ity is based on data for the gross domestic product, employment, and aver-
age weekly hours as described in labor-hour productivity.
Accuracy
There are no estimates of sampling or revision error for the unit labor
cost (ULC) data. However, because ULC are closely linked to produc-
tivity measures, ULC revision error is assumed to mirror that of produc-
tivity. In 19 of 20 cases, the second quarterly revision (four months after
the preliminary data) of labor hour productivity differs from the pre-
liminary index by 1.4 to 1.4 index points.
Relevance
Unit labor costs data indicate costprice pressures. When ULC increase
significantly, businesses may raise prices to maintain profit margins. Analo-
gously, when ULC increase slightly or decline, profit margins can be main-
tained with little or no price increases or even price declines. There is a
two-way street between ULC and prices, however, because ULC are af-
fected by cost-of-living wage increases made to compensate for price in-
creases. Moreover, in addition to ULC, prices reflect the demand for goods
and services, production costs other than ULC, such as purchased-materials
prices (see producer price indexes), interest rates, the impact of weather on
food, and the effect on energy prices of actions taken by the Organization of
Petroleum Exporting Countries to control oil production. Thus, ULC are an
important, but not necessarily a determining factor of price movements. As
in the case of labor-hour productivity, quarterly changes in ULC reflect short-

term movements in output associated with changes in economic activity
over the business cycle. Consequently, ULC movements over several quar-
ters should be observed to determine more basic trends.
Recent Trends
From 1995 to 2004, unit labor costs fluctuated within a range of 1.1 to
4.0 percent (Table 48.1). ULC changes were negative in 2002 and 2003.
Over the entire nine-year period, ULC increased at an average annual
rate of 1.3 percent.

Monthly.
VALUE OF THE DOLLAR 255
49
Value of the Dollar
The value of the dollar represents the foreign exchange price of the U.S.
dollar in relation to other nations currencies. It affects the competitive
position of U.S. goods in world markets. If the dollar rises in value, and
offsetting changes in the prices of exported and imported goods and
services are not made, U.S. exports become more expensive to foreign-
ers and imports become less expensive to Americans. If the dollar falls
in value and offsetting changes in the price of exported and imported
goods and services are not made, U.S. exports become less expensive to
foreigners and imports become more expensive to Americans. Measures
of the value of the dollar differ according to choices of groups of curren-
cies, weights, and weighting methodologies. The data are provided in
nominal values of market exchange rates and in real values of exchange
rates adjusted for price change.
Three nominal and three real value-of-the-dollar indexes are prepared by

the Federal Reserve Board (FRB). The six indexes are published in a sta-
tistical release (H.10) and in the monthly statistical supplement to the
quarterly FRB journal, Federal Reserve Bulletin (www.federalreserve.gov).
The indexes are provided daily (Web site only), weekly, and monthly.
The weekly data are published on Monday for the previous week. The
monthly data are published the first business day of the month for the
previous month. The monthly data for the real indexes are revised as
more accurate information becomes available.
255
Content
The six value-of-the-dollar indexes represent the average foreign ex-

change price of the dollar in relation to the currencies of a group of
nations. The nominal indexes are based on exchange rates used in actual
market transactions. Real exchange rates are nominal exchange rates
adjusted for price movements in the United States and abroad. Different
groups of countries are covered by the three nominal indexes: the broad
index, the major currencies index, and the other important trading part-
ner (OITP) index (Table 49.1).1 The list of countries included in these
indexes is re-evaluated annually.
The broad index is the most comprehensive of the three indexes. The
index is composed of the currencies of 25 countries, plus the euro cur-
rency area of 12 countries. These countries represent the major trading
partners of the United States. Individually, they were selected if they
accounted for at least 0.5 percent of U.S. nonoil merchandise imports or
U.S. nonagricultural merchandise exports in 1997 (merchandise refers
to goods as distinct from services). Collectively, these countries accounted
for 93 percent of U.S. nonoil merchandise imports and 91 percent of
U.S. nongold/nonmilitary merchandise exports in 2003. The base pe-
riod of the broad index is currently January 1997 = 100.
The major currencies index covers the subset of broad-index cur-
rencies that circulate widely outside the country of issue. The index is
composed of seven currencies, including the euro. Its movements sug-
gest shifts in the competitiveness of U.S. goods versus goods produced
by other industrial countries. It is also an indicator of financial market
pressures on the dollar. It accounted for 50 percent of U.S. nonoil mer-
chandise imports and 54 percent of nongold/nonmilitary merchandise
exports in 2003. The base period of the major currency index is cur-
rently March 1973 = 100.
The other important trading partner (OITP) index covers a subset of
the broad-index currencies that generally do not circulate or trade widely
outside the country of issue. The index is composed of nineteen curren-
cies from Latin America, Asia, the Middle East, and Eastern Europe. Its
movements suggest shifts in U.S. competitiveness in its trade with those
regions. It accounted for 43 percent of U.S. nonoil merchandise imports
and 37 percent of U.S. nongold/nonmilitary merchandise exports in 2003.
The base period of the OITP index is currently January 1997 = 100.
The value-of-the-dollar indexes are not seasonally adjusted.
Table 49.1
Value-of-the-Dollar Indexes: Country Coverage

(in order of U.S. trade importance in 2003, largest first)
Broad index
(26 countries including Major currencies index OUTP
Euro area) (7 countries) (19 countries)
Canada Canada
Euro area* Euro area*
China China
Japan Japan
Mexico Mexico
United Kingdom United Kingdom
Korea Korea
Taiwan Taiwan
Hong Kong Hong Kong
Malaysia Malaysia
Singapore Singapore
Brazil Brazil
Switzerland Switzerland
Thailand Thailand
Australia Australia
Sweden
India India
Philippines Philippines
Israel Israel
Indonesia Sweden Indonesia
Russia Russia
Saudi Arabia Saudi Arabia
Chile Chile
Argentina Argentina
Columbia Columbia
Venezuela Venezuela
*The Euro-area countries in order of U.S. trade importance, largest first, are: Ger-
many, France, Italy, Ireland, Netherlands, Belgium/Luxembourg, Spain, Austria, Fin-
land, Portugal, Greece.
Methodology
The weights used to combine the individual currencies in the value-of-

the-dollar indexes are calculated from the dollar value of internationally
traded manufactured, mineral, and agricultural goods. The weights are
revised annually based on new trade data.
The value-of-the-dollar indexes use geometric averaging of the individual
exchange rates. Geometric averaging, in contrast to arithmetic averaging,
treats relative exchange rate currency increases and decreases symmetrically.2
The weights are designed to reflect the international trade competi-

tiveness of U.S.-produced goods. They are based on each countrys bi-
lateral share of trade with the United States, as well as trade shares in
third countries where both the United States and a foreign country com-
pete for sales. For U.S. imports, a countrys weight is its share of U.S.
merchandise imports, excluding oil imports. For U.S. exports, two
weights are calculated. One weight represents each countrys share of
U.S. nongold/nonmilitary merchandise exports, and measures the com-
petition between U.S. goods and the countrys goods in the countrys
home market. The other export weight reflects competition in a third
country between imports of U.S. goods and goods from a competing
country. It is defined, formally, as the sum of the shares of U.S. exports
to third-market countries multiplied by the shares of the competing
countrys exports to the third countries in those countries imports. The
overall weights are weighted averages of these three weights.
The bilateral export weights exclude trade in military goods and gold,
and the bilateral import weights exclude oil. Military goods are not con-
sidered to be heavily influenced by trade competitiveness, because po-
litical and strategic factors dominate price differentials between U.S.-
and foreign-produced goods. Oil and gold are largely homogeneous prod-
ucts that are priced in world auction markets, which reflect global sup-
ply and demand rather than U.S. bilateral trade alone.
The weights are revised every year, based on updated data. For the
current calendar year, the weights reflect the trade flows of the year
before the immediately preceding year. Because trade data become avail-
able with a lag and are also subsequently revised, weights for the cur-
rent calendar year are subject to at least two revisions. The weights used
initially are based on trade data of two years earlier; they are revised
later when the data for the immediately preceding year become avail-
able, and they are revised again when the data for the calendar year
become available. For example, weights for the index calculations for
January 2006 will first be based on 2004 trade data; later in 2006, they
will be revised to reflect 2005 trade volumes; and in late 2007, they will
be revised to reflect 2006 trade flows. Weights may also be affected by
occasional revisions of past trade data.
Indexes of the real value of the dollar are based on differential move-
ments of consumer price indexes (CPI) in the United States and other
countries to convert nominal to real exchange rates. In order to compen-
sate for the late availability of the CPI for some countries, initial esti-
mates of the CPI values for the current month and recent months are
based on extrapolations of the most recent twelve-month in which data
are available for each country.
Accuracy
There are no estimates of sampling or revision error for the value-of-

the-dollar indexes.
Relevance
The value of the dollar affects the U.S. economy in several ways. It
influences the competitive position of U.S. goods and services in ex-
port and domestic markets, inflation, Federal Reserve monetary poli-
cies on interest rates, and stock prices (stock market price aggregates
and divided yields).
When the dollar is perceived to have become low in relation to its
intrinsic value compared with other currencies, the effect is thought to
raise U.S. production and prices, while when the dollar is considered to
have become relatively high, the effect is thought to lower U.S. produc-
tion and prices. For production measures, see industrial production in-
dex and gross domestic product; for price change measures, see the
But changes in the dollar are not transmitted to prices in a one-to-one
relationship. Part or all of the dollar changes may be offset by opposing
changes in export and import prices, as American and foreign exporters
try to maintain market shares rather than profit margins. Thus, a fall in
the dollar may be followed by a partial, but not fully compensating,
increase in U.S. export prices and a decline in import prices, while a rise
in the dollar may be followed by a partial decrease in export prices and
increase in import prices. The extent of these pass-throughs of price
changes that partly or fully offset changes in the value of the dollar can
be calculated in relation to import and export price indexes through the
preparation of supplementary trade-weighted exchange rates.
This is further complicated because income increases in the United
States generate more of an increase in U.S. imports than comparable
income increases in foreign countries generate in U.S. exports. This pat-
tern suggests that in order to maintain a stable surplus or deficit in the
balance of payments in the long run, continual devaluations of the U.S.
dollar would be required when the rest of the world grows at least as fast
(on a trade-weighted basis) as the United States.3
The value of the dollar also affects monetary policies by which the
Federal Reserve influences the economy. For example, large deficits in
the U.S. balance of trade are financed by foreigners who invest funds in
the United States. If the dollar declines or is expected to decline in the
future, this funding may be cut back, potentially leading to rising inter-
est rates, declining asset prices, and further downward pressure on the
U.S. dollar. In this situation, the Federal Reserve is faced with the pros-
pect of allowing greater increases in interest rates, which may raise the
likelihood of a recession. For the Federal Reserve, this international di-
mension complicates the development of appropriate policies.
In addition, the value of the dollar affects stock market price aggre-
gates and dividend yields through foreign investment in equities of U.S.
corporations. Expectations of a rising value of the dollar may induce
foreigners to buy U.S. stocks, which in turn tends to raise stock prices,
while expectations of a declining value of the dollar lessen the incentive
for foreigners to buy U.S. stocks (or heighten the incentive to sell U.S.
stocks), which tends to lower stock prices.
Because consumer price movements in the United States and other
industrialized countries are often in the same general range, the nominal
dollar indexes that are based on currencies of industrialized countries,
notably the major currencies index, are similar to the real dollar indexes.
By contrast, because industrializing countries typically have greater price
increases and sometimes encounter hyperinflation rates that no longer
occur in industrialized countries, the nominal and real indexes that con-
tain higher proportions of industrializing countries have increasingly
divergent movements. Generally, when the nominal and real indexes
have large differential movements, the real index is relevant for assess-
ing the long-term impacts of currency fluctuations on trade patterns.
Recent Trends
The three value-of-the-dollar indexes increased in nominal and in real

values with occasional interruptions from 1995 until 2002, and then typi-
cally declined in both nominal and real values during 200304 (Table
49.2). For the entire 19952004 period, in nominal values the OITP
index increased the most, the broad index showed the next largest in-
crease, and the major currencies index showed virtually no change. Also
Table 49.2
Value-of-the-dollar Indexes: Nominal and Real
Broad index Major currencies index OITP

(January 1997 = 100) (March 1973 = 100 (January 1997 = 100)
Nominal
1995 92.7 83.4 92.5
1996 97.5 87.2 98.2
1997 104.4 93.9 104.6
1998 115.9 98.4 125.9
1999 116.0 96.9 129.2
2000 119.4 101.6 129.8
2001 125.9 107.7 135.9
2002 126.8 106.0 140.6
2003 119.3 93.0 144.0
2004 113.8 85.4 144.0
Real
1995 86.9 81.0 104.1
1996 89.0 85.9 101.1
1997 93.7 93.2 102.1
1998 101.7 98.2 115.5
1999 101.1 98.0 114.2
2000 105.0 104.7 114.3
2001 111.1 112.2 119.0
2002 111.3 110.6 121.6
2003 104.6 97.7 123.3
2004 100.0 90.7 122.1
for the entire 19952004 period, the most striking difference between
the nominal and real values of the indexes occurred in the major curren-
cies index, in which the nominal index showed little change while the
real index increased, in contrast to the OITP index, in which the nomi-
nal values increased much more than the real values.

Notes
1. The three indexes were introduced in 1998. See Michael Leahy, New Sum-
mary Measures of the Foreign Exchange Value of the Dollar, Federal Reserve Bul-
letin (October 1998). This was updated in Mico Loretan, Indexes of the Foreign
Exchange Value of the Dollar, Federal Reserve Bulletin (Winter 2005).
2. For example, consider an exchange rate index that contains only two curren-
cies, with the two currency weights equal to one-half each. Suppose that one
currencys exchange value against the dollar doubles (i.e., appreciates by 100 per-
cent), while the other exchange value is cut in half (i.e., depreciates by 50 percent).
A geometric average of two exchange rates21/2 0.51/2 = 1implies no change in
the exchange rate index, while taking an arithmetic average(1/2 2) + (1/2 x 1/2)
= 1.25implies an appreciation of the dollar.
3. C. Fred Bergsten and John Williamson, Overview: Designing a Dollar Policy,
and Michael Mussa, Exchange Rate Adjustments Needed to Reduce Global Pay-
ments Imbalances, both in Dollar Adjustment: How Far? Against What? ed. C.
Fred Bergsten and John Williamson (Washington, DC: Institute for International
Economics, 2004), pp. 23 and 120.
Appendix:
Note on Sampling and
Nonsampling Errors in
Statistical Surveys
This description of sampling and nonsampling errors in statistical sur-

veys draws largely on explanations by the U.S. Bureau of the Census, to
which I have made certain additions. For example, see U.S. Census Bu-
reau, Housing Vacancy Survey: Source and Accuracy of Estimates,
Second Quarter 2004, p. 5, www.census.gov. I have also benefited con-
siderably from discussions with Thomas Jabine.
Sampling Error
Sampling error in statistical surveys (e.g., of households, businesses,

or governments) occurs because of variations in the estimated data
obtained from any one sample that was surveyed rather than from
the entire population. A sampling error can be calculated only for a
survey sample that is a probability sample (sampling errors cannot
be calculated for nonprobability samples). In a probability sample,
the survey respondents are fully representative of the entire popula-
tion being surveyed, such as of its demographic, economic, and geo-
graphic characteristics. A probability sample requires that each
member of the entire population has a known chance of being se-
lected as a survey respondent (referred to as random selection). The
calculation of a sampling error assumes that estimates of the differ-
ent samples conform to a statistical normal distribution (bell-shaped
curve) that has no biases or skewness.
The measure of sampling variability is referred to as the standard
error, which gives the probable error associated with the data estimate
that is based on a single sample. The sample estimate and its standard
263
error enable the preparation of confidence intervals, which are ranges

that would include the average results of all possible samples with a
known probability. For example, if all possible samples were selected,
each of the samples being surveyed were under essentially the same
general conditions and use the same sample design, and an estimate and
its standard error were calculated for each sample, then the approximate
confidence intervals are:
For one standard error, which is a confidence interval of approxi-
mately 67 percent, with a likely chance of occurring in two of three
cases, 67 percent of the intervals from one standard error below the data
estimate to one standard error above the data estimate would include the
average of all possible samples.
For 1.6 standard errors, the confidence interval is approximately 90
percent, with a likely chance of occurring in nine of ten cases.
For two standard errors, the confidence interval is approximately 95
percent, with a likely chance of occurring in 19 of 20 cases.
For three standard errors, the confidence interval is approximately
99 percent, with a likely chance of occurring in 99 of 100 cases.
Of course, as the confidence interval increases from one to three stan-
dard errors, the size of the probable error in the data estimate also in-
creases. Also, the average data estimate derived from all possible samples
may not be contained in any particular computed interval. However, for
a particular sample, one can say with specified confidence that the aver-
age data estimate derived from all possible samples is included within
the confidence interval.
Nonsampling Error
In contrast to sampling errors, nonsampling errors in statistical surveys

cannot be calculated from the sample itself. Examples of nonsampling
errors are definitional difficulties in the survey questions, differences in
the interpretation of questions, inability or unwillingness of respondents
to provide correct information, inability to recall information, errors made
in the collection or coding the data, errors made in processing the data,
errors made in estimating missing data, and failure to represent all units
of the entire population in the sample.
While nonsampling errors are more difficult to measure than sam-
pling errors, there are ways of assessing certain aspects of nonsampling
errors. For example: Sometimes special studies are undertaken to mea-
NOTE ON SAMPLING AND NONSAMPLING ERRORS 265
sure nonsampling errors, such as post-numeration surveys, and by

checking the sample survey data against comparable data such as from
administrative records of government programs. Another technique in
assessing nonsampling error is to analyze the nonresponse rate in a
survey to evaluate the characteristics of the nonrespondents. And in
the case of coding classifications such as of industries or occupations,
an independent verification of the sample is a check to see that the
coding was done correctly.
References
Administrative Office of the United States Courts, Statistics Division. Statistical

Tables for the Federal Judiciary. Quarterly.
Aizcorbe, Ana M., Arthur B. Kennickell, and Kevin B. Moore. 2003. Recent Changes
in U.S. Family Finances: Evidence from the 1998 and 2001 Survey of Consumer
Finances. Federal Reserve Bulletin. January.
American Bankers Association. Consumer Credit Delinquency Bulletin. Washing-
ton, DC. Quarterly.
Aten, Bettina H. 2005. Report on Interarea Price Levels. Bureau of Economic
Analysis, U.S. Department of Commerce. April 25. Available from the Bureau of
Economic Analyses, www.bea.gov.
Bergsten, C. Fred, and John Williamson. 2004. Overview: Designing a Dollar
Policy. In Dollar Adjustment: How Far? Against What? ed. C. Fred Bergsten
and John Williamson. Washington, DC: Institute for International Economics.
Board of Governors of the Federal Reserve System. Federal Reserve Bulletin.
Quarterly.
. Statistical Supplement to the Federal Reserve Bulletin. Monthly.
Broda, Christian, and David Weinstein. 2005. Are We Underestimating the Gains
from Globalization for the United States? Current Issues in Economics and
Finance, Federal Reserve Bank of New York. April; idem. 2004. Globalization
and the Gains from Variety. NBER Working Paper 10314. National Bureau of
Economic Research. Cambridge, MA.
Bureau of the Census, U.S. Department of Commerce. FT 900 Supplement. Monthly.
. Housing Starts. Monthly.
. Housing Vacancy Survey. Quarterly.
. Income, Poverty, and Health Insurance Coverage in the United States.
Annual.
. 2005. Alternative Poverty Estimates in the United States: 2003. June.
. Manufacturers Shipments, Inventories, and Orders. Monthly.
. Monthly Retail Trade and Food Services Survey, and Annual Benchmark
Report for Retail Trade and Food Services. Monthly and Annual.
. Price Index of New One-Family Houses Sold. Quarterly.
. Quarterly Revenue for Selected Services. Quarterly.
267
268 REFERENCES
Bureau of the Census, U.S. Department of Commerce, and U.S. Department of

Housing and Urban Development. New Residential Sales. Monthly.
Business. Monthly.
Bureau of Economic Analysis and Bureau of the Census, U.S. Department of Com-
merce. U.S. International Trade in Goods and Services. Monthly.
Bureau of Labor Statistics, U.S. Department of Labor. CPI Detailed Report. Monthly.
. Business Employment Dynamics. Quarterly.
. Employment and Earnings. Monthly.
. Job Openings and Labor Turnover. News Release. Monthly.
. Monthly Labor Review. Monthly.
. Multifactor Productivity Trends. Annual.
. PPI Detailed Report. Monthly
. 2001. Business Cycles Indicators Handbook.
. Consumer Confidence Survey. Monthly.
Employment and Training Administration, U.S. Department of Labor. Unemploy-
ment Insurance Weekly Claims Report. Weekly.
Executive Office of the President, Office of Management and Budget. 2004. Statis-
tical Programs of the United States Government: Fiscal Year 2005.
Fisher, Gordon. 1997 (revised). From Hunter to Orshansky: An Overview of (Un-
official) Poverty Lines in the United States from 1904 to 1965. Mimeo. This and
the following two papers can be obtained from Fisher at the Office of the Assis-
tant Secretary for Planning Evaluation, U.S. Department of Health & Human
Services.
. 1997 (revised). The Development of the Orshansky Poverty Thresholds
and Their Subsequent History as the Official U.S. Poverty Measure. Mimeo.
. 1995. Is There Such a Thing as an Absolute Poverty Line over Time?
Evidence from the United States, Britain, Canada, and Australia on the Income
Elasticity of the Poverty Line. Mimeo.
. 1992. The Development and History of the Poverty Thresholds. Social
Security Bulletin. Winter.
Fixler, Dennis J., and Bruce T. Grimm. 2005. Reliability of the NIPA Estimates of
U.S. Economic Activity, Survey of Current Business. February.
Frumkin, Norman. 2004. Tracking Americas Economy. 4th ed. Armonk, NY: M.E.
Sharpe.
Fuchs, Victor. 1965. Toward a Theory of Poverty. The Concept of Poverty. Wash-
ington, DC: Chamber of Commerce of the United States.
Garner, C. Alan. 1998. A Closer Look at the Employment Cost Index. Economic
Review. Federal Reserve Bank of Kansas City. Third Quarter.
Groshen, Erica L., and Simon Potter. 2003. Has Structural Change Contributed to
a Jobless Recovery? Current Issues in Economics and Finance. Federal Re-
serve Bank of New York. August.
Institute for Supply Management. ISM Manufacturing Report on Business. Tempe,
AZ. Monthly.
. ISM Non-Manufacturing Report on Business. Tempe, AZ. Monthly.
Kennickell, Arthur B. 2003. A Rolling Tide: Changes in the Distribution of Wealth
in the U.S., 19892001. Federal Reserve Board. September.
REFERENCES 269
Kowalewski, Kim. 2000. Personal Bankruptcy: A Literature Review. CBO Paper.

Congressional Budget Office. September.
Leahy, Michael. 1998. New Summary Measures of the Foreign Exchange Value of
the Dollar. Federal Reserve Bulletin. October.
Lee, Tae-Hwy, and Stuart Scott. 1999. Investigating Inflation Transmission by Stages
of Processing. Cointegration, Causality, and Forecasting: A Festschrift in Honor
of Clive W.J. Granger, ed. Ralph Engle and Halbert White. Oxford, UK: Oxford
University Press.
Liang J. Nellie, and Steven A. Sharpe. 1999. Share Repurchases and Employee
Stock Options and Their Implications for S & P 500 Share Retirements and Ex-
pected Returns. Finance and Economics Discussion Series no. 59. Federal Re-
serve Board.
Loretan, Mico. 2005. Indexes of the Foreign Exchange Value of the Dollar. Fed-
eral Reserve Bulletin. Winter.
McGuckin, Robert H., Ataman Ozyildirum, and Victor Zarnowitz. 2004. A More
Timely and Useful Index of Leading Indicators. The Conference Board. Octo-
ber.
Meyer, Peter B., and Michael J. Harper. 2005. Preliminary estimates of multifactor
productivity growth. Monthly Labor Review. June.
Modigliani, Franco, and Richard Sutch. 1996. Innovations in Interest Rate Policy.
American Economic Review. March.
Mortgage Bankers Association of America. National Delinquency Survey. Quar-
terly.
. Weekly Mortgage Applications Survey. Weekly.
Mussa, Michael. 2004. Exchange Rate Adjustments Needed to Reduce Global Pay-
ments Imbalances. In Dollar Adjustment: How Far? Against What? ed. C. Fred
Bergsten and John Williamson. Washington, DC: Institute for International Eco-
nomics.
National Agricultural Statistics Service, U.S. Department of Agriculture. Agricul-
tural Prices. Monthly.
Monthly.
National Research Council. 1995. Measuring Poverty: A New Approach, ed.
Constance F. Citro and Robert T. Michael. Washington, DC: National Academy
Press.
Office of Federal Housing Enterprise Oversight, U.S. Department of Housing and
Urban Development. House Price Index. Quarterly.
Orshansky, Mollie. 1963. Children of the Poor. Social Security Bulletin. July.
. 1965. Counting the Poor: Another Look at the Poverty Profile. Social Secu-
rity Bulletin. January. Reprinted in the Social Security Bulletin in October 1988.
Paquin, Paul, and Melissa Squire Weiss. 1997. An Analysis of the Determinants of
Personal Bankruptcies. Fall Church, VA: Capital One Financial Corporation.
October.
Perozek, Maria G., and Marshall B. Reinsdorf. 2002. Alternative Measures of Per-
sonal Saving. Survey of Current Business. April.
Ruser, John, Adrienne Pilot, and Charles Nelson. 2004. Alternative Measures of
Household Income: BEA Personal Income, CPS Money Income, and Beyond.
Available from the Census Bureau, www.census.gov.
270 REFERENCES
Ryscavage, Paul. 1999. Income Inequality in America. Armonk, NY: M.E. Sharpe.
Schreft, Stacey L., and Aarti Singh. 2003. A Closer Look at Jobless Recoveries.
Economic Review. Federal Reserve Bank of Kansas City. Second Quarter.
University of Michigan, Survey Research Center. Surveys of Consumers. Ann Ar-
bor. Monthly.
Visa U.S.A. Inc. 1996. Consumer Bankruptcy: Causes and Implications. July.
WEFA Group, Resource Planning Service. 1998. The Financial Costs of Personal
Bankruptcy. Burlington, MA. February.
Zarnowitz, Victor. 1999. Theory and History Behind Business Cycles: Are the 1990s
the Onset of a Golden Age? Journal of Economic Perspectives. Spring.
Index
Administrative Office of the United States Balance of payments, 1116, 18, 19, 20, 156,
Courts, 26, 28 157
Agricultural Adjustment Act (1939), 88 data accuracy, 14
Agricultural Prices, 88 data availability
American Bankers Association, 45 dates for, 11
American Housing Survey, 63, 119 sources, 11
American Samoa, 18 website, 11
Annual Benchmark for Retail Trade and Food data content, 1213
Services, 221 data methodology, 1314
Annual Survey of Manufacturers, 146, 180 data relevance, 1415
Assets. See Balance of payments; dollar value, 25960
Distribution of wealth; International import/export prices indexes, 142
investments, United States international investment, 158
Average weekly earnings, 36, 84, 85 defined, 11
data accuracy, 45 recent trends, 1516
data availability Balance of trade, 13, 1722
dates for, 3 data accuracy, 20
sources, 3 data availability
website, 3 dates for, 17
data content, 34 sources, 17
data methodology, 4 website, 17
data relevance, 5 data content, 1718
business sector productivity, 218 data methodology, 1920
distribution of income, 65 data relevance, 2021
government economic transactions, 100 balance of payments, 15
defined, 3 import/export price indexes, 142
recent trends, 56 defined, 17
Average weekly hours, 4, 710, 215, 25253 recent trends, 21
data accuracy, 8 Bank loans, commercial/industrial,
data availability 2325
dates for, 7 data accuracy, 24
sources, 7 data availability
website, 7 dates for, 23
data relevance, 9 data content, 2324
help-wanted advertising index, 115 data methodology, 24
defined, 7 data relevance, 2425
recent trends, 9 corporate profits, 59
271
272 INDEX
Bank loans, commercial/industrial (continued) Bureau of the Census (continued)

defined, 23 corporate profits, 58
recent trends, 25 distribution of income, 61, 63, 64, 69
Bankruptcies, personal, 2630 employment, 75, 78
data accuracy, 28 government economic transactions, 98
data availability home sales, new/existing, 117, 118
dates for, 26 house prices, new/existing, 122, 124
sources, 26 housing affordability index, 130
website, 26 housing starts, 132
data content, 2627 housing vacancy rate, 136, 137
data methodology, 28 import/export price indexes, 141
data relevance, 28 industrial production index, 145, 146
consumer credit delinquency, 46, 47 inventory-sales ratios, 159
defined, 26 manufacturers orders, 178
legislation, 2930n.1 personal income/saving, 197
recent trends, 29 poverty, 200, 202, 205
types of, 27, 29 retail sales, 221
Bankruptcy Abuse, Prevention and Consumer selected services revenue, 225
Protection Act (2005), 30n.1 unemployment, 244
Bankruptcy Reform Act (1978), 29n.1 Business Cycle Dating Committee, 175
BEA News, 92 Business Cycle Indicators, 36, 113, 171
Beveridge Curve, 170 Business Employment Dynamics, 163
Bureau of Customs, 19 Business profits/investment. See Corporate
Bureau of Economic Analysis (BEA) profits
balance of payments, 1314
balance of trade, 1718, 2021 Canada, 18, 77, 79
corporate profits, 56 Capacity utilization, 3134, 144
distribution of income, 64 data accuracy, 33
GDP price measures, 92, 93 data availability
government economic transactions, 96, 98 dates for, 31
gross domestic product, 103, 108 sources, 31, 32
international investment, 155, 156 website, 31
personal income/saving, 195 data content, 3132
poverty, 203 data methodology, 3233
producer price indexes, 209 data relevance, 33
Bureau of Labor Statistics (BLS) defined, 31
average weekly earnings, 3 recent trends, 3334
average weekly hours, 7 Commerce Clearinghouse, 63
balance of payments, 11 Company Organization Survey, 75
business sector productivity, 214, 216 Conference Board
consumer price index, 48, 50 consumer attitude indexes, 35, 36, 40
employment, 73, 76, 78 help-wanted advertising index, 11314
employment cost index, 84 leading/coincident/lagging indicators, 171,
government economic transactions, 98 177
import/export price indexes, 140, 141 Conference Board Help-Wanted Online Data
inventory-sales ratio, 160 Series, The, 114
job gains/losses, 163 Consumer attitude indexes, 3541
job openings/labor turnover, 167 comparison of, 3941
producer price indexes, 207, 209 data collection, 40
unemployment, 239, 244 household sample, 40
unit labor costs, 251 index content, 39
Bureau of the Census questions, 3940
balance of payments, 1314 recent trends, 4041
balance of trade, 1721 revisions, 40
capacity utilization, 32 seasonal adjustment, 40
consumer credit, 43 survey response, 40
INDEX 273
Consumer attitude indexes (continued) Consumer price index (CPI) (continued)

consumer confidence index (CCI), 3537, data relevance, 53
3941 bankruptcies, personal, 28
data accuracy, 37 business sector productivity, 218
data availability, 3536 distribution of income, 65
data content, 36, 3940 dollar value, 25960
data methodology, 3637, 40 employment cost index, 86
data relevance, 40 GDP price measures, 9394
defined, 35 gross domestic product, 109, 111
consumer sentiment index (CSI), 3741 import/export price indexes, 142
data accuracy, 39 unemployment, 245
data availability, 37 defined, 48
data content, 3738 legislation, 53
data methodology, 3839, 40 recent trends, 5354
data relevance, 40 Consumer purchasing power, 5, 198
defined, 37 See also Expenditures, consumer; Personal
data relevance, 40 income/saving
consumer credit, 43 Consumer Research Center, 35
defined, 35, 37 Consumption, 105t, 1067, 109
Consumer Confidence Survey, 36 Corporate profits, 5660, 105t, 107, 172
Consumer credit, 30n.2, 4244 data availability
data accuracy, 43 dates for, 56
data availability sources, 56
dates for, 42 website, 56
sources, 42 data content, 5658
website, 42 data relevance, 9, 5859
data content, 4243 bankruptcies, personal, 28
data methodology, 43 capacity utilization, 33
data relevance, 4344 defined, 56
consumer credit delinquency, 46 recent trends, 60
house prices, new/existing, 126 Cost-of-living index (COL), 48, 49
personal income/saving, 198 Council of Economic Advisors, 204
defined, 42 County Business Patterns (CBP), 75
recent trends, 44 CPI Detailed Report, 48
Consumer Credit Delinquency Bulletin, 45 Credit. See Consumer credit; Consumer credit
Consumer credit delinquency (CCD), 4547 delinquency
data accuracy, 46 Credit Union National Association, 43
data availability Currency. See Money supply; Value of the
dates for, 45 dollar
sources, 45 Current Population Survey (CPS)
website, 45 distribution of income, 63, 64
data content, 4546 employment, 78
data methodology, 46 help-wanted advertising index, 130
data relevance, 4647 housing vacancy rate, 137
defined, 45 poverty, 2023
recent trends, 47 unemployment, 244
Consumer price index (CPI), 4855 Cyclical/forecasting indicators. See Leading/
data accuracy, 5253 coincident/lagging indexes; PMI
data availability
dates for, 48 Data sharing between federal government
sources, 48 statistical agencies, xvxvii
website, 48 Debt, 14858, 235
data content, 4850, 8990 commercial, 2425
data methodology, 4, 5052, 124, 210, consumer, 2630, 40, 4347, 70, 126
25859 foreign, 21
chain index, 51, 145 government, 96102, 204, 245
274 INDEX
Department of Agriculture Department of Treasury

balance of payments, 14 balance of payments, 1314
farm parity ratio, 88, 89 balance of trade, 19
import/export price indexes, 141 consumer credit, 43
poverty, 202 government economic transactions, 98
producer price indexes, 209 interest rates, 14854
Department of Commerce international investment, 156
balance of payments, 11 personal income/saving, 197
balance of trade, 17 Diffusion indexes, 35, 37, 39, 188, 191, 192
corporate profits, 56 Distribution of income, 6167, 130
distribution of income, 60 data accuracy, 64
GDP price measures, 92 data availability
government economic transactions, 96 dates for, 61
gross domestic product, 103 sources, 61
home sales, new/existing, 117 website, 61
house prices, new/existing, 122 data content, 6263
housing starts, 132 data methodology, 6364
housing vacancy rate, 136 data relevance, 6465, 71
international investment, 155 distribution of wealth, 71
inventory-sales ratios, 159 employment, 81
manufacturers orders, 178 government economic transactions, 100
personal income/saving, 195 housing starts, 134
PMI, 191 poverty, 205
poverty, 200 defined, 61
producer price indexes, 209 recent trends, 6566
retail sales, 221 See also Employment; Personal income/saving
selected services revenue, 225 Distribution of wealth, 6872
Department of Defense data accuracy, 7071
balance of payments, 14 data availability
balance of trade, 1920 dates for, 68
Department of Energy sources, 68
capital utilization, 32 website, 68
import/export price indexes, 141 data content, 6870
producer price indexes, 209 assets, 6970
Department of Homeland Security, 13 liabilities, 70
Department of Housing and Urban data methodology, 70
Development data relevance
distribution of income, 63 distribution of income, 65
home sales, 117, 119 employment, 81
housing affordability index, 125 poverty, 205
Department of Justice, 27 stock market price aggregates/dividend
Department of Labor yields, 234
average weekly earnings, 3 defined, 68
average weekly hours, 7 recent trends, 7172
business sector productivity, 214, See also Personal income/saving
216 District of Columbia, 12, 18, 248
consumer price index, 48 Dollar value. See Value of the dollar
employment, 73, 76
employment cost index, 84 Earned Income Tax Credit (EITC), 205
import/export price indexes, 140 Earnings. See Average weekly earnings;
job gains/losses, 163 Distribution of income; Distribution of
job openings/labor turnover, 167 wealth; Employment cost index;
producer price indexes, 207 Personal income/saving
unemployment, 239, 246, 248 Economic growth, 20, 65, 81, 152, 219, 234, 245
unit labor costs, 251 See also Expansions; Gross domestic
Department of the Interior, 32 product
INDEX 275
Economic well-being Employment cost index (continued)

bankruptcies, personal, 2630 data content, 8485
consumer attitude indexes, 3541 data methodology, 8586
consumer credit delinquency, 4547 data relevance, 8687
distribution of income, 6167 defined, 4, 84
distribution of wealth, 6872 recent trends, 87
farm parity ratio, 8891 See also Unit labor costs, business sector
government economic transactions, 96102 Exchange rates, 142, 156, 25562
mortgage delinquency/foreclosure, 18587 Expansions, 4344, 59, 115, 146, 17172,
poverty, 200206 175, 177, 246, 248
Employee compensation, 105t, 107 See also Economic growth; Gross domestic
See also Employment cost index product
Employers Quarterly Tax Return (Form 941), Expenditures, 152
75 business, 56, 1045
Employment, 38, 7383 consumer, 3740, 43, 94, 99, 187, 19798,
data methodology, 78, 137, 215, 25253 223, 234
data relevance government, 96102, 105t, 106, 111
balance of trade, 20 household, 4855, 1045, 174
business sector productivity, 218 Export-Import Bank, 13
corporate profits, 58 Exports, 1120
government economic transactions, 100 dollar value, 25556, 25859
gross domestic product, 111 gross domestic product, 105t, 106
home sales, new/existing, 120 international investment, 158
housing affordability index, 131 producer price indexes, 208
inventory-sales ratios, 161 See also Import/export price indexes
manufacturers orders, 180
personal income/saving, 198 Farm parity ratio, 8891
defined, 3, 73 data accuracy, 90
establishment survey, 4, 8, 7376, 7983 data availability
data availability, 7374 sources, 88
data content, 74, 7980 website, 88
data methodology, 7476 data content, 8889
household survey, 7683 data methodology, 8990
data accuracy, 7879 data relevance, 90
data content, 77, 7980 legislation, 88
data methodology, 78 recent trends, 9091
survey comparison, 7983 Federal funds rate, 14854
component detail, 80 Federal Home Loan Mortgage Corporation, 151
coverage, 7980 Federal Housing Finance Board, 130, 148, 150
data accuracy, 8081 Federal Reserve Bank of New York, 14, 148,
data relevance, 8182 150, 151, 156
recent trends, 8283 Federal Reserve Board (FRB)
See also Help-wanted advertising index; balance of payments, 13, 15
Job gains/losses; Job openings/labor balance of trade, 1921
turnover; Unemployment bank loans, 2324
Employment and Earnings, 3, 7, 73, 76, 239 capacity utilization, 31
Employment and Training Administration, consumer credit, 4243
246, 248 consumer credit delinquency, 47
Employment cost index (ECI), 8487 distribution of wealth, 68, 70
data accuracy, 86 dollar value, 255, 25960
data availability gross domestic product, 111
dates for, 84 industrial production index, 144
sources, 84 interest rates, 148, 149, 151, 152
website, 84 international investment, 156
276 INDEX
Federal Reserve Board (FRB) (continued) Government economic transactions

personal income/saving, 196 (continued)
unemployment, 245 data content, 9798
Federal Reserve Bulletin, 23, 31, 68, 144, 255 data methodology, 98, 99100
Finance data relevance
bank loans, 2325 bankruptcies, personal, 28
bankruptcies, personal, 2630 gross domestic product, 111
consumer credit, 4244 defined, 96
consumer credit delinquency, 4547 recent trends, 100102
Interest rates, 14854 Great Depression, 59
mortgage delinquency, 8587 Gross domestic product (GDP), 99112,
stock market price aggregates/dividend 19396
yields, 22937 current dollars, 103, 107
Fisher ideal index, 93, 1089, 145 data accuracy, 110
Flow-of-funds accounts, 196 data availability
Foreign direct investments, 15657 dates for, 1034, 251
Foreign trade, 52 sources, 103
See also Balance of payments; Balance of website, 103
trade; Import/export price indexes; data content, 12, 5658, 160, 252
Value of the dollar demand components, 1036
Foreign Trade Zones, 18 income side, 1057
FT 900 Supplement, 17 product side, 1046
Full Employment and Balanced Growth Act supply components, 103, 1057
(1978), 53, 246 data methodology, 10810, 145, 189, 253
data relevance, 11011
GDP price measures, 9295, 145, 197 balance of payments, 15
chain-type price index, 9295 balance of trade, 20
data accuracy, 94 business sector productivity, 219
data availability capacity utilization, 33
dates for, 92 consumer credit delinquency, 46
sources, 92 consumer price index, 53
website, 92 demand components, 110
data content, 9293 dollar value, 259
data methodology, 9394 employment, 81
data relevance, 94 import/export price indexes, 142
bankruptcies, personal, 28 industrial production index, 146
business sector productivity, 218 interest rates, 152
distribution of income, 65 mortgage delinquency/foreclosure, 187
dollar value, 259 personal income/saving, 198
employment cost index, 86 producer price indexes, 211
import/export price indexes, 142 retail sales, 223
defined, 92 selected services revenue, 227
implicit price deflator, 9293, 95 supply components, 110
recent trends, 9495 unemployment, 245
See also Gross domestic product defined, 103, 195, 21315, 217
Gini index, 61, 62, 64, 65 chained dollars, 107, 109
Government. See Federal Reserve Board; current dollars, 107
Internal Revenue Service; and specific nominal GDP, 107
Bureaus/Departments real GDP, 103, 10712, 198
Government economic transactions, 96102, See also GDP price measures; Government
154, 204, 245 economic transactions
data accuracy, 99 Guam, 18
data availability
dates for, 9697 Hedonic price indexes, 52, 124, 210
sources, 96 Help-wanted advertising index, 11316
website, 96 data accuracy, 114
INDEX 277
Help-wanted advertising index (continued) Housing affordability index (HAI)

data availability (continued)
dates for, 113 data availability
sources, 113 dates for, 128
website, 113 sources, 128
data content, 11314 website, 128
data methodology, 114 data content, 12829
data relevance, 11415 data methodology, 12930
recent trends, 11516 home sales, new/existing, 120
Henry George effect, 123 house prices, new/existing, 126
Home sales, new/existing, 11721, 130 housing starts, 134
data relevance, 11920 mortgage delinquency/foreclosure, 187
housing starts, 134 defined, 128
mortgage delinquency/foreclosure, 187 recent trends, 131
mortgage loan applications, 183 See also Home sales, new/existing
defined, 117 Housing starts, 78, 13235
existing homes, 11821 data accuracy, 133
data accuracy, 119 data availability
data availability, 118 dates for, 132
new homes, 11718 data content, 13233
data accuracy, 118 data methodology, 133
data availability, 117 data relevance, 134
data content, 118 home sales, new/existing, 120
data methodology, 118 mortgage delinquency/foreclosure, 187
defined, 117, 119 defined, 132
recent trends, 12021 recent trends, 135
See also Mortgage loan applications Housing Starts, 132
House Price Index, 125 Housing vacancy rate, 13639
House prices, new/existing, 12227 data accuracy, 138
comparison, 12627 data availability
data relevance, 126 dates for, 136
recent trends, 12627 sources, 136
data relevance website, 136
housing vacancy rates, 139 data content, 13637
existing houses, 12526 data methodology, 13738
data accuracy, 126 data relevance, 13839
data content, 125 recent trends, 139
data methodology, 12526 Housing Vacancy Survey, 136
defined, 122 Humphrey-Hawkins Act (1978), 53, 246
new houses
data accuracy, 124 Import/export price indexes, 93, 109, 14043
data availability, 122 data accuracy, 142
data content, 12324 data availability, 140
data methodology, 124 dates for, 140
defined, 122 sources, 140
House purchase index, 18384 website, 140
Household income/expenditure. See data content, 14041
Consumer price index; Expenditures, data methodology, 14142
household; Income, household; Retail data relevance, 14243
sales value of the dollar, 259
Housing affordability index (HAI), 12831 defined, 140
data accuracy, 130 recent trends, 14243
278 INDEX
Imports, 1120 Interest rates

dollar value, 25556, 25859 data relevance (continued)
gross domestic product, 105t, 1067 consumer credit delinquency, 47
international investment, 158 dollar value, 25960
producer price indexes, 207 government economic transactions,
Income 99100
corporate, 56 gross domestic product, 111
domestic, 21, 108 home sales, new/existing, 120
government, 96103 house prices, new/existing, 126
household, 36, 39, 12831, 134 housing starts, 134
See also Average weekly earnings; international investment, 158
Distribution of income; Distribution mortgage delinquency/foreclosure,
of wealth; Employment; Employment 187
cost index; Personal income/saving; unemployment, 245
Poverty unit labor costs, 253
Income, Poverty, and Health Insurance defined, 107, 148
Coverage in the United States, 200 recent trends, 154
Industrial production index (IPI), 3133 types of, 14951
data accuracy, 146 Internal Revenue Service, 58, 63, 75
data availability International indicators
dates for, 144 balance of payments, 1116
sources, 144 balance of trade, 1722
website, 144 dollar value, 25562
data content, 145 import/export price indexes, 14043
data methodology, 14546 International investment, United States, 15558
data relevance, 146 data accuracy, 157
dollar value, 259 data availability
employment, 81 dates for, 155
government economic transactions, 100 sources, 155
inventory-sales ratios, 161 website, 155
manufacturers orders, 180 data content, 15556
defined, 144 data methodology, 15657
recent trends, 147 direct investments, 157
Inflation portfolio investments, 157
balance of trade, 18, 20 data relevance, 15758
capacity utilization, 33 defined, 155
dollar value, 259 recent trends, 158
gross domestic product, 111 See also Balance of payments; Balance of
home sales, new/existing, 119 trade
interest rates, 148, 152 International Monetary Fund (IMF)
unemployment, 24546 balance of payments, 13
See also Consumer price index; GDP price balance of trade, 19
measures; Producer price indexes international investment, 156
Institute for Supply Management, 188, 189, International Trade Administration, 13
191, 193 Inventory-sales ratios, 15962, 179
Inter-area price indexes, 49, 54n.2 data accuracy, 160
Interest rates, 21, 38, 105t, 14854, 174 data availability
data availability sources, 159
dates for, 149 website, 159
sources, 148 data content, 15960
websites, 14849 data methodology, 160
data content, 14951 data relevance, 8, 16061
data methodology, 130, 15152 corporate profits, 58
data relevance, 152, 154 defined, 159
bankruptcies, personal, 28 recent trends, 16162
INDEX 279
Investments. See Gross domestic product; Laspeyres index, 109

International investment, United States; Leading/coincident/lagging indexes (LCLg),
Stock market price aggregates/dividend 17177
yields data accuracy, 175
ISM Non-Manufacturing Report on Business, data availability
188, 191 dates for, 171
sources, 171
Job gains/losses, 16366 website, 171
data accuracy, 165 data content, 17274
data availability data methodology, 17475
dates for, 163 data relevance, 175, 177
sources, 163 average weekly hours, 8
website, 163 bank loans, commercial and industrial,
data content, 16364 25
data methodology, 16465 consumer confidence indexes, 40
defined, 163 consumer credit, 44
recent trends, 16566 employment, 81
Job Openings and Labor Turnover Survey, help-wanted advertising index, 115
115, 167 housing starts, 134
Job openings/labor turnover (JOLTS), 20, 36, industrial production index, 146
38, 165, 16770, 218 interest rates, 154
data accuracy, 169 inventory-sales ratios, 161
data availability producer price indexes, 21011
dates for, 167 stock market price aggregates/dividend
sources, 167 yields, 234
website, 167 unemployment, 249
data content, 16768 defined, 171
hires, 168 recent trends, 177
job openings, 168 Loans. See Bank loans; Consumer credit;
separations, 168 Consumer credit delinquency; Interest
data methodology, 16869 rates; Mortgage delinquency/
data relevance, 169 foreclosure; Mortgage loan applications
defined, 167 London Interbank Offered Rate (LIBOR), 150
recent trends, 170
See also Employment; Help-wanted Manufacturers orders, 17881
advertising index; Unemployment data availability
Johnson, Lyndon, 204 dates for, 178
Joint Economic Committee of Congress sources, 178
average weekly earnings, 3 website, 178
average weekly hours, 7 data content, 17879
employment, 74, 77 data methodology, 17980
unemployment, 239 data relevance, 8, 18081
defined, 178
Labor recent trends, 181
average weekly earnings, 36 See also PMI
average weekly hours, 710 Manufacturers Shipments, Inventories, and
business sector productivity, 21320 Orders, 178
employment cost index, 8487 Manufacturing and Trade Inventories and
help-wanted advertising index, 11316 Sales, 159
unit labor costs, 25154 Mexico, 77, 79
See also Employment; Job openings/labor Money supply, xiv
turnover; Unemployment Monthly Labor Review, 3, 7, 48, 73, 76, 84,
Labor force participation and unemployment, 140, 207, 214, 239, 251
24244 Monthly Retail Trade and Food Services
Labor Force Participation Rate (LFPR), Survey, 221
24244 Moodys Investors Service, 148, 150
280 INDEX
Mortgage Bankers Association of America, Office of Federal Housing Enterprise

182, 185 Oversight, 125
Mortgage delinquency/foreclosure, 18587 Office of Management and Budget, 98, 201
data accuracy, 186 Office of Thrift Supervision, 43
data availability Okuns Law, 245
dates for, 185 Overtime, 8, 9
sources, 185 Overall perspective on economic indicators
website, 185 (citation), xiii
data content, 18586
data methodology, 186 Paasche index, 109
data relevance Personal income/saving, 19599, 203
consumer credit delinquency, 46 data accuracy, 198
home sales, new/existing, 120 data availability
defined, 185 dates for, 195
recent trends, 187 sources, 195
Mortgage loan applications, 18284 website, 195
data accuracy, 183 data content, 19697
data availability disposable personal income (DPI), 196
dates for, 182 personal income (PI), 196
sources, 182 data methodology, 19798
website, 182 disposable personal income (DPI),
data content, 18283 19798
data methodology, 183 personal saving (PI), 198
data relevance, 183 data relevance, 19899
defined, 182 bankruptcies, personal, 28
recent trends, 18384 chained dollars, 198
Mortgages, 42, 44, 122, 126, 128, 15052, 154 consumer credit, 43
Multifactor Productivity Trends, 216 consumer credit delinquency, 46
current dollars, 198
National Agricultural Statistics Service, 88 employment, 81
National Association of Realtors (NAR), 118, home sales, new/existing, 120
119, 128, 130 house prices, new/existing, 126
National Bureau of Economic Research defined, 195
(NBER), 175 disposable personal income (DPI), 195
National Delinquency Survey, 185 personal income (PI), 195
National Opinion Research Center (NORC), recent trends, 199
70 disposable personal income (DPI), 199
National Research Council (NRC), 204 personal income (PI), 199
Net worth, 68, 71t, 72 Phillips Curve, 53, 24546
New Residential Sales, 117 PMI, 18889, 19194
Non-manufacturing business activity index data availability
(NMBAI), 18890, 191, 194 dates for, 191
data accuracy, 189 sources, 191
data availability website, 191
dates for, 188 data content, 19192, 193
sources, 188 employment, 191, 193
website, 188 industrial production index, 191, 193
data content, 18889 manufacturers deliveries, 191, 193
data methodology, 189 manufacturers orders, 191, 193
data relevance, 190 total purchased inventories, 191, 193
PMI, 194 data methodology, 19293
Nonaccelerating inflation rate of recent trends, 194
unemployment (NAIRU), 245 Poverty, 200206
Northern Mariana Islands, 18 data accuracy, 203
INDEX 281
Poverty (continued) Productivity, business sector, 21320, 251,

data availability 25354
dates for, 201 data relevance, 21819
sources, 200 distribution of income, 65
website, 200 defined, 21314
data content, 2012 labor-hour productivity
data methodology, 2023 data accuracy, 216
data relevance, 2035 data availability, 214
distribution of income, 65, 71 data content, 21415
PPI Detailed Report, 207 multifactor productivity
Price. See Consumer price index; Farm parity data accuracy, 218
ratio; GDP price measures; Gross data availability, 216
domestic product; Import/export price data content, 21617
indexes; Producer price indexes; Stock data methodology, 21718
market price aggregates/dividend yields defined, 213
Price Index of New One-Family Houses Sold, recent trends, 220
122 Profits. See Corporate profits
Primary credit discount rate, 151, 153t, 154 Proprietors income, 105t, 107
Prime rate, 150, 15254 Puerto Rico, 18, 248
Producer price indexes (PPIs), 20, 90, 93, 109, Purchasing Managers Index. See PMI
145, 20712
data accuracy, 210 Quarterly Financial Report (QFR), 58, 197
data availability Quarterly Revenue for Selected Services,
dates for, 207 225
sources, 207
website, 207 Real Estate Outlook: Market Trends and
data content, 2079 Insights, 118, 128
all-commodities index, 208 Receipts, government, 96102
stage-of-processing index, 2079 Recessions, 43, 59, 11112, 115, 146, 17172,
data methodology, 20910 175, 177, 246, 248, 260
data relevance, 21011 Recovery, 172, 175
business sector productivity, 218 Rental income, 105t, 107
distribution of income, 65 Retail sales, 22124
dollar value, 259 data accuracy, 22223
employment cost index, 86 data availability
gross domestic product, 111 dates for, 221
import/export price indexes, 142 sources, 221
stage-of-processing index, 21011 website, 221
unit labor costs, 253 data content, 22122
recent trends, 21112 data relevance, 8, 223
See also Consumer price index consumer credit, 43
Production consumer credit delinquency, 46
capacity utilization, 3134, 86, 17172, 2079 employment, 81
industrial production, 14447 mortgage loan applications, 183
inventory sales ratios, 15962 personal income/saving, 198
manufacturers orders, 17881 defined, 221
non-manufacturing business activity, recent trends, 22324
18890 Roosevelt, Franklin, 204
PMI, 19194
producer price indexes, 20712 Salary. See Average weekly earnings;
See also Gross domestic product; Industrial Distribution of income; Distribution of
production index; Producer price wealth; Employment cost index
indexes Sales, 172
282 INDEX
Sales (continued) Stock market price aggregates/dividend

See also Consumer price index; Producer yields (continued)
price indexes; Retail sales stock market price aggregates
Sampling and nonsampling errors in data accuracy, 233
statistical surveys, 26365 data availability, 22930
Saving, 40, 97 Surplus, government, 96102
See also Personal income/saving Survey of Construction, 124
Selected services revenue, 22528 Survey of Consumer Finances, 70
data accuracy, 22627 Survey of Current Business, 11, 17, 56, 92,
data availability 96, 103, 155, 195
dates for, 225 Survey of Plant Capacity, 32
sources, 225 Survey Research Center, 3738
website, 225 Surveys of Consumers, 3738
data content, 22526
data methodology, 226 Taxation. See Government economic
data relevance, 227 transactions; Gross domestic product;
defined, 225 Internal Revenue Service
recent trends, 22728 Trade. See Balance of trade; Import/export
Self-Employment Assistance programs, price indexes; Value of the dollar
24748
Service Annual Survey, 226 Unemployment, 7, 137, 23849
Spending. See Expenditures data relevance
Standard & Poors Corporation, 148, 150, bankruptcies, personal, 28
231, 23233 consumer credit delinquency, 46
Standard of living, 50, 158, 204 consumer price index, 53
State Tax Handbook, 63 gross domestic product, 111
Statistical Tables for the Federal Judiciary, 26 help-wanted advertising index, 115
Statistics of Income, 58, 63 house prices, new/existing, 126
Stock market price aggregates/dividend defined, 238
yields, 174, 22937 insured unemployment, 24649
data content, 23031 data accuracy, 248
Dow Jones Industrial Average, 231 data availability, 24647
Nasdaq Composite Index, 231 data content, 24748
New York Stock Exchange Composite data methodology, 248
Index, 23031 data relevance, 24849
Standard & Poors 500 Composite Price defined, 238, 246
Index, 231 recent trends, 249
data methodology, 23133 total unemployment, 23846
Dow Jones Industrial Average, 233 data accuracy, 24445
Nasdaq Composite Index, 232 data availability, 239
New York Stock Exchange Composite data content, 23944
Index, 232 data methodology, 244
Standard & Poors 500 Composite Price data relevance, 24546
Index, 23233 defined, 23839
data relevance, 23435 legislation, 246
consumer credit delinquency, 46 See also Employment; Job gains/losses; Job
dollar value, 25960 openings/labor turnover
defined, 229, 23334 Unemployment Insurance Weekly Claims
dividend yields, 23337 Report, 247
recent trends Unit labor costs, business sector (ULC), 214,
Dow Jones Industrial Average, 23536 25154
Nasdaq Composite Index, 23536 data accuracy, 253
New York Stock Exchange Index, data availability
23536 dates for, 251
Standard & Poors 500 Composite sources, 251
Index, 23536 website, 251
INDEX 283
Unit labor costs, business sector (ULC) Websites (continued)

(continued) balance of payments, 11, 17
data content, 25152 balance of trade, 17
data methodology, 25253 bank loans, 23
data relevance, 25354 bankruptcies, personal, 26
employment cost index, 87 business productivity, 214, 216
defined, 251 capacity utilization, 31
recent trends, 254 consumer attitude indexes, 36, 37
See also Employment cost index consumer credit, 42
United States Government. See Consumer consumer credit delinquency, 45
attitude indexes; Federal Reserve consumer price index, 48
Board; Government economic corporate profits, 56
transactions; Internal Revenue Service; distribution of income, 61, 63
International investment, United States; distribution of wealth, 68
and specific Bureaus/Departments dollar value, 255
University of Michigan, 35, 3738, 40 employment, 73, 76
employment cost index, 84
Value of the dollar, 140, 142, 156, 25562 farm parity ratio, 88
data accuracy, 259 GDP price measures, 92
data availability government economic transactions, 96
dates for, 255 gross domestic product, 103
sources, 255 help-wanted advertising index, 113
website, 255 home sales, new/existing, 117, 118
data content house prices, new/existing, 125
broad index, 25657 housing affordability index, 122, 128
major currencies index, 25657 housing starts, 132
other important trading partner (OITP) housing vacancy rate, 136
index, 25657 import/export price indexes, 140
data methodology, 25759 industrial production index, 144
data relevance, 25960 international investment, 155
balance of payments, 15 interest rates, 14849
balance of trade, 2021 inventory/sales ratios, 159
import/export price indexes, 142 job gains/losses, 163
interest rates, 152 job openings/labor turnover, 167
international investment, 158 leading/coincident/lagging indexes, 171
defined, 255 manufacturers orders, 178
recent trends, 26061 mortgage delinquency/foreclosure, 185
broad index, 26061 mortgage loan applications, 182
major currencies index, 26061 non-manufacturing activity, 188
other important trading partner (OITP) personal income/saving, 195
index, 26061 PMI, 191
Virgin Islands, 18, 248 poverty, 200
producer price indexes, 207
Wages. See Average weekly earnings; retail sales, 221
Distribution of income; Distribution of selected services revenue, 225
wealth; Employment cost index stock market aggregates/dividend yields, 229
Wealth, 158 unemployment, 239, 247
See also Distribution of wealth unit labor costs, 251
Websites Weekly Mortgage Applications Survey, 182
average weekly earnings, 3 Wholesale price indexes, 28
average weekly hours, 7 See also Producer price indexes
About the Author
Norman Frumkin, an economics writer in Washington, D.C., has a long-

standing interest in macroeconomic analysis, forecasting, and policies,
and in improving the quality of economic statistics. He has worked in
the U.S. government, industry, as an independent consultant, and has
taught courses on the use of economic indicators. He is also the author
of Tracking Americas Economy, 4th ed. (M.E. Sharpe, 2004).

(Econ, Econometrics) - Guide To Economic Indicators (4Th Ed)

Uploaded by

Copyright:

Available Formats

(Econ, Econometrics) - Guide To Economic Indicators (4Th Ed)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Econ, Econometrics) - Guide To Economic Indicators (4Th Ed)

Uploaded by

Copyright:

Available Formats

Guide to

Library of Congress Cataloging-in-Publication Data

Printed in the United States of America

The paper used in this publication meets the minimum requirements of

Anne Frances Feldman Frumkin and Joseph Harry Frumkin

Tables, Formulas, and Figure ix

1. Average Weekly Earnings 3

20. Gross Domestic Product 103

Appendix: Note on Sampling and Nonsampling Errors in

1.1 Average Weekly Earnings in Private Nonagricultural

22.1 Home Sales: New and Existing Houses 120

7.1 Capacity Utilization Rate 31

24.1 Housing Affordability Index 130

34.1 Leading, Coincident, and Lagging Composite Indexes 176

Guide to Economic Indicators is a reference book that gives concise

labor force participation on unemployment (see Chapter 47, Unem-

estimates are subject to sampling error, as noted in the above paragraph

Data Sharing and Confidentiality of Federal Government

A recent change affecting the production of federal government busi-

law to report, as distinct from voluntary surveys in which survey re-

Using This Book

Gross Domestic Product

HOUSEHOLD INCOME AND EXPENDITURE

Personal Income and Saving

Consumer Price Index

BUSINESS PROFITS AND INVESTMENT

INFLATION AND DEFLATION

Consumer Price Index

Gross Domestic Product

Selected Services Revenue

Government Economic Transactions

CYCLICAL INDICATORS AND FORECASTING

Leading, Coincident, and Lagging Indexes

Manufacturing Business Activity Index

Consumer Attitude Indexes

Average weekly earnings represent the money wages and salaries of

Where and When Available

Data on average weekly earnings of workers are prepared monthly by the

and commissions of nonsupervisory workers in other private nonagri-

Average Weekly Earnings in Private Nonagricultural Industries

Annual percentage change

weekly earnings would be if it were available. For further information on

Average weekly earnings data are based on job-related earnings of

From 1995 to 2004, average weekly earnings in current dollars in-

References from Primary Data Source

Average weekly hours represent the length of the workweek in private

Where and When Available

Data on average weekly hours in the workplace are prepared monthly

Average weekly hours measure time on the job, including straight-time

the average of full-time and part-time workers in private nonagricul-

Average Weekly Hours in Private Nonagricultural Industries

Average weekly hours is a sensitive barometer of labor demand. Em-

From 1995 to 2004, average weekly hours in all private nonagricul-

References from Primary Data Source

The balance-of-payments accounts are the most comprehensive measure

Where and When Available

The Bureau of Economic Analysis (BEA) in the U.S. Department of

There are no estimates of sampling or revision error in the balance-of-

The balance of payments reflects U.S. participation in world markets